ORNL-4254.txt

" -
ARTIN MARIETTA SYSTEMS LIBRARIES

IHI I! -.

: 3 -
3 445k 0358082 7 /A g 7

: ORNL.-4254
~ UC-80 — Reactor Technology
" MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT
'FOR PERIOD ENDING FEBRUARY 29, 1968 \.
 ORML TECHNICAL INFORMATIONM
' DIYISION
| R Y-12 TECHMICAL LIBRARY B
. N _ '+ Document Reference Section
' LOAM COPY ONLY
: . D¢ NOT tansfer this document o any other | (
! persom. }f you want cthers to see it, attach their ; :
names, return the document, ond the Library -
will arronge the loan as requested. v
UCN-1624 Sl ‘
' €3 3-80) LT Yy ]

C /ENERGY COMMISSION

1
gl
]
.‘_5
i
ot

Printed in the United States of America. Available from Clearinghouse for Federal
Scientific and Technical Information, National Bureau of Standards,
U.S. Department of Commerce, Springfield, Virginia 22151
Price: Printed Copy $3.00; Microfiche $0.65

LEGAL NOTICE

This report was prepared as an account of Government sponsored work. Neither the United States,

nor the Commission, nor any person acting on behalf of the Commission:

A. Makes any warranty or representation, expressed or implied, with respect to the accuracy,
completeness, or usefulness of the information contained in this report, or that the use of
any information, apparatus, method, or process disclosed in this report may not infringe
privately owned rights; or

B. Assumes any liabilities with respect to the use of, or for damages resulting from the use of
any information, apparatus, method, or process disclosed in this report.

As used in the above, ‘‘person acting on behalf of the Commission’’ includes any employee or

contractor of the Commission, or employee of such contractor, to the extent that such employee

or contractor of the Commission, or employee of such contractor prepares, disseminates, or
provides access to, any information pursuant to his employment or contract with the Commission,

or his employment with such contractor.

ORNL-4254
UC-80 — Reactor Technology

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT
For Period Ending February 29, 1968

M. W. Rosenthal, Program Director

R. B. Briggs, Associate Director
P. R. Kasten, Associate Director

AUGUST 1968

OAK RIDGE NATIONAL LABORATORY
Qak Ridge, Tennessee
operated by
UNION CARBIDE CORPORATION
for the
U.S. ATOMIC ENERGY COMMISSION

AT

3 445k 0358082 7

\ oy

This report is one of a series of periodic reports in which we describe briefly the progress of
the program. Other reports issued in this series are listed below. ORNL-3708 is an especially
useful report, because it gives a thorough review of the design and construction and supporting
development work for the MSRE. It also describes much of the general technology for molten-salt

reactor systems.

ORNL-2474 Period Ending January 31, 1958
ORNL-2626 Period Ending October 31, 1958
ORNL.-2684 Period Ending January 31, 1959
ORNL-2723 Period Ending April 30, 1959
ORNL-2799 Period Ending July 31, 1959
ORNL-2890 Period Ending October 31, 1959
ORNL-2973 Periods Ending January 31 and April 30, 1960
ORNL-3014 Period Ending July 31, 1960
ORNL-3122 Period Ending February 28, 1961
ORNL-3215 Period Ending August 31, 1961
ORNL-3282 Period Ending February 28, 1962
ORNL-3369 Period Ending August 31, 1962
ORNL-3419 Period Ending January 31, 1963
ORNL-3529 Period Ending July 31, 1963
ORNL-3626 Period Ending January 31, 1964
ORNL-3708 Period Ending July 31, 1964
ORNL-3812 Period Ending February 28, 1965
ORNL-3872 Period Ending August 31, 1965
ORNL-3936 Period Ending February 28, 1966
ORNL-4037 Period Ending August 31, 1966
ORNL-4119 Period Ending February 28, 1967
ORNL-4191 Period Ending August 31, 1967
Contents

INTRODUCGTION ...ttt e e b e e e e s
UMM A R Y .ot e ettt et et h e et et s b b ettt e es e e e ne s e

PART 1. MOLTEN-SALT REACTOR EXPERIMENT

L. MSRE O P E RA TIONS .ot oottt e e e et a1 e ekttt et et s ebe e e b e es 2eateas e s ebanneens
1.1 Chronological Account of Operations and Maintenance.............cccoccoeeeiiiniiiniiiiiie e
1.2 Operations ANALYSIS ... e e e

1.2.1 Reactivity Balance ..............ocooiiiiiiiiiiiii e s
1.2.2 Variations in Reactor Access Nozzle Temperatures ..........ccccocieviiiiii i e
1.2.3  Radiation Heating .........cooooiiiiiii i e e e e
1.2.4  Thermal Cycle HIStOTY .....oocoiiiiiii et e e e
1.2.5 239U Indication of Integrated POWer..............cocccoiiiiiiioiiiiiiee e e
1.3  Equipment Performance .............ccoiiiiiiiiii i e e
1.3.1  Salt PUmPS oo e e e e e
1.3.2  Heat TranS er . oo e e et
1.3.3  Salt SAmMPLEIS ..o e e e e
1.3.4 Control Rods and DIEives ........cciiiiiiiirii e e e
1.3.5  Radiator ENCLOSUIE .. .ot ettt
1.3.6  Off-Gas SYSEEMS ....oioiiiiiiii ettt e oot e S
1.3.7  MaAin BloWerS ..ottt e et
1.3.8  HeAOIS ..o e e e e U
1.3.9  Electrical SyStem......ooiii it e et e e et e
1.3.10 Salt Pump Oil SyStems......oviiiiiiiiieiie et
1.3.11 Cooling Water SYSTEM . ....ccoiiiiiiiioiiie ittt oo ettt et e e ns
1.3.12 Component Cooling SYSTEMS ......ccvivitiiiiii ettt
1.3.13 Containment and Ventilation ... e

2. COMPONENT DEVEL O PMEN T ettt e i e
2.1 Off-Gas SAmMPLert. ... oottt et et e bbb
2.2  Fuel Sampler-EntiCher ... e e
2.3  Decontamination STUAIES ... e e e
2.4 Study of Pin-Hole Camera for Gamma Source Mapping............ocooooiiiiniiinii
2.5 Freeze-Flange Thermal Cycle Tests ...
208 PUIMDS oottt ettt e b e a e

2.6.1 Mark 2 Fuel PUump ... et e e
2.6.2 MSRE Oil Pumps ..o et et e
2.6.3 Oil Pump Endurance Test............ocoviiiiiiiii e

1ii
iv

3. INSTRUMENTS AND CONTROLS ... ..o ettt et et e e 30
3.1 MSRE Operating Experience ..............ccoooiiiiniiiiieiiiicis e e 30
3.1.1 Safety System COmPONnents ..ottt e e 30

3.1.2  TREIMOCOUPLES .....ooiiiiiiiie e e e e 30

3.1.3 Other Instruments and Controls ...........c.cocceiiiiiiiiiiiiiii 31

3.2 Control System Design ... e 31

3.3 MSRE Neutron NOiS€ ANALYSIS ......ooocoiiiiiiiiieiiee ittt sttt e et e 32

3.4 Test of MSRE Rod Control System Under Simulated 233U Loading Conditions..................... 32

3.5 Analog Computer Studies of the MSRE System with 233U Fuel Loading ..........c..ccocooeeeine. 35

4. MSRE REACTOR AN ALY S S i ettt e e aeaaa e e e anes 36
4.1 INtTOAUCLION .oiiiiiiiiii e oottt e et e e 4ot e e e et 36

4.2 Simulation of Nuclear Excursion Incidents ............ccooiiiiiiciiiiiiiieee e 37
4.2.1 Uncontrolled Rod Withdrawal ..................coiiiiiiiii e e 37

4.2.2 Returmn of Separated Uranium to the Core .............coviiiiiiiiiiiiii e 40

4.3 Detection of Anomalous Reactivity Effects ...............oooiiiii 47

PART 2. MSBR DESIGN AND DEVELOPMENT

. D S G . i e ettt ettt et e 51
5.1 GENEIAL ..o oo e ettt e 51

5.2 FLOW DIAEIAM ...ttt ettt ettt e 51

5.3 PLant Layout ...oocoooiiiioiie e e e e 53

5.4 Reactor Vessel and COre ..........coooiiiiuiiiiiiiiie ettt e sr e e 55

5.5 Primary Heat Exchanger. ... e 61

5.6 Fuel Drain TamK ..o ettt e et b et e e e 61

6. REACTOR PHYSICS . ...ttt a s e e et e s e bbb ettt ess s s e et et e e ae e e e nanans e eaeeenn s 66
6.1 MSBR PhySiCs ANAlYSIS . oottt et e e e a e e 66
6.1.1 Reference REACTOT ... ...ccoiiiiiiiiei et ettt e e e e ein et 66

6.1.2  FUEL-CYCLE COSES ..ottt e e e e e et e e e e s et ettt e e e ee e e e e enee e e 70

6.1.3  Cell CalculationS .. ...c.ocviiiiiiiiie e e e e 70

6.1.4 Reactivity CoeffiCients ............cooiiiiiiiiii e e 71

6.1.5 Measurements of Eta for 233U and 235U in the MSRE ...........cooovoviiiiiecie e, 72

7. SYSTEMS AND COMPONENTS DEVELOPMENT ... .o e e e e 74
7.1 Noble Gas Behavior in an MSBR ... e 74

7.2  Sodium Fluoroborate Circulating Loop TeSt.........ccoccooiiiiiiiiiiiie e e 75

7.3 MSBR PUMPS ...oooiiiiiiiieeiieeete ettt ettt ettt ettt ettt e e et 75
7.3.1  Pump Program...........cooiiiiiii e e e 75

7.3.2  Fuel Salt PUMP.......oooiic e e, 76

7.3.3 Coolant Salt PUMP ........cccoooiiiiiiiiiiii et e e 78

7.3.4 Molten-Salt Pump Test Facility............coooiiiiiiii e, 78

7.3.5 Molten-Salt Bearing Program ... 79

7.3.6 Rotor-Dynamics Feasibility Investigation............ PRSPPSO UPPPTOTON 82

7.4 Remote MaintenanCe .....ooooiine e e e e e e e e 82
. MSBR INSTRUMENTATION AND CONTROLS ........ccecoiiiiiitie oo e 85

8.1 Analog Computer SEUGIES . ........coooiiiiiiiiii oo 85

. CHEMISTRY OF THE MSRE ..ot e e e, 88
9.1 Fuel Salt Composition and Purity................ccc.ooooiiiiii oo 89
9.2 MSRE Fuel Circuit Corrosion ChemiStIy .. ....cco.oiiiieii e oo 90
9.3 Isotopic Composition of Uranium in MSRE Fuel Salt........c..ocooomoiiioiiooeeoeooeeeee e, 93

. FISSION PRODUCT BEHAVIOR..........cooiiiiiiiiioeeee oo e 94

10.1 Fission Product Behavior in the MSRE ... 94
10.1.1 Fission Products in the MSRE Fuel . ..., 94

10.1.2  Fission Products in the MSRE CoVer Gas ..........ococomoieeeeeoeeeeoeeeoeoeooeeoeeoeeee e, 96

10.1.3 Deposition of Fission Products from MSRE Cover Gas on Metal Specimens .......... 9

10.1.4 Examination of MSRE Surveillance Specimens After 24,000 Mwhr ............................ 99

10.1.5 Hot-Cell Tests on Fission Product Volatilization from Molten MSRE Fuel........... 100

10.1.6  Miscellaneous TeStS ..........ccooiviiiiioiiieeee e e, 108

10.2 Fission Product Distribution in an MSRE Graphite Surveillance Specimen ........................... 115
10.3 Proton Reaction Analysis for Lithium and Fluorine in MSR Graphite ...................................... 119
10.4 Surface Phenomena in Molten SaltS ... e o, 125
. CHEMISTRY OF FISSION PRODUCT FLUORIDES...........ccoocoi ittt e, 129
11.1 Properties of Molybdenum Fluorides ..ot e, 129
11.1.1  Synthesis of MoF; and MOE oo, 129

11.1.2  Lithium Fluoromolybdates(IIL) ............ccooooiiiiitieeeeee e 130

11.1.3  Kinetics of MoF ; Behavior in 2LiF-BeF , .......c.ccc.cooooiiiiiininiii e 131

11.2 Mass Spectrometry of the Molybdenum FIuorides ..............ccoovoiovioees oo oo 134
11.3 Spectroscopic Studies of Fission Product FIuorides ..................coccooeomevmeooeeee oo 136
11.4 Preparation of Niobium Pentafluoride ... 137
. PHYSICAL CHEMISTRY OF MOLTEN SALTS ... .. oo e e 138
12.1 Thermodynamics of LiF-BeF , Melts by EMF Measurements..............c.coc.coovevoveesorerereeenon.. 138
12.2  Electrolysis of LiF-BeF , Mixtures with a Bismuth Cathode ..............ccc.ooocveiieiivreie . 140
12.3 A Review of Electrical Conductivities in Molten Fluoride Systems .............ccccovivvieeronn ., 141
12.4 Measurement of Specific Conductance in LiF-BeF , (66-34 mole %) .......c.ccccocvici v, 144
12.5 A Stirred Reaction Vessel for Molten Fluorides ...........ocooiiiiiioio e 146
12.6 The Chemistry of Silica in Molten LiF-BeF, . ) 146
12.7 A Silica Cell and Furnace for Electrochemical Measurements with Fused Fluorides ........ 149

12,8 Status of the Molten-Salt Chemistry Information Center ....... e 149
13.

14.

15.

16.

17.

18.

vi

CHEMISTRY OF MOLTEN-SALT REACTOR FUEL REPROCESSING TECHNOLOGY .................... 152
13.1 MSBR Fuel Reprocessing by Reductive Extraction into Molten Bismuth ... 152
13.2 Removal of Structural Metal Fluorides from a Simulated MSRE Fuel Solvent ...................... 155
13.2.1 Reduction of Structural Metal Fluorides ..o 155
13.2.2  Salt Filtration StudieS.. ...t st 157
13.3 Protactinium Studies in the High-Alpha Molten-Salt Laboratory ... 159
13.3.1 Protactinium Reduction by Solid Thorium in the Near Absence of Iron.................. 159
13.3.2 Reduction of Protactinium by Bismuth-Uranium Alloy ... 159
13.3.3 Reduction of Protactinium by Bismuth-Thorium Alloys ... 160
13.3.4 Two-Region Breeder Blanket CompoSition ..ot 160
13.3.5 Single-Region Fuel Composition ........ccccoiiiiiiiiiii i 161
BEHAVIOR OF BF ; AND FLUOROBORATE MIXTURES ... 166
14.1 Phase Relations in Fluoroborate Systems...........ccooooiiiiiiiiiiii U 166
14.1.1 The System NaF—NaBF4-KBF4-KF ................................................................................ 166

14.2 Nonideality of Mixing in Potassium Fluoroborate~Sodium (or Potassium) Fluoride
SYSEOIMS ..ottt ettt s et e et b b 167
14.3 Heat Content of NaBF -NaF (92.5-7.5 mole %)........cccccooiiiiiii 168
14.4 The Solubility of Thorium Metal in Lithium Fluoride—Thorium Tetrafluoride Mixtures ........ 168
14.5 Dissociation Pressure of BF , for the MSRE Substitute Coolant .........ooovevieiiieeee e 169
14.6 Chemical Themodynamics of the System NaBF NaF ... 170
14.7 Corrosion of Hastelloy N and Its Constituents in Fluoroborate Melts ... 171
14.8 Compatibility and Immiscibility of Molten Fluorides ... 171

PART 4. MOLTEN-SALT IRRADIATION EXPERIMENTS
MOLTEN-SALT CONVECTION LOOP IN THE ORR ...ttt s 174
15.1 Isotope Activity Balance (Loop 2) .....cooooiiiiiii et e s 175
15.2 Penetration of Fission Products into Graphite and Deposition onto Surfaces........................ 175
15.3 Studies of Surface Wetting of Graphite by Molten Salt...............coocoiiiiii 178
15.4 Design of a Third In-Pile Molten-Salt Loop ........ccooviiiiiiiiiiii i e 178
GAMMA IRRADIATION OF FLUOROBORATE ... e e et e, 180
PART 5. MATERIALS DEVELOPMENT

MSRE SURVEILLANCE PROGRAM ... e et e et e e acaans s e rsese e e aianes 183
171 General COMMENS ........c..oii et e ettt es ettt ees e ete et e e e m e ettt e e et e ebe e ee e e nnbesnseeraeareeas s 183
17.2 Examination of Hastelloy N Specimens from MSRE Surveillance Facility ...........c.cccc.ocin 184
GRAPHITE STUDIES ..ottt oottt e ettt e et e e et e e ea e et e e ses s o eabts e e e e s s enes 188
18.1 Procurement of Special Grades of Graphite ..., 188
18.2 Porosity Created in Grade AXF Graphite by Oxidation Pretreatment ....................coocoiin 189

18.3  X-Ray Studies 0n Graphite ......c.o.coooiiiiiiiii e ettt e 190
19.

20.

21.

vil

18.4 Gas Impregnation of Graphite with Carbon ..............cccooioiiiiiioe e, 191

18.5 Graphite Surface Sealing with Metals ...t 192

18.6  Graphite Irradiation Program ............ccc.coooiiiiiiiiiiiii e, 195

18.7 Nondestructive Testing StudieS ..........cccooiiiiiiiiiii e, 196

18.7.1 Graphite Ultrasonic Velocity Measurements ................cccoocioiiiiioiiieeesieeee . 196

18.7.2 Low-Voltage Radiography............coouiiiiii oo, 197

HASTELLOY N oottt 198

19.1 Influence of Strain Rate on the Fracture Strain of Hastelloy N ...........cocovvvvvin.. e 198

19.2 Status of Development of the Modified AlIOY .......ccccoooviiiiiei e e 201
19.3 Effect of Carbon and Titanium on the Unirradiated Creep-Rupture Properties of

NI-MO-Cr AlIOYS oottt ettt ettt e et e, 204

19.4 Electrical Resistivity of Titanium-Modified Hastelloy N ..o e, 205

19.5 Electron Microscope Studies of Hastelloy N.........coooiiiiiiiiee e 206

19.5.1 Phase Identification Studies in Hastelloy N ............ccccooiiiiiiiiiiiiiiieiiie e 206

19.5.2 Effect of Silicon on Precipitation in Hastelloy N............cccccccooiiiiiiiiiiiieie e, 209

19.5.3 Titanium-Modified Hastelloy N ...t e, 212

19,504 SUMMATIY ..ottt ettt e, 213

19.6 Diffusion of Titanium in Modified Hastelloy N................ e, 213

19.7 Measurement of Residual Stresses in Hastelloy N Welds ..........occooooiiiiiiieiiicieeecee e, 215

19.7.1 Experimental Results ..ot e 216

19.8 Application of the Narrow-Gap Welding Process to the Joining of Hastelloy N .................... 217

19.9 Natural Circulation Loops and Test Capsules ............cccooooeiiiiiiiiiiiiiii e 218

19.9.1 Fuel Salts .............ccco. ettt et et et e e h e e eae e e e ehaea e e e s St e e b b e e e ban e e e e e e seree e 218

19.9.2  Blanket SAItS ......ocoooiiiiiiiii e e e 221

19.9.3  Coolant SAIES .......ccccoiiiiiiiii e 221

19.10 Forced Circulation LioOP ..ot 226

19.11 Oxidation of Hastelloy N .........coooiiiiiiiniiiiiien bttt ta et et e et b ee e ara it nats et e eanennas 228

GRAPHITE-TO-METAL JOINING.........oeoiitiititiitieit et ee ettt e aee e et e e e e et aneen. 231

20.1 Graphite Brazing Development .........ccoooiiiiiiiiii e e e 231

20.2 Radiation Stability of Brazing Alloys of Interest for Brazing Graphite ............................... 234

20.3 Graphite-to-Hastelloy N Transition JOINt ........ccoocoiiiiiiiiiiiiiiiiie et e e 235

20.3.1  Conceptual DeSi@n . ..o e, 235

20.3.2 Heavy-Metal Alloy Development ............ccocooiiiiiiiiiiiiecnnn SRR UU TR 236

20.4 Nondestructive Testing Evaluation of Graphite-to-Metal Joints ...........cccoiiiiiiiiniiinn 238

SUPPORT FOR COMPONENTS DEVELOPMENT PROGRAM .........coviiiiiieeeeeeeeeeeeee e 240

21.1  Remote Welding ....coooiiiiiiiiiiiie ettt et et e e et e e et et et 240
22.

23.
24.
25.
26.
27.
28.

29.
30.

31.

viii

PART 6. MOLTEN-SAL T PROCESSING AND PREPARATION

MEASUREMENT OF DISTRIBUTION COEFFICIENTS IN MOLTEN-SALT-METAL

QY ST M S oo e e et et e e 241
22.1 Extraction of Uranium and Rare Earths from Fuel Salt of Two-Fluid MSBR’s ........................ 242
22.2  Extraction of Uranium from Single-Fluid MSBR Fuel...........c.cooo i 243
22.3 Experimental Procedure and Lithium ‘“LoSS™ ... e 247
PROTACTINIUM REMOVAL FROM A SINGLE-FLUID MSBR ..., 248
CONTINUOUS FLUORINATION OF MOLTEN SALT ... ittt 252
RELATIVE VOLATILITY MEASUREMENTS BY THE TRANSPIRATION METHOD ........................ 255
DISTILLATION OF MSRE FUEL CARRIER SAL T e, 258
PROTACTINIUM REMOVAL FROM A TWO-FLUID MSBR ..........ccoooiiiiiiie e 260
RECOVERY OF URANIUM FROM MSRE FUEL SALT BY FLUORINATION ... 264
28.1 Fluorination-Corrosion StUAY .. ..o e et et e e 264
28.2 Fission Product Behavior — Analytical Assistance Program.................ooin, 266
MSRE FUEL SALT PROCESSING ... e e e e 269
PREPARATION OF ”'LiF-233UF4 FUEL CONCENTRATE FOR THE MSRE .............ccccceiiiiii, 270
30.1  Equipment CHan@eS ... et e e e 270
30.2 Equipment and Process Status ...t 271

DECAY HEAT GENERATION RATE IN A SINGLE-REGION MOLTEN-SALT REACTOR ............ 275
Introduction

The objective of the Molten-Salt Reactor Program
is the development of nuclear reactors which use
fluid fuels that are solutions of fissile and fertile
materials in suitable carrier salts. The program is
an outgrowth of the effort begun over 18 years ago
in the Aircraft Nuclear Propulsion (ANP) program
to make a molten-salt reactor power plant for air-
craft. A molten-salt reactor — the Aircraft Reactor
Experiment — was operated at ORNL in 1954 as
part of the ANP program.

Our major goal now is to achieve a thermal
breeder reactor that will produce power at low cost
while simultaneously conserving and extending the
nation’s fuel resources. Fuel for this type of re-
actor would be 233UF‘4 or 235UF‘4 dissolved in a
salt that is a mixture of LiF and BeF ,. The fer-
tile material would be ThF , dissolved in the same
salt or in a separate blanket salt of similar com-
position. The technology being developed for the
breeder is also applicable to advanced converter
reactors.

A major program activity is the operation of the
Molten-Salt Reactor Experiment (MSRE). This re-
actor was built to test the types of fuels and ma-
terials that would be used in thermal breeder and
converter reactors and to provide experience with
the operation and maintenance of a molten-salt re-
actor. The MSRE operates at 1200°F and at atmos-

pheric pressure and produces about 7.5 Mw of heat.

The initial fuel contains 0.9 mole % UF 2 5 mole %
ZrF4, 29 mole % BeF ), and 65 mole % LiF, and
the uranium is about 33% %33U. The melting point
is 840°F.

The fuel circulates through a reactor vessel and
an external pump and heat exchange system. All
this equipment is constructed of Hastelloy N, a
nickel-molybdenum-chromium alloy with exceptional
resistance to corrosion by molten fluorides and
with high strength at high temperature. The reac-
tor core contains an assembly of graphite moderator
bars that are in direct contact with the fuel. The

ix

graphite is a new material having high density and
small pore size. The fuel salt does not wet the
graphite and therefore does not enter the pores,
even at pressures well above the operating pres-
sure.

Heat produced in the reactor is transferred to a
coolant salt in the primary heat exchanger, and the
coolant salt is pumped through a radiator to dis-
sipate the heat to the atmosphere.

Design of the MSRE started early in the summer
of 1960, and fabrication of equipment began early
in 1962. The essential installations were com-
pleted, and prenuclear testing was begun in August
of 1964. Following prenuclear testing and some
modifications, the reactor was taken critical on
June 1, 1965, and zero-power experiments were
completed early in July. After additional modifi-
cations, maintenance, and sealing of the contain-
ment, operation at a power of 1 Mw began in Jan-
uary 1966.

At the 1-Mw power level, trouble was experienced
with plugging of small ports in control valves in
the off-gas system by heavy liquid and varnish-like
organic materials. These materials are believed
to be produced from a very small amount of oil that
leaks through a gasketed seal and into the salt in
the tank of the fuel circulating pump. The oil va-
porizes and accompanies the gaseous fission prod-
ucts and helium cover gas purge into the off-gas
system. There the intense beta radiation from the
krypton and xenon polymerizes some of the hydro-
carbons, and the products plug small openings.
This difficulty was overcome by installing a spe-
cially designed filter in the off-gas line.

Full power, about 7.5 Mw, was reached in May
1966. The plant was operated until the middle of
July for about six weeks at full power, when one
of the radiator cooling blowers (which were left
over from the ANP program) broke up from mechan-
ical stress. While new blowers were being pro-
cured, an array of graphite and metal surveillance
specimens was taken from the core and examined.

Power operation was resumed in October with
one blower; then in November the second blower
was installed, and full power was again attained.
After a shutdown to remove salt that had acci-
dentally gotten into an off-gas line, the MSRE was
operated in December and January at full power for
30 days without interruption. The next power run
was begun later in January and was continued for
102 days until terminated to remove a second set
of graphite and metal specimens. An additional
operating period of 46 days during the summer was
interrupted for maintenance work on the sampler-
enricher when the cable drive mechanism jammed.

In September 1967, a run was begun which con-
tinued for six months until terminated on schedule
in March 1968. Power operation during this run
had to be interrupted once when the reactor was
taken to zero power to repair an electrical short
in the sampler-enricher.

Completion of this six-month run brings to a
close the first phase of MSRE operation, in which
the objective was to demonstrate on a small scale
the attractive features and technical feasibility of
these systems for civilian power reactors. We be-
lieve this objective has been achieved and that
the MSRE has shown that molten fluoride reactors
can be operated at temperatures above 1200°F
without corrosive attack on either the metal or
graphite parts of the system, that the fuel is com-
pletely stable, that reactor equipment can operate
satisfactorily at these conditions, that xenon can
be removed rapidly from molten salts, and that,
when necessary, the radioactive equipment can be
repaired or replaced.

The second phase of MSRE operation will be
operation with 233U fuel in place of 235U. A small
facility in the MSRE building will be used to re-
move the uranium presently in the fuel salt by treat-
ment with gaseous F,. Highly pure 2°3U will then
be added to the present carrier salt, and critical,
low-power, and full-power tests will be performed.

A large part of the Molten-Salt Reactor Program
is now being devoted to the requirements of future
molten-salt reactors. Conceptual design studies
and evaluations are being made of large breeder
reactors, and an increasing amount of work on ma-
terials, on the chemistry of fuel and coolant salts,
and on processing methods is included in the re-
search and development program.

For several years, most of our work on breeder
reactors has been aimed specifically at two-fluid
systems in which graphite tubes are used to sep-
arate uranium-bearing fuel salts from thorium-bear-
ing fertile salts. We think attractive reactors of
this type can be developed, but the core designs
we have been working on are complex, and several
years of experience with a prototype reactor would
be required to prove that graphite can serve as a
plumbing material while exposed to high fast-neu-
tron irradiations. As a consequence, a one-fluid
breeder has been a long-sought goal.

Two developments of the past year have estab-
lished the feasibility of a one-fluid breeder. The
first was establishment of the chemical stepsin a
process which uses liquid bismuth to extract prot-
actinium and uranium selectively from a salt that
also contains thorium. The second was the recog-
nition that a fertile blanket can be obtained with a
salt in which there is uranium as well as thorium
by reducing the graphite-to-fuel ratio in the outer
part of the core. Our studies show that a one-fluid,
two-region breeder can be built that has fuel utili-
zation characteristics comparable to our two-fluid
designs, and probably better economics. Since the
graphite serves only as moderator, the one-fluid re-
actor is more nearly a scaleup of the MSRE.

These features have caused us to change the em-
phasis of our breeder program from the two-fluid to
the one-fluid breeder. Work on both performed dur-
ing the past six months is described in this report,
but most of our design and development effort is
now directed to the one-fluid system.
Summary

MOLTEN-SALT REACTOR
EXPERIMENT

PART 1.

1. MSRE Operations

This report period was almost completely occu-
pied with a long run that began in September and
was still going at the end of February. When op-
erations were first resumed after the fuel sampler
mechanism was replaced, some difficulties were
encountered with a radiator door and a component
cooling pump. Once the long run was under way,
however, the only delay was occasioned by a wir-
ing failure in the sampler-enricher in November.

The very long run, without the complicating ef-
fects of drains and dilutions or fuel additions, af-
forded opportunity for very close comparison of
predicted and observed long-term changes in re-
activity. A gradual deviation between observed
and computed reactivity of —3.5 x 107°(% 8k/k)/
Mwhr was seen.

An experiment on the effect of minor variations
in fuel temperature, system pressure, and fuel salt
level on fission gas stripping extended over a two-
month period. There were several different indica-
tions that lower temperature, higher pressure, and
lower salt level increased the volume of gas cir-
culating with the fuel and delayed the stripping
of fission gases into the cover gas.

Component performance was generally quite good.

Two bearings on the main blowers were replaced
during a period at low power in January, and the
standby component cooling pump was not operable
during most of the run. However, neither of these
delayed the operation. The wiring failure in the
sampler-enricher interrupted high-power operation
for nine days, but it was not necessary to drain
the fuel.

During the six-month period, the reactor was
critical 89% of the time, and the integrated power
increased to 8581 equivalent full-power hours.

Xi

2. Component Development

The installation and preoperational testing of
the off-gas sampler were completed. Four samples
of reactor off-gas were removed for xenon isotopic
analysis.

The isolation chamber and drive assembly that
was removed from the sampler-enricher was exam-
ined and disassembled in a hot cell. After suc-
cessful decontamination, the major part of the unit
was returned for use as a spare. A proximity
switch was tested that promises to help prevent
cable snarls such as that which rendered the sam-
pler inoperative. A heated carrier utilizing molten
babbitt was built to keep samples hot in transit to
the analytical laboratory.

Tests of a pinhole gamma-ray camera for locating
radiation sources showed promise.

Thermal cycling of a prototype freeze flange was
resumed to lend confidence to predictions of fatigue
life.

The hot-test facility for the Mark 2 fuel pump was
prepared, and the rotary element assembly was
nearly finished.

3. Instruments and Controls

Only a moderate amount of maintenance on the
reactor systems was required. Five of the fifteen
relays in the rod scram coincidence circuit failed,
leading to a decision to replace them with a dif-
ferent type of relay. Water leaks in nuclear cham-
ber cables continued to give trouble, and two fis-
sion chambers and an ionization chamber had to
be replaced. A second failure occurred in a posi-
tion synchro on one of the control rods.

A few minor modifications were made, and some
design required for salt processing was completed.
Equipment and procedures for analysis of neu-
tron noise spectra were tested and used to obtain
data pertinent to reactor operation under various

conditions.
The adequacy of the rod servo control system
with 233U fuel was investigated. No need for
modification appeared.

4. MSRE Reactor Analysis

Reactor physics studies in support of planned
operation of the MSRE with 233U were extended
to help evaluate the nuclear safety of the system.
As was the case with the present 235U fuel load-
ing, the potentially most severe nuclear excursions
were associated with two hypothetical incidents:
reactivity addition by the simultaneous withdrawal
of the three control rods, and the gradual separa-
tion of uranium from the mainstream of circulating
salt, followed by its sudden resuspension and
rapid return to the core in concentrated form. Both
incidents were analyzed with the aid of a digital
kinetics program supplemented by analog simula-
tion. In the first incident, we found that rod scram
initiated by the action of the safety system would
limit the temperature-pressure rises in the nuclear
excursion to inconsequential proportions. This
same conclusion would apply in the case of the
second incident, unless the abnormal reactivity
loss represented by the uranium separation became
larger than about 0.95% 6k/k. This abnormal re-
activity loss would be easily detectable by routine
computer monitoring of the reactivity balance,
which should reveal any anomaly as large as 0.1%
ok/k.

PART 2. MSBR DESIGN AND DEVELOPMENT

5. Design

Developments in fuel processing and reactor core
configurations that enable a one-fluid molten-salt
reactor to be an efficient breeder have led us to
set aside our work on the design of a two-fluid
breeder reactor and to undertake studies of a one-
fluid breeder. The fuel for the one-fluid breeder
consists of fissile uranium and fertile thorium as
tetrafluorides dissolved in a lithium fluoride—
beryllium fluoride carmrier salt. We have been work-
ing on the design of a 2000 Mw (electrical) reactor
in which the fuel circulates upward around vertical
graphite bars in a reactor vessel and then com-
pletes the circuit through four pumps and four heat
exchangers. As in our designs for two-fluid reac-
tors, the reactor vessel, heat exchangers, and

Xii

pumps are installed in a heated and shielded cell.
Steam generating equipment is installed in cells
adjacent to the reactor cell, and sodium fluoro-
borate salt circulates between the primary heat
exchangers and the steam generators to transfer
the fission heat to supercritical steam.

Also communicating with the reactor cell are a
fuel processing cell, an off-gas disposal cell, and
a drain tank cell for storing fuel salt when the re-
actor is drained.

Basic piping layouts have been made for the fuel
and coolant salt systems. A system has been con-
ceived for cooling the drain tank by thermal con-
vection of sodium fluoroborate salt between coils
in the drain tank and air-cooled coils in a natural-
draft chimney outside the reactor building.

6. Reactor Physics

Neutronic calculations of a single-fluid molten-
salt breeder reactor, with fissile and fertile ma-
terials carried in the same salt stream, have shown
that breeding performance comparable with that of
a two-fluid MSBR can be achieved, provided the
core is properly designed to minimize neutron leak-
age. Breeding ratios of 1.05 to 1.07, fuel specific
power of 2 to 2.5 Mw (thermal)/kg, and annual fuel
yields of about 5% /year appear to be attainable with
fuel processing rates which probably imply fuel-
cycle costs less than 0.5 mill/kwhr (electrical).

Such a reactor would have a small negative over-
all isothermal temperature coefficient of reactivity
and a substantially negative prompt coefficient,
that is, ~ —3 x 10~ % 8k/°C, associated with a
change in salt temperature alone.

7. Systems and Components Development

The analytical model used to compute the steady-
state migration of noble gases to the graphite and
other sinks in an MSBR was extended to study the
effects of graphite surface area and surface coat-
ings on the xenon poison fraction. It was shown
that an 8-mil coating of material with a diffusion
coefficient of 10~ 8 ft*/hr would reduce the poison
fraction from 2 1/4% to the target value of 0.5%.

The alterations to and checkout of the test fa-
cility for operation with sodium fluoroborate were
completed, and a flush charge of 900 Ib of sodium
fluoroborate was added to the sump. The flush
charge is intended to remove residues of the Li-Be
salt previously circulated in the loop and will be

replaced after several days of operation. The loop
will be operated to study the pumping characteris-

tics of the salt and the problems associated with

control and monitoring of the salt composition.

A preliminary layout was made of a fuel salt
pump configuration applicable to the single-fluid
molten-salt breeder concept. Preliminary layouts
were made of a molten-salt pump test facility and
a molten-salt bearing tester suitable for the MSBE
salt pumps. Work was initiated on specifications
for the MSBE fuel salt pump. The rotor-dynamics
feasibility investigation was completed for the
long-shaft pump configuration that requires a mol-
ten-salt-lubricated bearing and was specified for
our two-fluid molten-salt breeder concept. Speci-
mens of cermet hard coatings which are plasma-
sprayed on Hastelloy N substrate were received,
and they are being evaluated as candidate mate-
rials for molten-salt bearings.

Studies of the problems associated with the
maintenance of the MSBR concepts were started.
The initial effort is to evaluate the problems
caused by scaleup of the general maintenance
system used at the MSRE. The three specific
areas under active study are (1) the application
of the portable maintenance shield concept to the
various cells of the MSBR, (2) remote welding for
vessel entry and component replacement, and (3)
replacement of the graphite moderator elements
of the core on a routine basis. The early develop-
ment of remotely operated vessel and pipe clo-
sures is important to the program, and the study
of remote welding as an initial approach is being
expedited.

8. MSBR Instrumentation and Controls

Analog computer studies of the dynamic behavior
of the Molten-Salt Breeder Reactor were begun. A
model of the two-fluid system was developed, and
several transient cases were investigated to de-
termine uncontrolled core behavior during reac-
tivity changes, fuel salt flow reductions, and sim-
ulated load losses. The results obtained lead us
to the tentative conclusion that the plant would
be inherently load following at the expense of
modest temperature changes. They also indicate
that it should be quite easy to accommodate rather
large load changes using a control system to main-
tain some desired temperature condition.

xiii

A model of the steam generator was also devel-
oped. The complexity of this model exhausts the
capacity of the ORNL analog computer, leaving no
equipment available for the simulation of the rest
of the plant. Alternative methods for studying the
dynamics of the entire Molten-Salt Breeder Reactor
power plant are being studied, with the most prom-
ising approach being one using a simpler linearized
model of the steam generator on a hybrid computer.

PART 3. CHEMISTRY

9. Chemistry of the MSRE

Chemical behavior in the salt, gas, oil, and
water systems has been under continuous surveil-
lance since the beginning of MSRE operations.

The results continued to show excellent materials
compatibility. There was, however, an unexplained
difference of about 0.02 wt % between the chemical
analyses for uranium and the uranium concentration
computed from operational data. The chromium
concentration reached a steady-state value of 85
ppm; this represents an insignificant amount of
corrosion, and the view has been advanced that
much of the chromium content of the fuel came

from the drain tanks.

Mass spectrometric analyses of the uranium iso-
tope distribution in the fuel promised to be useful
in rating the reactor output.

10. Fission Product Behavior

The fate of fission products in the reactor was
established in considerable detail.
usual behavior continued to be that of the more
noble metals such as Mo, Ru, Te, and Nb. These
left the fuel not only by depositing on walls but
also apparently as a smoke that was carried away
in the gas phase.

A new method of examining the concentration
profile of fission products in graphite from the
MSRE confirmed the results from the older method.
Concentrations of Li and F in the same graphite
were obtained by a highly sensitive method involv-
ing proton bombardment. Traces of these elements
penetrated deeply, but the amounts found, though
unexplained, were not chemically significant.

Samples of fuel from the MSRE were studied in
the hot cell; the same type of emanation of fission

The only un-
products was found as that previously encountered
in the MSRE pump bowl. This emanation was iden-
tified as having the form of aerosols, and pictures
of the particles were obtained by electron micros-
copy. Tests were initiated to explore the nature
of colloids in molten salts.

11. Chemistry of Fission Product Fluorides

Because of the peculiar behavior of fission prod-
ucts like molybdenum in the MSRE, an exploration
of the chemical behavior of molybdenum fluorides
and niobium fluorides was continued. Methods of
pteparing and characterizing MoF ,, MoF and
other compounds were perfected. The rate of auto-
oxidation and -reduction of MoF , in LiF-BeF,
melts was measured.

A mass spectrometric investigation of molybde-
num vaporization was extended to higher pressures
where the vapor-phase species were measured at
varying temperatures. The dimer Mo JF 1, Was en-
countered.

Spectrophotometric studies were attempted for
MoF ; and NbF , in LiF-BeF ,, and NbF, was syn-
thesized.

21

12. Physical Chemistry of Molten Salts

Electrochemical measurements have revealed the
solution thermodynamics of LiF-BeF , melts over
the whole composition range. Experiments on the
electrolysis of LiF-BeF , mixtures with a bismuth
cathode were initiated; these were of interest in
connection with extractive reprocessing of breeder
melts.

An improved method for balancing cell impedance
was employed in measuring the electrical conduc-
tivity of molten fluorides. The specific conduc-
tivity of LiF-BeF, (66-34 mole %) rises with tem-
perature and is about 1.6 mhos/cm at 510°C.

In the past the lack of vigorous agitation has
seriously handicapped work on heterogeneous
equilibria in molten fluorides; an improved stirred
vessel was therefore designed and constructed.
The inleakage of air to this vessel was scarcely
detectable. Such a vessel was used in unraveling
the chemistry of silica in LiF-BeF,. The use of
an overpressure of SiF, permitted electrochemical
cells involving LiF-BeF , to be constructed of
silica.

X1V

A Molten-Salt Chemistry Information Center has
been established and is operating successfully.

13. Chemistry of Molten-Salt Reactor Fuel
Reprocessing Technology

Laboratory-scale studies of the reductive extrac-
tion of protactinium, uranium, and rare earths con-
tinued to demonstrate the feasibility of reprocess-
ing two-region molten-salt breeders. As applied
to one-fluid breeders, the results were less favor-
able. The difficulty is that thorium tends to re-
duce before the rare earths.

The fluorination step for removing uranium in
reprocessing fuel leads to the accumulation of a
large amount of corrosion products in the form of
structural metal fluorides. Modes of reducing and
removing these products have been studied.

Experiments on the reductive extraction of prot-
actinium were carried out in a facility that allowed
realistic amounts of alpha-active *3!Pa to be fol-
lowed. The reductive extractions usually employed
bismuth alloys as the reducing and extracting me-
dium.

14. Behavior of BF3 and Fluoroborate Mixtures

A strong candidate as a secondary coolant was
the NaF-KF-BF , (47-5-48 mole %) mixture, which
had a melting point of 350 * 5°C. Experimental
phase diagrams for fluoroborates were revised to
be consistent with the heats of fusion of NaF and
KF. Using a copper block calorimeter, the heat of
fusion of the NaBF ,-NaF eutectic was found to be
31 cal/g, and the heat of transition was 14.7 cal/g.
The heat capacity of the liquid was 0.36 cal g~
°c~ L

The pressure of BF ; above fluoroborate mixtures
was measured; from these measurements the chem-
ical thermod ynamics of the system NaF-NaBF,
were elucidated.

PART 4. MOLTEN-SALT IRRADIATION
EXPERIMENTS

15. Molten-Salt Convection Loop in the ORR

During this report period we have completed ad-
ditional analyses of the fuel salt, graphite, and
metal in contact with fissioning fuel salt and cover
gas from the second in-pile molten-salt convection

loop. These data permitted completion of the iso-

tope activity balance.

Molybdenum, tellurium, ruthenium, and niobium
are almost entirely departed from the salt. These
elements showed no dominant preference for graph-
ite or metal but seemed to deposit on whatever
surface was available. Short-lived noble gases
appeared to have diffused appreciably into the
graphite, as shown by the presence of daughter
isotopes such as 89gt, '4%Ba, and others. How-
ever, the major proportions of these, and almost
all of the other alkali, alkaline-earth, and rare-
earth isotopes (all of which form relatively stable,
nonvolatile fluorides), were found in the salt.

Design of a third molten-salt convection loop
has been completed. Design features include the
use of a modified Hastelloy N containing titanium
for improved resistance to radiation-induced high-
temperature embrittlement, graphite surveillance
specimens in the core section to permit better
postirradiation determination of the interaction
of graphite and salt, and gas adsorption traps to
help identify the gas-bome fission product species.

16. Gamma Irradiation of Fluoroborate

Gamma irradiation experiments with sodium fluo-
roborate salt are being conducted in the central
channel of spent HFIR fuel elements to determine
the effects, if any, of such irradiations on the so-
dium fluoroborate and its compatibility with Hastel-
loy N.

One gamma irradiation experiment and an unir-
radiated control have been completed to date. Both
experiments used an NaF-NaBF | eutectic mixture
(normally 8-92 mole %) in a Hastelloy N capsule
containing a Hastelloy N corrosion test specimen.
The unirradiated experiment was operated for 840
hr (mostly at 600°C). The irradiation experiment
was operated for 533 hr in spent fuel element 34
(~8 x 107 t/hr at the start of the experiment, com-
pared with an estimated 5 x 107 r/hr in the MSRE
heat exchanger), and the salt accumulated an esti-
mated absorbed dose of ~1 x 10%* ev/g. The gas
space in the capsule was maintained at a nominal
temperature of 600°C. We were not able to monitor
salt temperatures in the lower part of the capsule
after thermocouples in this region were lost early
in the experiment. However, it is estimated that

XV

at least part of the NaF-NaBF , salt in the capsule
remained frozen (melting point ~ 380°C) for most
of the run, since water entered the container can
and soaked the magnesia insulating pad under the
capsule.

Salt from the irradiated capsule was discolored,
while that from the unirradiated capsule was en-
tirely white except for some small green crystals,
identified as Na 3CrF 6 found near metal surfaces.

Hastelloy N coupons exposed in the irradiated
and unirradiated tests exhibited negligible attack.

Analysis of residual gas from the irradiated and
unirradiated tests showed no trace of BF .

A second gamma irradiation capsule assembly
has been placed in operation, and we plan to ir-
radiate this experiment to a higher total dose than
the first experiment.

PART 5. MATERIALS DEVELOPMENT
17. MSRE Surveillance Program

We have removed two sets of surveillance speci-
mens from the core of the MSRE and one set of
metal specimens from outside the core. Metallo-
graphic studies did not reveal any evidence of cor-
rosion of the graphite and Hastelloy N by the fluo-
ride salt environment. The Hastelloy specimens
from outside the core were exposed to the cell en-
vironment and had an oxide film of about 0.002 in.
Testing showed that the mechanical properties of
the Hastelloy N changed during exposure, but
these changes are equivalent to those noted for
materials exposed to equivalent fluences in the
ORR. The reduction in the ductility seems to be
about saturated for the vessel, and the properties
are more than adequate for operation under design
conditions.

18. Graphite Studies

We continue to obtain samples of several grades
of graphite that are potentially applicable for use
in molten-salt breeder reactors. Some graphites
have been obtained from the Y-12 Plant for which
we have detailed information on the raw materials
and processing variables. These materials will
be included in all phases of our graphite work.

We are evaluating several techniques for surface
sealing graphite to obtain a low surface perme-
ability. Carbon coatings have been applied previ-
ously in a fluidized bed, and two new methods are
under study. One involves passing the carbona-
ceous gas through the wall of a graphite tube, with
the location of deposition being controlled by a
radial temperature gradient. A second method uti-
lizes a cyclic technique in which the sample en-
vironment is cycled between hydrocarbon mixtures
and vacuum. Samples coated by both techniques
are being evaluated. Sealing the graphite by the
deposition of molybdenum is complicated by the
need to minimize the quantity of metal present in
Our work shows that the
graphite can be sealed adequately by about 1 mil
of molybdenum if the graphite substrate has a

the core of the reactor.

small, uniform pore size of about 1 p.

We have designed an experimental assembly for
irradiating graphite in the HFIR. This facility
will allow us to irradiate small graphite cylinders
to a fluence of 4 x 1022 neutrons cm~ 2 year— ! at
700°C. We have run two short experiments in the
HFIR to check the design, have made the neces-
sary changes, and presently have two assemblies
in the HFIR. Changes in the specimen dimensions
will be measured, and techniques have been devel-
oped to measure the velocity of sound in the graph-
ite to determine changes in the elastic modulus.

19. Hastelloy N

The radiation damage in Hastelloy N continues
to receive considerable attention. Although the
strength at elevated temperatures is reduced by
irradiation, the reduction in the fracture strain is
of primary concern. We have found that the frac-
ture strain is dependent upon the strain rate, show-
ing a minimum fracture strain of about 0.5% at a
strain rate of 0.1%/hr. The titanium-modified
Hastelloy shows the same behavior, but the min-
imum fracture strain is about 3% compared with
0.5% for the standard alloy.

We are continuing the development of the tita-
nium-modified Hastelloy N. We have obtained from
commercial vendors about twenty-five 100-1b melts
and one 5000-1b melt of the modified composition.
The yield of fabricated shapes from the 5000-1b
melt was about 50%, a respectable value for stand-
ard mill practice. Studies on several small melts
show that both carbon and titanium contribute to
the strength of the modified alloy in the unirradi-
ated condition. The object of these studies is to

XV1

determine the titanium and carbon contents to give
optimum properties. Diffusion measurements have
been made to estimate how rapidly titanium can be
removed from the alloy by corrosion. The results
indicate that titanium diffuses less rapidly than
chromium and should not cause a significant in-
crease in the corrosion rate.

Electron microscopy of standard Hastelloy N has
shown that the large precipitates are basically car-
bides of the M [C type and that they contain sev-
eral percent silicon. The segregation of silicon
to the precipitates probably accounts for incipient
melting and weld cracking in the alloy. The sili-
con specification has been lowered to 0.1% max-
imum to minimize these adverse effects.

The residual stresses that develop during weld-
ing can cause subsequent cracking of the weldment
during service. We have developed a technique for
measuring these stresses in welded plates that
will allow us to determine welding parameters and
postweld heat treatments to minimize the residual
stresses. Measurements on Hastelloy N show that
the maximum residual stress is reduced from about
61,000 to 5000 psi by a postweld anneal of 6 hr
at 871°C.

We have several thermal convection loops in op-
eration to investigate the compatibility of struc-
tural metals with various fluoride salts. A loop
constructed of type 304L stainless steel and con-
taining a modified fuel salt has operated for 40,510
hr at a peak temperature of 677°C without incident.
Removable tabs of stainless steel indicate that
the corrosion rate is about 2 mils/year for the
first several hundred hours and then decreases,
consistent with the behavior expected for a dif-
fusion-controlled process. Analysis of the salt
indicates that only the chromium level is increas-
ing. A Hastelloy N loop containing the same salt
has operated without incident at a peak tempera-
ture of 704°C for 51,810 hr. Two loops constructed
of Hastelloy N and containing the potential coolant
salt NaBF ,-NaF (92-8 mole %) have operated for
about 3000 hr at a peak temperature of 607°C.
These loops are constructed so that samples can
be removed for weighing and salt samples taken
during loop operation. Chemical analysis of the
salt indicates that the chromium and iron levels
are both increasing at a rate consistent with a
diffusion-controlled corrosion process. Weight
changes of the removable specimens indicate a
maximum corrosion rate of a few tenths of a mil
xvil

per year. One of these loops contains specimens
of modified Hastelloy N (contains no iron), and it
was found that this material corroded at a lower
rate. Several other loops have been started very
recently to investigate the compatibilities of other
salts and Hastelloy N.

A forced-convection loop is in the final stages
of construction to investigate the compatibility of
Hastelloy N and NaBF  -NaF (92-8 mole %) under
realistic service conditions. This loop will have
a liquid velocity of 7 fps, a BF , control system,
and capabilities for sampling the salt during op-
eration.

Studies of the oxidation of Hastelloy N indicate
that silicon plays a critical role in the scaling re-
sistance. Silicon is usually present in air-melted
heats at a level of about 0.6%, and we would like
to reduce this further to improve the weldability.
Levels of less than 0.1% silicon can be obtained
routinely by vacuum melting. However, the heats
with lower silicon have higher scaling rates. The
further addition of titanium to the low-silicon ma-
terials does not cause any appreciable changes
in the scaling resistance. Although we have es-
tablished these trends, the oxidation resistance
in all cases is quite acceptable at 760°C.

20. Graphite-to-Metal Joining

We have been able to join graphite to Hastelloy N
in small sizes (1 in. in diameter) by direct brazing
with copper or with an Ni-Pd-Cr alloy. However,
the matching of surfaces to obtain good joints in
large pipe sizes requires extremely good control
over dimensional tolerances, so we have not been
able to make the direct joint consistently. How-
ever, joints can be made repeatedly when a molyb-
denum transition piece is used between the Hastel-
loy N and the graphite. A nondestructive testing
technique has been developed for determining
whether all areas of the joint are bonded. We have
investigated the effects of irradiation on several
brazing alloys that are potentially suitable for this
joint including Cu, Ni-Pd-Cr, Cu-Ni-Cr-Be, and
Cu-Ni-Ta-Be. The strength and ductility changes
indicate that all the alloys have acceptable proper-

ties after irradiation to fluences of the order of 102°

neutrons /cm 2.

Another approach for joining Hastelloy N to
graphite is a transition joint in which thin layers
of alloys having slightly different coefficients of

thermal expansion are brazed together with a very
ductile brazing alloy such as copper. Suitable
tungsten-base alloys presently exist for the joint,
and the development of molybdenum-base alloys
seems imminent.

21. Support for Components Development Program

We are setting up the equipment to evaluate sev-
eral welding processes that are potentially appli-
cable to the remote joining of Hastelloy N. The
aim of this study is to choose a welding process
that can be further developed for making joints re-
motely in a high radiation field.

PART 6. MOLTEN-SALT PROCESSING
AND PREPARATION

22. Measurement of Distribution Coefficients
in Molten-Salt—Metal Systems

Distribution of uranium, thorium, and rare earths
between selected molten fluoride salts and lithium-
bismuth solutions is being studied in support of
reductive extraction processes for MSR fuels. Data
obtained so far show that uranium can easily be
preferentially extracted from the fuel salt of a two-
fluid or a single-fluid MSBR. Separation of the
uranium from rare earths is good (separation factors
of at least 10%) with each type of fuel. In the proc-
essing of single-fluid MSBR fuel, uranium can be
separated from thorium by a factor of at least 10*.
Rare earths can probably be separated from thorium,
but more data are required to establish the optimum
conditions.

23. Protactinium Removal from

a Single-Fluid MSBR

The steady-state performance of a system for
isolating protactinium from a single-fluid MSBR
by reductive extraction was examined. Calcula-
tions based on available tentative equilibrium data
indicate that adequate protactinium isolation can
be obtained with a system consisting of an extrac-
tion column equivalent to about 12 ideal stages,
an electrolytic oxidizer-reducer, and a protactinium
decay tank having a volume of about 400 ft>. Re-
quired salt and metal flow rates are about 2 gpm.
xviii

The system was found to be quite sensitive to
minor variations in operating conditions, and sev-
eral methods for stabilizing the system have been
explored. Details will be revised as more accurate
data are available, but qualitative conclusions are
believed valid.

24. Continuvous Fluorination of Molten Salt

Studies are under way on protecting a continuous
fluorinator from corrosion by freezing a layer of
salt on the vessel wall. Operability of such a sys-
tem was demonstrated by countercurrently contact-
ing molten salt and an inert gas in the presence
of a frozen layer of salt in a 5-in.-diam, 8-ft-high
column. An internal heat source consisting of
Calrod heaters in a 3/4-in.-diam pipe along the cen-
ter line of the system was used to simulate the
volume heat source provided by fission product
decay in the molten salt. Frozen wall thicknesses
and temperature profiles in the frozen salt were in
general agreement with values predicted by rela-
tions based on radial heat transfer from a volume
heat source. Frozen wall thickness ranged from

0.3 to 0.8 in., depending on experimental conditions.

25. Relative Yolatility Measurements by the
Transpiration Method

Relative volatilities for several solutes were
measured in the temperature range from 900 to
1050°C using LiF-BeF , (90-10 mole %) as the
solvent. Values obtained with respect to LiF at
1000°C were about 0.04 for UF,, 25 for RbF, 95
for CsF, and about 2 for ZrF , (when present in
solution at a concentration of 0.083 mole %).

26. Distillation of MSRE Fuel Carrier Salt

Study of low-pressure distillation of MSRE car-
rier salt is under way in equipment which includes
a 48-liter feed tank, a 12-liter still, a condenser,
and a 48-liter condensate receiver. Two runs were
made using a still pot temperature of about 1000°C
and a condenser pressure of 0.06 to 2 mm Hg; the
total salt volume distilled during the runs was
about 60 liters. With the lower condenser pres-
sures, salt distillation rates of 1.2 and 1.5 ft >
day~! ft =% were observed at still pot temperatures
of 990 and 1005°C respectively.

27. Protactinium Removal from a Two-Fluid MSBR

Protactinium removal processes based on reduc-
tive extraction using liquid bismuth containing
thorium were analyzed to evaluate feasibility of
removing protactinium from the fertile stream of a
two-fluid MSBR. Calculations indicate that ade-
quate protactinium removal can be achieved with
a system consisting of an extraction column equiv-
alent to about three ideal stages, an electrolytic
oxidizer-reducer, a protactinium decay tank having
a volume of about 400 ft®, and a fluorinator for re-
moval of uranium from the decay tank. Required
fertile salt and metal flow rates are about 25 and
0.1 gpm, respectively, for a 1000 Mw (electrical)
reactor.

28. Recovery of Uranium from MSRE Fuel Salt

by Fluorination

Small-scale tests are being made with simulated
MSRE fuel salt to determine the effects of tempera-
ture, fluorine concentration, and fluorine flow rate
on the rate of uranium volatilization and the rate of
corrosion of Hastelloy N. Preliminary results indi-
cate that uranium can be readily removed from the
salt with fluorine at 450 to 500°C; however, cor-
rosion rates of up to 0.5 mil/hr can be expected
during the fluorination period.

Fluorination is also being considered for use as
the first step in a precise method for analyzing for
uranium in MSRE fuel salt. The UF ; produced
would be collected and decontaminated using the
NaF sorption-desorption method; then the UF ,-NaF
complex could be transferred to a low-level radia-
tion area for precision coulometric analysis of the
uranium. Decontamination factors for '°3Ru, °Nb,
and !*2Te have been greater than 10°; however, at-
tempts to achieve an adequate decontamination fac-
tor (greater than 10°) for iodine have been unsuc-
cessful so far.

29. MSRE Fuel Salt Processing

Modifications to the MSRE fuel processing facil-
ity have been completed except for the installa-
tion of the salt filter. Processing plans call for
fluorination, reduction, and filtration of both flush
and fuel salts.
Xix

30. Preparation of 7LiF-233UF‘ Fuel
Concentrate for the MSRE

Refueling and operating the MSRE with 233U fuel
in 1968 will require approximately 39.5 kg of 91.4%-
entiched ?*°U as "LiF-?*UF , (73-27 mole %) eu-
tectic salt. This fuel concentrate will be prepared
in cell G of the TURF building because of the ra-
diation from the high 232U (222 ppm) content of the
233y, Engineering design and installation of the
process equipment in TURF are essentially com-
pleted. Additional equipment has been designed
and built for drilling holes in the enrichment cap-
sules and for packaging the bulk charge of salt in
nine small salt cans. A shakedown run using
238UO3 was started January 15, 1968. By the
end of February, 90 to 95% of the uranium had been
converted to UF , by treatment with H ,-HF. Prog-
ress of the reaction has been difficult to follow in
the remote facility, and the conversion is slower
than expected, probably because of scaleup prob-
lems in the production equipment.

Equipment changes to eliminate operating prob-
lems encountered in the shakedown run are almost
completed.

31. Decay Heat Generation Rate in a
Single-Region Molten-Salt Reactor

Studies of heat generation by gross fission prod-
ucts and **®Pa in a one-region 2000 Mw (electri-
cal) MSR show that at equilibrium with continuous
processing these components are generating 289.4
and 0.74 Mw, respectively, in the fuel stream.
When noble gases are sparged from the circulating
fuel loop, the fission product decay heat decreases
to 257 Mw; if, in addition to sparging, the noble
metals are removed by reaction with reactor sur-
faces, the rate is 255.8 Mw. The value above for
233pa decay heat corresponds to a fuel stream
concentration of 0.256 g of ?3Pa per liter, the
equilibrium quantity when 233Pa is removed on a
3-day processing cycle. The fission product proc-
essing cycle time is 38 days. This processing
scheme requires decay storage of 190.5 kg of ?3°Pa
in the processing plant. At equilibrium, this 233Pa
produces 9.7 Mw of heat.

Calculations of the heat-generation rate at times
after reactor shutdown show that as much as 33%
less heat is generated in fuel from which both
noble gases and noble metals have been removed.
The difference between this rate and the gross rate
is not always this large, but it does average about
20% lower over the first year.
Part 1.

Molten-Salt Reactor Experiment

P. N. Haubenreich

The six-month period reported here was remark-
able for an unprecedented run that began in Sep-
tember and was still going at the end of February,
after more than five months. This run, coupled
with the very successful operation in the preceding
report period, successfully completed the phase of
the experimental program whose principal objective

was demonstration of reliability in sustained opera-
tion.

The first part of this report describes the experi-
ence with operation of the MSRE, development di-
rectly connected with the reactor, and analysis of
the planned operation with 33U fuel.

1. MSRE Operations

P. N. Haubenreich

1.1 CHRONOLOGICAL ACCOUNT OF
OPERATIONS AND MAINTENANCE

R. Blumberg R. B. Lindauer
J. L. Crowley C. K. McGlothlan
R. H. Guymon M. Richardson
P. H. Harley H. C. Roller

T. L. Hudson R. C. Steffy, Jr.
A. 1. Krakoviak B. H. Webster

At the beginning of this report period, the reactor
was down because of trouble with the fuel sampler-
enricher.! The sample latch had been retrieved
from the pump bowl, but the drive unit and isolation
chamber had not yet been replaced. This job was
completed during the first week in September while
the salt systems were being preheated for the re-
sumption of operation,

The first operation in mun 13 was the circulation
of flush salt for six days to permit testing of the

IMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 15.

new sampler mechanism. Also during this time the
coolant salt system was filled, drained for repairs
on the radiator door lifting mechanism, and again
refilled. The fuel loop was filled with fuel, and
full-power operation started on Friday, September
15. During the weekend, an oil leak shut down a
component cooling pump. Operation continued on
the second pump, but rather than start off what was
expected to be a long run without a standby com-
ponent cooling pump, we decided to shut down and
repair the ailing unit.

Repairs took only two days, and on September 20
the fuel system was refilled to start run 14. It was
more than six months before the fuel was again
drained.

Plans were to operate the reactor at high power
for several months in connection with several
studies. These included neutron irradiation of
Hastelloy N specimens in the core to higher flu-
ences than this alloy had ever before received,
measurement of uranium isotopic changes over a
period of substantial burnup to provide information
on cross sections, observation of long-term reac-
ORNL-DWG 68-3938

2
=
1240 ~ ] .
} REACTOR QUTLET TEMPERATURE
\
w4200 4f%r)-F——J — e
| | |
| | R
160 -
10
o NOMINAL FUEL PUMP BOWL PRESSURE ~
, o e e e e T - -
s .0 ! _firy_i\ o ,,.__{,,44__ |
2z 6 E[ T | o
al . R e -
s 1 L _

967

1968

Fig. 1.1. Outline of MSRE Operations, September 1967 to February 1968.

tivity behavior as fission products and plutonium
built in, and continued studies on fission product
distributions. The power history can be seen in
Fig. 1.1.

For the first two months of power operation, the
experimental objectives were pursued practically
without incident. On October 8 and 9, the power
was lowered to 6 Mw to permit a low-temperature,
low-salt-level experiment on gas ingestion into the
circulating fuel. On October 20 the power was
down briefly after an area power failure. Three
days at 10 kw, October 23 to 26, allowed the !3°Xe
to clear out of the fuel so that a reference meas-
urement of the reactivity balance could be made.
The only equipment problem was an oil leak that
shut down a component cooling pump, but opera-
tion continued without interruption on the standby
unit.

Power operation was interrupted in mid-November
by trouble with the wiring in the fuel sampler-en-
richer. Electrical leads between the outer and in-
ner containment boxes shorted out, leaving a sam-
ple stranded in the tube. The reactor power was

lowered to 10 kw while repairs were made. A con-
tainment tent was erected, and a hole was cut in
the outer box. This exposed the fault in the cable,
so it was not necessary to drain the fuel and go
into the inner box. Repairs were made, and tests
showed all circuits were operable except one non-
essential position switch.

While the power was down for the sampler repair,
an abnormal pressure drop developed in the fuel
off-gas line near the pump bowl. The cause was
not known, but to minimize the chance of salt mist
causing a worse blockage while the pressure was
being kept low for the sampler work, the fuel pump
was shut off. After two days, repairs permitted the
sample to be retrieved and the isolation valves to
be closed. The restriction in the gas line was then
blown clear, and nuclear operation was resumed.
For about 15 hr after the reactor was returned to
full power, indications were that the off-gas re-
striction was causing some flow to bypass through
the overflow tank. Then the pressure drop de-
creased and was not detectable for the next two
months.
From the beginning of run 14, the xenon effect at
full power had been about 0.32% &k/k — about what
it was in early power operation but up measurably
from the 0.27% 0k/k observed in runs 11 and 12.
The reason for this small shift was not apparent,
but there had been indications that the xenon strip-
ping was affected to a slight degree by fuel tem-
perature, liquid level in the pump bowl, and system
pressure. Therefore an investigation was started
into the effects of minor changes in these vari-
ables on the stripping of xenon and other fission
gases from the fuel. In order to lower the fuel tem-
perature without bringing the coolant salt below
1000°F at the radiator outlet, the power was first
reduced to 5 Mw. As can be seen from Fig. 1.1,
the experimental operation at various temperatures
and pressures extended over the two months from
the middle of December to the middle of February.

Beginning on January 22, the reactor power was
held at 10 kw for eight days while the temperature
and salt level were varied to observe reactivity ef-
fects in the absence of xenon. Small changes were
observed, reflecting variations in the amount of
bubbles circulating with the fuel salt (around 0.1
to 0.2 vol % normally).

Early in the 10-kw operation the main blower
bearings were inspected, and two were replaced
because of damaged balls. The filter in the cool-
ant off-gas line that had plugged a few weeks ear-
lier was also replaced while the power was down.

After 18 more days of 5-Mw operation, which com-
pleted the planned experiments on gas stripping,
the reactor was returned to full power. Analysis of
the data suggested that the system gas stripping
may have changed characteristics during the ex-

periments. Therefore, on February 29, the power
was again reduced to 5 Mw for a recheck of the
xenon under conditions outwardly the same as
those that had been tested in December.

While reactor operations were proceeding, prep-
arations were being made for the shutdown sched-
uled to begin late in March. These included pre-
paring the chemical processing system for fluorina-
tion of the salt to remove the uranium and design-
ing and building equipment for adding %*3U enrich-
ing salt to the fuel carrier left after the fluorina-
tion.

Operation was continuing at the end of the report
period, more than five months after the startup in
September. Statistics for the six-month period and
totals at the end are given in Table 1.1.

1.2 OPERATIONS ANALYSIS

1.2.1 Reactivity Balance

J. R. Engel

The continuous operation for over five months in
this period afforded an unprecedented opportunity
to study long-term changes in reactivity. The re-
activity balance under the simplest conditions (low
power and no xenon) was checked five times over a
period of substantial fuel burnup without any com-
plications of fuel additions or intervening fuel loop
drains and flushes. In addition, there was time to
vary operating conditions to observe effects on
xenon removal.

Table 1.1. Some MSRE Operating Statistics

September 1967—February 1968

Total Through Feb. 29, 1968

Critical time, hr
Integrated power, Mwhr
Equivalent full-power hours
Salt circulation, hr

Fuel loop

Coolant loop

3891 (89%)
21,823
3014 (69%)

4054 (93%)
4170 (95%)

10,909
62,130
8581

14,415
16,229

Effects of Gas in Fuel Salt. — Prior to the cur-
rent run (run 14) the reactor had operated outside
of a relatively narrow range of operating conditions
for only brief periods of time. The normal condi-
tions were 1210°F at the reactor outlet, 5 psig he-
lium overpressure at the fuel pump, and a narrow
range of fuel salt levels in the pump bowl. Previ-
ous deviations from these conditions showed that
at least temperature and pressure affected the
value of the residual term in the reactivity balance.
However, the effects had not been clearly defined.

During run 14, we performed a series of tests in
an effort to evaluate the effects of fuel system
temperature and overpressure and fuel pump level
on residual reactivity. In these tests the reactor
outlet temperature was varied between 1180 and
1225°F and the overpressure between 3 and 9
psig. Since the fuel pump level is restricted by
the pump hydraulic performance, only the normal
variation in fuel pump level was allowed. This
study was performed at a reactor power of 5 Mw to
provide the required latitude for the temperature
changes (see Fig. 1.1).

Small but significant variations in reactivity were
observed. The major part of the variations was
caused by changes in the '*5Xe poisoning in the
reactor, apparently induced by variations in the ef-
fectiveness of gas stripping caused by the param-
eter changes. Table 1.2 presents a summary of the
net observed '3°Xe poisoning terms at the various
conditions. In general, the xenon poisoning in-
creases with decreasing temperature and increasing
overpressure. However, the pressure effect prac-
tically disappears at the higher temperatures. Con-
versely, the temperature effect is smaller at the
lower pressures. The effect of fuel pump level is
much less pronounced than the temperature and
pressure effects, but decreasing level leads to
higher xenon poisoning for the range of levels in-
vestigated. 2

The details of the changes in '*3Xe poison were
obtained from the reactivity balance. In addition,
several samples of the fuel off-gas were obtained
under various operating conditions, so that the
136Xe/3%Xe ratio might be measured to verify
that the apparent xenon poisoning was in fact due

2 . . .

There is some evidence to suggest lower xenon poi-
soning at very low fuel-pump levels, where the helium
void fraction in the circulating loop increases substan-
tially.

to !35Xe. Preliminary results are in substantial
agreement with the poisoning measured by the re-
activity balance.

Aside from the variations in xenon poisoning,
there was other independent evidence whicn clearly
indicated changes in the effectiveness of removal
of gaseous fission products to the reactor off-gas
system. Poorer removal of the gases coincided
with the higher poison levels. The temperatures
of the off-gas holdup volume in the reactor cell
(L-522) and the particle trap are sensitive indi-
cators of the fission product concentration in the
off-gas stream. At constant system pressure,
where no changes in transit (decay) time are en-
countered, this concentration directly reflects the
effectiveness of the fission product stripping. The
observed temperatures at L.-522 and near the coarse
filtering medium at the entrance to the particle trap
are listed in columns 5 and 6 of Table 1.2.

Another effect observed during these tests was
significant variation in the amount of undissolved
gas in circulation with the fuel salt. These
changes were measured by the small reactivity ef-
fects at zero power with no '*3Xe in the system.
Additional qualitative support for variations in
void fraction was obtained during power operation
from spectral measurements of the inherent neutron-
flux noise and from temperature observations at the
reactor access nozzle. (See also Sects. 1.2.2 and
3.3.) The reactivity balance gave a reactivity loss
of 0.032% 6k/k between the condition with the
fewest circulating voids (1225°F and 3 psig over-
pressure) and that with the most (1180°F and O
psig). This implies a change of 0.15 to 0.2% by
volume in the void fraction between these two con-
ditions. No change in void fraction with fuel pump
level was detected, and there was no observable
pressure effect at 1225°F. However, at 1180°F
over half of the total change between the two ex-
tremes was due to the pressure difference. Col-
umn 7 of Table 1.2 shows the approximate magni-
tude of the additional circulating void fraction at
each condition (The void fraction at 1225°F and
5 psig is thought to be 0.1 to 0.15% by volume.)

The tests in this run indicated at least the qual-
itative nature of the effects of system temperature
and pressure on xenon poisoning and circulating
voids. However, they also demonstrated that the
effects are not necessarily completely reproducible.
During run 12 and earlier runs, the typical value
for !35Xe poisoning at 7.2 Mw was 0.27% Ok/k.
Table 1.2. Effects of Fuel System Pressure, Temperature, and Level on 135xe Poisoning at 5 Mw

Off-Gas System Temperature (OF)

Fuel Pump Reactor OQutiet Fuel Pump Net Xenon? Change from Minimum

Overpressure Temperature Level Poisoning Holdup Volume, Particle Trap, Circulating Void
(psig) (OF) (in.) (% Ok/k) Line 522 Yorkmesh Section Fraction (vol %)
5 1225 6.2 0.182 234 292 0
5 1225 5.6 0.187 228 272 0
1210 6.1 0.206 231 264 0.03
5 1210 5.6 0.219 229 260 0.03
5 1210 5.3 0.233 228 258 0.03
5 1180 5.7 0.336 221 0.11
5 1180 5.6 0.358 220 251 0.11
5 1180 5.3 0.374 219 246 0.11
9 1225 6.2 0.191 236 236 0
9 1225 5.6 0.201 232 231 0
9 1210 6.0 0.227 230 222 0.04
9 1210 5.6 0.232 228 224 0.04
9 1210 5.3 0.241 224 217 0.04
9 1195 5.9 0.263 221 214 0.10
9 1195 5.6 0.289 214 202 0.10
9 1195 5.3 0.346 204 180 0.10
9 1180 5.6 0.371? 193° 152° 0.18
9 1180 5.3 0.374" 195° 155° 0.18
3 1225 6.2 0.175° 187 244 0
3 1225 5.6 0.182° 193 227 0
3 1180 5.8 0.240° 185° 230° 0.10
3 1180 5.5 0.248" 188° 226° 0.10
3 1180 5.3 0.260° 189° 220° 0.10
5 1180 5.6 0.294° 217 265 0.11
5 1180 5.3 0.295° 216 255 0.11

fCorrected for drift in zero-power residual reactivity and for direct reactivity effect of differences in circulating
void fraction.

bThese data were taken after a period of operation at 10 kw, during which the pattern of xenon behavior changed
toward less pressure-temperature-level sensitivity.

®Partial restriction in off-gas line at fuel pump caused change in flow path and holdup time.
For the first four months of run 14, the character-
istic value for this term was 0.33% Ok/k. After the
period of low-power operation in January 1968, the
xenon term reverted to the old value. Since this
change occurred in the middle of the 5-Mw tests to
define xenon behavior, no explicit quantitative
evaluation can be made. This is illustrated by two
nominally identical tests made under conditions of
high xenon poisoning before and after this change.
Tests on December 23 to 25, 1967, and March 3 to
6, 1968, both involved operation at 5 Mw and
1180°F with 5 psig overpressure. The respective
xenon poisoning terms were 0.35 and 0.30% &6k/k.
Long-Term Changes. — Because of variations in
the xenon poisoning, the best data on the long-term
drift in residual reactivity are obtained at very low
power with no xenon present. In earlier operations,
these data were taken only near the beginning and
the end of the various reactor runs, and many runs
were relatively short in terms of fuel burnup and
net control rod movement. Comparison of data be-
tween runs was further complicated by the need to
compensate for the dilution effects of fuel drains
and flushing operations. All these results were
scattered within a relatively narrow band of reac-
tivity values with no particular trend in evidence.
During the six months of run 14, zero-power reac-
tivity data were collected on five separate occa-
sions. These data showed a very small but remark-

ably constant negative trend in residual reactivity
as a function of integrated power. The average re-
activity slope during run 14 was —3.5 x 10~ %%
ok/k)/Mwhr, Reexamination of earlier data showed
considerable scatter around this value but no real
inconsistency with it. Some of this scatter oc-
curred because significant scatter was encountered
between individual reactivity balances at some
conditions. Figure 1.2 shows the zero-power reac-
tivity balance results for the entire 235U operation.
Data points taken within a particular reactor run,
that is, after a reactor fill and before the subse-
quent drain, are connected by straight lines. Where
lines are drawn from two points at a given condi-
tion, the points represent the extremes of signifi-
cant reactivity balance variations at that condition.
At conditions where only trivial variations were
observed, only a single point was plotted.

If the negative reactivity slope observed in run
14 were applied to the entire power history, a total
negative deviation of 0.22% &k/k from the begin-
ning of operation would be expected. The fact that
a total change of this magnitude has not been ob-
served suggests the presence of errors that tend to
compensate for the negative drift observed during
operation. A possible source of such an error is
the corrections that must be applied for the dilu-
tion effects of reactor drains and flushes between
runs.

ORNL -DWG 68-3939

RESIDUAL REACTIVITY (%Ak/k)

20 24 28
INTEGRATED POWER (Mw hr)

32 36 40

60(x10°)

Fig. 1.2, Residual Reactivity at Zero Power.
A reactor flush results in a net transfer of ura-
nium (and other fuel constituents) from the fuel
salt to the flush salt. If the amount of uranium
transferred in such a flush is overestimated in the
correction that is applied to the reactivity balance,
the residual reactivity is shifted in the positive
direction — opposite to the direction of drift during
operation. An overcorrection of 2.6 kg in total ura-
nium would be required to compensate for a drift of
—0.22% 0k/k. To date, total uranium corrections
in the amount of 4.9 kg have been applied to the
reactivity balance; analyses of the flush salt when
it was last in the fuel loop indicated a total ura-
nium content of 5.0 kg. In applying the calculated
corrections, it is assumed that the total volumes
of fuel salt and flush salt remain constant. A gain
of 0.7 ft* in the actual flush salt inventory at the
expense of the fuel salt could also conceal the ex-
pected negative reactivity drift. However, the sys-
tem physical inventories indicate that if there has
been any shift at all, it has been to increase the
fuel salt volume at the expense of the flush salt.
Thus it does not appear that errors in corrections
between reactor runs are compensating for the
downward drift during the runs. The source of the
compensating effect has not yet been identified.

There are a number of possible causes for the
negative reactivity drift during any one run. How-
ever, none of these appears to be large enough to
explain the drift. For example, if the drift were
due to an error in power calibration, an error of
about 30% would be required. The data on the
buildup of 23U in the fuel (see Sect. 1.2.5) indi-
cate that the power calibration error is less than
10%. This is further supported by the rate of
239py buildup in the fuel and various fission
product studies. Another possible source of error
is nonuniform deposition of fission products in the
fuel loop. Even if all the nonvolatile nonsaturat-
ing fission products were deposited on the core
graphite (uniformly), the negative reactivity effect
would be less than one-half the drift rate in run 14.
Therefore the small negative trend in reactivity
remains unexplained at the present time. It is
worth noting, however, that the magnitude is well
within the allowable reactivity anomaly of 0.5%
Ok/k. (This limit is based on reactor safety
studies that placed conservative interpretations
on reactivity losses and their possible conse-
guences.)

1.2.2 Variations in Reactor Access
Nozzle Temperatures

J. L. Crowley

The reactor access nozzle is one of the two
places in the fuel loop where there is a gas-liquid
interface; the other is the fuel pump bowl. Varia-
tions in liquid level in the access nozzle, which
can be inferred from thermocouple readings, afford
some information on the gas circulating with the
fuel,

As shown in Fig. 1.3, the reactor access nozzle
(RAN) is a 10-in. pipe with a cooling-air jacket
on the outside. Inside is an air-cooled plug at-
tached to the closure flange. In the large plug is
a smaller nozzle and an air-cooled plug, providing
access to the sample assembly in the core. It was
intended that frozen salt plugs be established in
the annuli between the cooled plugs and nozzles
to prevent molten salt from coming in contact with
the metal ring seals at the access flanges. In ac-
tual operation, however, the indications are that it
is trapped gas in the annulus and not a frozen salt
plug which prevents liquid salt from rising to the
seal area.’

Two mechanisms transport gas to and from the
RAN annuli. Owing to the higher static pressure
at the RAN, helium is transferred from the RAN to
the pump bowl gas space as a solute. This effect
was previously noted on the Engineering Test Loop
in the operation of the forerunner of the reactor ac-
cess nozzle.* The mechanism which delivers gas
to the RAN is the collection of circulating bubbles
from the fuel passing by the lower ends of the an-
nuli. The equilibrium condition between these two
mechanisms determines the gas inventories and the
liquid levels in the RAN annuli. Changes in the
liquid levels can be detected by monitoring the
thermocouples attached to the RAN. In general it
has been noted that the RAN salt levels run higher
when the system operates at lower pressure or
higher temperature.

An example of the above effect is provided in
Fig. 1.4. This is a plot of two RAN thermocouples
(see Fig. 1.3 for location), the fuel pump helium

SMSR Program Semiann. Progr. Rept. Feb. 28, 1965,
ORNL-3812, p. 9.

MSR Program Semiann. Progr. Rept. July 31, 1963,
ORNL-3529, p. 40.
ORNL-DWG 68-5507

£ <

- 0

@ 0

URGLD

I I J I Hfifinfi J 1
7z A = 77 I
o e S == S e e
B ) A

D

[

TE
R-43B

FUEL SALT

COOLING AIR

1

10-in. ANNULUS

24/5-in. ANNULUS

I Y Y Y S R I sl B
N A S/ A 4
SN AN~ 7
B S P e O s ' O e O W A

Fig. 1.3. Reactor Access Nozzle Showing 10- and 21/2-in. Annuli,
ORNL—DWG 68—-5508

[ saLt |
TEMPERATURE ¢ —IT==

S O M S e
/TE R—43B /

e i e v e —— — —

TEMPERATURE
(°F)
©
e}
(@]
N

© T T ]

FUEL PUMP PRESSURE

. T T ]

6 7 8 ] 10 1t 12 13 14 15 16
DATE. BEGINNING FEB 6.1968

PRESSURE
(psig)
(6]

Fig. 1.4, Effect of Reactor Outlet Temperature and
Fuel Pump Pressure on Reactor-Access-Nozzle

Temperatures.

pressure, and the reactor outlet salt temperature
for a period of ten days in February. The test in
progress at that time was one to determine the ef-
fects of pressure and temperature on reactivity
while operating at 5 Mw power. The fuel pump he-
lium pressure was first lowered from 9 to 5 psig
with the reactor outlet at 1180°F. On February 10
and 11, the reactor outlet was increased to 1225°F,
while the fuel pump pressure was maintained at 3
psig. At the beginning condition, that is, high sys-
tem pressure and low salt temperature, the molten-
salt levels in the annuli are below the levels of
the two thermocouples. When the pressure was
lowered on February 6, RAN temperatures dropped
initially, indicating the lowering of the salt level
due to expansion of gas trapped in the RAN annuli.
Very soon, however, the molten salt rose to or
above the R43B thermmocouple location. Thus it
would appear that bubbles were bringing gas into
the annuli at a lower rate. The next changes, in-
creases in circulating fuel temperature on February
10 and 11, also resulted in rises in the molten-salt
level in the annuli. After the second temperature
increase, the level apparently rose over a period
of about two days, until it was near thermocouple
R7B. Thus the increase in circulating fuel tem-

perature also appeared to reduce the amount of
bubbles bringing gas to the annuli.

The variations in circulating gas concentrations
with changes in fuel temperature and pressure, in-
ferred from RAN temperatures, agree with the
changes indicated by the reactivity (see Sect.
1.2.1).

1.2.3 Radiation Heating
C. H. Gabbard

The temperature differences between the reactor
inlet and the lower head and between the inlet and
the core support flange continue to be the indicator
for possible sedimentation buildup within the reac-
tor vessel. These temperature differences agree
satisfactorily with previous data and indicate that
there has been no significant solids accumulation
within the reactor vessel. Table 1.3 shows a com-
parison of the data collected during run 14 with
those collected and reported previously.

The temperature distribution on the upper surface
of the fuel pump tank during run 14 was about the
same as that in run 12. As reported previously, ’
the temperatures are somewhat lower than in ear-
lier power runs, and the reason for the difference
has not been established.

Table 1.3. Power-Dependent Temperature Differences
Between Fuel Salt Entering, and Points on, the

Reactor Vessel

Temperature Difference (OF/MW)

Run No. Dates
Core Support Flange Lower Head
6 4/66—-5/66 2.11 1.54
11 1/67-5/67 2.14 1.50
12 6/67—-8/67 2.20 1.55
14 9/67-2/68 2.25 1.42

1.2.4 Thermal Cycle History
C. H. Gabbard

The accumulated themmal cycle history of the
various components sensitive to thermal cycle

5MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 22.
10

Table 1.4. MSRE Cumulative Thermal Cycle History Through February 1968

Thaw and
Component Heat/Cool Fill /Drain Power On /Off Thaw Transfer

Fuel system 9 40 66

Coolant system 7 13 62

Fuel pump 12 35 66 447

Coolant pump 8 14 62 127

Freeze flanges 100, 101, 102 9 36 66

Freeze flanges 200, 201 8 13 62

Penetrations 200, 201 8 13 62

Freeze valve
103 7 29 62
104 15 9 27
105 17 18 45
106 19 28 38
107 10 11 18
108 9 17 14
109 9 20 18
110 2 2 3
111 S 4 4
112 2 1 2
204 9 15 32
206 9 13 30

damage is shown in Table 1.4. Approximately 69%
of the design thermal cycle life of the fuel system
freeze flanges has been used to date. This com-
pares with a value of 63% of the design life that
had been consumed at the end of the previous re-
port period. Essentially no thermal cycle damage
has been accumulated on the fuel system flanges
since the fuel system was last filled on September
20, 1967, at the beginning of run 14.

1.2.5 236y Indication of Integrated Power

R. C. Steffy, Jr. J. R. Engel

The basic measurement of the nuclear power of
the MSRE is by the overall system heat balance.

The preponderant term in this balance is the heat
removed from the system by the secondary (coolant)
salt. Its accuracy is therefore strongly dependent
on the accuracy with which that term is evaluated.
Other indications of reactor power — radiator air
heat balances, fission product inventories, and nu-
clear instrument readings — have provided only
reasonable (£20%) confirmation of the heat-balance
power calibration. Since all these techniques
could involve substantial systematic errors, an ef-
fort has been made to check the power calibration
by an independent technique.

A potentially accurate standard as a monitor for
integrated power in a nuclear reactor is the change
in the isotopic composition of the uranium fuel.
The accuracy of this method depends on the exist-
ence of an isotope whose concentration changes
significantly and in a manner that can be accu-
rately related to energy production. The 236U in
the MSRE fuel is well suited to this purpose. The
fraction of 23U in the uranium mixture has in-
creased by a factor of 3 in the course of power op-
eration with the 235U fuel. Since 23%U is produced
only by parasitic neutron captures in 23°U and
since the rate of burnup of 236U is very low, the
net production of this isotope per fission event
depends primarily on the value of a (= 0./a,) for
235U in the MSRE neutron spectrum. Extension
of this relationship to total energy production de-
pends only on the effective energy yield per fis-
sion in this reactor.

The absolute rate of production of 236U in the
MSRE was inferred from several fuel samples in
which the uranium isotopic composition was meas-
ured. These data were combined with ‘“book’’ in-
ventories for total uranium and corrected for vari-
ous reactor reactions (drains, flushes, and fuel
additions) to obtain loop inventories for 23°U.
Since chemical results for total uranium have con-
sistently been within 1% of the ‘‘book’’ values and
the precision of the isotopic assays is high, the
accuracy of the 23%U inventory is probably within
+5%. Figure 1.5 shows the calculated 23°U inven-
tory plotted against the integrated power based on

ORNL-DWG 68- 5509

1000

TOTAL 238U IN SYSTEM (g)

20 30 40
INTEGRATED HEAT-BALANCE POWER (Mwhr]

Fig. 1.5. 236U Buildup in MSRE vs Power Production.

11

the heat balance measurements., The solid line is
a least-squares fit to the data points and has a
slope of 0.0110 t 0.002 g of ?3%U per megawatt-
hour. The theoretical slope, based on the ratios
of effective 235U neutron cross sections and fis-
sion energy yield in the MSRE, is 0.0109 g/Mwhr,
within 1% of the observed slope. The uncertainty
in the theoretical slope is probably less than 10%,
so the actual integrated power is probably within
10% of the value indicated by the heat balances.

1.3 EQUIPMENT PERFORMANCE

1.3.1 Salt Pumps

P. N. Haubenreich

The 1200-gpm pump that circulates fuel salt and
the 850-gpm pump that circulates coolant salt con-
tinued to run uneventfully. By the end of the re-
port period the fuel pump had run for 18,400 hr,
14,415 hr circulating salt and 3985 hr circulating
helium. The coolant pump had pumped salt for
16,229 hr and helium for 3082 hr, for a total of
19,311 hr. As reported in Sect. 1.3.6 and 1.3.10,
there is probably some oil leakage into the pump
bowls, but the rates must be very small since they
do not produce significant changes in the oil inven-
tories. The replacement rotary elements for both
pumps are seal-welded to prevent such inleakage,
but the oil problem is so minor that replacement is
not contemplated.

1.3.2 Heat Transfer
C. H. Gabbard

The heat transfer performance of the fuel-to-cool-
ant-salt heat exchanger continues to be monitored
by periodic evaluations of the heat transfer coeffi-
cient and by the heat transfer index.® The heat
transfer index was evaluated a number of times
during run 14 with the reactor at full power. How-
ever, only one additional heat transfer coefficient
measurement over a range of powers was made dur-
ing this report period. This measurement gave a
value of the overall heat transfer coefficient of

GMSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, p. 21,
605 Btu hr—! ft—2 (°F)~ !, compared with 586 Btu
hr=! ft=2 (°F)~ !, which is the average of the six
previous measurements. The average heat transfer
index increased during run 14 to 0.0376 Mw/°F,
which is also slightly higher than the previous
data. The apparent increase in heat transfer was
caused entirely by a recalibration of one of the
coolant salt flow elements, which increased the
indicated coolant salt flow rate and the calculated
heat removal rate at the radiator. (This increased
the calculated heat balance power by about 100 to
200 kw at full power.) Thus the conclusion is that
the heat transfer performance of the heat exchanger
has remained unchanged since the beginning of op-
eration.

1.3.3 Salt Samplers
R. B. Gallaher

During this six-month repotrt period, operation of
the sampler-enricher was resumed with a new isola-
tion chamber and cable drive assembly. Except for
the wiring failure that occurred in November, only
minor difficulties were encountered. A total of 70
sampling operations were made during the report
petiod, as follows:

10-g salt samples 45
50-g salt samples 16
Salt samples in freeze-valve capsules 2
Gas sample in freeze-valve capsule 1
Capsule exposure in gas space 5
Nickel rod exposure in salt 1

Three of the 10-g samples were of flush salt; the
others were of fuel salt, The 50-g samples were
taken for uranium isotopic analyses, determination
of oxide level or U3¥/U** ratio, and other special
analyses, as reported in Part 3 of this report. All
the sampling attempts were successful, except that
one 50-g capsule collected only 5 g of fuel salt.
The attempt appeared normal in every other respect,
and the explanation for the small amount of salt is
not known. Operations during the period brought
the total use of the sampler-enricher to 114 uranium

12

enrichments and 349 samples and special exposures.

Since the maintenance period that ended in Sep-
tember, the pressure transducer in the removal

valve buffer system has been sensitive to heat from
the illuminator, drifting by 25% of full scale when
the light is turned on. A reflective metal shield
was placed over the thermal insulation around the
transducer. This helped, but it did not eliminate
the effect.

The wiring fault appeared in the following man-
ner. On November 14, as a sample was being with-
drawn from the pump bowl, the 0.3-amp fuse in the
drive motor circuit blew. Subsequent tests showed
that the insert mode drew the normal 0.2 amp, but
in withdrawal the current rose to 1.0 or 1.5 amp.
Since the locked-rotor current is only 0.3 amp, the
high current indicated leakage. During further
tests, three circuits opened, disabling the motor
“‘insert’’ and ‘‘withdraw’’ circuits and the upper
limit switch.

The most likely location for the failure was be-
lieved to be at or between the outer and inner con-
tainment penetrations. This area was suspected
because of its small clearances and tight bends in
the cable. High-frequency capacitance measure-
ments with a Grid-Dip meter indicated that the
length of cable to the point of failure was consist-
ent with this guess. Therefore, after a nonflam-
mable plastic tent was erected over the samplet-
enricher to prevent spread of possible contamina-
tion, a 3-in.-diam hole was sawed in the cover
plate directly above the cable penetration into the
inner box (1-c area). The failure was found where
the wires were bent back down against the side of
the plug on the lower end of the cable between the
inner and outer boxes. A temporary connection
showed that everything inside the inner box was
operable. The sample capsule was retrieved, and
the isolation valves were closed at this time. The
damaged section of cable was abandoned, and a
new cable was installed having a penetration
through a 4-in. pipe cap welded over the sawed
hole in the top cover.

While the cap was being welded in place, a heavy
current evidently went to ground through another
cable that penetrated the top cover near the new
cap. The receptacle and plug were destroyed. Re-
pairs were made by cutting out and replacing this
penetration. As a result of this experience, an
isolation transformer and a fuse were added to each
of the three drive unit cables (see Sect. 3.1.3).
This allows one ground without interference with
operation.
After the repairs, all circuits were operable ex-
cept the upper limit switch. This switch normally
stopped the drive motor when the latch reached the
latch stop. Because the motor and circuit can
stand blocked-rotor current, the upper limit switch
was bypassed, and the motor is now turned off man-
ually when the position indicator shows the latch
is fully withdrawn.

Containment has been quite good. A minor re-
lease of activity to the stack occurred at least
once while area 3A was being evacuated. After all
disconnects in the sampler off-gas system were
checked and tightened, no further detectable re-
lease occurred.

During the report period the coolant salt sam-
pler was used to take five 10-g samples, bringing
the total to 65. No operating difficulties were en-
countered.

1.3.4 Control Rods and Drives

M. Richardson

The control rods continued to operate freely
throughout the report period. Tests in September
and late January showed no shift in position read-
ing and no appreciable change in rod drop times.
The only mechanical difficulty was with the fine-
position synchro transmitter on drive No. 2, which
developed an open circuit in January. This syn-
chro was one that was installed in May 1967, so
its service life was relatively short for some rea-
son as yet unknown.

1.3.5 Radiator Enclosure

M. Richardson

During the flush salt operation in September, at
the beginning of run 13, the outlet door tended to
jam because it was hanging crooked. The trouble
appeared to be in one of the housings for the
springs through which the cables lift the door. A
piece of some material, apparently weld slag, had
fallen in and jammed the piston in that housing,
preventing it from moving up. The piston in the
other housing was free. Both coil springs were
found to have become pemanently compressed,
probably because of subjection to high tempera-
tures in the startup period before hoods were in-
stalled over the doors. Since the springs were not

essential, they were replaced with pipe sleeves
that hold the pistons down.

The brakes in the lifting mechanism remained
adequate, limiting coastdown to about 3 in. The
door seals continued to provide adequate heat con-
tainment with the doors closed.

1.3.6 Off-Gas Systems

A. I. Krakoviak

The off-gas systems of both the fuel and coolant
system developed partial restrictions that caused
minor inconveniences but did not interfere with
normal power operation.

Fuel Off-Gas System. — A restriction became evi-
dent in the off-gas line somewhere between the
pump bowl and the junction of the overflow tank
vent with the 4-in. holdup line. The first indica-
tion appeared after 27 days of operation at 10 kw
following the sampler-enricher wiring failure of No-
vember 14. During this period the operational and
maintenance valves were open, and a 0.6-liter/min
helium purge was maintained down the sampler tube
to the pump bowl to prevent contamination of the
sampler by fission gases. The restriction was evi-
denced by an increase (0.5 psig) in the fuel pump
bowl pressure when the overflow tank vent valve
was closed during a routine return of salt from the
overflow tank to the pump bowl. After the wiring
repairs had been finished, the sample had been re-
trieved into area 1C, and the operational and main-
tenance valves had been closed, the restriction
was relieved by pressurizing the pump bowl to 6.0
psig and suddenly venting the gas into a drain tank
which was at 3 psig. The pressure drop then ap-
peared normal (<0.1 psi), but after three more days
of low-power operation, an abnormal pressure rise
(0.2 psi) was again observed when salt was being
returned from the overflow tank. Repetition of the
mild blow-through to the drain tank had little or no
effect this time. The line was not completely
blocked, however, and full-power operation was re-
sumed. At first, temperatures on the overflow tank
and the off-gas line (responding to fission product
heating) clearly indicated that there was enough
pressure drop at the pump bowl! outlet to cause
much of the off-gas to bubble through the overflow
tank and out its vent line. Then, after 15 hr at full
power, the bypass flow stopped, indicating that the
pressure drop through the restriction had decreased
significantly.
The pressure drop remained below the limits of
detection throughout the next nine weeks while the
reactor was operating at 7.2 Mw or 5 Mw. Then
after two days at 10 kw, the pressure again became
detectable and continued to increase over the next
six days at very low power. When power operation
was resumed at 5 Mw, temperatures indicated that
there was again bypass flow through the overflow
tank. The restriction increased during two weeks
of operation at 5 Mw, but the line never became
completely plugged. Then while the overflow tank
was being emptied, four days after the resumption
of full-power operation, the restriction partially
blew out, bringing the pressure drop again below
the limit of detection. The pressure drop remained
just at or below the limit thereafter.

The behavior was quite unlike the plugging that
occurred earlier in the same section of line as a
result of the overfill with flush salt,” mainly in
that it never completely plugged and there was a
suggestion of some effect of power level. Plans
were made to investigate this restriction in the off-
gas line at the next shutdown. Tools were prepared
like those used earlier to clear out the off-gas line
at the pump bowl. Preparations were made to re-
move the short flanged section of the off-gas line
for examination and replacement with a line instru-
mented with thermocouples.

Main Charcoal Beds. — The performance of the
charcoal beds was very good; none of the annoying
plugging problems experienced previously®:? ap-
peared during this report period. During run 13 and
the early part of run 14, with sections 1A and 1B
in service, the maximum pressure drop across the
beds in parallel was 3.0 psi.

The main charcoal beds were conservatively de-
signed to delay the krypton and xenon long enough
(7 days for Kr, 90 days for Xe) so that the major
radioactive constituent in the effluent gas would
be 10-year ®°Kr. The beds were designed to pro-
vide these holdup times when the off-gas flow was
equally divided between two sections in parallel,
but there were some indications during early opera-
tion that adequate holdup might be provided with
all the flow through only one section. Therefore,
in November a test of the capability of a single

"MSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, pp. 27-209.

8bid., p. 30.

9MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 28.

14

section was started, with only section 1A on line.
Only a very moderate rise in effluent activity ma-
terialized. Section 1A remained in service alone
for the remainder of the run with the exception of
a ten-day reactivity test at low fuel pump pressure,
when two beds were in service to permit lower
pressure.

Coolant Off-Gas System. — The coolant pump off-
gas system has experienced chronic plugging at the
small 50-u sintered-metal filter upstream of the
pressure control valve. This filter has plugged
regularly and has been replaced seven times over
the three years since the start of preoperational
checkout of the system. The plugging appears to
be due to liquefaction of oil vapors from the cool-
ant pump in the pores of the sintered metal. The
filter plugged again during this report period, and
the coolant pump pressure was controlled by vent-
ing through a line bypassing the plugged filter.? '°
The output from the coolant pump pressure control-
ler was switched from the normal control valve to
the vent valve to give satisfactory pressure control.

1.3.7 Main Blowers
C. H. Gabbard

The two main blowers, MB-1 and MB-3, continued
to run when needed and by the end of the report pe-
riod had accumulated 7361 and 6685 hr, respec-
tively, since they were rebuilt in the fall of 1966.
The main bearings on both blowers had to be re-
placed in January, but it happened that there was
no interference with reactor operation. The reactor
was being operated at 5 Mw on one blower, MB-3,
when routine monitoring showed the bearing vibra-
tion increasing from 0.8 to 1.5 mils. When an os-
cilloscope trace showed abnormal high-frequency
noise in the bearing, MB-1 was brought on line and
MB-3 shut down. A week later, the reactor power
was lowered to 10 kw as part of the reactivity ex-
periments; so for several days neither blower was
needed. Although the vibration meter on MB-1
bearings had shown no trouble during operation,
there was some periodic noise and roughness that
could be detected during a coastdown. Therefore,
while the power was down, the thrust bearing on
each blower was replaced.

In the MB-1 bearing, one ball had a relatively
large surface pit, accounting for the noise and

107pid., pp. 28-29.
MB~A

15

PHOTO 90647

MB-3

Fig. 1.6. Ball Bearing Failures from MSRE Main Blowers MB-1 and MB-3.

roughness. The MB-3 bearing was in much worse
condition, with one ball actually fractured into four
large pieces. The damaged balls from the two bear-
ings are shown in Fig. 1.6.

The MB-1 bearing was the one supplied with the
rebuilt blower in the fall of 1966 and had seen
6633 hr of service. The MB-3 bearing was a re-
placement, installed in March 1967, that had ac-
cumulated 4801 hr. Lubrication appeared to have
been adequate, so the relatively short service life
could not be attributed to improper lubrication
(which was the suspected cause in an earlier bear-
ing failure). Analysis of the bearing loading and
stresses revealed that according to accepted de-
sign correlations the expected service life was
only 5000 hr, about what was actually attained.
The original bearings were radial-type ball bear-
ings in which thrust load produced high stresses.
Therefore, they were replaced with angular-contact-
type ball bearings having a greater thrust capacity
and much longer expected life.

1.3.8 Heaters
T. L. Hudson

While the reactor system was being heated up for
run 13, heater FV-103 was lost by an open circuit.
This 1.5-kw bent-tubular-type heater normally sup-
plies about 200 w of heat to a 4-in. section of the

fuel drain line within the reactor vessel furnace,
between the freeze valve and the resistance-heated
section of the line. The purpose of the heater was
to help control the temperature profile through the
freeze valve. After the heater failed, tests with
flush salt showed that by proper adjustment of the
cooling air controls, the freeze valve could be
maintained reliably with thaw times in an accept-
able range. (Thaw time was around 12 min when
the reactor vessel was at 1180°F and the center
of the freeze valve was at 495°F.) Therefore this
heater was not replaced.

The last of six heating elements in heater HX-1
on the primary heat exchanger failed during the
previous semiannual report period. In December
a lead to the adjacent heater, HX-2, opened, thereby
reducing the output of this heater by 50%. A week
later a second partial failure further reduced the
output to only 33% of normal. Although these fail-
ures do not affect operations so long as salt is
kept circulating, the lack of heat in the two adja-
cent heaters will make it necessary to repair or re-
place them during the next shutdown.

1.3.9 Electrical System
T. L. Hudson

Power to the MSRE electrical system is supplied
from the ORNL substation by either of two 13.8-kv
power lines, a preferred line or an alternate.
Originally the two main blowers could not be op-
erated while the area was on the alternate feeder,
but rearrangement of feeders at the ORNL substa-
tica reduced the load on the altemate supply to
the MSRE. During this report period the reactor
was operated for the first time at full power while
on the alternate feeder. The occasion was a man-
ual switchover for 2 hr, while a defective insula-
tor was replaced at the substation.

A building power failure occurred on a fogyy
moming in October, when an arc developed between
a 13.8-kv line to the main transformer for the build-
ing and a parallel metal activator rod to the 13.8-kv
line fuse. The cause of the arc is unexplained:
Although the weather was foggy, it was not un-
usually wet. This fault caused the operation of
the preferred feeder overcurrent ground relay lo-
cated at the ORNL substation, which tripped the
feeder breaker before the line overcurrent relay lo-
cated at the MSRE could operate to prevent an au-
tomatic transfer to the alternate feeder. When the
MSRE was transferred, the alternate feeder over-
current ground relay operated and the alternate
feeder breaker also tripped. (To prevent an auto-
matic transfer to the alternate feeder on a similar
fault, an overcurrent ground relay was later in-
stalled at the MSRE.) After an interruption of 37
min, low-power nuclear operation was resumed on
emergency electrical power from the diesel gener-
ators. The damage was repaired, the spacing be-
tween the line and the activator rod was increased,
and normal service was restored after an interrup-
tion of 2 hr.

The three diesel generators at the MSRE have
proved to be quite reliable. They are started and
operated unloaded for about an hour each week.
Once a month they are tested under load. They
have never failed to start when required during a
power outage. Under emergency conditions they
are usually running within 2 min after a failure of
the normal power supply.

1.3.10 Salt Pump Oil Systems

A. I. Krakoviak

The lubricating oil systems for both salt pumps
operated continuously and without incident through-
out this six-month report period. The only prob-
lem was the recurring, gradual fouling of the water

16

side of the cooling coils on the oil reservoirs. By
December the temperature of the oil supplied to
the pumps was up to 150°F; so the water supply
was changed from tower water to cooler process
water. During about two months on process water,
the heat transfer improved to the point that tower
water again gave acceptable cooling.

Accumulation of shaft seal oil leakage from the
fuel pump was steady at about 12 cm3/day from
September through November. For the next two
months there was no measurable accumulation in
the oil collection tank. Then about February 10,
leakage began to collect again and averaged 6
cm3/day for the remainder of the month. Oil ac-
cumulation from the coolant pump shaft seal leak-
age also varied. In September and October it was
only 7 cm3/day. The rate increased in November
to 15 cm3/day, and in the last three months was
relatively steady at about 20 cm3/day. These ac-
cumulation rates are in the range observed in ear-
lier operation of the salt pumps.

During power operation, samples were taken at
weekly intervals, and they showed no significant
deterioration of the oil quality. To compensate for
these samples and shaft seal leakage, 2.8 gal of
oil was added to the fuel pump lubrication system
and 2.0 gal to the coolant pump system. Inven-
tories based on additions, sample removals, and
accumulation of seal leakage showed, during this
six-month report period, a net apparent gain in the
coolant oil system of 114 cm? and a net apparent
loss from the fuel oil system of 545 cm3. Because
of inaccuracy in the inventories and a slight sys-
tematic error as unmeasured losses in sampling
the fuel pump oil, the probable error in net change
is about 1200 cm®. Thus the apparent changes
are well within the probable error.

1.3.11 Cooling Water System

P. H. Harley

Operation of the cooling water systems continued
satisfactorily throughout the report period. The
only disturbance of normmal operation occurred when
a 6-in. cast-iron pipe supplying water for the build-
ing fire protection system and the potable water
system cracked underground just outside the build-
ing. To isolate the leak, it was necessary to shut
off both this line and the line supplying the process
cooling water system. Adequate cooling was main-
17

tained while the break was being repaired by shift-
ing some items from process water to tower water
and running a temporary hose connection from the
water main to the process water system. A dry fire
hose was laid from the nearby Nuclear Safety Pilot
Plant for possible emergency use.

Although leak tests in June 1967 detected no
leakage from treated-water components in the reac-
tor cell, a small leak apparently persists. Through-
out this six-month period, condensate collection in
the component cooling system averaged about 0.73
gpd. There was no accumulation of water in the

reactor cell sump. Measured losses from the treated-

water system averaged 1.5 gpd. About 0.75 gpd is
lost by evaporation from the degassing tank that
strips radiolytic hydrogen and oxygen from the
treated water. Thus the unmeasured loss agreed
closely with the condensate collection rate.

During a short period in December the treated-
water loss increased to about 4 gpd. The increase
was traced to a leak across a pressure-relief valve
which discharges to the waste system to prevent
excessive pressure buildup in the event the radia-
tion block valves close. The leaking valve was
replaced.

1.3.12 Component Cooling Systems

P. H. Harley

Oil leaks in the lubrication systems of the main
component cooling pumps were troublesome during
this period. In run 13, after six days of flush salt
circulation and two days of nuclear operation,
CCP-2 was shut down by low oil pressure, leading
to the decision to terminate the run (see Sect. 1.1).
Repairs consisted in capping a leaking drain line
and tightening several packing nuts and fittings.
The other blower, CCP-1, was used for the first
873 hr of run 14. At that time a serious oil leak
was indicated by low oil pressure and accumula-
tion of about 2 gal of oil in the condensate col-
lection tank connected to the blower containment.
The standby unit, CCP-2, was immediately started
up and operated without further difficulties through-
out the remainder of the period, 3036 hr.

The strainer in the discharge of the component
cooling pumps developed an excessive pressure
drop during run 13, and the screen was replaced
at the time of the repairs to CCP-2. The strainer
basket that was removed had been installed before

run 12. In 1800 hr of operation its 100-mesh
screen had become partially plugged with rubber
dust from the drive belts. The screen was damaged
in removal, and a replacement basket was made
using 16-mesh 0.023-in.-wire screen. This screen
showed no pressure buildup over the five months to
the end of the period.

Cooling air for the coolant salt freeze valves
was supplied most of the time by the service air
compressor, AC-3. Component cooling pump 3 was
kept in standby and used only when AC-3 was un-
dergoing programmed maintenance. Once while
CCP-3 was in service, a bearing seized, requiring
replacement of the bearing and drive belt. The
blower had been operated for a total of only about
200 hr since a similar bearing failure.!!

1.3.13 Containment and Ventilation

P. H. Harley R. C. Steffy, Jr.
Throughout all the operations in this period the
reactor and drain tank cells were held near -2 psig.

The air inleakage into these cells was about 15
scfd, less than the acceptable maximum by about a
factor of 5.

Inleakage rates are measured by a balance that
includes purge and exhaust flows and changes in
cell temperature and pressure. At the beginning of
runs 13 and 14, a familiar pattern was repeated.
The pattern consists of three stages:

1. In the first day after the cell is closed, while
temperatures are coming up to equilibrium, the
indicated inleakage rate is quite high, some-
times as much as 200 scfd.

2. For three days or so after the first stage, the
indicated inleakage stays up around 75 to 100
scfd. This high apparent rate is attributed to
gradual saturation of the cell atmosphere with
water vapor.

3. When the cell atmosphere becomes saturated at
the component cooling pump cooler, condensate
begins to appear there, and the indicated cell
leak rate breaks down shatply to the actual in-
leakage rate, usually 15 to 20 scfd.

11MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-41091, p. 29.
The cell atmosphere is maintained at about 3%

oxygen by supplying a continuous purge of nitrogen.

Oxygen analyzer readings can, in principle, be
used with other data to compute air inleakage.
However, assumption of no oxygen consumption in
the cell led to a calculated negative inleakage of
air. Therefore the inleakage measured by the
flows and pressure changes was assumed to be
correct, and the oxygen analyses were used to
compute oxygen consumption in the cells. Results
showed an apparent consumption of 3 to 4 scfd of
oxygen, presumably by reaction with some sub-
stance in the cells.

Moisture condensed in the component cooling
system (which is an extension of the reactor cell)
is drained once a day. Collection rates averaged
0.73 gpd over the five months of run 14 in this re-
port period. Daily amounts ranged from practically
none to several gallons when the temperature out-
doors (and in the special equipment room) was
changing.

Ventilation through operating areas of the reactor
building (and the reactor cell during maintenance)

18

is maintained by one of two stack fans. Stack fan
No. 1 is normally kept in service, with No. 2 on
standby. After almost continuous service since
lubrication system modifications in March 1966,
fan No. 1 developed roughness in a bearing and
the bearing was replaced in February 1968.

During maintenance work in September 1967, the
pressure drop across the roughing filters at the
ventilation stack increased to the point that stack
flow was significantly reduced. The filters in two
of the three parallel banks were replaced, and dust
filters were installed on inlets to ducts in all ac-
cessible areas to retard buildup on the roughing
filters. Pressure drop across the high-efficiency
particulate filters downstream of the roughing fil-
ters remained practically unchanged.

Activity released through the stack during the
six-month report period amounted to less than 0.25
mc of particulate activity and only 2.3 mc of io-
dine. Nearly half (1.0 mc) of the iodine was re-
leased during the maintenance work at the fuel
sampler early in September. The remainder came
from the sampler at various times.
2. Component Development

Dunlap Scott

2.1 OFF-GAS SAMPLER

R. B. Gallaher  A. N. Smith

The system developed for sampling and limited
on-line analysis of the fuel off-gas has been de-
scribed in previous progress reports.! During this
report period, the installation of the off-gas sampler
was completed, and the system was given a preop-
erational check, including leak tests, pressure
tests, and test operation of the instruments. Op-
erational use then began with the removal of sev-
eral samples for off-site analysis of xenon iso-
topic composition.

Prior to the removal of reactor gas samples,
three samples of a standard gas containing krypton,
xenon, and carbon tetrafluoride were trapped with
the refrigerated molecular sieve adsorber. The ad-
sorbed gases were swept into removable sample
bombs, whose contents were then analyzed to de-
termine the fraction of the sample passed through
the bed that was recovered. The fractions re-
covered were calculated to be 67, 76, and 86% for
the three samples. (The fraction recovered was
practically the same for Kr, Xe, and CF4.)

The molecular sieve and removal bombs were
used to remove four samples of reactor gas that
had been isolated for three to seven weeks in the
permanent isolation chambers in the off-gas sys-
tem. The first sample contained about 50% air,
possibly because of incomplete purging of the
system piping before the sample was removed.

No air was detected in subsequent samples.
Xenon isotopic ratios were measured, but the con-
centration of xenon was too low to obtain the de-

I MSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, pp. 41—43.

19

sired accuracy. Therefore, after the third sample,
the procedure was changed to obtain a higher con-
centration of gases in the bomb.

One sample was trapped directly from the re-
actor off-gas stream while the reactor was op-
erating at full power. When the sample was first
transferred to the removal bomb, the radiation
level in the containment box was over 100 r/hr,
but the radioactivity decayed rapidly, and the
sample was successfully removed after four days
of decay.

Several equipment problems were encountered,
but all could be remedied. During preliminary
testing, half the elements on one of the thermal
conductivity cells burned out for some unde-
termined cause. A new cell was procured and in-
stalled, and no further trouble occurred. Just be-
fore the first off-gas sample was removed from the
isolation chamber, one of the two pressure trans-
ducers failed. By modifying the procedure, it was
possible to operate without the failed element. It
was replaced, however, when the containment box
was opened for another reason. The box had to be
opened because of considerable difficulties with
the valve extension handles. When the extension
handles were first being installed, several of the
operating fingers broke off easily at welds. It
turned out that a piece of Inconel had accidentally
been used instead of stainless steel in making the
handles, and the welds were brittle. All the han-
dles were rewelded, using proper welding proce-
dures. Later, while the third off-gas sample was
being removed, a valve handle extension was lifted
up and improperly reengaged. The extension was
then damaged when an attempt was made to operate
the valve. The top of the containment box had to
be removed to repair the damage, and in this op-
eration three other extensions were damaged. A
slight modification was made in the top to improve
alignment between the valve handles and exten-
sions, and changes in the design of the extensions
were planned.

2.2 FUEL SAMPLER-ENRICHER

R. B. Gallaher

The tangling of the drive cable that led to the
reactor shutdown in August? apparently started
with the capsule or latch jamming at the lower
isolation valve (the maintenance valve). Then,
since the jam was undetected, the drive cable was
forced off the reel into the isolation chamber above
the valves, causing it to tangle and kink. The new
drive unit that was installed in September has a
latch of magnetic material, and during this report
period, a device was developed to sense the pas-
sage of the latch past a point in the tube. The
device is a commercially available unit that ac-
tuates a switch when a magnetic object moves into
and out of its field. The device was tested and
proved capable of sensing the latch inside the
sampler tube. The available unit is not resistant
to radiation damage and is therefore not suitable
for mounting inside the reactor cell. A design was
prepared to mount the switch about 10 in. below
the maintenance valve to indicate when the latch
has passed through both valves in the line. Thus
if the latch hangs up before reaching this point,
it should be detected in time to prevent enough
excess cable being unreeled to cause a serious
tangle.

Some additional information on one aspect of the
sampler trouble in August was uncovered when the
inoperative drive assembly and isolation chamber
(1-C assembly) were disassembled in a hot cell.
When the cover was removed in the hot cell, it was
not possible to obtain a clear view of the drive
unit. However, when the unit was removed from
the box, it was evident that the Teleflex drive
cable was not tangled in the gears as had been
supposed. In fact, there was little or no slack in
the cable between the reel cover and the hole in
the latch positioner through which the cable passed
to the isolation chamber below. The probable
reason that the cable could not be driven up or
down was found to be a sharp kink in the drive

2MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 15, 32.

20

cable. This kink was lodged in the '’/ -in.-diam
hole, about 5/8 in. from the bottom of the hole and
113'/1 ¢ in. from the top. The kink could not be
pulled easily through the hole, probably explain-
ing why the motor stalled and then could not start
the cable moving in either direction.

In the removal of the drive unit from the assembly
in the hot cell, most of the electrical insulation
was accidentally stripped from the lead wires. It
appeared that for some reason, probably radiation,
the insulation had become quite brittle. After dis-
assembly and decontamination the 1-C assembly
less the cable drive unit was returned to the re-
actor area. Replacement parts for the drive unit
were ordered so that a complete assembly could be
prepared as a spare.

At low temperatures, radiation from fission prod-
ucts can produce free fluorine in frozen fuel salt.
Because of concern over possible effects of fluo-
rine in samples intended for oxide or U% " analyses,
a carrier was designed and built to keep a fuel
sample at about 500°F from the time it is removed
from the sampler until it is unloaded at the ana-
lytical laboratory. The carrier uses molten babbitt
(85% Pb—10% Sb—5% Sn) as shielding and a heat
reservoir. Built-in electric heaters melt the bab-
bitt before the sample is loaded, and the heat of
fusion keeps the temperature from decreasing sig-
nificantly for about 9 hr. The sample is unavoid-
ably cooled below 400°F in the sampler after it is
removed from the pump bowl. The time required
to get the sample from the pump bowl to the car-
rier cannot be reduced much below 21/2 hr, but the
hot carrier prevents this period from being ex-
tended during shipment or storage.

2.3 DECONTAMINATION STUDIES

T. H. Mauney

Decontamination of the inoperative sampler-
enricher mechanism was used to develop further
the cleaning treatments tried previously on the
manipulator from the sampler-enricher.3

The unit was grossly contaminated with fission
products carried up from the pump bowl during
sampling operations, so that great care was neces-
sary in transporting and handling. The unit was

3Ibid., p. 40.
taken from the sampler-enricher into a specially
prepared open-bottomed lead coffin, bagged, and
stored in a spare cell at the reactor site until
plans could be worked out for decontamination.
These plans were complicated because a facility
had to be found that could contain the 3000-1b
coffin, with capabilities for removing the coffin
from the assembly, disassembling the mechanism,
and then decontaminating it. Such a facility was
found in the Fission Products Development Labo-
ratory of the Isotopes Division.

Some indication of the complexity of the dis-
assembly and decontamination job is given by the
photograph of the complete assembly (actually the
replacement) shown in Fig. 2.1. Disassembly be-
gan with the very difficult task of remotely un-
‘coupling seven tubing fittings and four bolts to
remove the upper plate. Fourteen bolts were then
removed from the cover plate. Removing the cover

TOP COVER 4
PLATE

Fig. 2.1. 1-C Area Unit from Sampler-Enricher (MSRE).

21

plate permitted some inspection of the drive unit,
but it was necessary to unbolt two more bolts to
lift out that mechanism. It was not practical to
attempt decontamination of the drive unit, es-
pecially since the drive cable was kinked and cut
off and the wiring insulation was damaged. All
the rest of the assembly was involved in the de-
contamination.

Decontamination was done in two stages: the
first in the Fission Products Development Labo-
ratory and the second in the Equipment Decon-
tamination Building. Table 2.1 lists the treat-
ments given in each stage and radiation levels
before and after. The readings shown before the
first stage were made 15 weeks after the last
sample was taken with the mechanism, and the
readings after the final stage were taken two
weeks later.

Table 2.1, Steps in Decontamination of 1-C Assembly

Stage 1 — Fission Products Development Laboratory
Initial radiation level (11-22-67), 100—500 r/hr

Initial contamination (smears), 5—1500 mr/hr
103Ru, 95Zr-Nb

Treatment (80—90°C spray, scrub with fiber brush)
1. Water, 30 min

. 5% nitric acid, 60 min

. Water, 30 min

. 5% oxalic acid, 30 min

. 5% Turco 4502, 30 min

. Water, 30 min

D AW

Final radiation level (12-1-67), 2—5 r/hr

3

Final contamination (smears), 1—-20 mr/hr L Ru,

952:-Nb

Stage 2 — Equipment Decontamination Building

Treatment (80-90°C)
1. Bab-O—oxalic acid, 30-min scrub with wire
brush
2. 0.40 M oxalic acid, 0.34 M hydrogen peroxide,
0.16 M citric acid, 30-min soak
3. Bab-0O, 30-min scrub with wire brush

Final radiation level (12-6-67), 200 mr/hr (1 tube
at 1 r/hr)

Final contamination (smears), 1—5 mr/hr

At the conclusion of the decontamination pro-
cedure, levels were low enough that the unit could
be worked on directly. Thus the decontamination
made available a spare component at substantially
less cost than a new one. In addition, it demon-
strated that the standard solutions used in the de-
contamination of stainless components are quite
effective on the kind of contamination present in
a salt sampler, giving an overall decontamination
factor between 500 and 2500.

2.4 STUDY OF PINHOLE CAMERA FOR
GAMMA SOURCE MAPPING

T. H. Mauney

Studies on the uses of a gamma-ray pinhole
camera were resumed. Although our original sur-
vey of the literature was not promising, informa-
tion was later turned up on a technique which is
capable of high-quality photographs. The objective
in the investigations reported here was to develop
the capability for making photographs that could
accurately map radiation sources in the MSRE re-
actor cell for maintenance and experimental plan-
ning.

The camera used in the studies is shown in Fig.
2.2. It consists essentially of a thick-walled lead
box with an aperture that allows gamma rays and
visible light to strike film at the other end of the
box. Sheet film is inserted in a slot using a
standard 4- by 5-in. film cassette. The aperture
is a thin section of aluminum, 1/8 in. in diameter,
through which gamma rays can penetrate, pierced
with a 0.016-in.-diam hole to admit visible light.
There is no shutter — exposures are long and are
satisfactorily controlled by other means. In the
tests described here, the lights were turned on and
off, and the gamma ray source was moved into and
out of the field. In a field application the film
would simply be inserted and removed.

Test photographs were made of a 20-curie
source, located in a well-lit shielded room. Some
were made with the camera only 12 in. from the
source, giving a radiation level at the camera of
110 r/hr, which is about the radiation level at the
top of the MSRE reactor cell. Other photographs
were made with the camera 15 ft from the source
and other objects, which is typical of distances in
the MSRE cells. Photographs were made with
various exposure times, with both x-ray film and

1 921r

22

ORNL-DWG 68-5510

/g in.DIAM

N
Y

L

_

e in.

0.016 in. DIAM

APERTURE
MATERIAL-ALUMINUM

FILM HOLDER
SLOT

Fig. 2.2. Pinhole Camera Used in Tests.

visible-light film. Data and results are given in
Table 2.2.

Two of the most significant photographs are
Figs. 2.3 and 2.4. Figure 2.3 is a photograph on
visible-light film exposed for 12 min. The source
is a Y -in. pellet inside a ¥ -in.-diam capsule, held
12 in. from the camera, in front of a scarred wooden
table leg. The poster is about 21/2 ft behind the
source. Visible details were reproduced clearly,
but the gamma radiation produced only a dim spot.
Figure 2.4 is a photograph of the same setup, but
on x-ray film exposed for 3 min. The source spot
shows up very clearly, and the visible scene was
reproduced well enough to identify the location
of the source relative to other objects. In tests
with the camera 15 ft from the source, the radia-
tion at the camera was only 0.5 r/hr, and the
Table 2.2. Data on Test Photographs with Pinhole Camera

Type of

Exposure Time

Test Distance Remarks
Film® Gamma Rays Light
B X ray 15 ft 61/2 min 0 Small spot, underexposed
C Visible 15 ft 10 min 10 min Good picture, no spot
D X ray 15 ft 10 min 0 Good spot
A X ray 12 in. 30 sec 0 Good spot
1 X ray 12 in. 30 sec 45 sec Good picture, no spot
2 X ray 12 in. 3 min 3 min Good picture, good spot — Fig. 2.4
3 Visible 12 in. 3 min 11/2 min Dim spot
4 Visible 12 in. 6 min 6 min Good picture; no spot
5 Visible 12 in. 12 min 12 min Good picture, dim spot — Fig. 2.3

4X-ray film was Kodak KK, ‘‘visible’” film was Kodak Tri-X.

source spot was swamped by visible light when
film was exposed simultaneously to light and
gamma rays. However, a very clear source spot
could be produced on x-tay film exposed only to
gamma rays. In this case an acceptable picture
was obtained by superimposing two negatives:
KK x-ray film exposed 61/2 min to gamma rays only
and Tri-X visible-light film exposed 10 min. The
source was hardly distinguishable on a composite
print, but its location could be determined fairly
well by comparison of the two negatives.

The results indicate that the pinhole camera can
be useful in surveying the MSRE reactor cell for
gamma-ray sources, and plans are to photograph
portions of the cell at the next shutdown.

2.5 FREEZE-FLANGE THERMAL CYCLE TESTS

F. E. Lynch

In the development of the freeze flanges for the
MSRE, a prototype flanged joint was subjected to
103 thermal cycles to determine the sealing and
distortion characteristics in simulated reactor
startups and shutdowns.* These thermal cycles
were typical of heating and salt-filling operations

‘MSR Program Semiann. Progr. Rept. July 31, 1964,
ORNL-3708, p. 180.

but were more severe than those encountered in
the MSRE. A satisfactory leak-tight gas seal was
obtained at both elevated temperature and room
temperature. This number of cycles was greater
than the number anticipated for the flanges in the
MSRE. Thus it was concluded that the MSRE
flanges should be satisfactory from the standpoint
of sealing and distortion.

Examination of the prototype flange during and
after the 103 cycles showed no cracks such as
would be produced by thermal fatigue. Predictions,
based on low-cycle fatigue analyses, were that
cracks should be expected at about 300 test cycles.
The permissible number of cycles of various kinds
on the reactor flanges was based on the same kind
of analysis, but the permissible number was set a
factor of 10 below the number at which cracking
would be expected.® Now, as reported in Sect.
1.2.3, the reactor flanges have reached 69% of the
permissible life. Although they may not reach the
permissible limit, it was decided to reactivate the
flange thermal cycle test. The purpose is to lend
further confidence in the fatigue calculations, and
plans are to continue thermal cycling the prototype

Sp. N. Haubenreich et al., MSRE Design and Opera-
tions Report. Part V-A. Safety Analysis of Operation

with 237U, ORNL-TM-2111, p. 70 (February 1968).
24

PHOTO 94689

Fig. 2.3. Test Photograph on Visible-Light Film. Tri-X ortho film, 192, source, 100 r/hr at camera, 12-min ex-

posure, 5-min development in x-ray developer.
25

PHOTO 94690

Fig. 2.4. Test Photograph on X-Ray Film. KK x-ray film, 192, source, 110 r/hr at camera, 3-min simultaneous

exposure to light and source, 8-min development.

freeze flange until it cracks or until the MSRE is
finally shut down.

The Freeze-Flange Thermal Cycle Facility con-
sists of an upper and a lower tank with intercon-
necting pipe containing the test flange. In the

course of the test, molten salt is oscillated be-
tween the two tanks. Salt is forced to flow to the
upper tank by pressurizing the lower tank with gas,
and it returns to the lower tank by gravity when the
pressure between the two tanks is equalized. A
SPRING-LOADED
RELIEF VALVE

VENT\ " /

//

~

!< TEST FLANGE

N

ORNL-LR-DWG 52043A

.
N

™~
EQUALIZING LINE FOR DRAIN-DOWN

| SYSTEM-OVERPRESSURE
REGULATING VALVE

| UPFLOW PRESSURE
REGULATING VALVE

Fig. 2.5. Freeze-Flange Thermal Cycle Facility.

schematic of the test facility is shown in Fig.
2.5. Figure 2.6 shows the cross section of the 5-
in. MSRE-type Hastelloy N flange.

The two tanks remain heated at all times, with
the molten salt in the lower tank when the system
is not in operation. At the start of a thermal
cycle, heaters adjacent to the flanges are turned
on, and the system is heated until the flange hub
reaches a preset temperature. When this tempera-
ture is obtained, contacts within a temperature
controller are closed, the lower sump is pres-

surized with helium, and a timer is simultaneously
started. Level probes control the flow of salt
between the two tanks by alternately pressurizing
the lower tank and equalizing the gas pressure.
This oscillating flow of the salt between the two
tanks continues until the preset time on the timer
elapses. The timer then turns off the pipe heaters,
and the salt drains to the lower sump tank. A
controlled cooldown is obtained by reducing the
setting on the pipe heaters and reenergizing them
for a period of time.
SOOI b
tin.” =l=0,030-ih. \
FCLLAA\:\(I?E m NJjeap WIDTH |

BUFFER
CONNECTION

{SHOWN ROTATED)

MODIFIED R-68
RING GASKET -~

FROZEN
SALT SEAL ] .

0.050-in,
[~ GAP
WIDTH \\

ORNL-LR-DWG 63248R2

H%s*in. R

v N

V’__fi_

1Y-in-R (TYP)

SLOPE 1:4

5-in. SCHED-40 PIPE

Fig. 2.6. Cross Section of 5-in. Molten-5alt Freeze Flange.

The tank temperatures are maintained between
1300 and 1350°F with corresponding pipe tempera-
tures of approximately 1300°F. The temperature at
the hub of the uninsulated flange at the time the
salt starts to oscillate is between 500 and 600°F.
Eight to nine hours are required fo preheat the pipe
to obtain this hub temperature. The salt is then
oscillated from one tank to the other for 5 hr.

Upon completion of the oscillation time, the
flanges are cooled down to approximately 130°F.

The original 103 cycles were completed using
helium as the cover gas, but when operation was
resumed we attempted to use argon. However, we
returned to helium when we found that the op-
erating temperature could not be maintained at the
desired level with argon. The history of the upper
bore and ring temperatures during a typical thermal
cycle using helium as the pressurizing gas is
shown in Fig. 2.7. A typical radial thermal gradient
during salt oscillation is shown in Fig. 2.8.
TEMPERATURE (°F)

1400

1200

1000

800

600

400

28

ORNL- DWG 68~ 5511

7

UPPER BORE
TEMPERATURE

$

o+—®

/

o T

/

[ J
/
®

\
\
\

l

PIPE HEATERS ON

L“ PREHEAT —————==

= OSCILLATING =

RING TEMPERATURE\
L
| Q@Q

PIPE HEATERS OF F—=~

0] 4q

8

12
TIME (hr)

16 20 24

Fig. 2.7. Upper Bore and Ring Temperature Characteristics of 5-in. MSRE Freeze Flange; Salt Oscillated with

Helium.
ORNL-DWG 68- 5512

1400 T }
1200 {——— %
1000 — ‘
) |
S s —x— SALT FREEZE TEMPERATURE
w 800 O (850°F)— v |
[aed
=]
&
14
«
s 600 |— —
e OSCILLATING WITH HELIUM °\L\.
UPPER BORE 1290 °F
ol bl
B
0 L J [
0 2 a 6 8 10 Iz

DISTANCE FROM CENTER LINE QF PIPE (in)

Fig. 2.8. Radial Temperature Distribution in 5-in.
MSRE Freeze Flange.

At this time, a total of 139 cycles have been
completed. The outer surfaces of the flange as-
sembly are visually inspected during and after each
cycle. A complete inspection of the flange was
made between cycles 103 and 104 when the fa-
cility was shut down. This inspection included
dye-penetrant inspection of the internal flange
face and bore surfaces as well as a dimensional
check of the bore. These complete inspections
will be repeated periodically through the duration
of the test.

2.6 PUMPS
P. G. Smith A. G. Grindell

2.6.1 Mark 2 Fuel Pump

As reported previously,® the Mark 2 fuel pump
was designed to give more salt expansion in the

*MSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, p. 64.

29

pump bowl, thus eliminating the need for the over-
flow tank used in the present MSRE pump instal-
lation. It also differs in having a longer unsup-
ported length of shaft and a different fission gas
stripper. During this six-month report period, work
continued on preparation of the pump and test fa-
cility for test operation with molten salt.” Instal-
lation of the pump tank in the test facility was
completed. An electrical furnace for preheating
was installed on the pump tank; various thermo-
couples and electrical preheaters were replaced

on the test system, the pressure measuring de-
vices on the salt flow venturi were replaced, and
the necessary thermal insulation was applied to
the test loop. Some delays were encountered in the
assembly of the pump rotary element, and the as-
sembly was not quite finished by the end of the
report period.

2.6.2 MSRE Oil Pumps

Shakedown of the second of two MSRE lubrica-
tion pumps that were refurbished® was completed,
and the two pumps were returned to the MSRE to
serve as spares.

2.6.3 Oil Pump Endurance Test

The oil pump endurance test’ was continued.
By the end of the period, the pump had run for
40,134 hr circulating oil at 160°F and 70 gpm.

7MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 45.

81bid., p. 46.

MSR Program Semiann. Progr. Rept. Feb. 28, 1962,
ORNL-3282, p. 55.
3.

Instruments and Controls

L. C. Oakes

3.1 MSRE OPERATING EXPERIENCE

The instrumentation and control systems con-
tinued to function well, and the failure rate of
components continued to decrease. Component
failures did not compromise reactor safety or cause
excessive inconvenience in reactor operation.

3.1.1 Safety System Components

J. L. Redford

Five more relays failed in the rod scram coinci-
dence matrix. This made a total of seven failures
in less than a year out of the 15 relays in the
matrix. These relays are 115-v ac relays operating
in a 32-v dc system. Relays more suited to the
service were procured and are ready to install.

Numerous false trips on channels 2 and 3 were
traced to intermittent false operation of core outlet
temperature switches. Replacing the switches
remedied the trouble.

Other than test scrams, the control rods were
scrammed only twice in the six months. The first
scram was caused by the building power failure in
October while the reactor was at full power. The
other occurred while the reactor was at 10 kw in
November and resulted from circuit testing in
search of a ground in the rod servo system.

One of the three ionization chambers began pro-
ducing a reverse current, and upon removal it was
found that moisture had leaked into the cable.

3.1.2 Thermocouples

C. H. Gabbard

As described in the last semiannual report?! there
has been very little change in the apparent ac-

30

curacy of the thermocouples in service on the salt
systems. There have also been remarkably few
failures due to breakage, shorting, grounding, or
apparent detachment from the surface being moni-
tored.

The thermocouples are Chromel-Alumel wires en-
closed in a %-in. Hastelloy N or stainless steel
sheath with magnesia insulation. Most junctions
are grounded. The sheathed thermocouple stock
was commercially manufactured to ORNL specifica-
tions, and the junctions and disconnect seals were
prepared at the MSRE by ORNL craftsmen.

A total of 1071 thermocouples are installed at
the MSRE. Of these, 866 are on the salt systems:
351 on the circulating loops and the temainder on
the drain tanks, drain lines, and freeze valves. Of
the 1071 thermocouples, only 9 have failed in more
than three years of service. Three others are un-
serviceable because of damage suffered during the
final stages of construction when repair was too
difficult to be worth while. A breakdown of the
failures is given in Table 3.1.

IMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 22—-23.

Table 3.1. Failures Among 1071 MSRE Thermocouples
Through February 29, 1968

Nature of Failure Number
Damaged during construction 3
Damaged during maintenance 2
ILead opened during operation 3
Abnormally low reading (detached?) 3
Bad disconnect in reactor cell 1

Total 12

3.1.3 Other Instruments and Controls

J. L. Redford

Two fission chamber failures occurred during the
period. One was because of a leak in the cable,
and the other has not yet been diagnosed.

The failure in the fine position synchro trans-
mitter on rod drive No. 2 appears to be an open
rotor winding, as in the previous failure reported
last period. With the transmitter in this condition,
the position indicator on the console tracks normal
changes but may lose 180° on rod scrams.

Miscellaneous failures included the air flow
switch on a radiator annulus blower. A new switch
with a smaller paddle will be tried. The power
supply to the single-point level switches in the
drain tanks failed because of a cooling blower fail-
ure. The blower was replaced, and the power
supply was repaired.

3.2 CONTROL SYSTEM DESIGN

P. G. Herndon

Further additions and medifications were made
to the instrumentation and controls systems as ex-
perience revealed the need or desirability of more
information for the operators, improved performance,
or increased protection. During the report period
there were 26 design change requests directly in-
volving instruments or controls. Seven of these
required only changes in process switch operating
points, sixteen resulted in changes in instruments
or controls, one was canceled, and the remaining
two were not completed. The more important
changes are described below.

To avoid further damage to wiring and equipment
resulting from short circuits to ground, isolation
transformers were installed in the power supply to
the motor control circuits of the fuel sampler-en-
richer cable drive motor. This permits the entire
circuit to float at 115 v above ground potential.
Before this revision, one control power conductor
was a grounded neutral. The floating circuit will
operate normally when only one point is grounded,
but a misoperation is likely to occur if two points
become grounded at the same time. To avoid two
simultaneous grounds, ground-detecting lamps were
installed, and all grounds will be corrected as soon
as possible.

31

The design of instrumentation and controls for a
fluorinated-fuel-salt filter in transfer line 110 be-
tween the fuel storage tank and the fuel drain tanks
was started. Eleven mineral-insulated, Inconel-
sheathed thermocouples are attached to the salt
filter and connected to the patch panel in the main
control room. Temperature interlocks are provided
to annunciate high temperatures in the filter gas
space and to stop the transfer of salt to the fuel
drain tanks by closing the fuel storage tank helium
supply valve if the temperature of the top flange
on the filter gets too high. A helium purge line is
also provided at the top of the filter to prevent salt
from contacting the flange.

Several additions to the fuel processing plant
helium supply system are designed to prevent an
accidental fill of the reactor during fuel transfer
operations: a pressure relief valve limits the
maximum pressure in the line supplying the fuel
processing plant to 50 psig, a capillary flow restric-
tor designed to the same specifications as FE-517
(the drain tank helium supply restrictor) limits the
maximum flow rate in the salt filter purge line to a
value less than 0.5 cfm, and a weld-sealed solenoid
block valve is installed in the filter purge supply
line. Safety-grade interlocks are designed to close
the purge supply block valve and the fuel storage
tank helium supply valve on an emergency drain or
fill-restrict signal. A safety-grade jumper is pro-
vided around these interlocks.

Plans are to use the decontamination cell to
store traps loaded with UF  from the fuel fluorina-
tion. To ensure that water does not accumulate
undetected in this cell, a bubbler-type liquid level
measuring system identical to that in other equip-
ment cells was designed for the decontamination
cell. A level indicator is located in the transmitter
room, and a high-level annunciator is provided in
the auxiliary control room.

The helium and standard gas supply systems on
the fuel off-gas sampler proved to be inadequate
during initial test operations. To obtain the pre-
cise flow control required for the purge and calibra-
tion operations and to provide better protection for
delicate components, a new gas supply system was
designed and installed. The volume of all lines
to each piece of equipment was reduced to a mini-
mum, special pressure regulators designed for use
in chromatograph systems were provided, and addi-
tional pressure-relief valves were installed. A
conductivity-type level detecting system was de-
signed for the off-gas sampler containment box.
The instrument provides an alarm if the box starts
to fill with water.

3.3 MSRE NEUTRON NOISE ANALYSIS

D. N. Fry R. C. Kryter
J. C. Robinson

The spectrum of inherent noise in the MSRE neu-
tron flux signal has been examined in some detail.
The data are collected through a special low-noise
instrument channel and are digitally recorded on
magnetic tape by the on-line computer. A program
has also been prepared to permit direct digital
analysis of the data tapes by the same computer
immediately after the data are collected.

A considerable amount of data was collected in
rmun 14 with particular emphasis on the series of
experiments at 5 Mw. The most prominent feature
of all the noise spectra is a peak at about 1 hertz
(see Fig. 3.1). In general, this peak increased in

32

amplitude under conditions where the fraction of
circulating voids increased. Figure 3.1 shows two
noise spectra which illustrate this effect. Other
variations were also observed, and detailed cor-
relations of the noise spectra with other reactor
data are being attempted.

3.4 TEST OF MSRE ROD CONTROL SYSTEM
UNDER SIMULATED 233U LOADING
CONDITIONS

S. J. Ball R. J. Steffy, ]Jr.
Because of the smaller fraction of delayed neu-
trons from 233U fission, the prompt power response
to a sudden reactivity change will be about three
times as large as for the present 235U fuel load-
ing.? Conceivably this increased high-frequency

Ibid., pp. 61—62.

ORNL-DWG 68-4047

1() N | - — — P S|
e = 1
e — T A
L 2] CURVE T
o A B ]
- REACTOR OUTLET TEMPERATURE (°F) 1180 14225
FUEL SYSTEM OVERPRESSURE (psig) 93 30

POWER SPECTRAL DENSITY (arbitrary units)

o}
o

FUEL PUMP LEVEL {in.}

FREQUENCY (Hz)

Fig. 3.1. Typical Effect of Reactor Operating Parameters on the Spectrum of Inherent Neutron Fluctuations in

MSRE. Reactor power = 5 Mw.
33

gain could cause enough overshoot following a
control rod adjustment that the rod servo controller
would hunt excessively. Thus it was desired to
determine in advance if any changes would be
necessary in the servo system to have it function
properly with 223U fuel.

A method was devised to test the capability of
the servo system for controlling the reactor with
233U fuel. Basically it consisted in modifying the
flux signal to the servo system so that the apparent
response to a rod change closely resembled that
from the 233U system. Appropriate resistors and
capacitors would be wired around existing elements
in the servo network (see Fig. 3.2) so that the flux
input signal would have a high-frequency gain
about three times normal but an identical steady-
state value.

Verification of the validity of this experiment
and its implementation were assigned to an MIT

Practice School team (P. J. Wood, leader, and
D. J. Roberts). The work reported below was
carried out mainly by them.

Before the experiment was conducted on the re-
actor, the method was tested by analog simulation.
The simulator used essentially the same model of
the MSRE that was used in the operator training
simulator3 except that only three delayed neutron
groups were used and that the external fuel circuit
was represented by a third-order system. Three
conditions were simulated: the MSRE with 235U
fuel and with 233U fuel, both with an unmodified
servo system, and the reactor with 235U fuel and
the servo system modified by the proposed RC net-
work. Figures 3.3 and 3.4 show the response of

3s. J. Ball, Simulators for Training Molten-Salt Reactor
Experiment Operators, ORNL-TM-1445 (April 1966).

ORNL-DWG 68-5513

c-201
O puf
I(
AN
A
10 meg
R-219
R-242
3 meg
AAA { :> TO FLUX DEMAND
CALCULATION
AMP.
1.5 meg
2 uf
FLUX SIGNAL
FROM COMP, ——
{ON CHAMBER
15k 200 uf 0.2 uf
AA 1 TO ROD CONTROL
?(')'Ak % = RELAYS
R-250 SES;O
—w—
{ meg
R~251
L
> R—-259
2t S0«

SYNCHRO
DEMODULATOR

Fig. 3.2. Schematic Diagram Showing Additions Made to Servo Circuit to Simulate 233y Fuel Loading.
ORNL-DWG 68-5514

POWER (Mw)

TIME (min)

Fig. 3.3. Analog Simulation of 0.5-Mw Load Changes.

75

+0.01% 84
7.0 - ————— —

L -0.01% 34
6 _

+0.01% Sk

POWER (Mw)
~
o

—001%8k
+001%84

C 025

ORNL-DWG 68- 5515

TIME (min)

Fig. 3.4. Analog Simulation of 0.01% Ok/k Changes.

the three systems to a 0.5-Mw load change and a
step reactivity change of 0.01% ok/k respectively.
It is apparent that the 235U + RC system closely

follows the 233U system response for fast changes.

Table 3.2 further illustrates the success of the RC
networks in making the 233U response approach
that of the 233U system. Thus the analog results
showed that the modification of the 235U system
response was apptoximately correct for high-
frequency changes (> 0.3 radian/sec) and that the

proposed experiment on the MSRE should indeed
test the servo system under conditions ‘close to
those existing with the 233U loading.

With confidence established that addition of an
RC network would provide a meaningful simulation,
the experiment was performed on the reactor. After
the proper RC elements were attached to the cir-
cuit, the system was perturbed small amounts in
three different ways: (1) a shim rod was inserted
~0.5 in., (2) the radiator load was changed by
Table 3.2. Comparison of Flux Response to 0.01% Ok/k
Step Increase in Reactivity in Three Systems on

Analog Computer

Difference Difference
Time After Between Between
Reactivity 233yy and 235y 233y and 235y + RC
Step (sec) ) (%)
1 33 8
2 26 9
3 22 6
5 10 4

0.5 Mw, and (3) the outlet temperature demand was
changed by 3°F. After each of these tests the sys-
tem was allowed to reach equilibrium and was then
returned to its original state by reversal of the
initial procedure. The ability of the controller to
correct for large changes was tested by a 2.7-Mw
power change (from 7.2 to ~4.5). In only one test,
the 3°F outlet temperature demand change, did the
regulating rod oscillate abnormally. When the de-
mand was increased, the rod oscillated for about
25 sec (six withdraws and six inserts) but then
steadied out without external interference. A
similar occurrence was not observed when the
temperature demand was decreased. The conclu-
sion was that at least when the reactor operates at
high power, the present rod servo system will be
adequate and will not introduce undue rod oscilla-
tions. Because the highest gain relative to the
servo dead band occurs at lower powers, the situa-
tion may not be as favorable, but minor adjustments
should be all that is required for 233U operation.

35

3.5 ANALOG COMPUTER STUDIES OF THE
MSRE SYSTEM WITH 233U FUEL LOADING

O. W. Burke F. H. Clark

As part of the analysis of the MSRE with 233U-
bearing fuel, a detailed simulation of the system
was set up on an analog computer. The simulation
included the period-scram circuits that cause the
control rods to scram after the period becomes
shorter than 1 sec. This could not be conveniently
included explicitly in the digital calculations be-
cause of the complex dependence of the scram
signal on the initial neutron level, the gamma
background, and the history of the flux and the
period as it approaches 1 sec. Results from these
analog studies were combined with digital calcula-
tions to describe the behavior of the reactor under
accident conditions. The findings of these coordi-
nated studies are described in detail in Sect. 4.2.

Some preliminary analog studies were made to
assess the performance of the servo controller with
233U fuel based on estimated values of instrument
time lags and rod-drive-motor inertial effects. It
appears that the controller will be quite adequate
when operating at the higher powers associated
with the temperature mode of control. There was
some indication of ‘‘hunting’’ at very low power
levels while operating in the flux mode of control.
These results support those obtained from the in-
situ test of the actual servo control system with
the modified flux input (see Sect. 3.4). Plans were
developed to measure control rod accelerations and
coastdowns experimentally. These will be in-
corporated in the analog simulation to more ac-
curately predict the performance of the control
system with 233U fuel.
4. MSRE Reactor Analysis

4.1 INTRODUCTION

B. E. Prince

Reactor physics studies in support of the future
operation of the MSRE with a 233U fuel loading
were extended to aid in evaluating the nuclear
safety of the system. Use was made of results of
previous computational studies of the neutronic
propetties of the system with 233U fuel, ! and
emphasis was given to analysis of some abnormal
situations which could conceivably lead to nuclear
excursions. To perform these studies, we used the
reactor kinetics—digital simulation code ZORCH, ?
developed for calculation of large transients in
power, temperature, and pressure in the MSRE.
The general approach taken in the safety studies
was similar to that used earlier for analysis of the
nuclear safety with the present ??3U-bearing fuel
salt.®'4 In the evaluation of nuclear safety for
the 233U loading, however, we have attempted
where possible to take advantage of experience
gained from both simulation studies and reactor
operation with the present fuel salt. Thus, from
the standpoint of magnitudes and rates of reactivity
addition and initial reactor conditions, some inci-
dents considered in the original safety analysis
could either be judged unrealistic or less severe
than others. As was the case with the present
fuel loading, the potentially most severe nuclear

1MSR Program Semiann, Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 50-62.

2C. W. Nestor, Jr., ZORCH — an IBM-7090 Program
for the Analysis of Simulated MSRE Power Transients
with a Simplified Space-Dependent Kinetics Model,
ORNL-TM-345 (September 1962).

3P. N. Haubenreich et al., MSRE Design and Opera-
tion Report. Part III. Nuclear Analysis, ORNL-TM-730,
pp. 122-56 (February 1964).

4S. E. Beall et al., MSRE Design and Operation Re-
port. Part V. Reactor Safety Analysis, ORNL-TM-732,
pp. 196—231 (August 1964).

36

excursions with a 233U fuel salt were found to be
associated with two postulated incidents. One
incident results from the sustained withdrawal of
the three reactor control rods at maximum rate with
the initial neutron level very low, near source
conditions. The other incident involves the
gradual separation of uranium from the main stream
of circulating salt during routine operation of the
reactor, followed by the rapid return of the uranium
in concentrated form to the core. We have performed
digital simulation studies of the consequences of
these incidents to determine both the inherent shut-
down capabilities of the system (through tempera-
ture-reactivity feedback) and also the effectiveness
of the reactor safety system in initiating control
rod scrams to suppress the excursions. Though the
latter is of principal concermn in these studies, it is
of interest to obtain the approximate relations be-
tween these reactivity addition incidents and the
inherent shutdown capabilities of the reactor,
without action of the safety-scram system. In addi-
tion to these simulation studies, in order to further
elucidate the nuclear safety consequences of
changing to a 233U fuel salt, we have compared
simulations of these incidents with ?33U and with
the present 235U fuel loading, the latter calculated
using updated values of the MSRE neutronic charac-
teristics 3 obtained since the earlier nuclear safety
studies. This comparison is described further in
the results summarized in the following sections.

A synopsis of the most important results of these
studies is also included in ref. 6 in context with
general description of the safety of the reactor
system with 233U.

B. E. Prince et al., Zero-Power Experiments on the
Molten-Salt Reactor Experiment, ORNL-4233 (February
1968).

6P. N. Haubenreich et al., MSRE Design and Opera-
tion Report. Part V-A. Safety Analysis of Operation
with 233U, ORNL-TM-211, pp. 40-~69 (February 1968).
4.2 SIMULATION OF NUCLEAR
EXCURSION INCIDENTS

B. E. Prince R. C. Steffy, ]Jr.

4.2.1 Uncontrolled Rod Withdrawal

In this hypothetical accident, a nuclear excursion
is produced by the sustained withdrawal of the
three control rods, with the reactor passing through
criticality when the rods are near the position of
maximum differential worth. The restrictions and
control interlocks which would prevent this type
of incident from occurring are described in ref. 6.
In this section, we will summarize the results of
digital simulations of the accident, including the
most severe conditions which would result without
consideration of the action of the safety system
and the effect of this action in initiating rod scram
and terminating the excursion.

To define initial conditions in the rod-withdrawal
incident, we have assumed that the reactor is
loaded with excess uranium equivalent in reac-
tivity to the worth of one control rod (2.75% &k/k).
The reactor is initially subcritical by the insertion
of all three rods, the fuel is circulating, and the
core temperature is uniform at 1200°F. The initial
fission rate is quite low, near a low limit deter-
mined by the strength of the inherent a,n source
in the 233U fuel (approximately 1 w). The three
rods are now assumed to be withdrawn in unison
at a speed of 0.5 in./sec. The reactor passes
through criticality when the rods are approximately
28 in. above the position of maximum insertion,
which is approximately 3 in. above the position of
maximum differential worth. In this region the rod
motion would correspond closely to a ramp addition
of reactivity of 0.093%/sec. Without consideration
of the action of the safety system, a short time
after criticality is reached, sufficient reactivity
would be added to produce a prompt-critical ex-
cursion. This would be terminated by temperature
feedback due to nuclear heating of the fuel. Then,
if the rod withdrawal were continued, reactivity
would be added at the above rate for approximately
16 sec after the time of criticality. The rate would
then decrease gradually until the rods reached their
upper limits. After the initial prompt-critical ex-
cursion, the power and temperature would increase
gradually to keep in step with the continued reac-
tivity addition.

37

These characteristics are exhibited by results of
digital calculations of the power-temperature-
pressure excursion, shown in Fig. 4.1. For these
calculations, initial steady-state conditions at a
power level of 1 w were assumed, and the rod
withdrawal was assumed to begin at time zero,
near the position of maximum differential worth.
Both the temperature of the fluid at the hottest
point in the reactor channel and the outlet tempera-
ture of the hottest channel are plotted in Fig. 4.1,
(As described in ref. 2, the digital model includes
a detailed numerical treatment of the axial convec-
tion of heat by fluid motion during the power tran-
sient.) The calculated pressure surge during the
period of very rapid heating, also shown in Fig.
4.1, results primarily from the effects of accelera-
tion of fluid in the outlet pipe. A description of
the mathematical model for the pressure rise calcu-
lation is included in ref. 3.

In order to illustrate the importance of the dif-
ferences in neutronics characteristics in changing
the loading from the present 235U-bearing salt to a
2337 salt, we also performed calculations similar
to those of Fig. 4.1 for the present fuel composi-
tion. To simplify this comparison, we assumed
that the reactivity addition rates were identical
(0.093% 6k/k/sec) and also that the initial neutron
levels were identical (1 w). In an accident simula-
tion for the 235U fuel, the reactivity addition cor-
responding to rod withdrawal would actually be
about 30% smaller than the above rate because of
the smaller worth of the control rods. However,
by making these assumptions, the nuclear excur-
sions can be compared simply on the basis of the
differences in delayed neutron fractions, tempera-
ture coefficients of reactivity, and prompt-neutron
generation time for the two fuels. The resulting
nuclear transients calculated for the 235U fuel
are shown in Fig. 4.2, It is evident that the rapid
portion of the transient occurs later in time with
235U than with the 233U fuel, since more reac-
tivity must be added to reach the prompt-critical
condition. As the reactivity addition continues
beyond this condition, the presence of the larger
amount of excess reactivity and the smaller tem-
perature coefficient of reactivity of the 235U fuel
lead to a larger temperature transient for the 235U
case. In other words, the inherent shutdown mech-
anisms with 233U begin to take effect sooner, must
compensate for less reactivity, and compensate
more effectively than with 235U,
38

ORNL—DWG 68—968

{800
1700 ///
1600 ’//

e MAXIMUM FUEL TEMPERATURE /

: 7

v

2 1500

<[

4

= / OUTLET TEMPERATURE OF

HOTTEST CHANNEL

—

1400 // /

1300 /]

1200 “

500

400
~ 20
[z}
a

3 300 w 19

= =

z z A

@ w 0

w

: e L

g 200 o 5

¢ L)

Q-

a 0

4 6 8
100 TIME (sec)
e ——
-/
0
4 6 8 {0 12 14 16 18

TIME (sec)

Fig. 4.1. Results of Uncontrolled Rod Withdrawal with No Safety Action. 233y fyel.

The calculations shown in Fig. 4.1 for the 233U ultimately be necessary to prevent overheating of
fuel indicate that the fuel temperature and core the core. To this aim, the safety system would
pressure rise incurred during the rapid portion of actuate a rod scram to limit the excursion. Scrams
the transient would be inconsequential in terms of would be actuated by any of three conditions:
conservative limits specified in the nuclear safety (1) high flux level (over 11.25 Mw, if the fuel pump
analysis (340°F and 50 psi).® However, it is clear is running, or 11,25 kw if the circulation is

that counteraction of the rod withdrawal would stopped), (2) positive periods less than 1 sec,
1800

1700

1600

1500

TEMPERATURE (°F)

1400

1300

1200
500

400

300

POWER {Mw)

200

100

39

ORNL—DWG 68-269

P,

~
/

MAXIMUM FUEL TEMPERATURE

/( QUTLET OF HOTTEST CHANNEL

/

— 20
(=]
‘@
a
~ 15
11]
N
& 40
Ll
o
?
a 5
w
o
a 0
6 8 10
TIME {sec)
8 {0 2 {4 {6 18

TIME {sec)

Fig. 4,2, Results of Uncontrolled Rod Withdrawal with No Safety Action. 235y fyel.
and (3) reactor outlet temperatures greater than
1300°F. In the incidents of the type considered
in these studies, the first two are the active mech-
anisms. In all our simulation studies, we have
assumed (1) that only two of the three control rods
drop on request, (2) a time delay of 100 msec
occurs between scram signal and the start of rod
drop, and (3) the rod acceleration is 10 ft/sec?.
Figure 4.3 shows the results of actuating the
rod scram due to increases in the neutron flux
above the 11.25-Mw level. This mechanism would
be effective in limiting the nuclear excursion to
inconsequential proportions. Actually, a safety
scram would occur earlier than indicated by Fig.
4.3, due to reactor periods smaller than 1 sec.
The digital calculations for this transient showed
that a 1-sec period is reached at approximately

40

1.4 sec, a 500-msec period at 2.4 sec, the 11.25-Mw

level at 5.3 sec, and the maximum power, limited
by temperature feedback, at 5.8 sec. Thus, calcu-
lations based on quite conservative assumptions
of the actual time of actuation of the period scram
showed that the power transient would be at least
two orders of magnitude below that shown in

Fig. 4.3.

4.2.2 Return of Separated Uranium to the Core

The inclusion in the nuclear safety calculations
of an incident involving the gradual separation of
uranium from the circulating salt and its rapid con-
centration and return to the reactor core is based
on the existence of two possible avenues for
chemical separation of uranium from the molten
fluoride salt.® One avenue requires the loss of
enough fluorine from the salt to ultimately result
in the production of metallic uranium. The other
requires the gross contamination of the salt with
enough oxygen or water vapor to result in the
production of ZrO, and finally UO,. Precautions
are taken to avoid contaminations, and evidence
accumulated from chemical analyses during opera-
tion with the present ?35U-bearing fuel salt has
indicated no trends in either of these directions
toward uranium separation. Although we expect
these trends to continue when the reactor is opet-
ated with 233U, the existence of these remote
possibilities motivates the analysis of a hypo-
thetical nuclear incident of this type.

To define the conditions for the incident, we
have assumed that, during routine operation of

ORNL-DWG 68—970

1230 } (
&
o 1220 MAXIMUM FUEL TEMPERATURE
> [
: —~L
14
: [
S 1210 ‘
oot
/(OUTLET OF HOTTEST CHANNEL
1200 ' L ] \

60
7 40
=
(v
ul
z
(@]
€ 20

. |
4 6 8 10 12 14 16 18
TIME (sec)

Fig.

4.3. Results of Uncontrolled Rod Withdrawal with Scram at 11.25 Mw. 233y fuel.
the reactor at some steady power level, uranium

is gradually removed from the main circulating
stream and is accumulated in low-velocity regions.
At the same time, the regulating rod is automati-
cally withdrawn to compensate for the reduction

in uranium concentration. We then assume that,
by some mechanism, this concentrated uranium is
suddenly dispersed throughout the 10 ft3 of fuel
salt in the lower head of the reactor vessel and is
carried upward through the channeled region of the
core by the fluid motion, producing a nuclear ex-
cursion. It is clear that the assignment of a single
magnitude or rate of addition of reactivity in this
incident is not possible, as was the case with
simulation of the uncontrolled rod withdrawal.
Thus, it was necessary to study the consequences
of the uranium separation incident as a function of
the amount of uranium (or reactivity) lost and also
of the initial reactor operating conditions. The
principal aim of the analysis was to compare the
magnitude of the reactivity loss which would be
required to produce severe rises in temperature

41

and pressure during the excursion associated with
the return of all the lost uranium with the magni-
tude of anomalous reactivity loss which could be
detected by routine computer monitoring of the
reactivity balance. As in the case of the uncon-
trolled rod withdrawal incident, the nuclear ex-
cursions were simulated both with and without
the safety scram of the control rods.

Although the magnitude of reactivity loss was
considered as a parameter in the analysis, for the
conditions defined above, the time-dependent
shape of the reactivity addition during the uranium
return would be largely determined by the fluid
dynamics of the core and by the spatial distribu-
tion of nuclear importance. Figure 4.4 shows the
time dependence of the reactivity calculated from
the flow velocities observed in the MSRE hydraulic
mockup vessel and normalized to Ak, the magni-
tude of the original reactivity loss. Equivalently,
Ak would be the reactivity effect of all the sepa-
rated uranium if it were uniformly dispersed through-
out the 70 ft3 of salt in the loop rather than con-

ORNL~—DWG 68—971

6
5
O
x
N
44
< / N \
-
=
=
23 ]
wl
o
L
>
= 2 + —
-
wl
@
1 / 1
0 2 4 6 8 10 12 14

TIME {sec)

Fig. 4.4. Time Dependence of Reactivity Addition Due to Sudden Resuspension of Uranium in Lower Head of

Reactor Vessel.
centrated in the salt moving through the core.
From Fig. 4.4, the maximum total reactivity which
could be added to the core is 4.9 Ako, and the
maximum rate of addition is 1.35 Ak /sec.

In choosing the initial reactor operating condi-
tions for this incident, we have assumed that the
lowest initial power level would be 1 kw and that
the maximum power would be 7.2 Mw. The former
value is about ten times smaller than the lowest
steady power at which the reactor would be rou-
tinely operated. The initial inlet fuel temperature
was assumed to be 1200°F in all the calculations.

For a typical calculated incident of this type in
which there is no safety rod scram, Fig. 4.5 shows
the transient behavior of the fuel salt temperature
at the hottest point in the core and also the tran-
sient salt temperature at the outlet of the hottest
channel. Here, the initial power level is 1 kw,
and the original reactivity loss Ak is 0.25% in
multiplication factor. Associated with the rapid
rise portion (t = 3.5 sec) of the curve for maximum
fuel temperature are prompt-critical excursions in
power and core pressure qualitatively similar to
those shown in Fig. 4.1. The maximum pressure
rise for this case was about 39 psi. In Fig. 4.5,
the temperatures ultimately reach a maximum and

42

then decrease as the concentrated uranium is again
removed from the core. The dependences of these
peak pressures and temperatures on the initial
power level and on the magnitude of Ako are shown
in Figs. 4.6 and 4.7. No safety rod scram was as-
sumed for these calculations. Figure 4.6 shows
that, for the case of Ak, = 0.25%, the worst condi-
tions for the pressure surge occur at low initial
power levels. This prompt rise in pressure results
mainly from expansion of the core fluid and con-
sequent fluid acceleration effects, which depend
on the rate of increase in the power level. At
higher initial power levels the nuclear heating of
the core fluid tends to slow the rate of rise in
power earlier in the course of the transient. The
dependence on Ako of the maximum temperature
rise of the outlet of the hottest fuel channel is
shown in Fig. 4.7 for the maximum and minimum
initial power levels assumed: 7.2 Mw and 1 kw.
For values of Ak larger than about 0.25%, the
temperature rise is nearly independent of initial
power. These cases correspond to reactivity addi-
tions well above prompt critical, in which the
graphite temperature change is relatively small
and the fuel temperature increase depends on the
integral of the power. At the opposite extreme, for

ORNL-DWG 68~ 966

1700
1600
MAXIMUM FUEL
TEMPERATURE / i \

w
e
w 1500
& F
g
14
w
£ \
Z 1400 OUTLET TEMPERATURE OF
= ’ HOTTEST CHANNEL

1300

/ N
2 4 6 8 10 2 14 16
TIME (sec)

Fig. 4.5. Temperature Excursion Coused by Sudden Resuspension of Uranium Equivalent to 0.25% 8k/k if Uni-

tormly Distributed. Initial power, 1 kw; no safety action.
43

ORNL —DWG 68-973
80

(o))
(@]

D
O
I

PEAK PRESSURE RISE (psi)
n
@]
/

0.001 0.01 OR 1 10
INITIAL POWER LEVEL (Mw)

Fig. 4.6. Effect of Initial Power on Peak Pressure Rise Caused by Sudden Resuspension of Uranium Equivalent
to 0.25% Ok/k if Uniformly Distributed. No safety action.

ORNL-DWG 68-972

600 120
500 [———— HOT-CHANNEL OUTLET 100
TEMPERATURE
g
400 80
w
< =
w
L Q
7] &
@ / w
W w
@ y, a
2 300 60 w
< o
x @
w
a — 0
=
G PRESSURE &
200 40
100 20
0 0
0 0.1 0.2 0.3 0.4 0.5

Ako, EQUIVALENT UNIFORM REACTIVITY (% A k/k)

Fig. 4.7. Effect of Magnitude of Reactivity Recovery on Peak Pressure and Temperature Rises During a Uranium

Resuspension Incident with 233U Fuel, No safety action.
44

small values of the reactivity (Ak, < 0.035%), the
reactor remains below prompt critical at the peak
of the reactivity addition, and delayed neutron
effects tend to control the response characteristics.
The dependence of the magnitude of the pressure
rise on Ako is also shown in Fig. 4.7. The curve
corresponds to an initial power level of 1 kw; on
the same scale, the pressure rise corresponding to
an initial power level of 7.2 Mw would be quite

small.

The preceding calculations take no account of
the control rod safety scram action in suppressing
the nuclear excursion. An indication of the in-
fluence of the initial reactor conditions and the
magnitude of reactivity addition on the action of
the safety scram system is provided by the curves
in Figs. 4.8 and 4.9. For Ak = 0.25%, Fig. 4.8
shows, as a function of the initial power level, the
relative sequence in time of achieving various
conditions during the transient: (1) a 1-sec period,
(2) a 1/2-sec period, (3) high neutron level (11.25
Mw), and (4) peak power conditions with only in-
herent shutdown by temperature feedback effects.
This figure indicates that a scram due to periods

TIME (sec)

ORNL- DWG 68- 5516

4
P [ T T TTT
TIME TO REACH PEAK POWER
| 1 ——
3 - \:fi‘ M
m
8 TIME TO REACH 11.25 Mw \T"FFLL
w2 et AR N o
s TIME TO REACH 0.5-sec PERIOD . q
a e [l B
{ |—] TIME TO REACH f-sec PERIOD ] T]]T i
O {1
0.00! 0.0t X! { 10

INITIAL POWER LEVEL (Mw)

Fig. 4.8. Effect of Initial Power Level on Times to
Reach Scram Conditions and Peak Power Conditions

with 233U Fyel. Ak, = 0.25%.

smaller than 1 sec would considerably precede the
conditions of high neutron level over all but the
highest indicated initial operating power. At these
high power levels, the smaller fractional increment
in power required to reach 11.25 Mw reduces the
time interval relative to that of the peak tempera-
ture conditions.

ORNL—-DWG 68-5517

N

Y

N\
N

NS

S~

AN

TIME TO REACH PEAK POWER —

TIME TO REACH 11.25 Mw

N

[/

\
T —

[ ——

TIME TO REACH 0.5-sec PERIOD
|

~—]

TIME TO REACH 1-sec PERIOD

|

0.4 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Dk, EQUIVALENT UNIFORM REACTIVITY (% Sk/k)

Fig. 4.9. Effect of Magnitude of Reactivity Recovery on Times to Reach Scram Conditions and Peak Power

Conditions with 233U Fuel. Initial power, 1 kw.

‘e
45

In an analogous fashion, Fig. 4.9 shows the
dependence of this time sequence on the magnitude
of Ak, for an initial power of 1 kw. Although the
times to obtain each condition following the start
of the incident are reduced, the time increments
between the curves tend to become constant as
Ak increases.

As would be expected, when the safety rod
scram is taken into account, the magnitude of
Ak required to produce substantial rises in tem-
perature and pressure is increased. Figure 4.10
shows the effect of the scram of two rods, initiated
by the 11.25-Mw level trip, in reducing the tem-
perature and pressure rises associated with large
reactivity additions. The curves shown correspond
to an initial power level of 1 kw. For the scales
shown, temperature-pressure rises corresponding
to 7.2 Mw initial power would be negligible. As
evidenced in Figs. 4.8 and 4.9, for the low initial
power levels, scrams due to short period would
actually precede the high neutron level condition.

Thus the results shown in Fig. 4.10 may be userd
as conservative upper limits on the temperature-
pressure rises for these reactivity magnitudes.
Based only on these limits, recoveries in uranium
corresponding to Ako £ 0.78% could occur without
exceeding the 340°F maximum temperature rise
criterion for limiting thermal stresses to safe
values.® In order to increase this level to a more
realistic but still conservative value, a detailed
simulation of the action of the short-period-scram
mechanism was required. These results are de-
scribed later in this section.

Simulation calculations comrresponding to the
results shown in Figs. 4.7 and 4.9 were also per-
formed for the present 235U fuel salt in order to
directly compare the responses of the MSRE to an
incident of this type with the two fuel loadings.
The normalized reactivity-addition curve shown in
Fig. 4.4 can also be used for the 235U fuel, since
the core hydraulics are unchanged and the spatial
distributions of nuclear importance are very nearly

ORNL-DWG 68~974

500 100
PRESSURE/
/

400 / 80
: /
[} _—
= —— >
W 359 “HOT- CHANNEL OUTLET ___| 60 2
@ /// TEMPERATURE o
(e

v

N 2
: / :
= w
~ a
s >
& 200 a0 2
a W
= / et
W a.
—

100 20

0 0
0.4 0.5 0.6 0.7 0.8 0.9

Aky, EQUIVALENT UNIFORM REACTIVITY (% &k/k )

Fig. 4.10. Effect of Magnitude of Reactivity Recovery on Peak Pressures and Temperatures During a Uranium

Resuspension Incident with 233U Fuel. Rod scrom at 11.25 Mw; initial power, 1 kw,
equal for each fuel. The results of these calcula-
tions are summarized in Figs. 4.11 and 4.12. It
can be seen from Fig. 4.11 that the temperature
rise corresponding to a given value of Ak is
higher for the 235U fuel salt, in approximately the
inverse ratio of the fuel-temperature coefficients
of reactivity for the two loadings.!'® Because of
the larger delayed neutron fraction for 23U, a
larger addition of reactivity is required to reach
the prompt-critical condition. This is evidenced
in Fig. 4.11 by the larger value of Ak required
to produce substantial nuclear heating, starting
from a low initial power level. In Fig. 4.12, it is
evidenced by the longer times following the start
of reactivity addition which are required to reach
the specified conditions. In general, however,
there are no important differences in the responses
in this nuclear incident which could have a deter-

ORNL—DWG 68 —5548

800 [ 160
HOT CHANNEL OUTLET
700 } TEMPERATURE 140
————————

600 | \ 120
[ ~
~:~J 500 100
s &
x SURE ]
L Z
g 400 80 @
i wut
< i x
1 ’ - D
w [4p]
2 g
= 300 60 o
— o

200 : 40

/1
100 // / 20
/ |
/ |
0 L 0
0 0. 0.2 0.3 04 0.5

Aky  EQUIVALENT UNIFORM REACTIVITY (% 8k/k )

Fig. 4.11. Effect of Magnitude of Reactivity Recovery
on Peak Pressures and Temperatures During a Uranium
Resuspension Incident with 235y Fyel. No safety

action,

mining detrimental effect on nuclear safety for
either fuel salt.

In order to further investigate the upper limits of
reactivity addition which could be tolerated in this
type of incident with 233U fuel, some analog simu-
lation studies of the MSRE were made.’ The
analog model included a representation of the
dependence of the period scram signal on the
initial neutron level, the gamma background, and
the history of the nuclear transient prior to scram.
In general agreement with the results shown in
Fig. 4.8, the analog calculations showed that, for
initial power levels above about 7 Mw, the high
neutron level trip would actuate the scram signal;
at less than 7 Mw, the period trip would actuate
the signal. For Ako between 0.9 and 1.2%, the
time to start a rod scram varied between 300 and
800 msec, with the shorter times corresponding to
higher Ak and low initial power.

Typical reactivity inputs in these cases are
shown in Fig. 4.13. Before a rod scram, the input
is very small and the resulting power transient is
also small. This implies that the excursions of
potential harm are those that could overcome the
available shutdown worth of the rods. In these
analyses, this was assumed to be 4.0%, correspond-
ing to the worth of two rods, dropping from an
initial position of 44 in., in the ‘‘shadow’’ of the
third rod over their entire length of travel. Thus,
reactivity additions large enough to cause a secon-
dary nuclear excursion correspond to Ako > 0.82%,
as seen from Fig. 4.13. For these large additions
with period scram, the power-temperature transients
were calculated both with the analog model and
with the digital program. Typical results are given
in Figs. 4.14 and 4.15. It may be seen that the
two models were in close agreement in calculating
power transients (Fig. 4.14) but that there were
some notable differences in the temperature calcu-
lations (Fig. 4.15). The analog model tends to
reach a peak temperature faster than the digital
calculation, but at a lower magnitude. These
models assume somewhat different mechanisms
for heat transport within the core. The analog
model uses a 27-lump representation of the MSRE
core, ® within which ‘“‘well stirred’’ conditions are

70. W. Burke and F. H. Clark, ORNL, personal com-
munication (January 1968).

8s. J. Ball and T. W. Kerlin, Stability Analysis of the

Molten-Salt Reactor Experiment, ORNL-TM-1070 (Decem-
ber 1965).
ORNL—-DWG 68-5519

8
7
6
5
B \
w
w4 \
= N\
P—

~_TIME TO REACH PEAK POWER |

//

TIME TO REACH {1.25 Mw
| N ]

\E

I I
TIME TO REACH 0.5-sec PERIOD

TIME TO REACH 1-sec PERIOD |

0 04 0.2 0.3

0.4 0.5 0.6 0.7 0.8

Akg ,EQUIVALENT UNIFORM REACTIVITY (% & /r//r)

Fig. 4.12, Effect of Magnitude of Reactivity Recovery on Times to Reach Scram Conditions and Peak Power

Conditions with 235U Fuel. Initial power, 1 kw.

assumed, implying immediate heat transport within
each lump. In contrast, the digital model assumes
no axial conduction or mixing.?

Peak hot-channel outlet temperatures as a func-
tion of both Ak, and initial power level are shown
in Fig. 4.16. The analog model indicated that a
Ak of about 1.07% would be required before the
tolerable temperature limits were exceeded. A
more conservative value of 0.96% was obtained
from the digital calculation. Both models indi-
cated that peak temperature was not very depen-
dent on initial power level, particularly for the
higher values of Ak,.

The magnitudes of the peak pressure rises calcu-

lated for these large reactivity additions with
period scram are shown in Fig. 4.17. These re-
sults also exhibit the inverse relation between

initial power level and peak pressure discussed in
the preceding sections. In order to obtain an upper
limit on the maximum pressure rise, an initial
power level of 100 w was chosen. Projection of
the calculated data to 100 w initial power shows
that a Ak of 0.95% could be tolerated without ex-
ceeding the conservative 50-psi limit in pressure
rise assigned in the safety analysis.

4,3 DETECTION OF ANOMALOUS
REACTIVITY EFFECTS

B. E. Prince

During routine operation of the MSRE, calcula-
tions of the complete reactivity balance are made
at 5-min intervals by the on-line digital computer.
ORNL~DWG 68-5520

| T — 1
| |
A/r0=|‘0%

o 5
5 / A/rO=OA816%\
: \
w
[
a - —= —
tJ
w)
2 i
>
= /T\
> -2 ‘ : —]
5 4 N
< ///£%=05% \\\
x

-3 | | | :
’ / | | \
) |
0 2 4 6 8 10 12

TIME (sec)
Fig. 4.13. Time Dependence of Total Reactivity Input

with Combined Uranium Resuspension and Rod Scram
Due to Short Period.

ORNL-DWG 68-5521

2000 ,
r MNALOG 4--DIIGITAL
1000 — —
500 — —]
200 ——— ‘
100 ’ EANALOG
i
50 | IIII h:
{ V/ \\
E] T
2 20
(r H
n N
z 10 A\ }
s = \ :
5 :x; \ . {_fi
L | _ -
)N
1.0 —4 . : \\\ -
- J -
—‘ 7
05 | T ~ ]
Nert”
0.2 1 : fl ‘
| |
0 2 4 6 8 10 12 14

TIME (sec)

Fig. 4,14, Time Dependence of Power Level for a
Uranium Resuspension Incident with Ako = 1.2% and
Initial Power = 7.5 Mw.

48

ORNL-DWG 68-5522

500 T
DIGITAL

 ANALOG/

O
(@]
(@]

n
o
o

|
Aky=12% |

o
o

MAXIMUM OUTLET TEMPERATURE CHANGE (°F)

100 |—- —

ANALOG/

¢
|

10 12 14

-100

|

|
z‘xk =0.9%
|

8
TIME (sec)

Fig. 4.15. Time Dependence of Hot-Channe! Outlet
Temperature, Calculated with Both Analog and Digital

Models. Initial power = 7.5 Mw.

Experience with these calculations, accumulated
during a substantial part of the operation with the
present 235U loading, has been described in ref 9.
At steady power, temperature, and pressure, we
have found that the variation in consecutive reac-
tivity balances is about +0.01% 6k/k. This value
represents the probable limit of precision in the
technique. Longer term variations in the residual
reactivity can be due either to systematic errors in
calculation of known reactivity effects or to
anomalous changes whose detection is important,
such as the hypothetical situation considered in
the preceding section. The long-term reactivity
variations which have been observed with the
present fuel salt have been very small. A sys-
tematic variation as small as 0.05% 6k/k would

9_]. R. Engel and B. E. Prince, The Reactivity Bal-
ances in the MSRE, ORNL-TM-1796 (March 1967).
ORNL-DWG 685523

PEAK HOT CHANNEL OUTLET TEMPERATURE (°F)

1800 T
ANALOG
MAXIMUM
1600 ALLOWABLE ] . fi
1400 -
1200
1800
! DIGITAL g
1600 MAXIMUM
ALLOWABLE — — —0— — _L__
INITIAL POWER
o 75 Mw
/ A 500 kw
o 100 kw
o 10O kw
{200 L. \ 1
0.7 0.8 0.9 1.0 1.1 1.2 1.3
Doy \To Bh/K)

Fig. 4.16. Dependence of Peak Hot-Channel Outlet
Temperature on Ako and Initial Power Level for Large
Reactivity Additions with Short-Period Scram.

be clearly observable by the computer-calculation
monitoring. A systematic long-term change larger
than 0.10% &6k/k should be easily separable from
the effects of any miscalculation of known reac-
tivity changes.

When the present fuel loading is changed to
233y, the fissile-uranium-concentration reactivity
effects will be larger,! by about a factor of 2.

This means that the reactivity balance calcula-
tions can be expected to be more sensitive to
variation in uranium content, For example, re-
moval of 1 g of 233U from the salt, by burnup or
any other mechanism, may be expected to produce
a reactivity reduction nearly four times that of 1 g
of 235U in the present fuel. An abnormal reactivity
loss of 0.10% &k/k would correspond approximately
to a 0.26% reduction in uranium concentration (of
isotopic composition to be used for the new load-
ing), which is total loss of about 80 g of uranium
from 70.5 ft3 of circulating salt. On the other

ORNL—-DWG 68-5524

400 }— T———
A
INITIAL POWER
350 ® 75 Mw
1 20Mw
A 500 kw
0100 kw
o
300 | (O kw

4 100 w(PROJECTED)’
250 +— /

200

o

PRESSURE RISE (psi)

ool /

MAXIMUM
50 b ——— i e —_——
ALLOWABLE
0 5
0.8 0.9 10 .4 i.2 1.3
A kg (%)

Fig. 4,17. Dependence of Peak Pressure Rise on
Ako and Initial Power Level for Large Reactivity Addi-

tions with Short-Period Scram.

hand, the upper limit of 0.95% &6k/k obtained in the
studies described in the preceding section cor-
responds to apptoximately 2.4% in uranium con-
centration (750 g of total uranium). Thus, there

is considerable latitude in establishing an admin-
istrative safety limit on anomalous reactivity ef-
fects to prohibit nuclear operation when the
residual term in the reactivity balance becomes
too large.
Part 2. MSBR Design and Development

R. B. Briggs

The primary objective of the engineering design
and development activities of the MSR program is
to design a molten-salt breeder experiment (MSBE)
and develop the components and systems for that
reactor. The MSBE is proposed to be a model of
a large power breeder reactor. Most of the design
and development effort is being spent on studies
of reference designs of 1000 Mw (electrical) and
larger power plants to provide the criteria and
much of the basis for design of the MSBE.

These studies were begun in September of 1965.
The results are reported in four semiannual prog-
ress reports through August 1967 (refs. 1—4) and
in ORNL-3996, Design Studies of 1000-Mw(e)
Molten-Salt Breeder Reactors. Based on the
studies, a two-region reactor with fissile and
fertile materials in separate fuel and blanket salt
streams was chosen as the most promising ap-
proach to the breeder. Factors related to fuel and
blanket processing were the most important con-
siderations in reaching this conclusion. A fluoride
volatility process was available for separating
uranium efficiently from fuel and blanket streams.
A distillation process had been conceived for
separating the "LiF and BeF_ salts from rare-
earth fission products in the fuel salt. Separation
of #33Pa from the blanket salt was not essential
to achieve breeding, although such separation
appeared to be possible and offered promise of
improving the breeding performance and the
economics.

TMSR Program Semiann. Progr. Rept. Feb. 28, 1966,
ORNL-3936, p. 172.

MSR Program Semiann. Progr. Rept. Aug. 31, 1966,
ORNL-4037, p. 207.

3MSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, p. 174.

4MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 63.

50

The design studies on the large two-fluid
breeder reactors reached a reasonable stopping
point in September of 1967. The design power for
the MSBE had been selected as 150 Mw (thermal) ,
the objectives and program for development of
that reactor had been summarized in ORNL-TM-
1855, and the programs for development of com-
ponents and systems had been outlined in ORNL-
TM-1855 and ORNL-TM-1856. Preparations were
being made to begin the conceptual design and
the development for the MSBE when two new de-
velopments caused us to delay that work and to
continue with the design studies of large power
plants.

One of the new developments was evidence that
233Pa and possibly the rare-earth fission products
can be separated from the mixed thorium-uranium
fuel salt of a one-fluid reactor by reductive ex-
traction methods employing liquid bismuth. The
second new development was the calculation that
the leakage of neutrons from the core of a one-
fluid reactor of modest size can be reduced to
desirable levels by increasing the ratio of fuel
salt to graphite to produce a greatly undermoderated
region in the outer nart of the core. By optimizing
the distribution of salt, the reduction in leakage
can be accomplished with a low specific inventory
of fissile material. These developments, when
combined with the greater simplicity and potential
reliability, make the one-fluid breeder a more
attractive concept than the two-fluid breeder,
which must rely on graphite piping in the core to
separate fuel and blanket streams.

Our current design studies of large one-fluid
breeders are aimed at describing those reactors
sufficiently for us to select a size and a design
basis for a one-fluid breeder experiment. De-
velopment work is limited to analyses in support
of the design and to experiments that are rzlevant
to one-fluid and two-fluid reactors.
5. Design

E. S. Bettis

5.1 GENERAL

E. S. Bettis R. C. Robertson

QOur version of a 1000 Mw (electrical) powes
plant based on the two-fluid MSBR was described
in considerable detail in our last semiannual re-
port.} We selected the modular concept in which
the plant contained four reactors, each with the
capability for generating supercritical steam
equivalent to 250 Mw (electrical), The core of
each reactor was designed to accommodate for a
reasonable time the dimensional changes produced
by irradiation of the graphite as those changes
were deduced from the most recent experimental
data. This led to the use of graphite fuel cells
in the form of reentrant tubes that were brazed to
metal pipes. These pipes were welded into fuel
supply and discharge plenums in the bottom of
the reactor vessel. The fertile salt filled the
interstices between fuel cells and a blanket re-
gion around the core.

This reactor presented two major problems. The
graphite in the core had to be replaceable for a
variety of reasons, and the most acceptable way
that had been conceived for replacing the graphite
involved replacing the entire core and reactor
vessel. Periodic replacement of the reactor vessel
appeared to be economical if the frequency were
governed entirely by radiation-induced dimensional
changes in the bulk graphite. There was, however,
concern that mechanical failure of individual
graphite cells or graphite-to-metal joints could
cause the replacement frequency to become ex-
cessive,

1MSR Program Semiann. Progr., Rept. Aug. 31, 1967,
ORNL-4191, p. 63.

The single-fluid reactor has the advantage that
the graphite functions only as moderator. In
principle, it can be present in the form of long
bars with no firm connections at top or bottom;
the bars can be removed individually or in groups;
the lifetime of the graphite should depend only
on the bulk changes in dimensions that result from
irradiation. There are also subsidiary advantages
related to the amount and arrangement of the
reactor equipment. These factors combined to
cause us to look seriously at the design of a
single-fluid reactor when there was evidence that
fuel processing could be provided to make a single-
fluid reactor a breeder.

In the beginning we expected that the reactor
would have to be a 20-ft-diam right cylinder or
larger in order for the neutron leakage to be ac-
ceptably low. A power output of 2000 Mw (elec-
trical) or greater per reactor then became neces-
sary in order to achieve a low specific inventory.
We believe that power plants of this size will be
common in the 1980’s when large molten-salt
reactors could be built, so we based our design
studies on a 4444 Mw (thermal), 2000 Mw (electri-
cal) plant. We have since found that zoning the
core permits one to obtain good breeding perform-
ance from 1000 Mw (electrical) and possibly from
smaller reactors. The design studies reported
here are for the 2000 Mw (electrical) reactor, but
the performance of 1000 Mw (electrical) and 2000
Mw (electrical) reactors is compared in the re-
actor physics section.

5.2 FLOW DIAGRAM

R. C. Robertson H. L. Watts

The flow diagram for a one-fluid reactor plant
with a capacity of 4444 Mw (thermal) and 2000 Mw
ORNL -DWG 68-4192

Ve

VENT
rr-oas|COOLING
HOLD UP
TANK
PRESSURE COOLING
COOLANT
SALT PUMP
FUEL FUEL VENT {;() 1/4:) 1;/()
PUMPS PUMPS ————n- SN —o.- SIS SRRt B S -
G (e 1 I l/ | |
S g = =B
f FROM CHEMICAL . L - -
- 3 PROCESSING 3 3 I ® [\ ' ®
NO.4) (-NO.3 NO.2H, IO O\ €
I Il
2 mmmmw 2 i
| STEAM SUPERHEATERS STEAM SUPERHEATERS

HEAT EXCHANGERS
114

\W“
iy u‘
REACT0R|H

1

HEAT EXCHANGERS
Ml
NO.1

lj

REHEATERS

TO CHEMICAL
PROCESSING

PRESSURE >
AND VENT

FUEL
DRAIN TANK

FREEZE VALVE

PROPORTIONAL FLOW VALVE

GAS SEPARATOR

FUEL

COOLANT

STEAM *

i//\ =

2" ~ - S
HEAT REJECT
STACK

TO CHEMICAL
PROCESSING

%% STEAM GENERATING UNIT NO. 2
#xx STEAM GENERATING UNIT NO. 3
xx% STEAM GENERATING UNIT NO. 4

Fig. 5.1.

'0
©
=z

PEEAROEEEG

#l

-

STEAM GENERATING UNIT NO. 1

PRESSURE
AND VENT
COOLANT SALT
DRAIN TANK
TEMP
°oF FLOW P
1000 200 f1¥/sec
1300 200
1300 50
1150 75
850 75
1150 75
1150 0-20
1150 16.2
850 16.2
1150 10

Flow Diagram for 2000 Mw (Electrical) Station.

o
z

PEEILREEE®

—

_{
° m
oz

)

850

FLOW

75 ft3/sec

10

256 x10° Ib/ hr
2.56x 10

1.26x 10
1.26)(106

4 f1%/sec

4
420,000 ft /min
420,000

(4]
Table 5.1. Estimated Properties of Fuel and Coolant Salts for One-Fluid

53

Breeder Reactors

Fu

el Salt at 1050°F Coolant Salt at 800°F

Composition, mole %

-1 0 ~—1

Specific heat, Btu 1b F

Density, 1b/ft3

Thermal conductivity, Btu ft—1 hr—1 Op—1

Viscosity, 1b ft™! hr!

Liguidus temperature, °p

BeF

29 LiF, ThF4, UF4 BF , NaF (48, 52)

(20, 67.7, 12, 0.3)

0.29 0.37
223 125
0.49 0.46
34 34

930 715

(electrical) is shown in Fig. 5.1. This diagram is
similar to those shown previously for the two-
fluid reactor with the obvious omission of the
fertile salt circuit.

The fuel salt for the one-fluid reactor con-
tains both fissile and fertile atoms in 7LiF-BeF2
carrier. The composition and our estimates of the
properties are shown in Table 5.1. In the flow
diagram this salt is circulated by four pumps
through a common reactor vessel. Each pump
circulates approximately 27,000 gpm of salt
through the reactor and a heat exchanger. The
salt enters the reactor at 1050°F and leaves at a
mean temperature of 1300°F,

The coolant salt — sodium fluoroborate — is
pumped in four heat transfer circuits, one for each
fuel circuit. Each pump circulates 53,000 gpm of
coolant salt through a primary heat exchanger and
through several superheaters and reheaters. The
coolant salt enters the primary heat exchanger at
950°F, flows to the superheater at 1150°F, and
leaves the superheater at 850°F. Since 850°F is
below the liquidus temperature of the fuel salt,
some 1125°F salt is bypassed around the super-
heater and mixed with the exit salt so that the
temperature of the coolant returning to the primary
heat exchanger is 950°F.

Each coolant salt loop has four superheaters,
making a total of 16 units in the plant. There are
two steam reheaters per coolant salt loop, a total
of eight units in the plant. Steam enters the
superheaters at 700°F and exits at 1000°F. Steam
enters the reheaters at 600°F and is heated to
1000°F for return to the turbine.

5.3 PLANT LAYOUT

C. E. Bettis H. M. Poly

C. W. Collins J. R. Rose

W. K. Crowley W. Terry
H. L. Watts

With the new reactor and the higher power, our
entire concept of the overall power breeder plant
has to be reevaluated. Such concepts as the
basic idea for the thermal shield and the double
containment are still valid, but we have devoted
only a little time to the layout of the new plant
and have not examined those details.

A plan view of a possible arrangement of the
various cells is shown in Fig. 5.2. The reactor
vessel, four heat exchangers, and four fuel pumps
are located in the reactor cell. This cell is
circular and is about 52 ft in diameter by 47 ft
deep. It has a thermal shield to prevent over-
heating of the concrete. It also has double con-
tainment, with a pump-back system whereby the
integrity of the containment is constantly moni-
tored.

Four steam-generating cells are located sym-
metrically in relation to the reactor cell. The
cells are approximately 33 ft wide by 46 ft long
by 43 ft deep. They contain only coolant salt and
steam and therefore have no need for a thermal
shield or double containment. These cells are
isolated from the reactor cell and from the high-
bay area by bellows seals around pipes that com-
municate with those areas.

The fuel drain tank is in a cell all its own.
This cell is below the level of the reactor cell
ORNL—DWG 68~ 4194A

FUEL SALT

INSTRUMENTATION
DRAIN TANK

CELL

155 ft - Qin.

COOLANT SALT
DRAIN CELL 43 fT-Oinx-——*. ’G—fi49 ft—=0in.—

COOLANT SALT
DRAIN CELL

33ft—Cin.
(BELOW GRADE)

26f1-0in.

el 71— REHEATER

33f1-0in. STEAM GENERATOR
129 ft-Oin. l
COOLANT SALT
PUMP
HOft-6in.
33f1-0in. (BELOW GRADE)

1 - CHEMICAL
| PROCESSING CELL

18 f1-0in.

\_7__l L l i_
!
OFF-GAS HEAT EXCHANGER | >
PROCESS CELL
REACTOR
- 62 ft-Qin. ————— =

|NSTRUMENT515§1,///
CELL

Fig. 5.2. Plon View of Cell Arrangement.

T — 1

vs
in order for salt to drain by gravity from the re-
actor into the drain tank. Double containment is
provided since radioactive salt is stored there.
The drain tank cell is about 28 ft wide by 49 ft
long by 38 ft deep. It is isolated from the reactor
cell by bellows seals around the communicating
salt lines.

The coolant salt is stored in a separate cell
about 26 ft wide by 43 ft long by 45 ft deep. This
cell needs little shielding and need be sealed
only sufficiently to prevent excessive heat loss.

The off-gas cell is approximately 18 ft wide,

63 ft long, and 43 ft deep. This cell must be
doubly contained, but the ambient temperature is
around 100°F, so it needs no electrical space
heaters or thermal insulation. Cooling of the gas
holdup tanks and the charcoal absorber beds is
done by water which comes from the plant feed-
water system.

The chemical processing cell is about 18 ft
wide by 63 ft long by 43 ft deep. This cell must
have double containment, but it too does not re-
quire space heaters or thermal insulation. In this
cell the pieces of apparatus are heated and cooled
individually.

On the cell plan we show two cells for instru-
mentation uses such as test points, junctions,
instrument air sources, etc. These cells have
approximate dimensions of 13 by 26 by 43 and
8 by 48 by 43 ft.

The off-gas, chemical processing, and instru-
mentation cells shown in Fig. 5.2 are meant to be
only representative since no serious design effort
has been put on them. The reactor, fuel drain
tank, and steam cells have been laid out suffi-
ciently to show the arrangement of the components
in some detail. These cell arrangements are
shown in Figs. 5.3 to 5.5.

Our arrangement of equipment for the one-fluid
reactor is based on one-pass upward flow of fuel
through the reactor vessel. This ““once-through’’
flow allows the reactor and heat exchangers to be
at the same elevation. The pumps are at the top
of the reactor and have drive shafts that may be
short enough to eliminate the need for molten-salt
bearings. The heat exchangers are mounted so
that they move and the thermal stresses are ac-
commodated without the use of expansion loops ot
expansion joints in the fuel piping. We have used
bellows in the coolant salt piping. We hope to be
able to do this, for the coolant pipes are 34 in. in
diameter, and expansion loops for such sizes are

55

very bulky. In the steam cell all components are
anchored solidly. Expansion in pipelines is taken
up by bellows in the pipes. The high-pressure-
steam and feedwater lines have large expansion
loops outside the cells to allow for dimensional
changes in those lines.

5.4 REACTOR VESSEL AND CORE

H. F. Bauman H. M. Poly
C. W. Collins W. Terry
W. C. George H. L. Watts

We have considered several concepts for the
single-fluid reactor vessel and core. Studies have
been made to try to attain the best combination of
fuel inventory, flow control, temperature control,
and ease of removal of the graphite from the re-
actor. A major complicating factor in the design
is the radiation-induced dimensional changes in
the graphite.

We spent considerable effort in an attempt to
employ reentrant concentric tubes of graphite for
the fuel passages in the core. By confining the
salt to such tubes and filling the spaces between
tubes with gas, the changes in salt volume caused
by the dimensional changes in graphite are in-
significant. This design, however, made necessary
some kind of metal-to-graphite joint that had to
be relatively leakproof, and these joints would
have to be broken to replace the graphite. Also,
the gas spaces in the core presented voids which
could fill with salt and greatly increase the re-
activity under certain conditions. Gravity flow
from the reactor vessel to the pump suction tank
was believed to be desirable, but the inventory
required to achieve this was excessive. The
concentric reentrant tube concept was abandoned
as having too many of the disadvantages of our
designs for two-fluid reactors and none of the
advantages.

The idea of a container filled with vertical
graphite pieces of some shape with fuel in the
spaces between graphite pieces and flowing in a
single pass upward through the vessel was
finally adopted for the basic design. With this
configuration, two possibilities for disposing the
graphite within the vessel come to mind. One
involves simply stacking graphite inside the
vessel, each piece contacting the neighboring
pieces and the entire bundle being finally re-
strained by the reactor vessel or by a cratelike
structure within the vessel.
ORNL-DWG 68-4190A

FUEL SALT

/T

INSTRUMENTATION
CELL

XXXXXX XA XXX

47 1t Oin.

HEAT

REACTOR EXCHANGER

43ft Oin.

L —INSTRUMENTATION
CELL

+—FREEZE VALVE

000006 OO0 0006 0G00I

/GROUND FLOOR LEVEL

CATCH PAN FUEL SALT
DRAIN LINE

EMERGENCY )
DRAIN LINE

38 ft Gin.

&— FUEL SALT
DRAIN TANK

Fig. 5.3. Elevation of Reactor and Fuel Drain Cells.

99
ORNL-DWG 68-4189A
FUEL SALT PUMP

//COOLANT SALT PUMP

REACTOR 3 — :Mfl“jj STEAM

PIPING

NN
HEAT
EXCHANGER

STEAM REHEATER
GENERATOR

Fig. 5.4, Elevation of Reactor and Steam Cells.

LS
,_._______
1

I
I
[
I
|
J

ORNL-DWG 68-4193

INSTRUMENTATION
CELL

Iln

o
‘ %
%

AN

| — |

REACTOR

CELL

i

STEAM CELL

Fig. 5.5. Plan View of Reactor and Steam Cells.

B
B
-.
%

STEAM
PIPING

k]

8S
This simple placement of the graphite might
suffice if there were no dimensional changes with
irradiation, but the graphite will shrink, the salt
fraction in the core will increase, and this in-
crease might be localized to create a rather large
pool of salt at some indeterminate and possibly
changing location within the core. Various
schemes were attempted to try to cause some
force (e.g., hydraulic pressure drop, buoyancy) to
compact the core as irradiation caused the pieces
to shrink., To date we have found no practical
method of overcoming this ‘‘crevassing”’ of the
core; we therefore prefer another design.

One way to be sure that the liquid volume will
increase uniformly within the core as the graphite
shrinks is to locate each piece of graphite at the
top and bottom of the reactor by a metal structure
that is dimensionally stable. The method of con-
fining the graphite at these locating points is
important. The attachment must present a minimum
of difficulty to removal of the graphite. It must
accommodate dimensional changes of the graphite
without undue restraint.

In addition to positioning the graphite, means
must be provided for supporting the considerable
load of the graphite in the core. When the reactor
is empty, this is a gravity load at the bottom of
the vessel. When the reactor is filled with salt
having approximately twice the density of the
graphite, the graphite floats and exerts a buoyant
force almost equal to the weight of the graphite
in the empty vessel. When the pumps are running,
there is an additional large upward force caused
by the pressure drop across the core.

We have investigated many configurations of
vessels with reinforced flat heads, dished heads,
and inverted dished heads. We tried circumfer-
ential fluid entry, through transition sections from
pipes to the reactor vessel. A design which
seems to offer the best embodiment of the de-
sired features is shown in elevation in Fig. 5.6.
Some important patameters for the reactor are
listed in Table 5.2.

The vessel is about 18 ft in diameter by 24 ft
high. It has a standard dished head at the bottom.
In the center is a 4-ft-diam manifold into which
the 24-in. outlet line from each heat exchanger
dischatrges, and the four streams mix in the plenum
formed by the dished head of the reactor vessel.

Mounted above the dished head is a flat sup-
port plate with perforated web reinforcing. This

59

Table 5.2, Design Parameters for One-Fluid Reactor

Thermal power, Mw 4444
Vessel diameter, ft 18.3
Vessel height, ft 24,5
Core diameter, {t 16
Core height, ft 20
Core volume, £t 3 4020
Average power density, kw/liter ~40
Fraction of liquid in core, %
Region 1 19
Region 2 17
Region 3 44
Reflector thickness, in. 12
Number of core elements 1760
Maximum velocity of salt in core, fps 13
Salt volumes, ft3
Core 1240
Reflector 25
Plena 590
Heat exchanger 320
Pumps 120
Piping 150
Total 2445
Fissile inventory, kg 1880
Fertile inventory, kg 90,000
Specific power, Mw (thermal)/kg 2.36

plate locates and supports the weight of the
graphite stringers comprising the center part of

the reactor core. The support plate is drilled on
an even square pitch, and nipples for receiving the
round ends of the moderator stringers are welded
into these holes. These nipples serve as orifices
for controlling flow and as sockets for locating
the graphite pieces.

The outer region of the core is mounted in
thimbles the same as the center region. Flow in
this outer region is greatly reduced over that
required in the center. This flow is obtained from
the pool below the center region of graphite
stringers and may be regulated by orificing the
flow channels through and around the graphite
pieces,

Near the top of the reactor vessel a square grid
is welded to the vessel. The squares in this
grid are large enough to contain nine of the core
pieces. This grid locates the top ends of the
pieces so that each group of nine is exactly
positioned within the core. The grid web is ¥, in.
60

ORNL-0OWG 68-41912

. ygftOInID——
- — H C7° F e —
16 ft Oin. DIAM CORE (38 UEL) COVER TRUSS
— |{t 4in. DIAM (13 % FUEL)— CONTAINMENT
6ft Gin. DIAM (18 % FUEL)*—-‘ SEAL WELDS
HOLD DOWN
PLATE
34t Oin.
FUEL TO -
PUMPS (4) ~= H
‘.—‘. k = i 1 “ I}
CONTROL ROD
N
REFLECTOR j\
GRAPHITE 24§t 8in.
PRESSURE 20 ft Qin.
VESSEL CORE
N
FUEL
ELEMENTS T
{8 ng
xfi§ —— J— S .- —
77 7%

Fig. 5.6. Elevation of Reactor Vessel,
thick. In order for the graphite to fit closely, the
top 16 in. of each piece is reduced in cross sec-
tion to permit its insertion past the grid. The
center piece of graphite is inserted last and
serves as a key to tighten the group. To remove
the graphite, the center piece must be removed
first.

With the sockets at the bottom and the grid
structure at the top of the vessel, the graphite is
supported freely and is retained in one radial
position. It is free to float in the salt and must
be restrained by some loading bearing structure at
the top. We have made the top of the vessel a
flat reinforced lid. This lid is held down by I-
beams clamped at the perimeter. A seal weld is
made for leak-tightness only. In each of the grid
openings above the graphite stringers, we have
a metal spider with a center rod which transmits
the buoyant and pressure forces to the reinforced
lid of the vessel. This spider also has provision
for orificing the flow of salt as it leaves the core
in order to help control the distribution of flow.

For nuclear reasons (discussed in the physics
section), the volume fraction of fuel in the core is
nonuniform. As currently conceived, the core has
three regions characterized by different salt
fractions. The central one-sixth of the core,
region 1, contains 19 vol % salt; the surrounding
one-third, region 2, contains 17 vol % salt; and
the outer one-half, region 3, contains 44 vol %
salt. The power production per unit volume of
salt in region 1 is nearly uniform, It decreases to
about 27% of the center value at the outer edge of
region 2 and to 3% of the center value near the
outer edge of region 3. We would like for the tem-
perature rise of the fuel that passes through the
core to be nearly uniform and equal to the speci-
fied mean temperature rise. This is desirable in
order to prevent overheating in the center of the
core and to conserve on total flow through the
reactor,

Many configurations of core pieces have been
examined in an attempt to achieve the desired
distributions of fuel volume and flow. The one
we presently like best is shown in Fig. 5.7. The
graphite element is 4 in. square with the edges
contoured and the center cored to obtain the speci-
fied salt fraction in each region. The design ap-
pears to eliminate places where salt could stag-
nate and overheat.

Present plans are to make the channels in re-
gion 1 of the core of uniform hydraulic radius and

61

to permit the salt to flow freely through all the
channels. In the regions 2 and 3 the channels
around the pieces, which are more difficult to
orifice, will be reduced in thickness as the heat
production decreases, and the channels through
the pieces, which are more easily orificed, will
be increased in size to obtain the desired fuel
fraction. By distributing the salt in this manner,
providing some orificing of the channels between
pieces, and carefully orificing the channels in
centers of the pieces, we hope to obtain a nearly
uniform radial pressure in the interconnecting
channels across the core. This should minimize
the cross flow between core regions at the ex-
pense of a small increase in total flow and de-
crease in overall temperature rise of the fluid.

We will have no firm design for the orifices or
good information about the flow distribution until
a study has been made of the flow in the reactor
core. This is in progress, but we have no re-
sults even of a preliminary nature.

5.5 PRIMARY HEAT EXCHANGER

C. E. Bettis H. L. Watts

We have not designed a heat exchanger for the
one-fluid reactor. What we have done is to scale
up the heat transfer surface from previous cal-
culations, base the length on the height of the
reactor vessel, and calculate the diameter re-
quired to provide the desired capacity. The heat
exchangers now have one-pass flow of both fuel
and coolant salt. The pumps are not integral with
the heat exchanger, although they can be made so
if further analysis shows it to be desirable.

5.6 FUEL DRAIN TANK

W. C. George H. M. Poly
H. L. Watts

We have spent a large part of our design effort
on a system for removing afterheat from fuel salt
in the drain tanks. With the modular plant design
of the two-fluid reactor, we proposed to cool the
drain tanks by reheat steam from the high-pressure
turbine. Since the likelihood of having more than
two of the four modules drained at any one time is
small, the supply of this cooling steam should be
dependable. With a single reactor, this is no
longer true, and steam cooling, while still possi-
ble, looks less attractive,
ORNL-DWG 68-4196A

i

T
|
|
~ 54,
|
|
|

SECTION A-A
FUEL VOLUME -18%

S

|
|

21 ft 6in. 21 f+ 6in.

R e TR
IN NS
\\

A "
N
\\ ‘
SECTION A-A
FUEL VOLUME ~13%
1Y% R

|

Y.
4
ot

{4+ Oin.

SECTION C-C
FUEL VOLUME -38%

[N W

NOTE : DIMENSIONS ARE IN INCHES SECTION B-B

Fig. 5.7. Configurations of Core Elements.
Another basic change to the drain system re-
sults from our arrangement of equipment for the
one-fluid reactor. A stopped pump no longer
drains the reactor. We presently see no reason for
draining the reactor quickly and would propose to
continue to operate at reduced power for some
time with one or possibly more pumps off.

We have one drain tank capable of holding and
cooling the entire salt volume (2247 ft*) of the
reactor primary system. This is a cylindrical
tank connected to the primary system through a
12-in, line. This line has a salt valve, at present
assumed to be a freeze valve, through which the
system drains into the tank. An additional 6-in.
line enters the tank near the top. This line is
connected through a freeze valve to a drain pan
in the reactor pit. In case of failure of a fuel
system component, the fuel can be drained into
the drain tank through this line. Salt is trans-
ferred to the reactor system by pressurizing the
drain tank with gas. Figure 5.8 is an elevation
drawing of this tank. Figure 5.9 is a plan of the
tank showing the manifolding of the cooling coils
which remove the afterheat from the fuel.

The cooling is done by U-tubes which extend
into the tank from headers located at the top but
below the dished head cover. These are ¥ -in.
tubes, 0.42-in. wall, with a separation of 3 in.
between the tubes in each U. The U-tubes are
attached to headers which are arranged on two
levels, one level running at right angles to the
other level. This arrangement results in coolant
tubes on 2% -in. square pitch throughout the volume
of the drain tank. The U-tubes are assembled in
modules, and these modules are connected to-
gether by bringing 1-in. lines through the dished
head of the drain tank.

There are many more coolant tubes in the drain
tank than are required to remove the heat. We did
this at least temporarily in order to reduce the
ligament dimension of the fuel to prevent hot
spots within the fuel volume. The coolant surface
in the drain tank can remove 300 Mw, while the
maximum load amounts to only 60 Mw. In Table
5.3 we show some design parameters of the drain
tank.

The top layer of coolant headers are part of
one autonomous coolant circuit, and the bottom
layer of headets are part of another circuit. Each

63

Table 5.3. Design Parameters for Drain Tank

Diameter of tank, ft 16
Height of tank, ft 20.5
Volume for fuel, ft° 2500
Volume of coolant in tank, ft3 278
Number of coolant circuits 2
Number of U-tubes per circuit 1540
Length of Ustube, ft 30
Velocity of coolant in tubes, fps 0.75
Overall U in tank, Btu hr™! ft~2 O ~! 90
Tank wall thickness, in. 11/4
Design pressure, psi 70
Number of modules of tubes per circuit 100
Tube diameter, in. 3/4
Tube wall thickness, in. 0.042

set of headers is piped to a finned-tube air cooler.
These coolers are located inside a chimney or
chimneys outside the building. We have not yet
decided whether to provide a separate chimney for
each coolant circuit.

The chimneys are equipped with dampers so
that the air coolers are maintained in a stagnant
hot ambient when not required to reject heat. In
fact, heaters are provided to keep the air coolers
above the melting point of the coolant salt at all
times. This system of air coolers, U-tubes in the
drain tank, and interconnecting piping is filled
with the fluoroborate coolant salt. The salt circu-
lates by natural convection, conveying the heat
from the drain tank to the air cooler and dumping
it into the atmosphere.

We have only preliminary designs of the air
coolers and chimneys. A manufacturer of such
chimneys says that a stack about 150 ft high and
about 30 ft in diameter will produce a pressure
drop of about 1 in. of water, which will be suffi-
cient to remove 60 Mw of heat from air coolets in
the base of the chimney.

By using this type of heat removal system, no
mechanical devices have to run to remove the
afterheat, and no reserves of water or coolants
other than air, flowing also by natural convection,
are required. As the afterheat load decreases with
radioactive decay, dampers in the chimney can
be closed to reduce the heat rejection rate.
64

ORNL-DWG 68-4188

- F} »
S i
S w
- 2
T
< > —u
T, g 3 S
x y A
= o
00 [&] p=4 uq._ = WH [%p)
- 1] o fia
5 i QO w
o &l > x 8 =}
o ul ov <L
O RS ._nn 2 %
3 X © Q-
i T /// > <z
ettt 1 £ TRPRANN ™ E<
T IT TI T U LTI TI3T [ -—— / >
/ s L L o8
. T T ITIII T Niseas m o nPUu
/ ’ g Ve — Qe
/ i ©
. \ >
e smu! . o
\ _ 4__fj_,_p‘$__l_\l_—fl~waam+
\ \ .II [TTT1
Ex g I
B \ 4 d__f,__fi,_+_f_fl:_,__P_fi+
YT ILITITITITTT
Vs |
. y "
w
a ’ p e ._n.
m ‘ \\ o
] e o < N o
© ’ TS TCCT 1 ”
e rryprp LT T N
p . L | FELLLTOO T | N
yd S o [ilssnka nn 1T ”
13nd -— = N\h\ltl- —— e e e g
+
’ +
Y
\ g
\ / s Lt J T

o
N
.

~

HEADERS (4) \//
AN
’ \\
; AN
N
N
=

{0-in. COOLANT PIPE

1%-in. COOLANT PIPES

A 13n4d

Fig. 5.8. Elevation of Fuel Drain Tank.
1-in. PIPE-VENT AND

PRESSURE LINE \fl

65

ORNL-DWG 68-4197

10-in. PIPE
i COOLANT HEADERS

= -

@—Q@CFRCUFT NO. 2
T

20 ft 6in.

12-in. PIPE \
SN
Il

3, ODX 0.042 WALL
U TUBES

1540 PER CIRCUIT

3080 TOTAL

i — I
— —_—
O Efl N g k_—?j = I lL8F j> CIRCUIT NO. 1
L_: - ) H _—%fi -
-@Sfi—:@:
,.o""““ “ ‘ ? CIRCUIT NO. 4
g LCIRCU!T NO. 2 i X 6
] T COOLANT
MHHIHHH T E_-;*_i—__fiii—__:__ e-mL:r:E o
—————————— .L\\’ .
HI‘H /FUEL SALT LEVEL FUEL SaLT
1 e -\
! PLENUM

| SKIRT

—e————— VESSEL

14 ft Oin.

DRAIN

Fig. 5.9. Plan View of Fuel Drain Tank.
6. Reactor Physics

A. M. Perry

6.1 MSBR PHYSICS ANALYSIS

O. L. Smith
W. R. Cobb H. T. Kerr

Physics analysis of the MSBR for the period
covered by this report was confined to studies of a
single-fluid machine. Results of the studies demon-
strated that the nuclear performance of a single-fluid
reactor could be made as attractive as that of the
40-w/cm?® two-fluid reactor discussed in previous
progress reports. Some of the results presented
here should be regarded as preliminary. In the dis-
cussion below, a reactor which includes all the most
recent design information is identified as the refer-
ence reactor. This designation is made for con-
venience of identification only. Several revisions
to this design are already being studied. For ex-
ample, it may be desirable to tailor the salt distri-
bution in the axial as well as the radial directic:.
Also, the reactor fuel cell described below is i:
tended only to show the approximate features ot an
MSBR single-fluid cell. Instead of having all of the
salt surrounding the graphite as depicted below,
final cell designs will almost certainly have part of
the salt interior to the graphite, partly to maintain
suitably low graphite temperatures and partly to
achieve the most advantageous self-shielding of
thorium resonance cross sections.

6.1.1 Reference Reactor

A schematic drawing of the most recent one-fluid
reactor core is shown in Fig. 6.1. The core is 16 ft
in diameter by 20 ft in length. Radially the core is
surrounded by 1 ft of graphite reflector. Above and
below the core are several structural regions which
support the core and transport salt., Since these

ORNL -DWG 68-5526

GRID AND PLENUM

|
!
J
|
[
|
|
|
|
!
}
|
19% i
|
i
f
l
|
f
|
l
|
|
f
|
|

—

]
|
|
F
|
1
|
|
f
|
1
CORE | CORE | _I_—~1ft GRAPHITE
|
|
|
i
|
|
|
|
|
|
|
I

CORE
17% | 44%
20 ft SALT ! SALT | SALT
[ L«— ~2 in. HASTELLOY
|
l
|
| L
' PLENUM
M
SALT |

Fig. 6.1.

SALT AND HASTELLOY

8 ft —————= |t ft

Reference Reactor Core.

regions contain various amounts of salt and graph-
ite in addition to Hastellecy N, they are also ef-

66
fective as blanket regions for the core. The reactor
is surrounded by a pressure vessel of Hastelloy

~2 in, thick. For reasons which will be described
below, the core contains 19% salt in the central
one-sixth of the volume, 17% in the next one-third,
and 44% in the outer one-half. The salt contains
67.68 mole % LiF, 20 mole % Bng, 12 mole %

ThF " and 0.32 mole % UF In addition to the salt
1nventory explicitly 1nd1cated in Fig. 6.1 (1745 ft3),
the external system (heat exchangers, piping, etc.)
contains 700 ft®. The total reactor power output is
4444 Mw (thermal) or 2000 Mw (electrical). The
average power density is 40 w/cm? of core volume.
Although the details of fuel processing are incom-
plete, it is currently believed that the salt process-
ing cycle time for fission product removal will be

~ 40 days, with a processing cycle time for protac-
tinium removal of ~5 days.

Figure 6.2 shows a plot of the total thermal flux
below 1.86 ev and the fast flux above 50 kev vs
radial position in the core midplane. Figure 6.3
shows the same information along the axis of the
core. The peak-to-average thermal flux ratio is
2.7 in the radial direction and 1.5 in the axial
direction, giving a total peak-to-average thermal
flux ratio of 4.0. The peak-to-average fast flux
(above 50 kev) ratio is 2.6 radially and 1.5 axially,
giving a total peak-to-average fast flux ratio of 3.9.

ORNL-DWG 68-5527

N

"\JHERMAL FLUX (£<1.86 ev)

(x10'%)

i, 10 e :

3

o \ }

5 8p—— \

g ~ ~«—CORE

5 ’

R F \ \\ REFL /
FAST FLUX

5 (£> 50 kev) \ \ VESSEL

L a4 L

: W \x

(@]

o

5

ut 2 l

=

120 160
RADIUS {cm)

Fig. 6.2. Fast and Thermal Flux Distributions in
Core Midplane.

67

The appearance of a radially nonuniform volume
fraction of salt in the core deserves special dis-
cussion because this feature produces a substantial
improvement in breeding performance over that of a
core with a uniform volume fraction of salt. A uni-
formly loaded single-fluid reactor can be designed
with a high breeding ratio (e. g., 1.07) simply by
making the reactor so large that leakage is negligible.
However, the large inventory required to fuel such a
reactor greatly penalizes the yield. On the other
hand, if one introduces a nonuniform loading with
relatively small salt volume fractions in the center
of the reactor (™17 to 19%) and larger volume frac-
tions (~44%) in the outer region, small leakages
may be achieved with modest size cores. With this
salt distribution the neutron energy spectrum is
harder in the outer region than in the center, the
neutron source density near the surface is reduced,
and hence leakage is reduced. At the same time
the harder spectrum and the relatively large in-
ventory of salt in the outer region enhance neutron
captures in the thorium resonances, The nonuniform
core may be characterized by saying that the central
part of the core acts like a core and that the outer
part, containing the same materials but in different
proportions, acts like a blanket. Consequently a
single fluid is effectively made to serve a dual
purpose. (It should be pointed out that all refer-

ORNL-DWG 68-5528

S LT T T
8 I l _THERMAL FLUX
o J | (£<1.86 ev)
£ 10 L. 4‘» gl — t_.. —— e -
Q
» i ) \» \
g |
£ S T k 7 '
3 | | } |
[l ! i
= 6L _L ~ .,_\ A\
> |
w “~FAST FLUX '
z 41— — (£>50kev) — } 1 -
o |
@ | ' ‘
S \ |
N
=

) ]

O 100 200 300 400 500 600 700 800

DISTANCE FROM REACTOR TOP (cm)

Fig. 6.3. Fast and Thermal Flux Distributions Along

Reactor Axis.
68

ences to the core include both regions. Specifically,
reported power densities are averages over the two
zones.) Comparison of the highest-yield one-fluid
2000 Mw (electrical) uniform core with the highest-
yield nonuniform core shows an increase in yield
from 4.0 to 5.5%, as well as an increase in breeding
ratio from 1.048 to 1.068.

The optimum salt volume fraction distribution so
far as yield is concerned has a value of 14% in the
central one-sixth of the core volume. However, for
that core the thermal flux, and hence the specific
power, is strongly peaked in the center of the core,
being ~ 80% higher than shown in Fig. 6.2. Heat
removal from such a strongly nonuniform heat source
distribution is difficult. By raising the salt fraction
in the central one-sixth of the core to 19%, the flux
in the center of the reactor can be flattened and the
peak flux reduced as shown in Fig. 6.2. The yield
is reduced at most 0.3%. The considerable simplifi-
cation of heat removal problems afforded by the
flattened source distribution seems to justify the
small penalty on yield.

Also, since the fast damage flux distribution
closely follows the thermal flux distribution, re-
ducing the peak thermal flux as much as possible
i- desirable from the standpoint of maximizing the
useful lifetime of the graphite in the center of the
core. Present calculations indicate that the graph-
ite in the maximum flux region of the reference re-
actor will have a replacement lifetime of from 2 to
3 years if the graphite lifetime is limited to a
fluence of 3.0 x 1022 avt (E > 50 kev).

It is instructive to compare the reference reactor
with a variety of other designs. An OPTIMERC
calculation determined the maximum yield attainable
in each case. The comparison, in terms of relative
changes in yield and breeding ratio, is shown in
Table 6.1. All variations are expressed relative to
the reference reactor.

Comparison of cases 1 through 5 shows that the
reference reactor has a higher yield than any uniform
core and has the optimum diameter of the nonuniform
cases, The optimum axial length as well as the
desirability of a tailored salt distribution in the
axial direction are under investigation,

Comparison of cases 1, 6, and 7 indicates that 12
mole % ThF4 is nearly the optimum percentage of
this salt component. Larger mole percentages give
marginally improved nuclear performance while
having markedly poorer thermal properties (particu-
larly a higher melting point).

Comparison of cases 1 and 8 shows some incen-
tive for choosing a 2000 Mw (electrical) design
over a 1000 Mw (electrical) design. However,
despite the improved yield of the 2000 Mw
(electrical) reactor, the performance of the 1000
Mw (electrical) reactor is good. The 1000 Mw
(electrical) reactor may eventually be selected
as the reference design, depending upon the out-
come of studies currently under way to formulate
design criteria,

Case 9 shows that the reflector is of marginal
value so far as neutron economy is concerned. The
reflector is of small value because the outer part of
the core functions as a blanket. If the reflector
should prove to be a complicating or expensive part
of the reactor it could be removed without great
penalty to the yield. However, it may be necessary
or desirable in any case to use the reflector to
shield the reactor vessel against neutron irradiation.

Anticipating the possibility that design problems
may somehow necessitate a certain volume of salt
outside the core and hence in a largely parasitic
region, case 10 is intended to show that the intro-
duction of a substantial inventory of salt outside the
core does not in fact greatly impair the performance
of the system. To a large extent this salt pays its
way by serving as blanket material and allowing a
reduction in the amount of salt in the outer part of
the core itself.

Finally, case 11 is presented to show the kind of
nuclear performance that can be achieved in a re-
actor much smaller in size and power output than
the reference reactor. In contrast with the 40-w/cm3
power density of the reference reactor, the power
density in case 11 is 25 w/cm3. In case 11 the
volume fraction of salt in the central half of the core
is ~14%, and in the outer half it is ~70%. The
strongly nonuniform volume fraction of salt in the
core of the small system minimizes neutron leakage
and affords a good breeding ratio and a modest
yield. The unflattened peak thermal flux at the
center of the small reactor is very nearly the same
as the flattened thermal flux in the central one-sixth
of the core of the reference reactor. Thus, it ap-
pears possible in a reactor the size of case 11 to
approximate rather closely the peak fluxes, flux
distributions, and breeding ratio of the reference re-
actor, with somewhat greater departures from the
core average power density and the yield of the
larger machine.

Having presented the principal features of the
one-fluid reactor, it is worth while to compare it
69

Table 6.1. Comparison of Other Configurations with the Reference Reactor®

Reactor Description

Change in Change in Yield

Breeding Ratio (%)
1. Reference reactor, 20X 16 ftb nonuniform core, 0 0
2000 Mw (electrical)
2. Similar to reference reactor but with 20-ft core - 0.001 --0.55
diameter
3. 25X 25 ft core, uniform salt volume fraction of -~ 0.007 ~-1.7
0.127, 2000 Mw (electrical)
4. 20X 20 ft core, uniform salt volume fraction of ~0.02 ~1.5
0.14, 2000 Mw (electrical)
5. 15X 15 ft core, uniform salt volume fraction of - 0.047 ~2.4
0.18, 2000 Mw (electrical)
6. Similar to reference reactor but with 10 mole % ~— 0.006 ~0.4
ThF4
7. Similar to reference reactor but with 8 mole % —~0.017 —~1.4
ThF4
8. Similar to reference reactor but with 14 X 14 ft — 0.004 - 0.8
core and power output of 1000 Mw (electrical)
9, Similar to reference reactor but without radial — 0.003 - 0.3
reflector
10. Similar to case 9 but with 4 in. of pure salt around —0.001 - 0.6
core in radial direction
11. Similar to reference reactor but with 10 X 10 ft core - 0.02 -4.0

and power output of 250 Mw (electrical)

dA11 the reactors were optimized to obtain the maximum yield,

bHeight and diameter,

with the two-fluid reactors considered in ORNL-
3996 and in previous reports, Table 6.2 gives
pertinent information for both systems. Data are
included for two versions of two-fluid breeders for
1000 Mw (electrical) power plants. One plant has a
single reactor vessel with average and maximum
power densities in the core of 80 and 160 kw/liter
respectively. The other plant contains four 250 Mw
(electrical) reactor modules, each with average and
maximum power densities of 40 and 80 kw/liter
respectively. This modular plant, of considerably
degraded breeding performance but longer life of
graphite under irradiation, was finally selected to
be the reference design for the two-fluid breeder
plant, primarily on the basis of cost and practicality
considerations relative to replacing the graphite,.

Since the two-fluid reactors are for 1000 Mw
(electrical) systems, we should compare them with
a 1000 Mw (electrical) one-fluid reactor. The
2000 Mw (electrical) one-fluid reference reactor is
included in Table 6.2 for completeness, The
numbers are not all consistent, but the reactors
should generally be compared on the basis of the
maximum power density (life of the graphite) in the
core. It should be possible to replace the graphite
in a one-fluid reactor more frequently than in a two-
fluid reactor for the same cost. On the basis of
economic considerations, then, the one-fluid reac-
tors shown in the table are more nearly comparable
with the modular version of the two-fluid reactor.

From the table it is clear that, although the two
designs are rather different, the overall nuclear per-
70

Table 6.2. Comparison of the Characteristics of Two-Fluid and Single-Fluid MSBR’s

Two-F luid MSBR Single-Fluid MSBR

250 Mw 1000 Mw 1000 Mw 2000 Mw
(e lectrical)a (electrical) (electrical) (electrical)
Core height, ft 10 12.5 13.7 20
Core diameter, ft 8 10.0 9.7 11.32
Blanket thickness, ft 1.5 1.5 2.0 2.34
Core power, Mw (thermal) 555 2222 1812 3646
Blanket power, Mw (thermal) 410 798
Average core power density, 39 80 64 64
kw/liter
Peak-to-average power ratio in 2.0 2.0 2.0 2.0
coreb
Graphite replacement life, yearsC 3.4 1.7 2.1 2.1
Specific fuel inventory, kg/Mw 1.04 0.73 1.06 0.94
(electrical)
Neutron balance, neutron captures
per neutron absorbed in fissile
material
Fissile material ( 233y + 235y) 1.000 1.000 1.000 1.000
Moderator 0.033 0.032 0.041 0.045
Carrier salt 0.075 0.067 0.052 0.053
Fission products 0.031 0.032 0.021 0.023
Leakage 0.005 0.001 0.014 0.011
Breeding ratio 1.069 1.071 1.068 1.068
Annual fuel yield, % /year 5.0 7.4 4.8 5.5
Fuel doubling time, years 14 9.4 14.4 12.6

40One-quarter module of a 1000 Mw (electrical) plant,

Assumed average ratio maintainable over life of graphite,
€Allowable dose = 3.0 X 1022 nvt > 50 kev, plant factor = 0.8.

formance of the two systems is quite similar. The
principal differences appear in the details of the
parasitic losses in the neutron balances of the two
systems. However, as shown in the table, the
relatively modest differences essentially cancel.

6.1.2 Fuel-Cycle Costs

In all the calculations described above, emphasis
was placed upon maximizing the yield of the various
systems. Details of the salt processing scheme

have not yet been fully determined, and hence it is
premature to discuss the fuel-cycle costs in any de-
tail. However, it is currently felt'that fuel-cycle
costs less than 0.5 mill/kwhr will be achievable
with the liquid-metal extraction techniques which
are being investigated for the one-fluid system.

6.1.3 Cell Calculations

A number of calculations have been performed to
determine the nuclear properties of a single-fluid
71

MSBR fuel cell. The calculations are based on the
cell shown schematically in cross section in Fig.
6.4. This cell contains a central graphite log ~4.7
in. in diameter surrounded by a region of salt oc-
cupying 20% of the total cell volume. Conclusions
based upon calculations using this cell must be re-
garded as tentative for several reasons, First, the
core of the reference reactor actually contains three
different salt volume fractions and hence three dif-
ferent cell designs. Second, an MSBR cell will
probably contain a portion of the salt in passages
through the graphite, perhaps in a central cylindrical
passage. Third, a cell ~4 in, in diameter is proba-
bly required in order to maintain sufficiently low
graphite temperatures. Despite these shortcomings,
it is felt that the characteristics of the cell
presented here will be a useful and reasonably ac-
curate first approximation to those of actual cells,
Figure 6.5 shows flux plots of the total fast flux
above 50 kev and total thermal flux below 1.86 ev
as a function of radial position in the cell. The
curves ate shown relative to the cell average fast
flux and cell average thermal flux respectively,
The detailed distributions shown for the fluxes in
Fig. 6.5 are in fact superimposed upon the gross
distributions shown in Fig. 6.2. Table 6.3 shows
the ratio of the average ftux in the salt or graphite
region to the cell average flux for each of several
energy ranges above thermal,

ORNL-DWG 68~5529

GRAPHITE

Fig. 6.4. One-Fluid MSBR Fuel Cell.

GRNL -~ DWG 68-5530

—_—
-

>
3%
>E g
228 obys SRV I S
LEV CELL AVERAGE THERMAL )
Twy FLUX (£ < 1.86 ev) ’ i
F7 oo v r 1 / N
1.2 _fi—j—]__r T l
Y I O
WX CELL AVERAGE FAST J
>3 = /FLUX (£ > 50 kev) !
258 10 k- L— A R R
|
224 o
w
- | | 1
0.9 F— ——— - = 4~ == —
| ; | T SALT\
- — ————GRAPHITE— | —5‘4 ]i
0.8 | | 1 | Il
0 ! 2 3 4 5 6 7
RADIUS {cm)

Fig. 6.5. Fast and Thermal Flux Distributions in One-
Fluid MSBR Cell.

6.1.4 Reactivity Coefficients

A number of isothermal reactivity coefficients
were calculated using the reference reactor and the
cell of Fig. 6.4. These coefficients are summarized
in Table 6.4. The Doppler coefficient is primarily
that of thorium. The graphite spectral coefficient is
positive because of the competition between thermal
captures in fuel, which decrease less rapidly than
1/v, and thermal captures in thorium, which decrease
nearly as 1/v with increasing temperature. The
total spectral component includes the Doppler term,
the graphite spectral term, and the thermal spectral
effects of all important moderators in the salt. The
resonance density component represents the effect
of decreased self-shielding of thorium with de-
creasing salt density. The total density component
includes all effects associated with changes in the
density of both salt and graphite, including the reso-
nance density term. The prompt salt component in-
cludes all reactivity effects associated with an iso-
thermal core temperature change except the graphite
spectral and graphite density components, the latter
of which is negligible. The last entry in Table 6.4
combines the spectral and density effects into a
total isothermal reactivity coefficient.

Several important conclusions can be drawn from
these coefficients. First, the total power coeffi-
cient is negative. Second, both the total density
72

Table 6.3. Flux Disadvantage Factors in Fast and Epithermal Ranges
Group Energy Range ¢g2‘aph1te /¢ cell ¢ salt /(]5 cell

1 0.821-~10 Mev 0.967 1.132

2 0.0318~0.821 Mev 0.985 1.060

3 1.234-31.82 kev 0.999 1.004

4 0.0479~1.234 kev 1.006 0.977

5 1.86—47.9 ev 1.008 0.969

Table 6.4. Isothermal Reactivity Coefficients

of the Reference Reactor

Reactivity Coefficient,

Component LQ’_‘_ (per OC)
k 0T

X107
Doppler -~3.6
Graphite spectral +1.9
Total spectral - 0.26
Resonance density ~1.2
Total density - 0.26
Prompt salt ~2.4
Total ~0.49

and the total spectral reactivity coefficients are
negative. Third, and perhaps most important, the
prompt coefficient of the salt, which largely deter-
mines the fast transient response of the system, is a
relatively large negative coefficient and should af-
ford adequate reactor stability and controllability.
Studies of the actual dynamic performance of the
system are in progress.

6.1.5 Measurements of Eta for 233U and 235U
in the MSRE

G. L. Ragan

Isotopic analysis of two MSRE fuel samples
differing in depletion of the 235U or 233U fuel can
be used to measure the value of 5 for the principal

fuel isotope. Samples have been taken for this
purpose during operation with the 235U loading, and
others will be taken during operation with 233U, A
precise measurement for the latter would be of in-
terest to the MSBR program since, as discussed by
Perry,! the uncertainty in the effective value of 5
for 233U contributes the major uncertainty in the
breeding ratio of molten-salt breeders. The meas-
ured 7 (233U) will be quite comparable with that for
proposed MSBR designs because of the similarity,
as demonstrated by Prince,? of the spectra of MSBR
and MSRE. For both 23°U and 233U loadings of
MSRE, comparisons between measured and calculated
values of 7 provide tests of calculational methods
and cross-section data. The first experiment, with
235y, will check procedures and equipment; the re-
sult will test the validity of the method, and may,
indeed, constitute a useful check on the cross sec-
tions of 235U,

The basic idea of the measurement, discussed in
terms of 233U for concreteness, is as follows:
Absorptions in 233U are measured by observing the
decrease in 233U concentration in the salt over a
period of time at power; captures in 233U are meas-
ured by observing the increase in 23*U concentra-
tion (corrected for absorptions in 234U itself).
These concentrations, and the changes in them, are
measured relative to that of 238U, whose neutron
absorption cross section, and hence concentration
change, are far smaller than those of either 233U

1A. M. Perry, “Influence of Neutron Data in the Design
of Other Types of Power Reactors,? to be published in
The Proceedings of the Second Conference on Neutron
Cross Sections and Technology, Washington, D, C.,
March 47, 1968.

’B. E. Prince, MSR Program Semiann, Progr, Rept,
Aug. 31, 1967, ORNL.-4191, p. 50.
or 234U, These relative concentrations can be
measured very precisely, yielding a precise value
for the conveniently defined cross-section ratio

Using a well-established? value for v, one gets

Of
n=v—=v(l-y).
oa
An expression has been developed for the cross-

section ratio y in terms of the quantities actually
measured by the mass spectrometer and the ef-
fective absorption cross sections of 234U and
238, For the mass spectrometer measurements,
one includes: (1) the measured current ratios
(e. g., that in the 233U channel to that in the 233U
channel); (2) the instrument calibration factors
(e. g., the relative current sensitivities of channels
233 and 238); (3) interchannel cross-coupling fac-
tors (e. g., the spurious current in channel 234 due
to the breadth of the 233 peak); and (4) variation
in these instrument factors between measurements
on initial and final samples. The ratios of the ef-
fective cross sections of 234U and 238U to that of
233y are involved because of the small but signifi-

3See, for example, J. R, Stehn et al., Neutron Cross

73

Sections, BNL-325, second ed., Suppl, 2 (February 1965).

cant corrections for depletion of 234U and 238y
during the course of the experiment. An extensive
error analysis? has been performed for both the
233y and 235U cases. The uncertainties which
appear to limit the attainable precision are those of:
(1) ratio of current in another channel to that in
channel 238 (assumed uncertainty +0.01%); (2) ef-
fective absorption cross sections of 234y, 236U,
and 2380 relative to those of the principal fuel iso-
topes 233U and ?3°U (assumed uncertainty +10%).
On this basis, it appears that an experiment in-
volving 2% depletion of 33U should measure y to
1+3.3%, corresponding to an uncertainty of +0.008

in the effective value of 5 (?33U) in the MSRE
spectrum. The corresponding errors for 2% depletion
of 235U are 11.4% in y and $0.007 in (335U). To
these uncertainties owing to the measurements
themselves, one should add an allowance for the
uncertainty in . Using presently accepted values?
for these [ +0.32% for v(?33U) and +0.25% for
v(235U) ], one estimates an overall uncertainty of
+0.011 (fortuitously coinciding) for both 5 (2" 3U)
and 5 (233U), as measured in the MSRE spectcrum.
This value may be compared with our present esti-
mate! of £0.015 for the uncertainty in (233U), in a
typical MSBR spectrum, attributable to uncertainties
in neutron cross sections,

4A. M. Perry and G. L. Ragan, Measurement of T](233U)
in the MSRE (in preparation).
7. Systems and Components Development

Dunlap Scott

The programs concerned with migration of noble
gases, the technology of sodium fluoroborate, and
the early development of pumps for molten-salt
reactors were continued during the period. A
study was begun of remote maintenance for MSBR’s,
including the problems involved in the remote weld-
ing of pipes and vessel closures. Studies of the
effectiveness of decontamination of small com-
ponents for subsequent maintenance and of the
use of a gamma-ray camera for locating radioactive
hot spots, both of which are pertinent to the
maintenance plan for an MSBR, are reported in
the MSRE section of this report.

7.1 NOBLE GAS BEHAVIOR IN AN MSBR
R. J. Kedl

The results of '35Xe poisoning calculations
were presented for a specified two-fluid reactor
concept in the last semiannual progress report
(ORNL-4191). They showed that the target
poison fraction of 0.5% was attainable. Since
then, the proposed core has increased in size in
order to lower the power density. In addition, the
proposed fuel cell geometry has changed to signif-
icantly increase the graphite surface area per
unit volume of core. These factors will increase
the 135Xe poisoning, and it now appears that the
attainable poison fraction will be between 1 and
2%. As a result, consideration is being given to
graphite with a very low-permeability coating.

Calculations were made to determine the effect
of coating the graphite with a thin layer of low-
permeability graphite. As a reference design,
essentially the same reactor concept was used as
described in ORNL-4191, but with the core size
increased to yield a mean power density of 20

74

kw/liter. The fuel cell geometry was also
changed somewhat to give a higher specific sur-
face area. The various xenon migration parame-
ters were chosen to obtain an !35Xe poison frac-
tion of 2.25% (high end of the attainable range)
with no coating, and then the calculations were
extended to include the effects of coatings. The
results of these computations appear in Fig. 7.1.
In the coating the void fraction available to xenon
was made variable in such a way that it changed

ORNL- DWG 68— 5531

SIGNIFICANT PARAMETERS:
REACTOR POWER - 556 Mw(THERMAL)
CORE POWER DENSITY =20 kw/liter
BULK GRAPHITE DIFFUSION COEFFICIENT -

1073 $12/hr
BULK GRAPHITE AVAILABLE VOID—10%
CIRCULATING BUBBLE SURFACE AREA ~ 3000 ft2

COMPOSED OF "ONCE THRU" BUBBLES AND
NO "RECIRCULATING" BUBBLES

MASS TRANSFER COEFFICIENT TO BUBBLES—
2.0 ft/hr

FUEL CELL GEOMETRY —CONCENTRIC ANNULUS

XENON DIFFUSION COEFFICIENT IN COATING
(#12/hr)

O AVAILABLE VOID IN COATING (%)

3
L
=
o 00
— - -6
o2 10 3
<1 \
g \
g \
2 F\ -7
o) 10 1
£ | h—
>(-l<, “ \
n
52 \ \ .
R R 078 03
10~ O
0
0 5 [0} 15 20

COATING THICKNESS (mils)

Fig. 7.1. Effect of Coated Graphite on 135% e Poison

Fraction.
by one order of magnitude when the diffusion co-
efficient changed by two orders of magnitude.
The diffusion coefficient of 10— 3 ft2/hr for the
bulk graphite represents graphite believed to be
readily available. It is important to notice that
the bulk graphite must be improved to a diffusion
coefficient of less than 10~7 ft?/hr before there
is a significant effect on the poison fraction,
whereas an 8-mil coating of 10~ 8-ft2/hr material
would reduce the poison fraction to the target
value of 0.5%.

7.2 SODIUM FLUOROBORATE CIRCULATING
LOOP TEST

A. N. Smith P. G. Smith
Harry Young

The alterations necessary to convert this test
facility, formerly called the Fuel Pump High-
Temperature Endurance Test Facility,’ to operate
with sodium fluoroborate were completed. The
work included installation of (1) equipment for
automatic control of BF, flow through a dip leg
in the pump bowl and helium flow to the pump
shaft annulus, (2) an access nozzle for removing
salt samples from the pump tank, (3) a thermal
conductivity cell to monitor the composition of
the off-gas stream, and (4) equipment for con-
trolled ventilation of the loop enclosure and in-
strument cubicles with provisions for venting the
discharge above the building roof.

Instruments and components were checked out
for proper operation, and the system was leak
tested. A flush charge of about 900 1b of sodium
fluoroborate was added to the sump, and this
salt is now being circulated. The flush salt,
which is intended to remove residues of the
lithium fluoride—beryllium fluoride salt previously
circulated in the loop, will be drained after sev-
eral days of circulation, and a second batch will
be added. The salt will then be circulated for
several months for investigation of

1. pumping characteristics of the salt, including
minimum overpressure necessaty to prevent
cavitation,

2. problems associated with control and moni-
toring of the salt composition.

IMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 95.

75

Salt samples will be used for direct indication
of salt composition. Devices that will be evalu-
ated for monitoring the composition indirectly are
a themmal conductivity cell, which will provide an
indication of the BF , partial pressure in the off-
gas stream, and a cold-finger unit, which will
check for changes in the liquidus temperature of
the salt. The cold-finger system is currently
under design and will not be available during
the initial period of circulation. Tentative cali-
bration data on the thermal conductivity cell
indicate that a change in BF | partial pressure of
3.5 mm Hg will result in a change in signal of
2.5% of the indicator scale. This sensitivity is
enough to detect any significant changes in the
salt composition.

7.3 MSBR PUMPS

A. G. Grindell L. V. Wilson
P. G. Smith C. E. Bettis
H. C. Young C. K. McGlothlan

7.3.1 Pump Program

The present emphasis on studying the single-
fluid MSBR has influenced the pump program.
The number of hydraulic designs has been re-
duced from 3 required for the two-fluid concent
to 2 (see Table 7.1). Preliminary study in.:icates
the practicability of using a sump pump configura-
tion similar to that of the salt pumps for the

Table 7.1.
2000 Mw (Electrical) MSBR

Pump Requirements for

Fuel Coolai.:
Number required 4 4
Design temperature, °F 1300 1300
Capacity, gpm 24,000 53,000
Head, ft 200 200
Speed, rpm 890 890
Specific speed, N 3170 3830
Net positive suction head 25 50

required, ft

Impeller input power, hp 7500 7000

Molten-Salt Reactor Experiment (MSRE) and of
using pumps of essentially one mechanical design
and one drive design.

The emphasis on the single-fluid concept has
not, however, altered the pump program objec-
tives? and the desire for the strong participation
of the United States pump industry in design,
development, production, and operation of pumps
for the Molten-Salt Breeder Experiment (MSBE).
Our present plan for securing industrial participa-
tion in the pump program, in brief, consists of
the following items:

1. Interested pump manufacturers will be re-
quested to submit proposals for designing the
MSBE fuel salt pump to our specifications.

2. After evaluation of these proposals, selected
pump manufacturer(s) will be asked to design the
fuel salt pump and to estimate the cost and
schedule to produce a prototype. The cost esti-
mate will include whatever development program
the pump manufacturer proposes and his partici-
pation in the ORNL pump program. The pump
manufacturer will also be requested to provide
cost and schedule estimates for producing the
required salt pumps for the Engineering Test
Unit (ETU) and the MSBE.

3. After evaluation of the completed designs,
selected pump manufacturer(s) will be requested
to produce the prototype fuel salt pump, complete
his development program, and participate in the
ORNL pump program.

4. The prototype fuel salt pump will be sub-
jected to molten-salt testing in the Fuel Salt
Pump Test Facility with the participation of the
pump manufacturer(s).

5. The salt pumps for the ETU and the MSBE
will be produced by the pump manufacturer(s).

6. With the participation of the pump manu-
facturer(s), proof tests of the ETU and MSBE salt
pumps will be performed in the Fuel Salt Pump
Test Facility.

7. Adequate support, in which the pump manu-
facturer(s) will participate, will be provided to
sustain the continued satisfactory operation of
the salt pumps in the ETU and the MSBE.

We will make design studies and analyses of
the salt pump configuration and its relationships
to the reactor system to become thoroughly

2MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 96.

76

familiar with the pump problems. The studies will
be used to provide insight and direction to the
pump design and development for the industrial
participant(s) as well as the Laboratory.
Information is now being assembled for use in
writing specifications for the fuel salt pump.

7.3.2 Fuel Salt Pump

In our present design of the single-fluid breeder,
the pumps are near the top of the reactor cell.
The length of the shaft can be reduced consider-
ably from that of the long-shaft pump? proposed
for our design of two-fluid reactors; this shorten-
ing presents an opportunity to eliminate the
molten-salt bearing required in the long-shaft
pump.

Figure 7.2 shows a preliminary layout of a con-
cept of the fuel salt pump for the single-fluid
reactor. Although the capacity is much larger,
the pump is similar in configuration to the MSRE
salt pumps. The drive motor is mounted in a
fixed position on top of the concrete shielding,
and the bearing housing is recessed into the
shielding to reduce the shaft overhang. The pump
shaft is mounted on two pairs of preloaded oil-
lubricated ball bearings. The impeller is over-
hung about 61/2 ft below the lower bearing. A pre-
liminary study indicated a first shaft critical
speed greater than 1200 rpm.

Internal shielding of the bearing housing is pro-
vided by a steel or Hastelloy N shield plug cooled
by an organic liquid, possibly the same as the
bearing lubricant. To reduce deflections of the
shaft at the impeller caused by hydraulic radial
forces on the impeller, the pump casing is of
either the double-volute (as shown) or the vaned
diffuser design to minimize these forces. Both
casings have the added advantage of greater
structural strength than a single-volute or a
vaneless diffuser. At the design temperature of
1300°F, where the allowable stress for Hastelloy
N is 3500 psi, this advantage is of considerable
importance to the designer.

The bearing housing is designed for replace-
ment by semidirect maintenance, either in situ or
in a hot cell after the entire rotary element has
been removed from the reactor system.

3MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 98.
77

ORNL-DWG 68-5532

MOTOR

|_——-———BELLOWS

L,

COUPLING

BEARING AND
SEAL ASSEMBLY

14 ft 2in.

CONCRETE
SHIELDING

PUMP TANK
8ft OD

| NORMAL OPERATING
_ LEVEL

TTT——————— IMPELLER
2 ft 6in, diam

Fig. 7.2. MSBR Fuel Salt Pump Concept. Preliminary layout of short-shaft pump.
The drive motor will be installed in a fixed
position, and the pump will be permitted to move
freely with the thermal expansion of the reactor
system. For power transmission a floating
coupling or universal joints will be used to ac-
commodate the approximately 1/8-in. horizontal
and vertical displacements of the pump that are
anticipated during operation at temperature. The
transmission device will also have design features
to accommodate the approximately 5-in. vertical
and 11/2-in. horizontal movements that occur be-
tween room temperature and reactor cell design
temperature; we do not presently anticipate
operating the pump at room temperature.

The pump tank provides volume to accommodate
the anticipated thermal expansion of the fuel salt
at off-design conditions. It is almost completely
decoupled hydraulically from the flowing salt in
the impeller and volute passages by (1) labyrinth
seals installed in the pump casing around the
pump shaft and on the periphery of the casing
and (2) bridge tubes that connect the volute to the
inlet and outlet nozzles attached to the pump
tank. The bridge tubes also eliminate structural
redundancies between the pump tank and the
volute and its supporting structure.

In a noteworthy fashion the hydraulic decoupling
serves to minimize the changes that may occur in
the pump tank liquid level when several pumps are
being operated in parallel and one of them stops.
Consider, if (1) the gas volumes of the salt pumps
being operated in parallel are interconnected,

(2) the salt volume in each pump tank is connected
directly to its pump suction, and (3) all pumps are
being supplied from a common plenum in the re-
actor, then the following events would occur when
one pump stops: The level of salt in the tank of
the stopped pump would try to increase by an
amount equal to the velocity head at the pump
suction plus the head loss in the suction line
from the common supply to the pump tank. This
change in level would be ten or more feet, which
would represent an unwanted increase in pump
shaft length. Also, unless there is sufficient
reserve salt volume in the other pump tanks to
supply the increased salt requirement of the
stopped pump, the system fluid would in-gas when
the salt level in the tanks of the operating pumps
is lowered to the level of the pump suction. How-
ever, by connecting the liquid in the hydraulically
decoupled version of the pump tank to a point in
the reactor plenum where the velocity changes

78

very little when one pump is stopped, and by
making the pressure drop in this connecting line
very low for the salt flow returning from the tank
to the plenum, the level change in the pump tanks
probably can be held to about 2 ft.

Although not shown in Fig. 7.2, the upper por-
tion of the pump tank and its internal structure
will be cooled by a small bypass flow of the fuel
salt to remove beta and gamma heat. A study
using an existing heat transfer program (SIFT-
TOSS) is in progress to determine the temperature
distribution in the pump shaft and shaft casings.
The first case being solved assumes full-power
gamma heating on one half of the shaft casing and
no gamma heating on the other half., The results
of the first case will indicate the next case to be
considered.

A preliminary investigation has been made of
the capacity of the fuel salt pump and its speed
on various pump parameters such as impeller
diameter, net positive suction head required
(NPSH), hydraulic efficiency, and shaft diameter.
The results are shown in Fig. 7.3.

7.3.3 Coolant Salt Pump

There is sufficient similarity in the design head,
speed, temperature, and power characteristics of
the fuel and coolant salt pumps (see Table 7.1)
that it appears feasible to interchange the rotary
elements (except for the impellers) and the drive
motors. Interchangeability could lead to con-
siderable savings in the MSBE salt pump program.

7.3.4 Molten-Salt Pump Test Facility

The design was initiated for a molten-salt pump
test facility which will provide adequate non-
nuclear testing capabilities for the MSBE salt
pumps. Basically, the facility will consist of
the necessary piping system, heat sink (preferably
to air), variable flow restrictor, flow measuring
device, salt storage tanks, preheaters, and con-
trols and instrumentation. The following is a
partial list of the criteria for the facility:

Maximum temperature 1400°F
Maximum cover gas pressure 150 psig
Maximum flow 10,000 gpm
Maximum head loss 200 ft

Maximum heat sink capability “~100,000 Btu/min
79

ORNL —DWG 685533
— 338
— — — 205 | PUMP GAPAGITY

—ee—-1{6.9 {1000 gpm)
100
%0 b TresSTmm LYDRAULIC EFFIGIENGY (%)
f. =
50
\
\ il
40 \“\
N
N ) . NET POSITIVE SUCTION
RN P HEAD REQUIRED (ft)
™ e
30 \‘:fi < J
~
/ TSRS
1R} -
P 7\ J IMPELLER DIAMETER (in)
Ve o~
00 4 7(
7~ -
P <
- e
~
e
Al e Jon
T = (T[T T SHAFT DIAMETER (in)
REACTOR POWER - 2000 Mw{ELECTRICAL)
TOTAL SALT FLOW —435,000 gpm
PUMP HEAD —200 ft
0
600 800 1000 1200

PUMP SPEED {rpm)

Fig. 7.3. Effect of Capacity and Speed of MSBR Fuel 5alt Pump on Yarious Pump Parameters.

The facility including test pump will be ap- that of the MSRE salt pumps, we plan to pursue
proximately 60 ft high, if the long-shaft pump is the molten-salt bearing program vigorously, There
used, and approximately 50 ft long. It is proposed is need to establish whether such bearings can be
to locate the facility in Building 9201-3. used in a pump. In addition, the bearing materials

Preliminary design studies have been directed may be applicable to the plugs and seats of valves
to the sizing of a salt-to-air heat exchanger and for molten salt.
investigating flow characteristics of the salt Bearing Materials. — The Metals and Ceramics

piping loop.

Division is preparing a program in which candi-
date bearing materials will be subjected to fuel

7.3.5 Molten-Salt Bearing Program and coolant salt tests. Hard-surface coatings on

Hastelloy N substrate will be studied. The

Although the single-fluid reactor concept may various methods that will be tried for applying
permit the use of a short-shaft pump similar to the coating to the substrate include plasma spray,
multiple coating by chemical vapor deposition,
and brazing,

Sixty-four hard-coated specimens® consisting of
plasma-sprayed cermet on Hastelloy N substrate
were received from Mechanical Technology In-
corporated. Four different coatings are repre-
sented: (1) tungsten carbide bonded with 7 to
10% cobalt, (2) 25% tungsten carbide—7% nickel—
68% mixed W-Cr carbides, (3) 40% pure tungsten
carbide—50% tungsten carbide with 12% cobalt
binder—10% molybdenum, and (4) 75% chromium
carbide—25% Ni-Cr alloy binder.

These coatings will be evaluated by the Metals
and Ceramics Division on the results of (1) chemi-
cal compatibility tests with molten salt, (2) ther-
mal cycling tests in molten salt, (3) x-ray studies
of coating structure made before and after molten-
salt tests, (4) surface roughness measurements
taken before and after molten-salt tests, (5) de-
termination of coating adherence to Hastelloy N
substrate, and (6) metallographic studies. Coat-
ings which have promise will be evaluated further
in actual molten-salt bearing operation.

Bearing Tests. — The materials that have suf-
ficient promise will be fabricated into bearings
and tested in molten salt. The first molten-salt
tests will be conducted in an existing facility
which can accommodate a 2-in.-diam by 2-in.-long
bearing.* The facility was last operated with
potassium, and its refurbishment to molten salt
operation is being studied.

The layout of a facility for testing relatively
large molten-salt bearings® is complete. The
facility utilizes such existing items from the
MSRE pump program as the Mark 1 prototype pump
tank and rotary element, the structure from the
water pump test stand, and a high-temperature
molten-salt drain tank. The tester will accommo-
date a 6-in.-diam by 6-in.-long bearing and will
be operable at speeds to 1800 rpm, temperatures
to 1400°F, and bearing loads to 500 1b,

The tester will provide basic information and
help resolve some of the problems of design, de-
velopment, fabrication, and operation of molten-
salt bearings. The major of these problems is
proof testing the selected bearing materials and
their mounting arrangement should they have a

4MSR Program Semiann. Progr, Rept. Aug. 31, 1967,
ORNL~-4191, p. 100.

80

coefficient of thermal expansion significantly
different from that of Hastelloy N,

A cross section of the tester is shown in Fig.
7.4. It consists of the aforementioned existing
pump modified by removing the impeller, volute,
and other structures inside the pump tank to make
room for the test journal, bearing and mount, and
bearing loading device. The journal is attached
to the end of the shaft by means of a conical
mount that accommodates relative thermal ex-
pansion between these two pieces by sliding
along the conical sutfaces and at the same time
maintains concentricity between the outside of
the journal and the shaft center line.

The bearing consists of four tilting pads
mounted on a ring made of the same material as
the pads. The ring, in turn, is mounted on a
structure of leaf springs having radial flexibility
sufficient to permit relative thermal expansion
between the ring and structure without overstres-
sing the members. The radial stiffness, however,
is sufficiently high to transmit the bearing load
easily. The bearing is loaded by means of a
hydraulic cylinder acting on the lever that pene-
trates the tank through a bellows and bears
against the leaf spring structure., The entire
bearing assembly is supported on slender rods that
permit the assembly to move with almost zero
redundancy upon application of the bearing load.
Lugs are provided to resist the bearing torque.
The tester will be driven by a low-torque motor
to reduce damage in the event of bearing seizure.
We expect the bearing to operate predictably in
molten salt, so no attempt will be made to meas-
ure such functional characteristics as operating
clearance and attitude angle.

The tilting pad bearing configuration was se-
lected for the tester because of its ability to
better accommodate those conditions that cause
bearing damage and lead subsequently to failure.
Damage is most likely to result from (1) the
presence of crud or particulate matter in the lubri-
cating salt, (2) thermal distortion of the bearing
surfaces or loss of the alignment between the
journal and bearing, or (3) mechanical disturbance
of the bearing surfaces caused by starting and
stopping the journal., To a greater degree than
most bearing configurations, the tilting pad is
capable of self-cleaning should any particulate
matter pass into the bearing clearance. It can
also tolerate a relatively high degree of misalign-
ment.
ORNL-DWG 68-5534

MOTOR\\\\\\\\;

[

COUPLING
\

=i

BEARING HOUSING

HYDRAULIC CYLINDER

LOAD CELL
LOADING ARM\
3ft 7in.
PIVOT
TANK
(3 ft Oin, OD) - -
.

TILTING PAD
BEARING - - -
(FLEXING MOUNT)

JOURNAL
(CONICAL MOUNT) S

‘ TO DRAIN TANK

Fig. 7.4. Large-Scale Molten-Salt Bearing Tester.
7.3.6 Rotor-Dynamics Feasibility Investigation

An analysis of the rotor-dynamic response* of
a long-shaft pump has been completed by Mechan-
ical Technology Incorporated. The final report is
expected soon.

7.4 REMOTE MAINTENANCE

Robert Blumberg P. P. Holz

A fundamental criterion in the design of a
molten-salt breeder reactor is that maintenance
will be done in a safe and economical manner.
Attainment of this objective requires close sur-
veillance of the reactor design during the con-
ceptual stage so that considerations of the basic
problems of component maintenance and replace-
ment are included.

We are now establishing the broad outlines of
the maintenance system design and development
program. Initially, our approach will be based on
the maintenance system used at the MSRE, which
consists in disconnecting and replacing com-
ponents by manipulation of expendable long-
handled tools through a movable maintenance
shield. The experience with this maintenance
system gained at the MSRE provides a basis for
the anticipation and evaluation of problems in the
much larger MSBR.

Several factors which influenced the evolution
of the MSRE maintenance system and also apply
to the MSBR are

1. the nature of the tasks it had to perform (un-
scheduled repair and replacement of a large
number of different components),

2. the physical character of the reactor plant
(basically a chemical processing plant sepa-
rated into several rooms, roughly according to
function),

3. the economics of other approaches to the
maintenance problem, generally involving the
use of electrically controlled equipment oper-
ated completely remotely and limited to ex-
actly prescribed functions.

However, in transposing to the MSBR we recog-
nize several differences that cannot be dismissed
by simply scaling up from the MSRE. In essence,
the MSBR maintenance system development pro-

82

gram will be concerned with applying MSRE tech-
niques to a larger reactor and to overcoming
problems caused by the differences.

There are three specific areas under active con-
sideration: (1) the application of the portable
maintenance shield concept to the various cells
of the MSBR, (2) remote welding for vessel entry
and component replacement, and (3) replacement
of the graphite moderator elements of the core on
a regular basis. Factors involved in each of these
areas will influence the reactor equipment layout
and component design, so that an early resolution
is thought necessary.

MSBR Maintenance Shield

The application of the portable maintenance
shield concept to the MSBR is important in that
this basic piece of equipment will be used for
almost all remote maintenance operations. A
practical, workable design must be achieved be-
fore we can confidently look at in-cell problems.
The design must take into account the shield-
block layout, the method of gaining access through
the containment, the layout of components in the
cell, and any physical obstructions on the working
floor level. Layouts of various shield block and
containment membrane arrangements have been
started to determine the effect on remote mainte-
nance procedures. Also of interest in this general
area are experiments to be conducted to measure
the residual radiation levels in the reactor cell
of the MSRE. This information will be used to
estimate levels that will be encountered in the
MSBR during maintenance operations and to de-
duce whether they will impose limitations on
manipulations through the maintenance shield.

Remote Welding

We propose to use remote cutting and welding as
the means for disconnecting the salt piping for
component replacement and for gaining access in-
to the reactor vessel for graphite moderator re-
placement. The early development of this capa-
bility is important in establishing the maintenance
concept as well as in establishing the space re-
quirements in the plant layout. Figure 7.5 shows
schematically the equipment involved in the process
for welding the horizontal piping.
3\%

83

SE()
e

ORNL-DWG 68-5535

/4

/ROOF SHIELD BLOCKS

5\

2\7 TYPICAL CUTTING AND
WELDING SITE~__
C
REACTOR
VESSEL 4
T \ HEAT
EXCHANGER K
d, 8 dy
s o * ‘f
13

Fig. 7.5. Schematic Yersion of Remote Welding Components.

Item 1 is a vehicle which traverses the circum-
ference of the pipe and has the capability of cut-
ting and preparing the pipe for welding, welding
the pipe, and then inspecting the joint.

Item 2 consists of the long-handled tools which
insert and support item 1, provide some minor
control or adjustment over the various processes,
and provide routing for the service lines.

Item 3 is the maintenance shield for the MSBR.

Item 4 is the equipment which has the primary
control of the welding, cutting, and inspection
processes.

Item 5 is whatever is used to support the com-
ponents and position the ends of the piping for
welding. The rollers shown are schematic. The

design of component supports must include con-
sideration of the weld joint pre-positioning re-
quirements,

The development effort will be primarily con-
cemed with an assembly consisting of items 1 and
4 and the connecting cables. Activity thus far
has been limited to generating broad outlines of
the overall program and to a search for informa-
tion on the state of the current technology. Sig-
nificant and applicable work has been found in
the area of automatic butt welding of pipes in a
horizontal position.

Two firms build welding equipment that is
capable of completing good welds automatically
in horizontal piping, but these require manual
installation and direct monitoring. These two
firms are Liquid Carbonic Corporation of Chicago,
Illinois, a subsidiary of General Dynamics Corpo-
ration, and North American Aviation, Inc. Both
systems use tungsten inert-gas welding and a
control system that operates on the usual welding
parameters to achieve automatic welding. There
are some differences in the detailed electrical
and mechanical design, in their capabilities, and
possibly in their state of perfection. While this
equipment represents an advanced state of the art
for automated welding, no existing system fully
meets our specific requirement for remote welding,

We plan to evaluate the use of such equipment
in the production of the high-quality welds re-
quired in a nuclear system and then to proceed
with solving the problems of weld joint design,
remote alignment and positioning of the compo-
nents, and remote inspection techniques which
are necessary for providing confidence in the
weld quality.

Replacement of Graphite Moderator Elements

Because of radiation-induced damage to graphite,
some of the moderator elements of the reference

84

design may have to be replaced every few years.
The planning and preparation for this task are
important parts of the remote maintenance pro-
gram and involve four problem areas. These are

1. a miscellaneous area concemed with the
general setup of the maintenance equipment,
with the provisions for aftetheat removal, and
with the method for disconnecting auxiliaries,

2. the process for remote cutting and rewelding
of the vessel closure,

3. the handling and storage of the very large
vessel lid,

4. the handling of new and of radioactive moder-
ator elements.

Because of the large amount of radiation and
contamination, item 4 presents the greatest prob-
lem. A number of methods are being evaluated
that range from use of a huge charge-discharge
machine to a simple long-handled grapple ap-
proach. In handling the contaminated graphite,
we have the option of breaking the moderator
bars into smaller pieces and loading them into
a transport cask below the reactor shield, or of
bringing them up into a container above the shield
and disposing of them from there.
8. MSBR Instrumentation and Controls

L. C. Oakes

8.1 ANALOG COMPUTER STUDIES

O. W. Burke
F. H. Clark

S. J. Ditto
R. L. Moore

Analog computer studies of the dynamics of
MSBR systems were begun. These studies are
expected to form the basis for the determination
of a suitable control scheme as well as the safety
requirements of the plant. The basic approach is
to first simulate in as great detail as possible the
various individual subsystems to determine the
behavior of these subsystems during postulated
transient conditions. These studies will then
form the basis for judging the validity of simpler
models so that the entire plant can be simulated.

An approximate space invariant model of the
reactor kinetics yielded the following equations
for precursors, C,, and power, P:

d —_
— P(¢) = (il P(y+ YA, CLD,
dt
e Bipey -, € e
= Ci0 =1 P() -4 i(t)_;c‘ (0
e“/\iTL
+C(t-T1) -

The term C (¢ — 7,) dictates the need for a trans-
port lag generator for each delay group if they are
to be faithfully simulated. Since only two transport
lag generators are available, it is necessary to use
approximate methods of computing the transient
delayed neutron precursor concentration using
equations involving only two transport lags. Two
such methods were compared: (1) the six groups
were reduced to two effective groups by use of the

85

method of Cohen and Skinner,? and (2) the lag

term of the two groups with the shortest time
constants was neglected; for the two groups having
the longest time constants, the lag terms were ap-
proximated by

dC;
Ct=1)=C( -1 5 ;

and the two other delay groups were treated with
the transport lag devices.

Cases were run with both models, and the results
showed comparable and acceptable performance.

Preliminary studies of the dynamics of two-fluid
MSBR dynamics were made to determine the initial
reactor response to certain load and reactivity
transients in the absence of any control action.
The durations of some of the transients were short
compared with the transit time of the salt in the
fuel circulation loop and therefore yielded no
significant information regarding overall stability.

Figure 8.1 shows the results of the simulation of
the insertion of approximately 0.75% 6k/k as a step
while the reactor was operating at design point
power level of 556 Mw [thermal power of a single
module of the four-module 1000 Mw (electrical)
two-fluid breeder plant]. Although the power
reached a peak value of about 775 Mw in 0.375 sec,
the fuel temperature at the core outlet only in-
creased by about 30°F. After the initial peak the
power settled to about 625 Mw, at which time the
run was terminated, since further changes would
be at a rather modest rate which we considered
quite amenable to routine control measures.

IE. R. Cohen and R. E. Skinner, ‘‘Reduced Delayed
Neutron Group Representations,’® Nucl. Sci. Eng. 5,
291-98 (1959).
86

ORNL-DWG 68-5536

806 1400
Qi FUEL TEMPERATURE AT CORE QUTLET —

w

- o
s / \ .
2 / \ cr
P

@
<
Y POWER | z
&}
o
& \\ l =
| B
0.5
j sec
556 l 1 900
Fig. 8.1. Response to Step Reactivity Addition of 0.746% &k/k.
ORNL-DWG ©8-5537
10

° \] S N S B N
- FUEL TEMPERATURE AT CORE OUTLET ©
- o
9t . >
S \ | 2
I o <
uw ° POWER s€c W
= o
o¥ | =
a — FLOW W

0 L l 900

Fig. 8.2. Response to Flow Coastdown.

Figure 8.2 shows the results of a run simulating
the coastdown of fuel salt flow without changing
coolant salt flow rate, again with initial conditions
at design point. The coastdown assumed was an
exponential decay from 100 to 10% flow rate with a
time constant of about 6 sec. In this case the
power decreased to approximately 12% in about
50 sec and then began to rise slowly as the effects
of the excess cooling by the heat exchanger began
to reduce the mean fuel temperature in the core. At
this time the temperature of the fuel leaving the
core was about 1340°F and was rising at a rate of
less than 1°F /sec.

The curves of Fig. 8.3 show the response of the
reactor and primary loop to a simulated loss of
load. In this case the temperature of the coolant
salt entering the primary heat exchanger was in-
creased from its design value of 850°F to 1100°F

at a rate of 25°F/sec. This appears to the primary
heat exchanger as a complete loss of load at the
rate of 10%/sec, since the coolant salt leaving the
heat exchanger at the beginning of the transient
has not had time to retum to the heat exchanger
before termination of the study. Thermal lags in
the heat exchanger and core caused the power to
lag behind the load, but the response appears
quite good. These results lead us to the tentative
conclusion that the plant would be inherently load
following at the expense of modest temperature
changes. They also indicate that it should be
quite easy to accommodate rather large load
changes using a control system to maintain some
desired temperature condition.

Although it had originally been planned to ex-
tend these studies to include investigation of the
details of the required control characteristics, the
87

ORNL—- DWG 68-5538

POWER (Mw)

556 , , T 1400
e [T T T
| ‘ ™
Lfi.fik FUEL TEMPERATURE AT CORE OUTLET <
| L .
2 5
sec K
14
START o
FUEL TEMPERATURE AT =
HEAT EXCHANGER OUTLET =
0 ' R 900

Fig. 8.3. Response to Load Reduction of 10%/sec.

shift of emphasis from the two-fluid to the one-
fluid system has caused us to set aside studies
of the two-fluid reactor for the present. Our next
studies will be aimed at a reference design of a
single-fluid MSBR.

Simulation of the entire plant will require addi-
tional analog computer equipment to simulate
transport lags in the piping of the secondary salt
loop. An order has been placed for two lag genera-
tors, with delivery expected about the middle of
the 1968 calendar year.

The steam generator has been modeled for ex-
amination of its dynamic behavior on the analog.
Considerable simplification of the system equa-
tions has been undertaken. Further, an attempt
will be made to cause dynamic variations which
are model dependent rather than real to occur in
times short compared with times of physical in-
terest. We will do this by (1) forbidding unrealisti-
cally rapid adjustment of control devices and (2)
possibly rescaling time. Such procedures are often
used to adapt the computer to a problem which is
actually beyond its nominal capability. We cannot,

of course, be assured that the analog computer will
perform satisfactorily in this mode. Accordingly,
work is continuing toward programming the problem
for hybrid computer solution.

It should be observed that this problem is nearly
identical to a problem done by the British CEGB
and Electronic Associates on a hybrid computer,
and their experience demonstrates clearly the
need for a machine with the capabilities of a
hybrid computer to handle this problem properly.
These results, unfortunately, are unavailable to
us because of proprietary reasons.

The model itself has been linearized, a procedure
we consider appropriate for control, if not for
safety, purposes. The heat transfer coefficient is
now permitted to be a function of the velocity of
both fluids. Analog controls are provided to permit
variation of salt velocity, water pump speed in-
directly affecting initial pressure, initial tempera-
ture, and throttle opening. The throttle is treated
at all times as a critical aperture. This treatment
causes decoupling of effects downstream of the
throttle.
Part 3. Chemistry

W. R. Grimes

The chemical research and development effort in
close support of the MSBR program includes, as
described in this chapter, a variety of studies. A
major share of this effort continues to be devoted
to the immediate and anticipated problems of the
operating Molten-Salt Reactor Experiment (MSRE).

Sampling of the MSRE fuel and coolant salts,
interpretation of the analyses for major and minor
constituents of the melt, and investigation of the
distribution of fission products as revealed by
specimens exposed to the pump bowl gases con-
tinued as routine. The results of salt analyses

continued to show excellent materials compatibility.

Significant fractions of several fission products
continued to appear in the pump bowl gas space.
The fate of fission products in the reactor was
established in considerable detail. The only un-
usual behavior continued to be that of the more
noble metals such as Mo, Ru, Te, and Nb. These
left the fuel by depositing on the walls, and pos-
sibly also as a smoke that was carried away in the
gas phase. Since the chemistry of the volatile
fluorides of these fission products may have rele-
vance to future aspects of the MSBR program, in-

vestigation of the chemical behavior of their fluo-
rides was continued.

The prospects for application of fluoroborate
coolants in molten-salt reactor technology con-
tinued to appear favorable and has justified a
continuing investigation of the NaF-NaBF , and
NaF-KF-BF, systems. As innovations in MSBR
design increasingly favor the development of a
single-fluid system, efforts to devise efficient
chemical reprocessing methods have become more
significant. Laboratory studies of recovery of
protactinium and removal of lanthanide fission
products from the core salt by reductive extraction
into molten metals continue to show promise and
were actively pursued.

Solution thermodynamics and electrochemistry of
LiF-BeF, melts were investigated in connection
with chemical behavior and extractive reprocessing
of breeder reactor melts.

Development studies in analytical chemistry have
been directed primarily toward improvement in
analysis of radioactive samples of fuel for oxide
and uranium trifluoride and for impurities in the
helium gas from the MSRE.

9. Chemistry of the MSRE

R. E. Thoma

Chemical behavior in the salt, gas, oil, and
water systems of the MSRE has been the subject
of a continuous surveillance program since the
beginning of MSRE operations. Surveillance is
accomplished partly by on-line instrumental analy-
sis. The condition of salt systems, however, is

appraised preponderantly from the results of chemi-
cal analysis. Numerous samples are obtained
therefore as part of this program. They afford a
means for determining the stability and compati-
bility of salt-metal systems, the condition and
service life of coolants and lubricants, and the
composition, purity, and oxidation-reduction po-
tentials of the fuel salt. In addition, analyses of
the fuel salt are performed regularly to measure
isotopic composition of the fissile material in
order to compute fuel burnup rates. It has been
the purpose of this program to apply the experience
gained during the period when the MSRE contained
235(_238( fuel salt to operation of the reactor in

the near future with 233U fuel salt and subsequently

to the design of molten-salt breeder reactors.

Results of previous MSRE materials analyses
were summarized in the last report of this series.
In the intervening period, the MSRE has operated
continuously for its longest period of power gen-
eration uninterrupted by drain and flush operations.
This period of some 25,500 Mwhr has afforded an
unprecedented opportunity to determine the extent
to which a program of routine analyses is relevant
to reactor operations.

1

9.1 FUEL SALT COMPOSITION AND PURITY

It became evident at the onset of MSRE power
operation that the on-line reactivity computation
was about ten times as sensitive for detection of
small changes in uranium concentration as individ-
ual chemical analyses of the fuel salt could be
anticipated to be. Therefore, until more precise
methods for the determination of uranium are de-
veloped, the chief function of individual analyses

1MSR Program Semiann, Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 102-8.

must come from their indication of long-term trends
in fuel composition and from their application to
studies of chemical changes in the flush salt.

Routine MSRE operations alter the composition
of the fuel salt principally by the consumption or
addition of ?*SUF, and by intermixing fuel and
flush salts. In current operations the fuel salt has
been circulated in the reactor for more than five
months without alteration of its composition by
intermixing fuel and flush salts. The average
composition of the fuel salt during runs 13 and 14
is compared with previous values in Table 9.1.
These results reinforce the prior inference that
the circulating fuel salt would exhibit excellent
chemical stability in nuclear operations and indi-
cate that estimates of fuel composition, as com-
puted from operation data, are in satisfactory agree-
ment with analytical results. Final evaluation of
operating power levels and residual reactivity de-
pend upon such concurrence.

Statistical analyses of the chemically determined
values for uranium concentration show that control
limits have not been exceeded. However, within
these limits a disparity between computed and
analytical values has developed, one in which the
results of chemical analyses indicate that the con-
centration of uranium in the fuel salt is currently
lower by ~0.02 wt % than that computed from op-
erational data. This trend is suggested by com-
parison of the average values of uranium as deter-
mined from chemical analysis with those computed
from operation analysis (Table 9.2).

The results of individual chemical analyses of
uranium in the MSRE fuel salt are compared with

Taoble 9.1. Average Composition of MSRE Fuel Circuit Salt in Mole %

- Number of

Run No. LiF BeF2 ZrF4 UF4 Samples
Design 65.00 29.17 5.00 0.83

FP-4 63.36 + 0.57 30.65 £ 0.58 5.15 £ 0.12 0.825 £ 0.011 22
FP-5-7 63.29 £ 0.72 30.70 £0.70 5.19 £0.13 0.824 1 0.011 14
FP-8 65.84 £2.49 28.57 £2.12 4.82 +0.35 0.771 * 0.058 8
FP-9 64.17 £ 0.10 30.04 £ 0.14 4,99 £0.19 0.794 £ 0.011 4
FP-10 64.65 T 0.45 29.64 1 0.48 4,92 £0.06 0.791 1 0.010 10
FP-11 64.30 £ 0.90 29.89 1 0.81 5.01 £ 0.13 0.803 £ 0.018 41
FP-12 64.46 1 0.83 29.86 1 0.77 4.88 £0.14 0.794 + 0.021 26
FP-13-14 64.07 *0.95 29.95  0.95 5.15 £ 0.13 0.819 £ 0.015 44

Table 9.2. Comparison of Analytical and Computed
Values of Uranium in the MSRE Fuel Salt

Percent Uranium

Run Number
No. Book Analytical Akg U Mwhr of
(av) {(av) Samples
x 10°
5-7 4.623 4.629 £0.026 +0.3  15.6 14
8 4.605 4.632 £0.011 +1.35 4,4
9 4.591 4.603 10.031 +0.60 2.5 4
10 4.573 4.609 £0.020 +1.80 10.5 10
lla 4.569 4.570 £0.018 +0.05 12.5 35
lib 4.579 4,571 £0.019 -1.40 3.0 6
12a 4.561 4.548 £0.022 —~0.65 4.5 18
12b 4.590 4.572 £0.032 ~0.90 1.5 8
13-14 4.564 4.543 10.032 ~1.05 25.5 44

the nominal book value for uranium as shown in
Fig. 9.1. The actual trend in these values is more
clearly evident from the control chart shown in
Fig. 9.2. The slope of the points shown here was
computed from a least-squares program using both
weighted (where individual data points were as-
signed values of 4/0? — dashed line, Fig. 9.2)
and raw values (solid line, Fig. 9.2) of uranium
concentration. The results show that a real dis-
parity exists between values of uranium concentra-
tion in the fuel salt as computed from operational
analysis and chemical analysis. The raw values
indicate a difference in chemical and book value
of 1.28 kg in total uranium. A similar disparity
has also been noted in a gradual decline in the
computed reactivity which has been observed in
runs 13 and 14 (this report, Sect. 1.2.1) and which
seems to have proceeded from the beginning of
power operation, a decline which would account
for an apparent difference of 0.95 kg in total
uranium.

We find no chemical basis for this unexplained
difference between the anticipated values for ura-
nium concentration of the fuel salt and the experi-
mentally determined values. The reality of the
difference seems to be unquestionable and is found
in corresponding magnitude in reactivity computa-
tions. We conclude therefore that it will be neces-
sary to reevaluate the method by which the con-
centration of uranium in the fuel salt is computed.

90

The overall purity of the MSRE fuel salt, as
judged from the concentrations of oxide and chro-
mium, has been maintained during the present
report period. An innovation was made in the
method of handling samples of fuel salt obtained
for oxide analysis or for measurement of the con-
centration of trivalent uranium in fuel salt (this
report, Sect. 2.2). This change is the use of a
heated carrier to maintain the temperature of salt
specimens after isolation at 200 to 300°C and
thereby to ensure that the fluorine recombination
reaction can proceed sufficiently rapidly to prevent
evolution of free fluorine from frozen salt. This
technique presumably obviates the possibility that
nascent fluorine might remove oxide from the salt.
After this new method of handling salt specimens
was employed, the oxide concentration in one salt
sample (FP-14-53) was found to be 58 ppm, not
significantly greater than that found in samples
before the heated carrier was used. We conclude
that this result confirms the results of previous
oxide analyses and indicates that the oxide con-
centration of the fuel salt has remained at about
50 ppm since the beginning of MSRE operations.

9.2 MSRE FUEL CIRCUIT
CORROSION CHEMISTRY

In previous power operations of the MSRE, the
concentration of chromium in the fuel salt has re-
mained constant during periods of salt circulation.
Until experiments were initiated in December 1967
to examine the relationship of residual reactivity
to temperature, salt level in the pump bowl, and
cover gas pressure, the concentration of chromium
in the fuel salt during run 14 was 72 £ 8 ppm.
After power was reduced from 7.2 to 5.0 Mw and
reactor outlet temperature from 654 to 638°C, we
found that the chromium level of the fuel salt had
risen to approximately 85 ppm. Other conditions
which were subsequently imposed during these
tests included additional variation in operating
temperatures within the range 638 to 663°C and
variations from 5.0 to 9.0 psi in cover gas pres-
sure. The results of current chemical analyses
indicate that concentration of chromium in the
fuel salt has reached a new steady-state value
of 85 ppm.

While the cause of the recent 13 ppm increase
in chromium in the fuel salt is not known, only
temperature changes during otherwise steady and
URANIUM {(wt %)

4.700

4.680

4.660

4,640

4.620

4.600

4,580

4.560

4.540

4.520

4.500

4.480

4.460

4.440

4.420

91

ORNL-DWG 68—-55639

Fig. 9.1. Comparison of

MEGAWATT HOURS (x103)

Analytical and Computed Yalues of Uranium in the MSRE Fuel Salt.

| ’
' -
j -~-- COMPUTED VALUES
o ANALYTICAL VALUES
o S I
EL-\"!) i (T% r_
l 0
3 [
!
L. —
v\ A
[ 'fi S ,
i'r TYR‘ i S
| | a1 Y
o) \\41\
] L
EREERE
1966 - 1967 -—% 1968 — =
B I I A I I
L |
5 10 15 20 25 30 35 40 45 50 55 60

65
92

ORNL—-DWG 68—-5540

4.73
4.71
469 —
| = —
4.67 J
o]
° o
° T — 0o ° o2 mJ /NOMINAL BOOK VALUE
5 o o 0 o
$ 0o — TR
g ° b o FROM RAW DATA
2 g ° 8 ° o ©
pd o . — S o0
<4 463 ] ~ O
. E”}—fl-—— R e e ] e SV 4 St P Sy
o ° FROM WEIGHTED DATA—" ~da olo
o o T
o] © O 0
o S~ —
4.0t el o—1 5 4o
____F__F__.__ ___________.____-_____________0____&__<
o } '1
4.59
1l
4.57 — | L
O
o]
| o
4.55
0 5 10 15 20 25 30 35 40 45 50 55 80 65

MEGAWATT HOURS (x103)

Fig. 9.2. MSRE Fuel Salt — Normalized Values for Uranium Analyses. Wt % U (analyzed)/wt % U (book) X 4.65.

continuous power operation seem to be without
precedent. The observed rates of corrosion in the
MSRE have been significantly lower than predicted
from thermodynamic data and diffusion theory and
might have been expected to produce chromium in
the fuel salt at rates which would have caused the
salt now to contain 250 to 300 ppm of Cr?*. The
temperature dependence of the equilibrium Cr° +
20F , = CrF , + 2UF, is too low to explain this
recent increase of chromium concentration.

It has been postulated that one of the principal
reasons for the unexpectedly low values observed
is that the metal surfaces of the fuel circuit have
been covered with a film of the noble-metal fission
products Nb, Mo, Tc, and Ru about 10 A thick.
Results of electron microprobe analysis of the
metal surveillance specimens? removed from the
MSRE in May 1967 lend support to this view in

that they did not reveal any change in chromium
concentration below a depth of 10 g, the limit of
measurement.

Possibly the temperature changes imposed re-
cently caused spallation of the noble-metal fission
products from the hotter regions of the fuel loop,
behavior which could occur if the thermal coeffi-
cients of expansion of the noble-metal fission
product film and base metal alloy are unequal.
Chromium from the freshly exposed alloy surface
would then be available for reestablishment of its
steady-state activity, If this sort of mechanism
is indeed operative, similar behavior should occur

2Furnished through the courtesy of C. Crouthamel and
associates, Chemical Engineering Division, Argonne
Naticnal Laboratory, Argonne, I1L
93

on resumption of power operation after p«:riods
when the reactor is drained and cooled. Actually,
slight increases in chromium concentration of the
fuel salt have been observed each time such
circumstances have occurred.

The significance of the mechanism proposed here
is that it requires assignment of previous changes
in the chromium concentration of the fuel salt to
the circuit system rather than to the drain tanks. If
the total amount of chromium represented by this
increase, +47 ppm from the outset of MSRE opera-
tions, were leached uniformly from the fuel circuit,
it would correspend to removal of chromium from a
depth of 0.283 mil, or to an average corrosion rate
of 1.972 x 102 mil per 1000 hr of operation with
salt circulating in the fuel circuit.

9.3 ISOTOPIC COMPOSITION OF URANIUM IN
THE MSRE FUEL SALT

We have obtained mass spectrometric analyses3
as part of a program of routine analysis of the MSRE
fuel salt. Results of the isotopic analyses have
been accumulated in order to evaluate the potential
applications of such analyses to the computation
of power generation in the MSRE. The results ob-
tained from the beginning of power operations with
the MSRE are summarized in Fig. 9.3. The data
shown here indicate moderately good correlation of
the spectrochemical determination of 235U concen-
tration with calculated values. The analytical re-
sults show a continuous average bias of + 0.6% as
compared with calculated values.

The results of spectrochemical analyses shown
in Fig. 9.3 indicate that the concentration of 23U
in the fuel salt increases as anticipated, essen-
tially linearly with power generation. Small changes
in the 235U /2380 ratio which have occurred as

3Perforrned by R. E. Eby, Analytical Chemistry Divi-
sion, ORNL.

235U was depleted from or added to the fuel salt
have little effect on the relative concentration of
238y, Results of the 235U concentration analyses
may be used effectively therefore to compute the
output of the reactor. Results of such computa-
tions are described in Sect. 1.2.5.

ORNL-DWG 68-~5544

33.6+
33.4 !p\\fi Rl//c ]
N
33:2 KO AN 235 T S
\ II NN
N
N i \NOMINAL
N\ | AN
33.0 S . S
N \ ] N
\\ | N\
N N i AN
\{ 1 N N
32.8 \\ } Ny . _—
N |
N
N
32.6
® 0.5
z
Q~
N P
S 0.4 -
234, /
ZL0-/0\0——0——-(::* P °
0.3
236y,
0.2 //O/
O.1W
66.5 o
R/’O-ifu//
66.0 | ¥ — ]
—
65.5
0 0 .20 30 40 50 60

MEGAWATT HOURS (x 103)

Fig. 9.3. Isotopic Composition of Uranium in the MSRE
Fuel Salt.
10. Fission Product Behavior

10.1 FISSION PRODUCT BEHAVIOR
IN THE MSRE

S. S. Kirslis F. F. Blankenship

The reasons for interest in the behavior of fis-
sion products in the MSRE have been discussed
in previous reports. ! ~® The principal practical
concern continues to be with the possible depo-
sition of the noble-metal fission products (Mo,

Tc, Ru, Te, and Nb) in the graphite core of an
MSBR. Neutron economy would suffer if too large
a fraction of these species possessing moderate
neutron cross sections were deposited in the high-
flux region. Information on the concentrations of
fission products in MSRE fuel salt is of immediate
interest in planning the volatility processing of
the current charge of MSRE fuel salt.

In the current report period, tests similar to
those previously carried out were continued, with
emphasis on checking and improving the experi-
mental techniques. In addition, several hot-cell
tests were carried out on the nature of gas-borne
activities above molten fuel salt samples from the
MSRE. The following sections will report in some
detail on

1. the completion and evaluation of fission product
analyses from the second set of graphite and
metal specimens from the core of the MSRE,

2. analyses of MSRE fuel salt sampled by an im-
proved technique,

3. analysis of the pump bowl cover gas,

lp. R Kasten, Graphite Behavior and Its Effect on
MSBR Performance, ORNL-TM-2136, p. 4.1.

2S. S. Kirslis, MSR Program Semiann. Progr. Rept.
Aug. 31, 1966, ORNL-4037, pp. 16566,

3S. S. Kirslis and F. F. Blankenship, MSR Program
Semiann. Progr. Rept. Aug. 31, 1967, ORNL-4191,
p. 116.

4. an investigation of contamination problems
with metal specimens exposed in the pump
bowl,

5. hot-cell tests on the quantity and chemical
form of fission products found in the gas phase
above the surface of MSRE fuel salt,

6. miscellaneous tests.

10.1.1 Fission Products in the MSRE Fuel

The possibility was suggested in the previous
report* that the occasional high values and wide
scatter of measured concentrations of noble-metal
activities in fuel salt samples may have been due
to contamination of the ladle sampler by activities
present in the pump bowl gas phase. To eliminate
this possibility, salt samples were taken using
freeze-valve capsules of the type previously used
for sampling the pump bowl gas phase.’

The evacuated 20-cc capsule was sealed by a
fusible plug of Li ,BeF, which melted when the
capsule was lowered below the fuel level in the
pump bowl, allowing the fuel salt to fill the evac-
uated capsule. The enclosed salt sample was
thus protected from contamination from outside.
In the analytical hot cell, the end of the inlet
capillary was plugged with wax, and the exterior
of the capsule was leached with acid to remove
contaminating activity. The capsule was cut into
several sections with a tubing cutter, the bulk of
the salt was removed from each section, and the
emptied sections were leached with acid to re-
move adhering salt. The salt, after powdering
and mixing, and the interior leach were analyzed

4be'd., p. 121, The suggestion was by W. R. Grimes.

5s. S. Kirslis and F. F. Blankenship, MSR Program
Semiann. Progr. Rept. Feb. 28, 1967, ORNI1.-4119,
p. 139,
95

Table 10.1. Fission Products in MSRE Fuel Samples

Sample No. FP14-22 FP14-20 FP14-30 FP14-63 FP14.66
Sampling date 11-7-67 11-3~67 12-5-67 2-27-68 3-5-68
Sampling method Ladle Freeze-valve Freeze-valve Freeze-valve Freeze-valve
capsule capsule capsule capsule
Fission
Isotope Yield Disintegrations per Minute per Gram®
Mo 6.06 8.15 x 1010 2.22 x 10° 8.5 x 108 3.20 x 108 3.22 x 108
1321¢ 4.7 8.9 x 10° 8.19 x 108 1.26 x 10° 2.78 x 108 ~6.2 x 108
1291¢ 0.35 1.86 x 108 <1.07 x 108 <2.92 x 107
103Ru 3.0 3.74 x 10° 1.40 x 108 4.9 x 107 2.54 x 107 <2.2 x 107
106Ry 0.38 ~2.7 x 108 2.38 x 107 <6.4 x 108 <3.2x 10° <1.6 x 107
?5Nb 6.2 <108 4.2 x 107 7.0 x 10° 3.4 x 10° ~2.5x 10°
%57: 6.2 1.28 x 1011 1.19 x 10*1 1.29 x 1011 1.48 x 101! 1.21 x 101!
131y 3.1 6.64 x 101? 5.51 x 101° 3.41 x 1010 4,46 x 1010
39gr 4.79 9.15 x 10! 8.15 x 1010 8.46 x 10'° 9.15 x 1010 9.07 x 101°
143ce 5.7 1.69 x 10! 1.41 x 1011 1.45 x 101 1.86 x 10*! 9.17 x 101°
147Ng 2.7 6.75 x 10'° 4.34 x 1010
140, 6.35 1.59 x 1011 1.16 x 1011 9.30 x 1010 1.68 x 10!
239Np 5.96 x 1011 5.36 x 10! 6.2 x 10%! 4,70 x 101}
1g 1.41 x 101! ~9.1 x 10%°

4All activities calculated back to sampling time.

radiochemically. The inlet capillary was dis-
solved and analyzed as a separate sample.

The results of the radiochemical analyses are
reported in Table 10.1, in which each value repre-
sents the sum of the activities found in the salt,
the interior leach, and the capillary dissolution,
and is reported in disintegrations per minute per
gram of sample calculated back to the time of
sampling.

The concentrations of the fission products with
stable fluorides (rare earths, alkalies, alkaline
earths, and zirconium) were in satisfactory agree-
ment with previous analyses of samples taken by
the ladle method, However, the concentrations
of the noble-metal fission products were at least
an order of magnitude lower than in previous sam-
ples. In the first sample, FP14-20, the bulk of
the noble metals in the total sample were found

in the dissolution of the inlet capillary. It was
suspected that noble-metal activities adhering to
the outside of the capillary tip were covered by

the drop of wax used to seal the tip before leach-
ing the outside of the capsule. In subsequent
samples the outside of the capillary tip was sanded
to remove adhering activity before applying the
wax; only a small fraction of all activities was
then found in the capillary.

The leach of the interior of the emptied capsule
usually contained an appreciable fraction of the
total noble metal in the capsule, occasionally as
much as half. Most or all of the °>Nb was in this
leach. Since the salt sample contacted the interior
of the nickel capsule in the molten state for only
about 15 min, these observations indicate that
noble metals deposit rapidly from molten salt on
a clean nickel surface. It is therefore possible,
particularly in the case of ?°Nb, that the salt sam-
pled was previously depleted in noble metals by
deposition on the exterior of the capsule. This
difficulty can be avoided experimentally in several
ways: (1) the capsule or the capillary can be con-
structed of or plated with a noble metal, (2) only

a long capillary with a salt seal at its tip need be
submerged in the fuel salt, or (3) the salt seal can
be made of a high-melting salt like LiF, allowing
the capsule exterior to saturate with noble metals
before the sealing salt dissolves.

Deposition on the capsule exterior probably does
not deplete the salt sampled of noble metals by
more than a factor of 2, except in the case of
95Nb. However, it leads to another analytical
difficulty which may help to explain the consider-
able scatter of the noble-metal data in Table 10.1.
The exterior of the capsule is leached repeatedly
with 2 ¥ HNO , at 85 to 95°C until the last leach
contains less than 1% of the gamma activity of
the first leach. Noble metals, however, tend to
teplate on nickel from the leaching solution.
Normally 15 to 20 leaches are required to reach
the 1% level. Therefore, some noble metal prob-
ably remains on the exterior of the capsule when
leaching is halted. Since the noble-metal con-
centrations inside the capsule are very low, a
slight remaining contamination of the exterior
might account for much of the activity found in
the inside leach. On this basis, the lower figures
in Table 10.1 are probably more nearly correct.

In any case the scatter of the noble-metal con-
centrations in the fuel salt samples is of little
practical consequence since the values are so
small, representing on the average less than 1%
of the quantities which would have been found if
none had left the salt phase. By contrast, the
fission products with stable fluorides show con-
centrations within 30% of their calculated values,
although scatter exceeds the claimed 20% ana-
lytical accuracy.

In summary, it is felt that the estimates of noble-
metal concentrations in fuel salt are much more
reliable from freeze-valve capsule samples than
from ladled samples. Previously reported data
should be discounted. For salt-seeking species,
both sampling methods are equally satisfactory,
and previously reported data should be valid.

The planned fluorination of the fuel charge in

the MSRE to remove UF  is simplified by the
fact that only 1% or less of the noble-metal fis-

96

sion products remain in the fuel. Special trapping
techniques are required to separate volatile noble-
metal fluorides from UF .

10.1.2 Fission Products
in the MSRE Cover Gas

The analyses of seven gas samples taken in the
mist shield region of the MSRE pump bowl by the
freeze-valve capsule technique have been thor-
oughly discussed in previous reports. =2 The
results indicated that a small amount of fuel salt
mist and an appreciable share of the total noble-
metal nuclides produced by fission were by some
mechanism injected into the cover gas phase (see
Table 10.3). Uranium concentrations in the gas
samples were usually higher than those calculated
from the amounts of salt mist indicated by °5Zr
analyses.

Only a single cover gas sample, FP14-67, was
taken in this report period, since experimental
time was preempted by other high-priority work.

A freeze-valve capsule of slightly different design
(Fig. 10.1) was used, and the sample was taken
during 5-Mw operation under otherwise normal con-
ditions.

The analytical results, together with those of
several previous runs, are shown in Table 10.2.

10.1.3 Deposition of Fission Products
from MSRE Cover Gas on Metal Specimens

Tests in which small metal specimens were ex-
posed to the gas phase and fuel phase in the pump
bowl gave the first qualitative indications of un-
usual volatilization and plating behavior of the
noble-metal fission products. High concentrations
of noble-metal activities were found on specimens
exposed to cover gas or fuel, while fission prod-
ucts possessing stable fluorides showed much
lower activities. Results have been reported pre-
viously from a large number of tests run with sev-
eral types of metal specimens (Hastelloy N, Ag,

81bid., pp. 138—41.

7MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL~4191, pp. 11619,

85. 8. Kirslis and F. F. Blankenship, Reactor Chem.
Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229,
pp. 8~10,
ORNL-DWG 68-6079

3/4-in. - OD NICKEL TUBE 0.020 -in. WALL
i " '
¢

SR USSR INTRY

AL RABRLRRY
AW

SO TYSSLY AASSARTY LW

Yg-in. Ni TUBING

\\

L Vg -in. WELD ROD

_ 35— FLARED CcuP
L'Z BeF4 g

Fig. 10.1. Freeze-Yalve Capsule for Sampling Pump

Bowl Cover Gas.

Ni, Au, and stainless steel) under a variety of
reactor operating conditions and exposure times.
The results of these tests were disappointing
from the standpoints of obtaining quantitative in-
formation and of showing variations in deposition
under different test conditions. The reproduci-
bility of results was poor, with variations between
duplicate runs often as high as an order of magni-
tude. Deposition on the different metals was not
significantly different. Exposure time had little
effect; samples exposed for 1 min collected as
much activity as those exposed for 10 min.
Changes in reactor operating conditions had little
effect on deposition. Even samples exposed sev-
eral hours after stopping the fuel pump or two days
after draining the fuel out of the reactor showed
as much deposition of noble metals as normal
samples. The addition of beryllium to reduce the
fuel caused no significant change in deposition.

97

The only variable which had a definite effect on
gas phase deposition was a reactor shutdown of
two weeks or longer. Then decreases in noble-

metal deposition by an order of magnitude were

observed.

In the current report period, efforts were con-
centrated on trying to identify the experimental
factors causing difficulty. Comparative runs were
made in which the specimens (Ag, Hastelloy N,
and stainless steel) were fluorinated, hydrogenated,
or left in the normal slightly oxidized condition.
Exposure times were 1 and 10 min. The six tests
gave practically undistinguishable results. Depo-
sition of noble metals was approximately halved
on the fluorinated specimens compared with the
others, but this small difference is probably not
significant.

The last two tests identified the basic difficulty.
In the first of these, the metal specimens were
lowered in the sampling pipe to a position 2 ft
above the pump bowl. During the 10-min exposure,
the usual helium flow of 200 cc/min was passed
down the tube. The test assembly was then with-
drawn and packaged for shipment to the analytical
laboratory in the usual way. Approximately the
same amounts of noble-metal nuclides were de-
posited on the test specimens as in noirmal ex-
posures in the pump bowl.

The second test consisted in hanging a standard
test assembly on the cable latch, leaving it there
for one day as in the customary test procedure,
removing it with the sampling cubicle manipulator,
and packaging it for shipment and analysis in the
usual way. Again the amounts of noble-metal ac-
tivities found on the specimens were nearly as
great (within a factor of 1 to 10) as in normal pump
bowl exposures.

Clearly the contamination of the samples during
the various manipulations involved in the test pro-
cedure before or after the actual pump bowl expo-
sure was responsible for most of the observed ac-
tivities. This difficulty was not identified sooner
because of the astonishing reproducibility of the
contamination. Variations of several orders of
magnitude would be expected rather than the ob-
served zero to one order. This uniformity suggests
that most of the contamination may have occurred
while the test assembly remained for a day on the
latch in the highly contaminated 1C area. The
mechanism by which the contamination spreads
from the pump bowl to the various surfaces is un-
Toble 10.2. Fission Products in MSRE Cover Gas as Determined from Freeze Valve Capsules

Experiment No. FP10-11 FP10-22 FP1142 FP1146 FP11-53 FP12-7 FP12-26 FP14-67
Sampling date 12-2766 11167 4-11-67, 02:49  4-18-67, 02:28 5-2-67, 10:43 6-2-67, 06:50 7-17-67, 06:03 3668, 06:03
Operating time, dayse 14.5 off, 12.6 on 14.5 off, 27.7 on 65 on, 1.5 hr off 14 off, 72 on 14 off, 86 on 92.3 on, 42,5 off 46 off, 23 on 13 at 7 Mw, 6.6 at 5 Mw
Nominal power, Mw 7.4 7.4 0 7.2 7.2 0 7.2 5.0
Be addition No After 5.5 After 8.40 ¢ No No No After 37.8 g No
Features Regular Regular Pump off 1.2 hr Regular Helium bubbles Power off 42.5 days Regular Regular
Accumulated Mwhr 13,600 16,200 27,900 29,100 31,700 32,650 36,500 62,770
Yield b
Isotope (%) Disintegrations per Minute in Total Sample
%Mo 6.06 2.04 x 10! 1.36 x 101! 1.05 x 1011 2.31 x 101! 1.57 x 1011 2.74 x 101! 3.07 x 1011
103py 3.0 3.80 x 10° 2.63 x 10° 2.51 x 10° 4.64 x 10° 1.12 x 101@ 4x10° 1.18 x 101°
10624 0.38 ~6.7 x 107 7.74 x 107 8.2 x 107 9.49 x 107 4.03 x 108 1.7 x 108 5.03 x 108
1320 ~g.7 5,73 x 101° 5.08 x 1010 1.15 x 1011 3.35 x 101! 1.88 x 1011 4.17 x 107 3.16 x 109> 1.21 x 1011
12914 0.35 7.98 x 108 3.51 x 10° 2.17 x 108 6.6 x 108
93Nb 6.2 <3.26 x 107 2.09 x 108 6.45 x 108 1.3 x 10° 1.05 x 1010 2.26 x 10%? 3.52 x 10° 2.18 x 10°
#5zr 6.2 <2.9x 10° <2,2 %10’ <4.4 x 107 ~2x 10 1.8 x 108 8.64 x 10’ 2.98 x 107 2.14 x 108
140p, 6.35 2.75 x 108 3.48 x 108 6.16 x 108 4,01 x 108
131y ~3,1 9,75 x 10° 2.03 x 107 5.7 x 107 9,81 x 108 8.63 x 10° 1.67 x 1010 3.11 x 10'°
89gr 4.79 2.13 x 108 4,08 x 10° 3.72 x 10° 8.35 x 108 3.71 x 10° 1.70 x 10°
111ag 0,019 5.5 x 108 1.33 x 108
14lce ~6.0 4.79 x 108 1.83 x 108
1440 ~6.0 4,17 x 107 1.71 x 108
235y« 3.86 0.55 59 9.2 23 25 28

fDuration of previous shutdown and of continuous operating time just before sample was taken,

bDisintegrations per minute calculated to the time of sampling or of previous shutdown.

[~ . .
Micrograms in sample.

86
certain but probably involves rubbing of the cable,
latch, samplers, and manipulators against each
other.

These observations make it necessary to discard
any quantitative interpretation of the pump bowl
exposure tests. The source of the contamination
is still principally the gas phase above the fuel
in the pump bowl, so that the qualitative conclu-
sion that noble-metal activities predominate in
this gas remains valid. Also the conclusions from
test FP11-50, in which graphite and metal speci-
mens were at least partially protected from con-
tamination by a perforated metal screen,® probably
remain sound. These observations on contami-
nation confirm the explanation given for the high
concentrations of noble-metal fission products
in fuel salt sampled by the open ladle technique.

A capsule with a sliding sheath has been de-
signed which will protect metal or graphite sam-
ples from contamination except when the test as-
sembly is in the pump bowl.

10.1.4 Examination of MSRE Surveillance
Specimens After 24,000 Mwhr

In a previous report, 10 a summary was presented
of the more significant results from the examina-
tions and analyses of the graphite and Hastelloy
specimens which were exposed to fissioning molten
salt for 24,000 Mwhr in axial positions in the MSRE
core. Since then, the analyses have been com-
pleted, and the middle graphite bar (Y-7) was also
sampled by a different method and analyzed for a
few isotopes with a germanium diode gamma spec-
trometer. The same specimen was also examined
with an electron probe in an attempt to detect sur-
face contamination and by proton bombardment to
determine the concentration profiles of lithium and
fluorine. The significance of these additional re-
sults will be briefly discussed.

Completion of Analytical Results. — The con-
centration profiles of °°Mo, *32Te, !%3Ru, ?°Nb,
957, 89S, 14%Ba, and 235U in the exposed graph-
ite specimens were described in the last report. °
The completion of analytical work on these nu-
clides caused no significant changes in the pro-

9MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL~4101, pp. 131—35.

1016id., pp. 121—28.

99

files given. All the samples were also analyzed
for 1°°Ru. Without exception, the 1°®Ru values
were a factor of about 20 lower than the °3Ru
values, so that the concentration profiles have
identical shape.

In addition, some selected samples were ana-
lyzed for lllAg, 137CS, 13 lI, le, I“Ce, 144Ce,
and 14’Nd. The !!!Ag profile behaved like 193Ru,
the activity dropping steeply through four orders of
magnitude in 10 mils. The !37Cs behaved in an
unusual manner, as it also did in the 7800-Mwhr
exposure. Its activity dropped two orders of mag-
nitude in 10 mils and then began slowly rising;
1311 behaved similarly. It has been suggested
that both these species can diffuse in graphite.
The '4’Nd, 1#4Ce, and !*!Ce exhibited fairly
straight plots of the logarithm of the activity vs
penetration distance, and the slopes were each
quite similar to that for 1*°Ba. This behavior
does not agree with the theory that the slope
should vary inversely with the half-life of the rare-
gas precursor. The half-lives are ‘‘very short,”’
less than 1 sec, 1.7 sec, and 16 sec, respectively,
for the precursors of !*’Nd, !*4Ce, 4!Ce, and
140Ba. Yttrium-91, with a 10-sec krypton pre-
cursor, gives a slope less steep than '*°Ba.

With its very short-lived xenon precursor, it is
puzzling how **47Nd got into the graphite at all.
From the fact that the three rare earths had simi-
lar concentration gradients, it may be guessed
that they entered and diffused in the graphite as
rare earths rather than as xenons.

For each of the species which gave straight
plots of log activity vs distance, the slopes were
steeper for the top graphite sample than for the
sample in the middle of the core, although the
graphite types were supposedly identical. Both
of these samples showed distinctly steeper slopes
than the bottom sample, which was of the more
porous ‘‘lattice stock’ variety.

The concentration profile plots revealed an inter-
esting indication of the difference in permeability
of the same piece of graphite from different sides.
For the top graphite sample, the concentration
profiles of all nuclides, without exception, were
higher for penetration from one of the narrow sides
than for penetration from any of the other sides.

Sampling by Sanding. — Five graphite blanks
were obtained during the course of milling the ir-
radiated samples by milling a clean piece of the

- same type of graphite on the same milling machine.
It was very disturbing that analyses of the blanks
sometimes showed more activity (particularly for
99Mo, '93Ru, !%6Ry, and ?5Nb) than the irradiated
sample just previously milled. Confidence was
destroyed in all analyses giving lesser values than
the blanks. Nevertheless, these analyses behaved
in a very consistent manner, resulting in smooth
concentration profiles.

To check the validity of the sampling method, a
remaining stored piece of the irradiated middle
graphite bar (Y-7) was handled by a sanding method
that permitted sampling from the inside out (see
Sect. 10.2). The concentration profiles obtained
tailed off toward the center of the specimen in a
manner very similar to that observed with the
milled specimens. Absolute comparisons between
the two sets of data usually agreed to within better
than an order of magnitude. Since the area of
graphite sampled by the two methods differed by
a factor of about 30, local variations in graphite
density and permeability might explain the dif-
ference in results.

The results of the sanding method restored much
of our confidence in the milling method of sampling.
Certainly the reality of the tails of the concentra-
tion profiles was established. The simplest ex-
planation of the high blanks is that the blank
graphite specimen somehow became contaminated
inside the hot cells, where the milling was done.

Electron Probe Examination. — A small sample
of the irradiated middle bar graphite (Y-7) was
mounted for electron probe examination for pos-
sible identification of any surface deposits. The
sample was shipped to the Argonne National Lab-
oratory, where an instrument for examining radio-
active samples was available. The electron probe
examination, which could be used to within 1 or
2 u of the edge of the graphite, detected no im-
purities. Typical detection limits were 0.02 wt %
for tellurium and 0.04 wt % for uranium and tech-
netium.

Lithium and Fluorine Concentration Profiles in
Irradiated Graphite. — A proton bombardment method
for determining the concentration profiles of lith-
ium and fluorine in irradiated graphite is described
in Sect. 10.3. The results for the middle graphite
bar (Y-7) indicated that the molar concentration of
lithium exceeds that of fluorine at all points in-
side the graphite. This observation proves con-
clusively that the fission products observed in
irradiated graphite do not permeate as fluorides.

100

They probably are present either as carbides or
in the metallic form.

Fission Product Distribution in the MSRE. — Re-
sults have been reported above for the quantities
of fission products in the fuel salt and in the cover
gas of the MSRE. The previous report!? gave
values for the quantities deposited on graphite
and on Hastelloy N on May 5, 1967, after 32,000
Mwhr of power operation. In addition, the total
inventory of each nuclide at this date is avail-
able from a recently developed computer program
for calculating the total amount of each isotope
from the operating history of the MSRE. An ap-
proximate material balance for May 8, 1967, based
on these data is given in Table 10.3. It differs
from the balance given in the previous report®?2
because of the revised concentrations of noble
metals in the fuel salt.

The balances are good for nuclides with stable
fluorides which remain in the salt phase; fair
balances were obtained for 132Te, °°Nb, and
99Mo; and '°3Ru is poorly accounted for.

To enter the cover gas data into a material bal-
ance, the quantity of each species lost per day
(observed concentration times volume of cover
gas flow per day) must be compared with the pro-
duction of that species by fission per day. As
discussed in the previous report, 3 it is likely
that the gaseous concentrations measured in the
mist shield region are higher (by a small factor)
than those in the bulk of the cover gas. This will
tend to yield high material balances for species
like °°Mo, whose concentration in the gas phase
is large.

10.1.5 Hot-Cell Tests on Fission Product
Volatilization from Molten MSRE Fuel

It has been established that the gas phase above
the fuel in the MSRE pump bowl contains appreci-
able concentrations of fission products, particu-
larly the noble metals. However, no hard informa-
tion exists regarding the chemical and physical
nature of the volatilized species and the mech-
anism of volatilization. In view of the potential

"1bid., pp. 126—27.
121bid., p. 128.
B1bid., p. 118,
101

Table 10.3. Approximate Fission Product Distribution in MSRE After 32,000 Mwhr

Inventory
Isotope in MSRE Percent in Percent on Percent on Cover Gasb
(dis /min)® Fuel Graphite Hastelloy N (%/day)
x 1017
%6 7.91 0.94 10.9 40.5 77
132
Te 5.86 0.83 10.0 70.0 66
103py 3.36 0.13 6.6 14.9 40
95Nb 4.40 0.044 36.4 34.1 5.7
957, 6.00 96.1 0.03 0.06 0.14
895, 5.02 77.0 0.26 33
131y 4.00 64.0 1.0 16

%This total inventory was calculated from the power history of the MSRE.

b These values represent the percentages of daily production rate lost to the cover gas per day.

practical importance of the volatilization phenom-
enon as a method of removing noble-metal fission
products from the fuel circulation system of an
MSBR, it is desirable to investigate these more
basic questions.

There are serious limitations on the types of
tests which can be carried out in the existing
MSRE pump bowl facility. However, it was noted
that two of the cover gas samples taken after re-
actor shutdown (FP11-42 and FP12-7 in Table
10.2) contained sizable concentrations of noble-
metal activities. This indicated that fission was
not essential to the volatilization process and
suggested the possibility of studying volatilization
from a molten sample of MSRE salt in a hot cell,
where experimental possibilities would be much
less limited.

First and Second Hot-Cell Tests. — The experi-
mental plan was originally designed to distinguish
between the volatile fluoride and metallic colloid
hypotheses'* on the nature of the volatilizing spe-
cies. We planned to measure the activities in the
gas phase above the surface of molten MSRE salt
when (1) pure helium was slowly passed over the
quiescent surface, (2) a mixture of 5% H , in he-
lium was similarly passed, (3) pure helium was
bubbled through the salt, and (4) the H,-He mix-

Y41bid., p. 110,

ture was similarly bubbled. The plan was based
on the assumptions that only high-valent fluorides
would volatilize in procedure 1, since gentle he-
lium flow would provide no agitation to suspend
metallic particles in the gas phase; that H, in
procedure 2 would reduce volatile fluorides at the
melt surface and prevent them from leaving the
fuel salt; and that metallic particles should be
suspended in the gas phase by the bubbling gas
in procedures 3 and 4.

Two hot-cell experiments of this type have been
carried out: one with a sample of MSRE salt 35
days old and one on the day after sampling. The
50-g salt sample was contained in a flanged gas-
tight stainless steel reaction vessel (Fig. 10.2)
fitted with a gas inlet tube, through which a dip
tube line could also be inserted, and a large tube
through which the ‘“‘probe’’ tubes were inserted to
with 7, in. of the molten-salt surface. A side tube
from the large tube provided an alternate path for
gas flow. The gas inlet tube and the alternate line
were connected to valves on a simple gas manifold
(Fig. 10.3) carrying two other valves which were
connected to a vacuum pump and to the gas supply
line.

The tank helium was purified by two titanium
sponge traps at 600°C, one near the tank outside
the cell and the other just before entering the gas
manifold. The second titanium trap was added
after the first test showed signs of fuel hydrolysis.
SWAGELOK FITTING

Ya-in.-OD GAS INLET LINE

YT 22T T 2T T 2T T 2T IZZ T ZIZTITZ I ZZ 2T 22T
N AT OO OV TS P O OO

ey

7/
v

v/

)
)

L LMY

ORNL-DWG 68-6080

Y/4-in.-0D "PROBE" TUBE

SWAGELOK UNION, ¥gin. TO Y4 in.

Y4-in.-0D ALTERNATE LINE

SWAGELOK FITTING
SODA-LIME
3/g-in.-OD TUBING

NoF PELLETS
FELT-METAL FILTER

SALT SURFACE
509 MSRE FUEL SALT

REACTION VESSEL

45— TUBE FURNACE

Fig. 10.2. Hot-Cell Test Reaction Vessel.

The traps were not heated when the 5% H,—95%
He mixture was used in order to avoid TiH, forma-
tion. We then depended on the efficient adsorptive
drying action of degassed titanium sponge. The
flow of gas was measured outside the cell with a
low-flow-range rotameter.

To avoid exposure of the fuel salt sample to air
as much as possible, the top of the usual 50-g
sampling capsule was cut off just above the salt
level, and the capsule was placed in the flanged
reaction vessel without fracturing the solidified
lump of salt. A cap was placed on the probe tube
line, and the capsule was evacuated while the re-
action vessel was heated to 200°C. A sheathed
Chromel-Alumel thermocouple wired to the reaction
vessel was used to indicate and control the fur-
nace temperature. During the evacuation, every
joint in the system was leak tested with acetone
and a Hastings gage. When no more leaks could

be found and the gage reading dropped below

10 y, the system was filled to 1 psig with puri-
fied helium, and the cap on the large tube was
slightly opened to provide a helium flush of about
15 cc/min.

The reactor was then heated to 600°C, the cap
was removed, and the probe tube was quickly in-
serted. The Swagelok fitting on the probe tube
had previously been attached at a level such that
the lower end of the probe was within 7, in. of
the molten-salt surface. The only exit for the
helium flow was up through the probe tube, so all
the gas sampled passed within %’4 in. of the molten-
salt surface. Because of the resistance to flow
of the Feltmetal filter in the probe tube (Fig.
10.2), the system pressure usually rose to 2 to 3
psig at a helium flow of 10 to 15 cc/min. In the
first run, with 35-day-old salt, the flow was con-
tinued for 1 hr. In the second run the duration
of flow was decreased to 30 to 40 min.
PRESSURE GAGE
0-30 psia HOT CELL
WALL

103

ORNL-DWG 68-6081

TO HASTINGS
GAGE AND
VACUUM PUMP

l

}g /ALTERNATE LINE
< <

PRESSURE GAGE
0-30 psia

Ti TRAP
600°C

(.~ ROTAMETER
0-100 cc/min

NEEDLE VALVE
="
—><} m He TANK

PRESSURE REGULATOR

o 5% H
2’
\_/ 95 % He TANK

X PROBE TUBE

Ti TRAP

600°C /
GAS INLET LINE

REACTION VESSEL

MSRE SALT —

FURNACE M

7%,

Fig. 10.3. Complete Apparatus for Hot-Cell Tests.

The first probe tube was then removed, and the
large tube was capped loosely. The titanium
traps were cooled, and the 5% H,—95% He mixture
was passed through the system to flush out the
pure helium. The flow of gas was adjusted at
10 to 15 cc/min, and the second probe was in-
serted, run for 1 hr, and removed. With the large
tube capped, a Swagelok union on the gas inlet
line was opened, and the dip line was placed in
position so that its end was about 3.5 in. below
the melt surface. The alternate line was open
during this operation to provide flushing flow over
the salt. With the gas inlet line attached to the
dip line, the cap on the large line was slightly
opened, the alternate line valve was closed, and
a bubbling flow of 5% H,~95% He was started at
10 to 15 cc/min. The third probe was then in-
serted, run for 1 hr, and removed. The gas supply
was converted back to pure helium, the titanium
traps were heated, and a fourth probe was run for
1 hr with helium bubbling at 10 to 15 cc/min. In

the bubbling runs the pressure usually had to be
increased gradually to about 5 psig to maintain
the 10- to 15-cc/min flow rate.

In the first run the pressure in the reactor was
released at the end of the bubbling tests by loos-
ening the Swagelok fitting on the probe. This re-
quired all the gas in the long line to the outside
manifold to bubble rapidly through the dip tube
as the pressure decreased to atmospheric. It is
estimated that the bubbling flow was about 200
cc/min for 2 min at the end of these tests. In the
second run, it was realized that the pressure
across the dip tube could be equalized by opening
the valve on the alternate line. By this means the
pressure was relieved without a sudden violent
bubbling through the dip tube.

Samples of the fuel salt in the reaction vessel
were taken with small ladle-type samplers after
each test with helium flow.

The probe tubes in these tests were 1/“-in.-diam
nickel tubes containing a 1-in.-long empty section
at the bottom, a 7% -in.-thick Feltmetal filter
(100% retention of particles larger than 4 p), and
1-in.-long sections filled with NaF and soda-lime
pellets. In the analytical hot cell, the ends of
each tube were plugged with small rubber stoppers,
and the outside surface was leached with acid un-
til the last leach contained less than 1% of the
gamma activity of the first leach. The tubes were
then cut into a bottom section, a Feltmetal section,
an NaF section, and a soda-lime section. Each
sample was separately dissolved, and the solution
was analyzed for 12 fission products and 23°U.

The analytical results of these tests are given
in Tables 10.4 and 10.5, with all activities cor-
rected back to the time of salt sampling. Gamma
scans of the samples from the first test showed
that the predominant activity was '3'I. This be-
havior was undoubtedly due to fuel salt hydroly-
sis, indicating some water impurity in the reaction
vessel atmosphere. Much less !3!I was found in
the samples from the second test, in which the
leak testing was more thorough and the various
experimental operations were performed more
smoothly with the benefit of experience from the
first run. However, the salt surface appeared
green and scum free after both tests, so the de-
gree of hydrolysis could not have been excessive
in the first test. Since a 35-day-old sample was
used in the first test, the short-lived °°Mo and
132Te activities could not be detected in many
of the samples. Even where values are reported,
they are not reliable.

The results of these tests are very instructive
in spite of the wide scatter in results within each
run and between runs. The following comments
apply to both tests but more specifically to the
second test, whose results showed less scatter.
With gas passing slowly over the quiescent molten
fuel surface, appreciable quantities of all the ma-
jor fission products were found inside the probes,
with the noble metals predominating as in anal-
yses of MSRE cover gas. There was no difference
in volatilization behavior when H, was added to
the sweep gas. If the noble metals had been pres-
ent as high-valent fluorides, they should have been
reduced to nonvolatile states by H,. Thus the
noble metals are probably present in the gas phase
as a colloidal suspension of tiny metallic parti-
cles. The presence of activities such as ®°Sr,
957, 140Ba, 141Ce, and 235U (in the relative
proportions found in fuel salt) inside the probes

104

showed that the fuel salt was also present in the
gas phase as a colloidal mist. Simple vaporization
of fuel salt could not account for either the quan-
tities or the relative proportions of the activities
found in the probes. This observation has dis-
turbing implications regarding the proposed distil-
lation separation of fission products from the bulk
components of fuel salt, since appreciable quan-
tities of fuel salt would volatilize as mist as well
as vapor.

For each nuclide the quantities deposited in the
bottom empty section of the probe and on the filter
section were similar (more so for the second test).
About 10% of these quantities were usually found
in the NaF and soda-lime sections beyond the
filter. The gaseous suspensions of noble metals
and of fuel salt are thus surprisingly stable, and
at least 20% is composed of particles less than
4 p in diameter.

When helium or 5% H,—95% He was bubbled
slowly through the melt (second test), the amounts
of activities found in the various sections of the
probes were very similar to the amounts when the
gases slowly passed over the melt. Thus the gas
phase above the melt appears to be as easily
‘‘saturated’’ with the colloidal suspensions by
the quiescent surface as by slow bubbling. (This
observation is contraindicated by a laboratory
mockup experiment in which the slow bubbling
of helium through molten Li ,BeF , produced visible
droplets of salt on a horizontal metal plate 1 cm
above the salt surface.) However, when the gases
were bubbled rapidly through the melt for 2 min at
about 200 cc/min (first test), the amounts of all
activities in all sections of the probes increased
by factors of 10 to 1000 compared with the slow-
flow cases. Therefore the amount of fume or
mist formation is greatly enhanced by turbulent
gas-fuel contacting.

The readiness with which gaseous suspensions
of noble metals and of fuel salt are formed above
the highly radioactive fuel melt invalidated the
original presumption that only fission products
with volatile fluorides could leave the melt under
the gentle sweep conditions. However, the evi-
dence from the hydrogen gentle sweep runs and
particularly from the hydrogen bubbling run in
the second test is nearly incontrovertible that
volatile fluorides are not involved in a significant
way in causing the noble-metal fission products
to become gas bome. A second line of argument
Table 10.4. First Test of Volatilization of Fission Products from MSRE Fuel

Disintegrations per Minute per Total Sampleb

Flow Conditions Sample® 103, 106, 951 990 1325, 129, 95, 1414, 2%y
(g total)

10—15 cm>/min pure He B 7.11 x 10 4.87 x10° 1.80 x 10° Low Low 1.99 x 10 1.70 x10®  6.72 x 107 6.20
over surface FM 4.30x10% 3.59x10° 1.3 x10° Low Low Low ~1.5 x 105 1.1 x 10° 0.064
NaF  8.14x10* 1.58x10* <2.3x10° Low Low Low  <6.9x10* <1.7x10* 0.086
SL, 6.70 x 10* 1.00 x 10* <5.9 x 10* Low Low Low  <2.2x10* <83x10° 0.027

10-15 cm®/min 5% H, B 3.77 x 107 3.01 x 10° 9.53 x 108 Low Low 7.45 x10% 1.53x10®  1.96 x 108 22.7

over surface FM 4,22 X 105 3.50 x 104 1.6 x 10° Low Low Low 1.3 % 106 1.43 x 10° 0.157
NaF  3.39 x10* 5.76 x10° ~3.7 x10° Low Low Low 5.3x10°  3.73 x 10° 0.071
SL 1.47 x 10*  3.41 x10® ~9.0 x 104 Low Low Low 1.6 x 10°  2.08 x 10° 0.066

10-15 cm®/min 5% H, B 3.85 x10° 4.7 x 10’ 0 Low Low ~1.9x 107 5.8 x10'% 5.18 x10!°  7096.00

bubbling through salt FM 2.45 x10° 1.7 x 108 0 Low 7.8 x 10° 6.2 x 107 2.86 x 101 2.35 x 10!'%  3532.00
NaF 7.5x 107 7.6 x10% 7.15x107 ~1.1x10'! 7.8x10° 4.2x10%° ~1.2x10% 9.00 x 10° 0.143
SL 1.73x 107 ~1.5x 10° 1.67 x 108 ~4.2 x10% ~3.4 x 10° 1.9x10% 3.31x10°  4.36 x 10° 0.149

10—15 cm>/min pure He B 2.14 x10® 1.5 x107 1.72x10'% ~1.2x10'! 5.9x10!° 2.2x107  3.26 x10°  2.70 x 10° 414.00

bubbling through salt  FM 1.52x 10° 1.65x10%® 7.84 x 109 ~1.9x10'? 9.8 x 1010 3.8x107 1.25%x10° 1.17x10'°% 1720.00
NaF  1.28x10® 9.8x10%° 3.53x107 ~2.7x10' ~1.4x10'0 6.0 x10% 8.8x10° 6.24 x 10° 0.297
SL ~6.4 x107 V6.3 x10% 7.73x 107 ~8.5x10'° 5.6 x10° 3.3x10% 2.00x10% 7.75x 10° 0.367

B is bottom empty 1 in. of probe, FM is section including the Feltmetal filter, and NaF and SL are the sections packed with NaF and soda lime

respectively.

PActivities calculated back to fuel sampling time: 10-23-67, 0944 AM.

SOT
Table 10.5. Second Test of Volatilization of Fission Products from MSRE Fuel
. Disintegrations per Minute per Total Sianmple',J Total U
Flow Conditions Sample
IOSRu 106Ru 95Nb 99M° 132Te 129Te 141ce 144ce 952r 131I BQSr 14UBa (yg)
10 cc/min pure He over B ~7.24 x 108 ~2.46 x 10° 9.10x 10°  1.88 x 108  7.38 x 107 2.5x10%  2.07 x 10° 2.58 x 10° 2.59 x 10° 2.75 x 108 3.64 x 108 ~4.6 x 10° 0.06
surface FM 3.21 x 107 6.63 x 107 6.40x 10°  1.09x 102  3.81 x 107 5.4 x 106 5.51 x 106 9.56 x 10° 9.06 x 108 1.50x 10'%  1.20x 10° ~1.1 x 108 0.18
NaF 4.63 x 10 1.86 x 107 <2.8x10° 1.90x 107 244 x10% <4.1x10° 5.47x10° 1.03 x 10° 5.29 x 10° 1.77 x 10° 1.03 x 10° <1.6 x 10° 0.14
SL 2,46 x 10° 1.64 x 107 8.7x10° 2,78x 107  3.53 x 10° 5.5%x 105 7.69x 10% 1.78 x 105 ~8.5x 10° 3.92 x 108 1.34 x 10° 1.7 x 10° 0.04
10 cc/min pure He over B 1.29 x 107 2.03 x 107 2.9%10% 1.94x107 7.51x 107 3.6x10% 1.39x10° 2.77 x 10° 1.27 x 10° 3.47 x 107 2.99 x 10° 2.6 x 10° 0.09
surface FM 2,51 % 107 4.60 x 107 4.8x 10%  3.34x 107 3.13x 107 5.4 x 10%  5.16 x 10° 7.84 x 10° 6.23 x 108 6.29 x 108 1.20 x 108 9.02 x 10°% 0.44
NaF 1.62 x 107 7.23 x 105 ~2.6x10° 2.63x 107 1.73 x 107 2.5%x10%  2.31x 108 3.20 x 10° 7.38 x 10° 1.26 x 10° 1.20 x 107 ~8.7 x 10° 0.58
SL 1.04 x 107 7.94 x 10° 1.4x 10%  1,04x107 1.18x 107 1.3x10%  1.24 x 10° 1.98 x 10° 1.95 x 10° 7.75 x 108 3.50 x 10° ~3.1x 10° 0.27
10 cc/min 5% H, over B 1.78 x 107 3.23 x 108 1.8x 10%° 1,66 x 107  2.64 x 108 6.2x10%  3.40x 10° 4.85 x 10° 1.69 x 10° 8.24 x 107 3.03 x 105 6.8 x 10° 0.16
surface FM 1.37 x 107 5.00 x 10° ~9.6 x 10° 8.68x 10%  3.08 x 107 3.2x10% 5,43 x 108 7.66 x 10° 4.50 x 108 6.43 x 108 6.96 x 10° 1.1 x 106 0.61
NaF 3.56 x 108 1.48 x 107 2.5%x10% 9,90x10% 1.57x 107 2.2%x10°  8.33x 10° 5.66 x 10° 4.18 x 10° 1.98 x 10° 6.70 x 106 1.10 x 10° 0.26
SL 8.39 x 106 3.05 x 10° 1.5x10% 7.98x 1085  7.36 x 10° 1.2x10%  1.23x 108 1.81 x 10° 1.09 x 108 8.51 x 108 2.04 x 10° 2.8 x 10° 0.07
10 cc/min 5% H, bub- B ~1.7 x 107 ~4.9 x 108 0 5.29 x 107 2.35x 107 2.3x10%  2.47x 108 1.60 x 108 1.36 x 108 1.71 x 10° 3.36 x 107 3.15 x 108 29.9
bling through salt FM 1.47 x 108 1.06 x 107 3.2x 107  1.06 x 10° 6,07 x 107 4.1x10% 5,26 x 107 1.71 x 107 5.29 x 10° 3.13 x 10° 2.54 x 107 2.95 x 107 2.77
NaF 5.23 x 108 1.07 x 108 1.3x10%  4.44x10° 2.94x10% ~8.1x10%5 1.26x 106 2.00 x 10° 1.11 x 106 7.03 x 10° 1.71 x 10° ~8.1x 105 0.28
SL 7.55 x 10° 2.16 x 108 1.76 x 105 3.49x 107  3.88 x 106 7.9% 105  1.07 x 10° 1.79 x 10° 7.11 x 10° 3.30x 108 2.11 x 10° ~3.0x 105 0.05
10 cc/min pure He B 6.59 % 10° 2.24 x 108 4.06 x 107 8.32x10° 8.02x107 ~2.9x10% 5.64x10° 7.98 x 10° 2.46 x 10 1.48 x 108 3.07 x 10° ~g.1 x 10° 0.26
bubbling through salt FM 6.85 x 107 1.09x 107  ~540x 107 1.22x10%  2.99 x 107 8.8 x10% 7.22x10° 9.76 x 10% 4,64 x 10° 3.83x10'%  1.71 x 107 1.81 x 108 0.75
NaF 6.80 x 10° 1.59 x 108 3.12x10%  1.14x107  3.50x 10° 1.73 x 10° 5.87 x 10% 1.44 x 108 6.52 x 10° 2.42 x 10° 3.1 x 10° 0.20
SL 8.18 x 10° 1.71 x 107 1.82x10°  5.05x10° 1.47x10%° <7.9x105 1.42x10° 2.53 x 10° 8.29 x 10° 1.91 x 10° 1.42 x 108 3.2 x 10° 0.29
Fuel salt, dis min~! g~} $7.5x10%  <£6.8x10° ~10° 1.18x 10° 1,71 x 108 8.64 x 10! 7.30x 101° 1.22x 10t 3.11x10'°  6.82x10%° 1.55x 10!

%B is bottom empty 1 in. of probe, FM is section including the Feltmetal filter, NaF and SL are the sections packed with NaF and soda-lime respectively.

bActivities calculated back to fuel sampling time: 12-6-67, 8:45 AM.

901
to the same effect may be based on the demon-
strated readiness of fuel salt particles to become
suspended in the gas phase. If fuel salt so easily
becomes gas borne, the same should be true of
noble-metal particles in the fuel. To account for
the preferential volatilization of noble metals, it
may be necessary to postulate that the colloidal

particles are concentrated at the salt-gas interface.

Although the hot-cell tests provide strong evi-
dence as to the nature of the gas-borne activities,
the question of mechanism remains unanswered.
The formation of gaseous colloids from the qui-
escent molten surface may not be explained by
normal physical processes. One possibility is
that recoils from beta emissions near the surface
of the molten salt may cause the ejection of tiny
particles of fuel salt and of noble metals into the
adjacent gas phase. Another suggestion!? is that
differences in thermodynamic contact potentials
between metals, salt, and the gas phase would
tend to eject metal particles from the salt phase
into the gas phase. It has also been suggested!®
that the bursting of very tiny gas bubbles, per-
haps formed by radioactive decay, might cause
aerosol formation.

Third Hot-Cell Test. — A third hot-cell test was
designed to (1) confirm previous results using sim-
ilar probes, (2) determine whether the suspended
metal or fuel particles in the gas phase were elec-
trically charged, (3) attempt to determine the size
of the particulates in the gas phase by examining
with an electron microscope a filter through which
the exit gas was passed, (4) test the diffusion be-
havior of the gaseous activities by observing the
distribution of activities deposited on the inside
of a 1/4-yln.-diam vertical tube whose open end was
near the fuel surface and whose top end was
closed, and (5) obtain an indication of particle
size distribution in the gas above molten fuel by
passing the gas through a long copper tube and
measuring the distribution of activities down the
length of the tube. It was also intended to carry
out several of these tests at intervals over the
course of a month to determine the effect of ac-
tivity decay on the volatilization process. How-
ever, the furnace used to heat the reaction vessel

151 etter from Jere Nichols to F. L. Culler, Jan, 8,
1968.

16personal communication from J. Braunstein, Re-
actor Chemistry Division.

107

burned out after the completion of the first se-
quence of these experiments, ruining the reaction
vessel in the process.

Samples from these five runs were delivered to
the analytical laboratory, and results are not yet
available.

Fourth Hot-Cell Test. — The burned-out furnace
in the hot cell was replaced, and a fourth hot-cell
test was started with a fresh 50-g sample of MSRE
fuel salt, FP14-69. Because of experimental dif-
ficulties and an intervening weekend, the first two
runs of this test were carried out five days after
the salt was sampled. The sample was kept at
200°C during this period to avoid radiolysis.

The first run contained a probe tube of the usual
type and two electrodes for measurement of the
charge on the particulate matter in the gas phase
above the molten fuel. This test will not be de-
scribed in detail since analytical results are not
yet available.

In the second run the test assembly was a stain-

less steel rod to which were attached six 1/8-ir1.-
diam copper screens spaced at levels 1/2, 1, 1.5,
2.5, 3.5, and 4.5 in. from the molten-salt level.
The copper screens were of the type used to hold
electron microscope specimens and had previously
been coated with a thin (600-A) vaporized layer of
carbon. The gas exiting from the reactor passed
at 15 cc/min for 40 min through a 1/2-in.-diam. tube
which surrounded the rod holding the screens.
With this experimental arrangement the exiting gas
traversed the screens at a linear velocity of about
0.5 cm/sec, with good opportunity for the particu-
late matter to deposit on the screens.

The 1/2—ir1.—diam tube and the stainless steel rod
were then removed from the reactor, and the six
copper screens were carefully loosened from the
rod and dropped into small plastic bottles. Through
the plastic bottle, no sample read more than 300
mr/hr of gamma activity, so that the screens could
be examined in the electron microscope in Building
3019. Gamma scans of the five lower samples
showed high !3!1 activities (~108® dis/min per
sample) which masked the probable presence of
other gamma activities. The !3!I contamination
of the topmost sixth sample was lower (4.6 x 10°
dis/min), so that the presence of °°Mo (3.1 x 10°
dis/min), 1°3Ru (1.7 x 10° dis/min), ?5Nb (2.8 x
10° dis/min), and '*°La (5.9 x 10* dis/min) could
be detected. These activity readings were taken
7.0 days after the salt was sampled. The high
108

1311 activities on most of the samples are again
ascribed to fuel sample hydrolysis by traces of
water in the reactor. The degree of hydrolysis
was probably very slight, since special care was
exercised to make the system leak tight and since
the system was evacuated to a pressure below

10 p with the fuel salt at 200°C before the runs
were started.

At this writing, the three topmost copper screens
(at 2.5, 3.5, and 4.5 in. above the molten salt) have
been examined in the electron microscope. A siz-
able quantity of particulate matter was observed
on the screens, often covering several percent of
the areas viewed. There appeared to be three
discrete particle size ranges: very fine particles
35 to 180 A in diameter, medium-size particles
1000 to 2000 A in diameter, and occasional large
particles more than ten times larger than the
medium-size particles (Fig. 10.4). Most of the
area of the deposited material was represented by
the fine-size particles.

A curious circular pattern of particles was often
seen, particularly on the two topmost screens
(Fig. 10.5). An exact circle was outlined by
medium-size particles. The interior of the circle
was filled with randomly deposited medium and
fine particles. The area density of the fine par-
ticles was the same inside and outside the circles,
but the density of medium-size particles was at
least ten times as great inside than outside the
circles.

At high magnification, it can be seen that many
of the medium-size particles are transparent (Fig.
10.6). The small particles can be seen inside
the outline of the medium-size spots. It is pos-
sible that the fine particles are deposited on top
of the larger particles, but most of the larger spots
are pale gray compared with the darker color of
the small particles. Also a faint halo is visible
around many of the larger spots.

Most of the medium-size spots appear to be
round. Several of hexagonal shape can be dis-
cerned. Figure 10.4 shows a number of cubical
shapes. Some small- and medium-size triangles
(usually very dark) are seen in Fig. 10.6. A dark
triangle is often visible at the edge of a large
pale spot.

These observations suggest that the medium-
size spots, particularly those in the circular pat-
terns, are thin films rather than thick three-dimen-
sional particles. Microscopists who have looked
at the circular patterns say they are very similar

to patterns obtained when small droplets of solu-
tion evaporate on a flat surface. It is possible
that MSRE fuel salt particles on the screens may
pick up atmospheric moisture over the course of
several days exposure to air before examination.
The most abundant phase in fuel salt, Li BeF ,
is known to be deliquescent. The solution so
formed could spread over the carbon film to a thin
layer. When this film is evacuated and heated by
the electron beam, the water could evaporate and
form the characteristic circular patterns. Close
examination of Fig. 10.6 shows that many of the
smaller spots are also pale in color. These may
represent only partial solution. The darker spots
with no pale areas may be materials which do not
dissolve.

An electron diffraction pattern (Fig. 10.7) is
shown of the material in the area in the light
square of Fig. 10.8. This type of pattern is char-
acteristic of salts (like the fuel salt) but is not
the type of pattern obtained from graphite or metals.
An attempt is being made to identify positively the
material in diffraction patterns of a number of dif-
ferent areas on the screens.

The remaining three samples will be examined
with the electron microscope, and care will be
taken with future samples to avoid exposing them
to atmospheric moisture.

The observations reported are of considerable
significance, since they represent the first hard
direct evidence that colloidal particles are to be
found in gas sweeping slowly by the surface of
MSRE fuel salt. The diffraction pattern identifies
part of the volatilizing material as a salt, probably
fuel salt. In addition, we now have some idea of
the particle sizes we are dealing with. Most of
them are extremely small — of a size which could
well be ejected into the gas phase by beta recoil.

It will be desirable to expose some of the elec-
tron microscope screens in the MSRE pump bowl
gas space and in the access tube as soon as pos-
sible. Plans have already been made to expose
some in the MSRE off-gas line.

10.1.6 Miscellaneous Tests

Two tests which do not fit into previous cate-
gories will be reported below.

Surface Salt from the MSRE Pump Bowl. — A
number of observations in previous tests, partic-
ularly the finding of high concentrations of noble-
metal fission products in the gas phase of the
109

PEM-137-H-5

Fig. 10.4. Electron Micrograph of Particles in Gas Flowing over MSRE Salt. 88,000x.

110

- EM—139-—H-6

Fig. 10.5. Circular Patterns Frequently Found on Screens in Gas Above MSRE Salt. 32,000x.

134

Fig. 10.6. Electron Micrograph Showing the Transparency of Some Particles. 96,000x

10.7. Electron Diffraction Pattern of Particulate Matter Above MSRE Salt.

Fig.

PEM-200-8967

Fig. 10.8. Area of Deposit Giving the Diffraction Pattern of Fig. 10.7.

MSRE pump bowl, suggest that the noble-metal
fission products may be highly concentrated at
the fuel-gas interface and may even be incorpo-
rated in any surface scum or foam floating on the
salt. There is basic and practical interest in
finding an answer to this question. Considerable
thought has been given to the experimental prob-
lem of sampling the fuel surface itself. The two
major problems are contamination of the sample
by the gas phase (or by deposition from the fuel
phase) and dilution of the sample by salt from
below the surface. Several rather elaborate de-
vices have been designed but have not yet been
tried.

By accident, we may have obtained a good sur-
face salt sample by a method which would be dif-
ficult to duplicate intentionally. Sample FP14-60
was to have been a 50-g sample for the third hot-
cell test. It was taken in the usual large nickel
capsule with windows near the top to admit the
salt. When the sample arrived at the hot cell,
it was found that the capsule contained only 5 g
of salt. Although the normal 50-g sampling pro-
cedure was used, it appears that the bottom of the
capsule windows must have been exactly at the
fuel salt level. If it had been higher, no salt
would have entered the capsule; if it had been
lower, the capsule would have filled completely.
It is likely that the ripples in the fuel surface
occasionally lapped a little salt into the capsule.

Unfortunately the capsule must have been con-
taminated inside and out with noble metals from
the gas phase, and on the outside by deposition
from the fuel melt. Since the sample was there-
fore not an ideal one, it was decided to handle it
by the procedure used for ladled 10-g samples and
to compare the results with those from ladled 10-g
samples. Large differences in activities would be
apparent.

The 50-g capsule was therefore cut off about 1
in. above the bottom, and the salt was removed by
shaking in the Wig-L-Bug (as for ladled 10-g sam-
ples). Salt samples were then weighed out and
analyzed for the usual radioactive nuclides and
for Ni, Cr, and Fe. Radiochemical analyses in
disintegrations per minute per gram, corrected
back to the time of sampling, were 1.49 x 10'1
for °9Mo, 4.31 x 101° for 132Te, 9.04 x 107 for
103Ry, 4.07 x 102 for 1%%Ru, ~4.7 x 102 for ®5Nb,
1.40 x 10! for °5Zr, 8.77 x 101° for #9Sr, 1.09 x
101! for '%1Ce, 9.26 x 101° for 1*%Ce, 8.78 x 101°

114

for 1*%Ba, and 5.96 x 101° for '3!I. When these
figures are compared with the values in Table 9.2
of the previous report, 17 the value for **Mo is
higher than all except one, !32Te is higher than
the average by a factor of 3, '%3Ru is less than
two and greater than four values, '°°Ru is higher
than all values by an average factor of 2, ?°Nb

is less than all values by an average factor of
0.2, and the remaining salt-seeking species have
similar values. The results of analyses of Ni, Cr,
and Fe were not excessively high: 152, 90, and
159 ppm respectively.

These results indicate no large concentration
effect for noble metals or metals of construction
at the fuel surface. The data for *°Nb in fact
indicate a depletion effect. These indications
should be regarded as suggestive rather than con-
clusive, since it has not been proven that the
sample analyzed was a surface salt sample and
since the comparative method used assumes with-
out proof that the different sample containers were
contaminated to a similar degree by noble metals
in the gas phase and in the fuel phase.

Deposition of Noble Metals on Nickel from Cover
Gas and from Fuel Salt. — In a number of previous
pump bowl tests, the deposition of noble metals
on the stainless steel cables used to suspend fuel
sampling capsules was examined to determine
whether deposition was greater at the fuel-gas
interface than in the cover gas or in the fuel.
While there was much scatter in these data, the
deposition on the interface region was usually
higher (by a small factor) than on the higher or
lower regions. In an earlier section, it was pointed
out that this type of observation was made uncer-
tain by the contamination problem.

In a special test with an unrelated objective,
FP14-55, a nickel rod inside a perforated nickel
basket was lowered into the pump bowl so that the
middle of the rod was at the level of the fuel salt
surface. This provided a sample of metal exposed
to the interface region in a container which of-
fered some protection against contamination by
handling.

The nickel rod was cut into a gas-phase region,
an interface region, and a fuel-phase region. The
position of the gas-fuel interface was determined
by examination of the rod with a low-power hot-

17MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 120,
cell microscope. The activity on each section
was leached off, and the leaches were analyzed
radiochemically. The gas-phase and interface
regions showed very similar depositions of noble
metals. The deposition of noble metals on the
liquid-phase region was higher by factors of 30
for °*Mo, *°3Ru, and '°%Ru and 5 for !32Te and
95Nb. Since the deposition was so disparate be-
tween gas~ and fuel-exposed regions, the deposi-
tion on the interface sample should have been
heavily weighted by the portion of this sample
which was submerged in fuel. The results suggest
that the visual determination of the fuel-gas inter-
face position on the rod was at fault. The dif-
ference in surface appearance (drop-shaped water-
marks and a dark deposit) that was taken to
indicate the salt level may actually have repre-
sented a region well above the salt level. In this
case, the results given represent merely the dif-
ference in deposition from the gas phase and from
the fuel phase. They are markedly different from
the previous (questionable) results on stainless
steel cables, where similar deposition of noble
metals from both phases was observed.

10.2 FISSION PRODUCT DISTRIBUTION IN AN
MSRE GRAPHITE SURVEILLANCE SPECIMEN

D. R. Cuneo
H. E. Robertson

F. Dyer
L. Bate

In a previous report, '8 determinations of fission
product distribution in MSRE surveillance speci-
mens after 24,000 Mwhr were reported. Samples
were obtained by milling off layers of the graphite
surfaces in a plane parallel to the longitudinal
axis of a 0.47 by 0.66 in. cross section bar. The
specimens were about 4.5 in. long. The resulting
powder was dissolved and analyzed radiochemi-
cally for selected fission products. Near the
surface, layers as thin as 1 mil were milled off as
samples; subsequent samples at greater depths
were as thick as 10 mils. Generally, a total
depth of about 50 mils was sampled from the four
sides of each specimen. The following inherent
uncertainties in this sampling procedure were
recognized: (1) some contamination was certainly
carried by the steel milling surfaces to the next

185, s. Kirslis and F. F. Blankenship, MSR Program

115

Semiann. Progr. Rept. Aug. 31, 1967, ORNL~4191, p. 121.

sample from the preceding (and hotter) sample;
(2) with the milling apparatus used, it is possible
that the amount of graphite removed per cut was
not uniform over the length of the specimen; and
(3) it was difficult to be sure that the amount of
graphite removed was uniform over the width of
the specimen.

In order to determine if the ‘“tails’’ or concen-
trations of fission products observed at consider-
able depth in the specimens were real, the follow-
ing sampling scheme was devised. The rectangu-
lar-shaped specimen was sawed longitudinally at
midplane. A small (0.25- to 0.30-in.-diam) core
was then drilled from the cold inner (new) surface
to the outer (hot) surface. The hot surface of this
small graphite core was glued to a cold graphite
coupon, which in turn was glued to a precision-
ground tool steel piston. The piston fits into a
3.5-in.-diam holder. This assembly is shown in
Fig. 10.9. A dial-indicating micrometer allows
reading changes in the piston vertical position to
direct readings of 0.1 mil. The samples are ob-
tained by describing a figure 8 motion on a clean
surface of emery paper resting on a precision
lapping plate. The resulting powder is taped in
place on the emery paper. This procedure, first
devised by Lonsdale and Graves,'? has been de-

194, K. Lonsdale and J. N. Graves, *Diffusion of
Thorium in Pyrolytic Carbon Coatings,’” pp. 18=35 in
Coated Particle Fuels Research at General Atomic,
Oct. 30, 1964—April 30, 1965.

GRAPHITE
0

Fig. 10.9. Apparatus Used for Sampling of Graphite
Core. View on right shows location of graphite core
cemented to floating piston. View on left shows entire

assembly.
116

scribed in detail elsewhere,?? We believe that
use of this apparatus certainly eliminates the un-
certainties described above for the milling opera-
tion type of samples. The weak point in this pro-
cedure for these surveillance specimens is that
only a small portion of the specimens is sampled
when compared with milling samples.

The taped-on samples are transferred to a gamma-
ray spectrometer for determinations of fission
product concentrations. This approach to the prob-
lem was begun at a time when the surveillance
samples had cooled beyond determination of short-
lived isotopes of Mo, Te, ], and Ba. Itis felt
that these short-lived isotopes can also be de-
termined satisfactorily by counting the entire
powdered sample, as we have done with the long-
cooled samples, for longer-lived nuclides.

The first core-drill sample was taken from the
portion of specimen Y-7 from which some 60 mils
of the surface under consideration had been re-
moved earlier by milling, 28 Small samples (1 to
10 mils) were removed from this core by grinding,
starting at the cold (midplane of the specimen)
end. These samples were counted using an Nal
crystal that had only sufficient resolution to allow
good counting data for °5Zr, 137Cs, and !**Ce.
(For subsequent samples a germanium diode was
available, thus allowing greater resolution and
determination of more nuclides.) Part of the data
obtained from these grinding samples are reported
in Table 10.6. The last entry in the table, 70 mils
from the original sutface, was obtained from a
sample taken 10 mils deep from the surface left
following the milling. From this depth (10 mils)
on to the ““new’’ surface, the concentrations of
the three nuclides dropped, as much as an order
of magnitude. We have no explanation for this
anomaly. However, it is clear that the tails ob-
served from the milling-type samples'® are real
since we see measurable concentrations extending
to midplane in Table 10.6.

The data in Table 10.6 show little change in
the '37Cs concentration with depth, This nuclide
has a 3.9-min gaseous precursor (1*7Xe), which
allows considerable time for diffusion in the
gaseous state.

2OR. B. Evans [II et al., **Recoil of Fission Products
in Pyrolytic Carbon,’® Gas-Cooled Reactor Program
Semiann. Progr. Rept. Sept. 30, 1965, ORNL~3885,
pp. 131—409.

Table 10,6. Fission Product Concentrations Varying
with Depth in Specimen Y-7 as Determined

from First Grinding of Specimen

Depth from Atoms of Nuclide per Cubic
Original Centimeter of Graphite
Surface®

(mils) 57; 137¢s 144ce
235 ax10t!  sx10'® 2.5x10!2
{midplane)
200 1x10'?  6x10!? 4% 1012
150 6x 10?2  7x10l4 2 x 1012
125 1x10'%  gx10t? 5x 1012
100 2.5x10'%  1x10!3 5x 1012
90 2x10'%  1x10'%  6.5x10!?
80 2.5 x 10 1x10!% g x 1012
70 2x 1013 1x 1013 2x 1033

aOriginal surface is that before cuts were removed by
milling operation.

To check the possibility that zirconium was pro-
duced in situ from uranium contained in the
graphite before its exposure to the salt, a corre-
sponding piece of unirradiated material was sub-
jected to activation analysis. Analysis showed
the natural uranium content of the specimen to
be <0.06 ppm, whereas about 70 ppm would be
required to yield the amount of ®3Zr found at
about 100 to 125 mils from the original surface.

Next, a core (305 mils in diameter) was drilled
through a portion of the Y-7 sample which had
not been milled. Initially, 251 mils were removed
from the 467-mii-thick specimen in the following
increments: one 2Z-mil sample, nine 1 mil each,
five 5 mils each, five 10 mils each, one 15 mils,
two 25 mils each, and two 50 mils each. At a
later date, the remaining 216 mils were sampled
by grinding ten increments 1 mil each, five 5 mils
each, five 10 mils each, three 25 mils each, and
one 56-mil portion. Partial results of the gamma-
spectrometer counting data are shown in Figs.
10.10 and 10.11. We believe that the graphite
core sutface represented by the left ordinate in
each figure is that surface which was exposed
directly to free-flowing fuel salt in the reactor
and that the surface represented by the right
117

19 ORNL-DWG 68—-6082
10 = R {'— R — 7‘*7’ - —][ I ;"77 I — [ A 7%;.77f7 T
——— S S e — =
st —|— — o | T
— —1— — e —_— T e —— .—‘L. {
2 — _ — — R — — —_— _ —

124046 I — - S
E — ] f——— L+ ’\ ’fi }_ - li )
§ | S 1 S O
£ | |
| _ _ 7
[ o — :QI: A e

\/ R — ML ,_;) T i /_,.___ t‘d'
| 95Nb i I 95, / ;"'v/ /-

N A
i L = A R

20 30 40 50 60 70 80 70 0 50 40 30 20 10 0
DISTANCE FROM SURFACE (mils) DISTANCE FROM SURFACE (mils)

Fig. 10.10. Concentrations of Nuclides vs Depth in Graphite Starting at Surface in Contact with Free-Flowing
Fuel Salt (a) and with Stagnant Fuel Salt (b).
118

ORNL-DOWG 68-6083

DISTANCE FROM SURFACE (mils)

10
5
] . 137cs .
R L n '4ce
f—" o 1440,
-
1016 10
13755 1
- .
5 L] —a -l
o - - ]
\ 137, L
Y 141ce v
2 AL R __'L/ 1 Az N 2‘;‘/‘/
Y - -l ——— o Y W —Ae A
15 \ " AT - ‘ 1 ‘ —A _, — 7 ] s
® {0 i . sy iy rY : - P
S 5 PN 141 —
5 o\e AN __~Ce /
5 [ NS
" . .
e o\ e A
> \ A I
§ ta & 144, — & J
° 10 \\ o — 7 L
N o . I |
1
\(\ ——
—e— 1
° TN < 144 .l
—— F\ ——— —— ce I
/
10>
V4
/
5 _J._
7[
/
2 4
02 L2 b
0 10 20 30 40 50 60 70 80 70 60 50 40 30 20 {0

DISTANCE FROM SURFACE (miis)

Fig. 10.11. Concentrations of Nuclides vs Depth in Graphite Starting at Surface in Contact with Free-Flowing

Fuel Salt (a) and with Stagnant Fuel Salt (b).

ordinates was exposed only to stagnant fuel salt
which found its way through slits available where
adjacent pieces of graphite were stacked together.
However, this portion of Y-7 was left unmarked
when stored since it was intended for permeability
studies. From these curves and data not shown
representing sampling to the midplane of the core,
we observe the following. In Fig. 10.10, the °>Nb
follows a similar pattem from one surface to the
other, ending at the back (or stagnant) surface at
a greater concentration by a factor of about 15 to
20. The °3Zr concentration gradient is consider-
ably steeper on the back side of the specimen,
with no detectable amount between 15 and 85 mils.

Ruthenium-103 was found from the ¢‘hot’’ salt
surface to midplane, but only to a depth of 25 mils
from the opposite surface. Ruthenium-106 fol-
lowed the shape of the 1°3Ru curve to about 50
mils and then showed a drop at 65 to 70 mils;
thereafter it was below a value of 1 x 1012, Start-
ing from the opposite surface (Fig. 10.105) it
dropped below limits of detection at 15 mils. The
103,106Ry values are similar, mil by mil, when

we consider that the fission yield of 193Ru is
0.38% and that of 1°®Ru is 2.9%. Cerium-141 has
an 18-min barium and a 3.7-hr lanthanum precursor,
while !*44Ce has only short-lived ones. These
different origins of the two isotopes may help
explain the marked differences in penetration in
graphite as shown in Fig. 10.11; the melting point
of barium is 725°C?! and operating temperature of
the graphite is believed to be 650 to 750°C. If
the ordinates in Fig. 10.10b and 10.115 represent
a graphite surface exposed to stagnant fuel salt,
then the higher surface concentrations and lesser
depth of penetration for several of the nuclides
may have occurred as follows. If this surface had
contact only with a limited amount of salt for a
long period of reactor operation, then penetration
by any nuclide observed, except '*7Cs, which
probably penetrated from the opposite surface,
would be limited by fission products available.
However, at the end of reactor operation, barren
salt circulated through the reactor for a very short
time (relatively) did not have sufficient contacting
time with this surface to remove fission products
as thoroughly as at the opposite free-flowing salt
surface.

In Table 10.7 are given details of the distribu-
tion of seven nuclides starting from the graphite
surface in contact with free-flowing salt and ex-
tending to a depth where ~100% of all nuclides
have been found, These data present in tabular
form the data represented graphically in Figs.
10.10a and 10.11a. Since these data and those
reported previously by Kirslis'® were obtained by
quite different techniques and analytical proce-
dures, it is interesting to compare the results, as
seen in Table 10.8. While there are apparent dif-
ferences in results from the two methods, it ap-
pears likely that the method we have described,
that is, grinding of a small graphite core and dry
graphite powder gamma spectrometry, yields usable
results. The small core samples, of course, have
the distinct possibility of including a crack,
which could markedly affect the results.

In Table 10.9 the data for the total atoms of in-
dividual nuclides, surface to midplane, are given
for the second sampling (representing the first
half of this graphite core) and the third sampling
(representing the second half). We see that with
the exception of the large difference for ?°Nb, the
total nuclide contents of the opposite halves of
the core are very nearly the same. This is some-
what surprising in view of some of the large dif-
ferences noted for nuclide concentrations per unit

21Handbook of Chemistry and Physics, 47th ed., by
R. C. Weast, Chemical Rubber Co., 1966.

119

of volume that we found by comparing data in the
left and right halves of Figs. 10.10 and 10.11.
However, in most cases, 90 to 95% of each nuclide
was found in the first 10 mils from the graphite
surfaces.

Another (the third) core was drilled from speci-
men Y-7. It was sampled by grinding throughout
its entire length of 470 mils. Table 10.10 shows
a comparison of total atoms of each nuclide per
square centimeter of graphite surface from one
surface through to the opposite surface for the
second and third cores. Reasonable agreement is
found for six of the seven nuclides; the discrepancy
for 141Ce between the two comes about because
of low concentrations found in the first 10 mils
of the third core. Table 10.11 gives comparisons
of concentration profiles for seven nuclides
through the second and third cores.

By the hand grinding of small cores from graphite
surveillance specimens, we can expect to obtain
representative results for the distribution of fission
products in the specimen. Short-cooled specimens
of graphite from the MSRE may prove too radio-~
active for hand grinding, and methods of perform-
ing this operation remotely are being investigated.
The attractive features of this method include
savings by direct gamma spectrometry of the
graphite powder rather than dissolution of the
much larger milling samples and the necessary
radiochemical separations of each nuclide. The
grinding procedure has also allowed determinations
of Li* and F~ on each freshly exposed surface of
the core.

10.3 PROTON REACTION ANALYSIS FOR
LITHIUM AND FLUORINE IN MSR GRAPHITE

R. L. Macklin E. Ricci
J. H. Gibbons T. H. Handley
D. Cuneo

Consideration of the proton-induced reactions
"Li(p,n) and '°F(p,ay) suggested that they might
be used to measure the concentration of these
target nuclides in graphite as a means to determine
the extent to which MSR fuel salt components
penetrate into the graphite under irradiation. Here-
tofore, none of the standard analytical techniques,
including neutron activation, has shown much
promise for this application at the few ppm (1074
wt %) level.
Table 10.7. Distribution of Fission Products in MSRE Sample Y-7 as Found by Second Core Grinding Starting at Graphite Surface
in Contact with Free-Flowing Salt and Extending to a Depth of 201 Mils

Total
Distance Percent Percent Percent Percent Percent Percent Percent
from of Total Cumulative of Total Cumulative of Total Cumulative of Total Cumulative of Total Cumulative of Total Cumulative of Total Cumulative
Surface  Zrin 7% ?57r SNbin % ?°Nb  !®Ruin % !%%Ru  !%6Ruin % '%®rRu  137csin % !137cs  lcein % !%lce %%Cein % !*4Ce
(mils) This Cut This Cut This Cut This Cut This Cut This Cut This Cut
02 73.9 73.9 80.3 80.3 82.2 82.2 85.3 85.3 12.5 12.5 38.3 38.3 75.7 75.7
3 8.8 82.7 10.2 90,5 9.7 91.9 4.3 89.6 1.9 14.4 15.8 54.1 8.7 84.4
4 2.6 85.3 1.6 92.1 1.3 93.2 1.4 91.0 0.2 14.6 3.6 57.7 0.9 85.3
5 1.1 86.4 1.3 93.4 1.2 94 .4 1.5 92.5 0.7 15.3 5.1 62.8 1.0 86.3
6 1.1 87.5 1.1 94.5 0.9 95.3 1.2 93.7 0.6 15.9 3.3 66,1 0.8 87.1
7 0.5 88.0 0.5 95.0 0.4 95.7 0.6 94.3 0.5 16.4 2.7 68.8 0.5 87.6
8 0.8 88.8 0.7 95.7 0.5 96.2 0.6 94.9 0.5 16.9 1.8 70.6 0.6 88.2
9 0.3 80.1 0.4 96.1 0.3 96.5 0.6 95.5 0.5 17.4 1.4 72.0 0.3 88.5
10 2.0 91.1 0.2 96.3 0.1 96.6 0.3 95.8 0.4 17.8 1.0 73.0 0.2 88.7
11 0.3 91.4 0.2 96.5 0.2 096.8 0.3 G65.1 0.5 18.3 1.7 74.7 0.3 89.0
46 95.7 98.2 98.4 99.1 31.0 88.2 91.9
126 98.5 99.4 99.2 100+ 58.7 95.4 94,5
151 98.7 100+ 99.4 100+ 66.9 96.5 94.9
201 99,3 100+ 99.6 100+ 87.6 08.7 96.9

0zl
121

Table 10.8. Comparison of Results for Nuclide Concentrations in MSRE Graphite Surveillance Specimen Y-7
as Obtained by Milling and Grinding Methods

Concentrations are atoms of nuclide per cubic centimeter of graphite

Sample Depth from Surface (mils)

Nuclide Method?

2 5 10 25 30 40
95Nb AP 1.3 x 107 2.7 x 1013 1.3 x 1013 1.8 x 1014
B 1x 1010 1.8 x 1015 3% 104 6 x 1013
957, A 2.3%x 105 1.4 x 1014 3.9% 103
B 6 x 1015 8.5 x 1013 3.5x 10!3
1030y A 3% 1010 2.3 x 104 1.1 x 104 4.6 x 1013
B 2% 106 1.3 x 10%° 3x 104 4% 1013
106y A 1.4 x 1015 1.8 x 104 7.4 x 1013 2.8 % 1013
B 6x 1015 5% 10%% 1.3 x 1014 2x10%3
137¢s AP 1.7 x 1016 1.3x 103 6% 1014 1.3x 1013
B 1.2 x 1016 2.2 x 1013 1.5 x 1015 1x 103
141ce A 3.8 x 1015 2.5%10%% 1.5 x 1015 4 x 1014
B 8 x 1015 2.2 x 1013 7.5 x 1014 2.1 x 104
144ce A 7.7 x 1015 1.5% 10%5 6.1 x 1014
B 2x 105 1 x104 3.5% 1013

“Method A: samples milled from 0.66-in.-wide by 41/-in.~long specimen in May 1967; samples were dissolved and
nuclides determined radiochemically. Method B: Samplés ground from surface of 0.3-in.~diam core of graphite in
December 1967; dry powder samples were analvzed by gamma spectrometer using a germanium diode.

bSpecimen different but similar to Y7.

Table 10.9. Total Atoms of Nuclides per Square
Centimeter Found by Core-Drill Sampling from Surface
to Midplane (Second Sampling) and Midplane to
Opposite Surface (Third Sampling)

Second, Third,
Nuclide Hot Salt Surface  Stagnant Salt Surface
x 10717) (x 10™17)
95Nb 3.47 72,
955, 0.73 0.75
103p, 3.35 4.43
106 1.0 1.17
1374 7.75 10.1
1414, 1.6 2.13

144, 1.14 1.22

122

Table 10,10. Total Atoms of Nuclide per Square Centimeter from One Surface to Opposite Surface

for Two Cores Removed from MSRE Surveillonce Specimen Y.7

95Nb 9521‘ 103Ru 106Ru 137Ru 14 ICe 144Ce
Total atoms per
square centimeter
Second core 1.5x 107  7.5x10'® 7.8x10'7 2.2x10'7 1.8x10'® 3.9x10'7 23x10!7
Third core 9.0x10'%  6.0x10'% 7.3x10'7 s51x10!7 1.1x10'® osx10'® 1.7x10'7
Second: third 1.7:1 1.25:1 1.1:1 1.:2.3 1.:6:1 4.1:1 1.35:1
core ratio
Table 10.11. Comparison of Fission Product Concentration Profiles for Two Samples Core Drilled
from MSRE Graphite Surveillance Specimen Y-7
%5Nb Very similar concentration profile in both cores and found in every sample to midplane from all surfaces
957, Profile curves similar in shape for both cores, with slightly steeper slopes for third core curves. No
detectable amount beyond 150 mils from first surface of third core; found all the way from first surface
of second core. No detectable amount beyond 4 mils from second surface of third core; some quantity
at 100 mils from second surface of second core
10320 Second core: found to persist to midplane (233 mils) from first surface and to depth of 25 mils from
opposite surface. Third core: much steeper drop; 7 X 1013 atoms/cc at 5 mils from first surface vs
1.5 % 10!3 for 5emil depth in second core. Disappeared at 10 mils from both surfaces of third core
106p4 Similar behavior in both cores: found to about midplane from first surface of both cores and no deeper
than 15 to 20 mils from opposite surface of both cores
137Cs Similar concentration profiles for both cores, with slightly higher values for second core from 10 mils to
midplane (233 mils)
1"'ICe From 10 mils to midplane of both cores, concentrations are quite similar; however, for third core there
is an unexplained flatness (and even lack of detection in some samples) from 10 mils to the surfaces
144ce Second core: found in every sample, first surface to midplane (233 mils), with increase in concentration
from 100 mils to midplane; not found deeper than 15 mils from opposite surface. Third core; not de-
tectable below 15 to 20 mils from either surface
For 7Li the high neutron yield and relatively low  ¢m (~0.33 mil) and produce no “Li(p,n) reaction
threshold of the 7Li(p,n) reaction (1.881 Mev) offer neutrons at greater depths. The major neutron back-
a unique identification with high sensitivity. The ground (mostly from cosmic rays) can be con-
Be(p,n) threshold occurs at 2,059 Mev, and the veniently determined by a companion bombardment
12¢, 13C, and '°F thresholds still higher. Pro- at 1.881 Mev.
tons of 2.06 Mev penetrating the surface of graph- The nuclear reaction !°F(p,a)'®0*(y)1°0 gives
ite lose energy by Coulomb interaction with the a 6,14-Mev gamma ray with a very high yield. The

electrons. They reach the Li threshold in 8 x 10™* energy is large compared with gamma rays from
fission products. The reaction has recently been
used to study fluorine penetration of Zircaloy by
Starfelt et al.?2? A survey of other proton-induced
reactions in graphite and in the molten salt used
for reactor fuel indicated that fluorine penetration
in the graphite moderator of the MSRE might also
be determined to a very high degree of sensitivity.

Prominent proton resonances in the '°F(p,ay)
reaction occur near 0.5, 1.5, and 2 Mev. Protons
of such energies would penetrate graphite to a
depth of 0.001 to 0.005 cm (a few tenths to 2 mils).
For smaller depths a detailed study of the gamma-
ray yield with varying proton energy can be ana-
lyzed to give the concentration as a function of
depth.?? The chief background expected is from
13C(p,y) in the graphite, giving gamma rays near
9 Mev. The incomplete absorption of these in the
spectrometer gives a flat background near 6 Mev.

Protons at 1.88 Mev from the ORNL 3-Mv Van de
Graaff were collimated to a 0.32-cm-diam spot.
Steady currents from 1/2 to 1 pa were collected at
the sample and integrated to monitor the number
of protons in each run. A 7.5 x 7.5 cm Nal(TI)
scintillation counter was placed 4 to 225 cm from
the sample to record the gamma-ray spectrum.

A long counter??® was used to measure neutron
yield and was placed straight ahead from the pro-
ton beam at a distance of from 11 to 35 cm from
the sample. A sketch of the apparatus is shown
in Fig. 10.12.

Thin standards used were 180 pug/cm? and about
14 pg/cm? of LiF evaporated on aluminum. A
thick LiF crystal was used as a standard also,
after evaporating a thin conductive layer of silver
on three of its faces to avoid electrostatic prob-
lems from the proton charge buildup. Possible
bias in our calibration for fluorine concentrations
is estimated at 10% to allow for the effect of
anisotropic emission of gamma rays near 90°. For

lithium concentration the uncertainty is 20%, largely

because we could not use 1/r? scaling to greatly
reduce the dead-time corrections (due to high count
rate) in counting the standard. In the case of the
MSRE sample (Y-7) mounted on an iron cylinder

for lapping, an additional large correction for
neutron transmission through the iron raises the
possible bias for lithium to 25%.

22%. Moller, L. Nilsson, and N. Starfelt, Nucl. Instr.
Methods 50, 270 (1967).

235 . 0. Hansen and J. L. McKibben, Phys. Rev. 72,
673 (1947).

ORNL-DWG 68-2540

SHIELD

PARAFFIN

GRAPHITE SAMPLE
PROTON \
BEAM\ /IRON

TO AMPLIFIER
AND SCALER
TO AMPLIFIER FOR
AND MULTICHANNEL NEUTRON
ANALYZER COUNT
FOR

GAMMA SPECTRA

Fig. 10.12. Plan View Diagram of the Experiment.
Protons from the ORNL 3<Mv Van de Graaff accelerator
were focused onto the surface of the sample in vacuum.
Facilities for measuring the charge, suppressing elec-
tron emission, etc., are not shown. The sample was
viewed by a gamma-ray spectrometer near 90° to the

o
beam and by a neutron counter near 0",

Figure 10.13 shows the gamma spectrum ob-
tained using the “LiF target. Also shown is the
spectrum from a sample of beryllium metal. The
gamma-ray yield from the beryllium is about 1/300
of that from fluorine and occurs mostly at a higher
energy (7.48 Mev).

Three samples of graphite were studied. A
piece of clean graphite gave the background
spectrum indicated in Fig. 10.14. A comparison
with the standards indicated that a fluorine con-
centration of about 1 ppm could be detected in a
10-min run. The lithium detection limit with the
neutron counter used was about 0.1 ppm.

A sample of graphite (Y-5, grade CGB, bar 635)
exposed to nonradioactive molten salt for nine
months was studied, as well as one from the MSRE
(Y-7, withdrawn in May 1967). Successive surface
layers of these samples were ground (lapped) away
to measure the fluorine concentration at succes-
sive, deeper levels; since the back face of sample
Y-7 had also been exposed to the molten salt,
concentrations were studied as this face was ap-
proached from the interior of the sample. The re-
moved material is to be analyzed for fission prod-
ucts and uranium.
ORNL-DWG 68-2544

10,000 =
S000 r—_—'
| Qe
— T Sy

® it 19 (p,a7)
2000 A R — 7 per)
& i)
s &
M | e
%
— | 6.13 MeV _o—
500 —
1 | i
z | | e
5 I
2 \
a 200 [ I
K
S 400 Lfi\ i
é t &8
> T
g
Z 50
5 F:P __H,M
-
wi
1 .
7.48 MevV—
2l - l
\ L |
0 100 200 300 400 500 600

PULSE HEIGHT CHANNEL NUMBER

Fig. 10.13. Pulse-Height Spectra in the 7.5- by 7.5-
em Nal(TI) Spectrometer from Samples of 19 and "Be
for Equal Proton Bombardment. The yield from TLi
above 0.478 Mev {channel 37) was negligible in com-

parison.

The sample exposed to nonradioactive MSRE
fuel salt (Y-5) showed surface contamination of
1.1 ug/cm?. This was removed by a 5 x 10~ cm
lapping. Since the proton beam is capable of
evaporating or sputtering surface layers, the sam-
ple was bombarded for more than 2 hr with a ¥ -pa
beam before lapping. The fluorine contamination
of the surface was decreased about 5%. Acci-
dental surface contamination was a problem in
handling these samples, since the "LiF standards
used in the same apparatus had up to a million
times higher concentration. Discrimination be-
tween thin surface layers and material distributed
in depth was accomplished by varying the proton

124

energy. Figure 10.15 shows the concentrations of
lithium and fluorine found at depth in sample Y-5.
The initial very high lithium value near the surface
did not persist in the further measurements and is
not considered typical of the bulk graphite. It may
represent material originally in a small fissure
about 8 x 10~* cm deep.

The MSRE moderator sample (Y-7) showed about
3 r/hr (surface) fission product activity, requiring
special handling. A 1.8-cm lead filter was used
in front of the spectrometer to reduce the overali
counting rate from the radioactivity. Surface
fluorine measured about 3.7 pg/cm?. Concentra=
tion as a function of depth is shown in Fig. 10.16.
To determine whether this fluorine concentration
was due to salt penetration of a crack fortuitously
crossing the sample, a microscopic examination
and an acetone test were performed. Neither test
showed any evidence of a crack. (Incipient cracks
have normally led to fracture during the coring
procedure used in cutting the sample from a larger
piece of graphite. Sample Y-7 showed no fractures
in the section we studied.)

The 7Li concentrations found are summarized in
Fig. 10.17. The lithium analysis was developed
during the course of the experiments, so data are
not available very close to the first surface
studied.

The sample exposed at the center of the reactor
for nine months (Y-7) shows over 100 times the
lithium and fluorine content found in the control
sample (Y-5) given identical treatment except for
irradiation. Since graphite is expected to with-
stand up to 20 times heavier irradiations before
replacement for structural reasons, it will be im-
portant to study penetration after longer reactor
exposures.

In the unirradiated sample the ratio of lithium
to fluorine (~0.43) is probably consistent with
that expected for the 7LiF molecule (7/19) in
view of possible errors in normalization. The
ratio expected from the fuel salt composition is
1/6. The nonstoichiometric ratios found (particu-
larly in the irradiated sample) seem suggestive
of ionic diffusion.

The fluorine concentration in the irradiated
sample (Y-7, Fig. 10.16) follows the inverse of
the depth from the front or back face fairly closely,
although the anomaly near 0.008 cm (3 mils) from
the rear face seems real. (The anomaly is also
seen in the lithium data, Fig. 10.17, and may
correspond to an inhomogeneity in the graphite.)
125

ORNL-DWG 68-2542

ENERGY (MeV)
41.2 2 3.0 39 4.8 5.8 6.7 7.6 86 95 10.5
10 —T T ] = T — I i 7,
a " | - |
5 Y
s
* [ ]
oy e
"t oo,
2 ‘ .‘)) L
3 ¢ ;
o= . i =
% > . ‘ » I .--.vh ) : lI ]
* ® ] o ne e,
b .. ry '.i o %S " “1 “‘-. .‘! t :
(: l \‘\ 2%, 9%% T:'fl H4ppm °F ‘I.!D'J. o |
3 N O I S e el ey I Poes, |
o 3
o | | I *
10° | £ 1y e
- i RiE e
5 ! I I | hi%
| M L ] e |
| | | ‘ . |
| L] »
2 ] | | b pd
I ; Pl \
| i - | .
10! | | | | 1 ]
90 140 190 240 290 340 390 440 490 540 590

CHANNEL NUMBER

Fig. 10.14. Pulse-Height Spectra from Samples of Clean Graphite (Dashed Lines) and Graphite Containing a
Trace of Fluorine. The peaks at 2.367, 3.150, and 3.150 — 0.511 Mev are attributable to ]2C(p,}/) reactions, the

spectrum near 9 Mev to 13C(p,)/).

The lithium concentration in the irradiated
sample is not proportional to the fluorine, rising
from the 7/19 ratio typical of “LiF near the sur-
face to 6/1 at the center. The minimum lithium
concentration appears to be nearer to the first
(outside) face of the sample than to the center.
Only the molten-salt flow velocity was reportedly
much different at the two faces.

10.4 SURFACE PHENOMENA IN
MOLTEN SALTS

H. W. Kohn F. F. Blankenship

The possibility that colloids or metallic sols
are involved in the corrosion product and fission
product behavior in the MSRE fuel has led to a
study of colloids in molten salts. When the MSRE
fuel was initially treated with hydrogen and beryl-
lium during purification, sols as well as larger
particles of reduced nickel and iron were evidently
produced. Some of these have remained suspended
throughout operation of the reactor and presumably
constitute the iron and nickel content regularly

reported in fuel analyses. Also, since beryllium
treatments in the reactor did not decrease the
chromium concentration, part of the chromium in
the fuel is presumably present as a metallic sol.
In general, the sols would be expected to exist as
alloys rather than as pure metals.

The more noble of the fission product metals,
because of the reduction potential reflected by
the UF , content, are expected to exist in the
MSRE as elemental metals. Most of such elements
plate out on walls as metal (or possibly as carbides
on graphite, in the case of molybdenum and ni-
obium). (An appreciable portion also appears to
leave with the off-gas.) However, to the extent
that they do remain in the fuel, they are probably
present as sols.

Again these sols are undoubtedly alloys, and
some may contain little or no structural metal.
There is evidence (see Sect. 10.1) that sols con-
taining fission products leave the melt, but a
corresponding loss of structural metals has ot
been noted.

As to the fate of these sols, we have obtained
evidence from laboratory experiments that there
ORNL-DWG 68-2543F

200 o FLUORINE
1Y, a LITHIUM
100
50
|._
T
© 20
w
=2
% 10 =%
E TN
e —
g 50— 1\}
- \
o \
S A
E 20 — [\ \
W \ \
S 10 \ X
A) ~
3 e i
N ~
\ ™~
N\
\\
0.2 \‘\
\\\
~
0.1 >
0] 2 4 6 8 10
DEPTH (mils)
I | l | l ]
) 0.005 0.01 0.015 0.02
DEPTH (cm)

Fig. 10.15. Lithium and Fluorine Concentrations in a
Sample (Y-5) of Unirradiated Graphite Exposed to Molten
Fluorides for Nine Months. The initial high lithium
value near the surface did not persist in further measure-

ments.

is marked tendency for flotation to occur. We
studied samples from a fuel solvent melt to
which large amounts of structural metal fluorides
had been added and then reduced with hydrogen
and zirconium. On melting a coarsely ground
portion of this material in glass, a dark scum
accumulated on the surface of the quiescent melt.
When the scum was removed and discarded, the
remaining salt had a decreased structural metal
content. The results are shown in Table 10.12,

126

ORNL-DWG 68-2544R

S o MEASURING INWARD FROM FIRST -
s00 ——F N SURFACE =
s A MEASURING OUTWARD TOWARD M-
\\i SECOND SURFACE . T
200 N
; ° \o \l
I 100 Ay
v s i
= 50
% SR
_g [)N
20 A
g \0\?
& 10 i
- CENTER OF I
5 NI samPLE | HH
L {
2
, il

o1 2 5 10 2 5 10 2 5 100 2 5 1000
DEPTH (mils}
L L | 1 | 1 ! | | 1 J
1032 5 1022 s54w0'2 5 a°
DEPTH (cm)
Fig. 10.16. Fluorine Concentrations in Graphite

Sample Y-7, Exposed to Molten Fuel Salt in the MSRE
Reactor for Nine Months. Measurements were made as
the sample was ground away in layers progressing from
the first surface to the center (open circles)and then from
the interior toward the second surface (closed triangles).
The distances shown are as measured from the nearest

surface exposed to molten salts.

This suggested that the black scum was finely
divided metal that had somehow floated to the
surface.

Two subsequent trials with salt from the same
source were performed with helium bubbling
through the melt in a test of flotation. The scum,
shown in Fig. 10.18, formed in greater abundance,
and the evidence for flotation was as shown in
Table 10.,13.

Table 10.12. Structural Metal Content Before and

After Removal of Scum from Melt

Makeup Analysis After Treatment

(ppm) (ppm)
Fe 286 158
Cr 445 47
Ni 4680 218

ORNL-DWG 68-2545R

500
© MEASURING INWARD FROM FIRST SURFACE HH
A MEASURING OUTWARD TOWARD SECOND  [T]
SURFACE
200 T
&
A \
100 \
X
- =\
5 50 YR
w
2 b
- T
2 20 Hl =+d
g t
Q
o
< 10
FJ ~O
5
.
2
1
10 2 5 10 2 5 100 2 5 1000
DEPTH (mils)
L L s L L ! L [ i ]
0.005 001 002 005 04 0.2 05 1.0

DEPTH (cm)

Fig. 10.17. Lithium Concentrations in the Graphite
Sample (Y-7) Exposed Nine Months in the MSRE Reactor.
The significance of the different symbols is as described
for Fig. 10.16. The distribution does not appear to be
symmetrical, and a nonuniformity near 0.008 cm from the
second surface is suggested. The fluorine measurement

(Fig. 10.16) indicates a similar minimum.

There was a possibility that the black material
included finely divided ZrO the oxide that pre-
cipitates from fuel solvent. Since there was no
significant segregation of zirconium, we con-
cluded that flotation had removed predominantly
metals.

The conditions for the formation of a similar
scum at the liquid surface of the pump bowl in the
MSRE are good. However, attempts to detect it
directly have been inconclusively unsuccessful.
Most of the material collected in the laboratory
experiments on flotation seems to be larger than
colloidal in size (>0.1 p), but the sol particles are
probably behaving in the same manner as the larger
particles.

There is evidence that sol particles escape
from radioactive MSRE fuel to form an aerosol
(see Sect. 10.1). The mechanism for this implaus-
ible behavior remains unknown. One suggestion
is that because the gas above radioactive fuel is
conducting, a contact potential between the liquid
salt and gas phases impels charged particles from

127

PHOTO 90934

Fig. 10.18. Flotation in Fuel Solvent.

the interface. Another suggestion is that recoil
is responsible. Bursting bubbles may have an
effect, but the results of hot-cell tests (see
Sect. 10.1) with quiescent fuel from the MSRE
suggest that bubbles may not be necessary.
The whole question of the formation and sta-
bility of colloids in molten salts has been virtual-
ly untouched in the literature. Accordingly, ex~
periments in nitrate melts have been initiated
(because of the convenience of working with
nitrates) to characterize such colloidal behavior.
Table 10.13. Segregation of Reduced Structural Metals
by Flotation

Analyses in weight percent

Fe Cr Ni Zr

Second melt

Top 0.147 0.012 0.31 9.79

Bottom 0.008 <0.004 <0.006 10.2
Third melt

Top 0.074 0.030 0.42 9.19

Middle 0.013 <0.009 <0.2 10.1

Bottom 0.010 <90.003 0.12 11.3

Colloidal dispersions of silver metal in molten
potassium-sodium nitrate eutectic have been pre-
pared by uv photolysis. Colloidal gold has been
prepared in sodium-lithium and sodium-potassium
nitrates by reduction of gold chloride and also
by supersonic dispersion of particulate gold from

128

exploded wires. Silver iodide in potassium-
lithium nitrate melts has also been prepared.
Transmission and scattering spectra are being
measured.

Another surface-related behavior that has been
studied is foam formation. Foam formation in
nitrate melts was induced by adding aluminum
nitrate nonahydrate. The foam was first formed by
bubbles of nitrogen oxide from the decomposition
of aluminum nitrate. When helium was bubbled
through the melt, the nitric oxide foam gave way
to a helium bubble foam which, if undisturbed,
lasted for about 15 min. Similiar behavior was
encountered some time ago on an addition of
thorium fluoride to a nitrate-fluoride melt.2* In
both cases we could hypothesize that the foam
was stabilized by finely divided oxide. In view
of the rare occurrence of molten-salt foams, we
have conjectured that a finely divided wetted
solid may be a requisite for stable foam formation
in molten salts.

24R. A. Strehlow, private communication,
M. Chemistry of Fission Product Fluorides

11.1 PROPERTIES OF MOLYBDENUM
FLUORIDES

C. F. Weaver H. A. Friedman
D. N. Hess

The synthesis and characterization of molybdenum
fluorides and a study of their reactions in molten

2LiF.BeF ,''? were continued. The synthesis of
ruthenium fluorides was initiated.

T1.1.1 Synthesis of MoF ; and MoF

The first step in the synthesis involved forming

a solution of MoF , and MoF by reacting molybdenum

metal with an excess of MoF _ refluxed in a Pyrex
container such as that shown at the right in Fig.
11.1. Commercial MoF _ was purified with respect
to HF before use by distilling it over NaF in an
apparatus similar to that shown at the left in Fig.
11.1. Hydrogen fluoride is known to catalyze the
reaction 2MoF _ + 38i0, - 3SiF, + 2MoO;* failure
to remove HF resulted in several explosive ruptures
of the Pyrex equipment, while, in the absence of
HF, the MoF6 was contained for weeks without

an observable pressure increase. As much of the
apparatus as possible was flamed under vacuum be-
fore use, because these fluorides are very reactive
toward moisture. After the molybdenum metal was
completely reacted with the excess MoF , the
system was evacuated and the temperature slowly
increased. During this procedure the excess MoF

IC. F. Weaver, H. A. Friedman, and D. N. Hess,
““Behavior of Molybdenum Fluorides,’’ Reactor Chem.
Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4220.

2C. F. Weaver and H. A. Friedman, ‘‘Behavior of
Molybdenum Fluorides,”’ MSR Program Semiann. Progr.
Rept. Aug. 31, 1967, ORNL-4191, pp. 142—44.

3A. J. Edwards, R. D. Peacock, and R. W. H. Small,
J. Chem. Soc. 1962, pp. 448691,

129

left the system, leaving a residue of MoF .. Upon
further heating, part of the MoF _ sublimed to the
neck of the Pyrex flasks, and part disproportionated,
producing MoF ; and MoF , (which then left the
system). Experiments in which the terminal tem-
peratures were 200° and 150° both produced MoF ,
as a tan residue in the bulb and MoF as a bright
yellow deposit in the neck of the flask. The
ratio of MoF ; to MoF ;| was higher at 200° than at
150° An experiment which was heated to only
100° produced the MoF  in the neck of the con-
tainer but left a viscous residue with a molyb-
denum-to-fluorine ratio slightly higher than MoF ..
This residue did not solidify upon cooling to room
temperature and storing for several days.

The MoF , and MoF ;| were both identified by x-ray
diffraction and by chemical analysis. Their dif-
fraction lines are listed in Table 11.1. Those
for MoF , were identical to those obtained by
D. E. LaValle et al.*'% in earlier synthesis and
structure studies. The pattern for MoF ; corresponds
to that calculated® from structural data in the
literature.® The analytical results’ confirming the
stoichiometry are shown in Table 11.2. The im-
purity levels determined by emission spectroscopy 8
are also shown in Table 11.2. The optical prop-
erties of these substances and of NbFsg were de-
termined to provide a rapid and convenient means

“D. E. LaValle et al., J. Am. Chem. Soc. 82, 2433—
34 (1960).

5 . .
Personal communication, R. M. Steele, Metals and
Ceramics Division.

6Pattern calculated by G. D. Brunton, Reactor
Chemistry Division.

7Analyses performed by E. C. Lynn and Dave Canada,
Analytical Chemistry Division.

SAnalyses performed by J. A. Carter et al., Analytical
Chemistry Division.

9Samp1e prepared by L. M. Toth by vacuum distilla-
tion of impure NbFs, Reactor Chemistry Division.
MONEL VACUUM PRESSURE GAGE

SODIUM FLUORIDE TRAP—]

el
2

L.—WATER JACKET

i f

ORNL—DWG 6711720

TO VACUUM OR HELIUM

Z ; :
//%II/IIIIIIII/I/I/AN FL

2 AAVIIN,

P2
77, SwRmRRRY

\.\S

/|

V22

N

A

U
li l ‘L—WATER
: JACKET

Fig. 11.1. Apparatus for the Synthesis of Molybdenum Fluorides.

for future identification; they are summarized in
Table 11.3.

The absorption spectrum of MoF is being in-
vestigated.!® Preliminary results show absorp-
tion in the region of 1260 nm and an intense ultra-
violet shoulder near 340 nm.

s P cooperation with Jack Young, Analytical Chem-
istry Division.

11.1.2 Lithium Fluoromolybdates(ill)

The relative stability of MoF, at temperatures
of reactor operation and the fact that it can produce
both molybdenum metal and volatile molybdenum
products upon disproportionation suggested that
behavior of Mo® ¥ in MSRE-type solvents should
be investigated. Examination of quenched melts
along the join 2LiF.BeF ,-MoF ; have disclosed
two previously unknown compounds shown to be-
131

Table 11.1. X-Ray Diffraction Lines®

Table 11.2. Analyses of the Molybdenum Fluorides

MoF3 M(:-F5
Powder Pattern Debye-Sherrer
20 1001/1, 20 Intensity
23.3 100 11.10 w
31.8 25 16,73 m
34.4 13 19.65 w
39.3 5.4 20.64 m
41.9 8.2 22.31 s
47.5 8.1 24.31 VW
51.7 12 25.11 vw
54.9 3.2 26.53 ms
55.3 6.5 27.49 vw
60 4.7 28.61 vw
61.6 4.4 31.11 VW
66 2.1 32.64 w
68.2 1.9 33.81 vw
72.6 2.7 37.22 w
74.3 1.1 39.55 w
77.3 2.5 42.51 VW
81.3 1.4 45.37 ms
47.35 vw
48.77 vw
51.51 VW
53.37 VW

aCopper target.

long to the system LiF-MoF .. Their optical prop-
erties and x-ray diffraction patterns have been
determined and are given in Tables 11.4 and 11.5
respectively. Attempts to determine the stoichiom-
etry and melting behavior of these phases have
been complicated by corrosion of container ma-
terials such as nickel and copper. However, the
formulas are certainly close to SLiF .2MoF , and
LiF.MoF ,, and the compounds have been so labeled
in Tables 11.4 and 11.5. These efforts are being
continued with the following goals in mind:

1. Definite determination of the formulas and
melting behavior of these compounds.

2. Determination of the liquidus values and
primary phases along the join 2LiF.BeF ,-
MoF ..

3. Analogous studies of the behavior of niobium,
ruthenium, and technetium,

MOF3 MoF5

Wt %

Mo 62.7 49.6

Mo (calcd) 62.7 50.2

F 37.9 49.9

F (calcd) 37.3 49.8
Ppm

Al <200

B 500

Be 10

Ca <100 <500

Cu 50 3000

Li 500 100

Mg 50 <100

Mn 30

Na 300 500

Si 300 200

w <500

Table 11.3. Optical Properties of the Molybdenum

Fluorides and Nicbium Pentaffuoride

MoF

3 Uniaxial positive

N =1.592
N_=1.624
Golden vellow

MoF Biaxial

Optical angle = 90°
Na= 1.520

Ny = 1.548

Lemon yellow

NbF Biaxial negative
Optical angle = 50°
N, =1.484

N'y = 1.516

11.1.3 Kinetics of MoF3 Behavior in 2LiF-BeF2

Previously reported work? indicated that MoF
may be stable for many days or disproportionated
to molybdenum metal and MoF, in a few hours in
the temperature range 500 to 700° depending on
the temperature and pressure. It was further shown
Table 11.4, Optical Properties of the Lithium

Fluoromolybdates

132

LiF-MoF3 Uniaxial positive
N, =1.512
N_.=1.524
5LiF-2MoF3 Biaxial negative
N, =1.472
N'y = 1.490

Optical angle ~ 60°

Table 11.5. X-Ray Diffraction Powder Patterns

SLiF2MoF LiF-MoF ,

20 1001/1, 20 1001/1,
20.2 100 19.3 29
21.4 72 21.3 100
25.9 34 26.9 37
32.9 33 30.1 4
35.2 6 33.3 27
35.8 11 35.0 11
40.1 21 38.2 4
40.9 10 40.0 12
41.6 12 42.3 10
43.5 23 44.3 19
44.7 3 44.6 12
48.7 52.9 16
51.1
52.8 18
55.1 7
57.1 4
61.1 6
63.1 4
68.9 6

that the corrosion of the container material by
MoF , was much greater if molten 2LiF-BeF2 was
present and that copper was much superior to
nickel as a container material. The reaction

MoF; (1 mole %) + 3UF; (4 mole %) > Mo + 3UF
was observed in molten 2LiF-BeF2 at 500°C,

These observations suggest that Mo® * plays a
central role in the behavior of molybdenum in the

4

MSRE and that its kinetic behavior should be
quantitatively investigated in the ppm range.
Initial conditions were selected to provide a
simple reference system which could be made
progressively more complex and more analogous
to the MSRE. The first experiment was made with
a copper container, a 2LiF-BeF‘2 melt, and a
helium flow rate of 8 liters/hr per kilogram of
melt. The MoF , was added to the molten 2LiF.
BeF, in the form of a previously fused mixture of
LiF-MoF3 (75-25 mole %). Filtered samples
were taken over a period of several hours, as
shown in Table 11.6 under ‘‘first run.”” The three
types of analyses”'3:1! for molybdenum gave es-
sentially the same answers. The wet method
provided the least scatter and indicated that,
after an initial mixing period, the concentration
of molybdenum in the melt did not change over a
23-hr period and that there was no difference be-
tween the filtered and unfiltered samples. The
second experiment was conducted in the same way
except that after 2 hr at 500° the temperature was

11AnalySeS performed by E. I. Wyatt et al., Analytical
Chemistry Division.

ORNL-DWG 68-5542

%

/

20

103¢ (ppm™)

80
TIME (hr)

{00 120 140 160

Fig. 11,2, Removal of Mo3+ from Molten 2LiF-BeF2.
133

Table 11.6. Kinetic Behavior of Mo> ' in 2LiF-BeF,

Length of
Temperature Time After Molybdenum Molybdenum by Molybdenum by Copper by Copper by
of Sample Type of Addition of by Wet Spectroscopic Activation Wet Spectroscopic

(OC) Sample Molybdenum Salt Chemistry Analysis Analysis Chemistry Analysis

(hr) (ppm) (ppm) (ppm) (ppm) (ppm)
First Run
500 Filtered 0.15 620 300 638,621 500
Filtered % 810 300 845,896 30
Filtered 2 820 700 770,908 500
Filtered 6 810 700 836,819 100
Filtered 23 810 500 752,743 100
Unfiltered 23 800 500 846,839 100
Second Run

500 Filtered 2 690 700 113 100
500-~700 Filtered 2.5 510 400 64 70
700 Filtered 3.5 440 300 183 200
700 Filtered 7 390 300 50 50
700 Filtered 24 190 100 208 100
700 Unfiltered 24 250 200 81 70
7009 Filtered 27 <10 <100 41 70
700% Unfiltered 27 20 <100 396 500

%Reduced molybdenum in melt with Hz'

raised to 700° In this case there was a definite
decrease in the molybdenum concentration in the
filtered samples. The unfiltered samples con-
tained more molybdenum than the filtered ones,
suggesting that metallic molybdenum was present
and that it was being excluded by the filter. To
further confirm that the metallic molybdenum was
not passing the filter, the sample was reduced
with an excess of hydrogen. No molybdenum was
detected in the filtered reduced sample but was
found easily in the unfiltered sample, suggesting
that part of the metallic molybdenum was sus-
pended in the melt but not passed by the filter.
The copper analyses do not appear to have a
trend with time. The scatter of the wet copper
analyses was greater than expected from the
analytical procedure and is probably an artifact

of the sampling technique, which involves a copper

filter. Unfiltered samples of 2LiF-BeF, taken in
a nickel tube with and without Mo® © were not

found to contain copper. The absence of a copper
concentration increase with time and the level of
concentration both indicate that the decrease in
molybdenum concentration was not caused by MoF
reduction by copper.

The effect of the helium flow rate (I\/IoF6 removed
rate) on the kinetic behavior of Mo® ' in molten
2L'1F‘-BeF2 was investigated in runs 3 and 4. The
data in Table 11.7 and Fig. 11.2 indicate that the
molybdenum was removed from the 2LiF.BeF , by
a second-order process, the most likely event be-
ing

2M0F3 = Mo + MoFGT

The rate constant k in the relation

1 1\1
k=<——— — (1)
C C t
Table 11.7. Kinetic Behavior of Mo® ' in 2LiF-BeF,

Temperature Length of Time After Molybdenum by
of Sample Temperature Is Raised Wet Chemistry
Co from 500°C (hr) (ppm)
Third Run
500 0 806
690 % 663
700 1%, 570
700 47, 465
700 20 240
700 44 140
700 68 112
700 140 53
Fourth Run
500 6 158
700 A 137
700 11/2 98
700 2%, 73
700 19 43
700 431/6 16
700 677, 14

(c = concentration of Mo® ) has the value 1.2 x
10~* ppm~—! hr=? for a helium flow rate of 3/4
liter/hr at 700°and 1.2 x 103 ppm~! hr~! for
12 liters/hr at 700°C.

If one makes the steady-state assumptions for
reaction (1) that the concentration of MoF . was
constant and very low compared with the concen-
tration of Mo® ', then k takes the form af/(b + df),
where a is the rate constant for the forward reac-
tion, b is the rate constant for the back reaction,
dis a constant, and f is the flow rate. If b << df,
that is, if the back reaction is unimportant, then
the rate expression reduces to

c
d a ,

—_— = — =

dt d

2

which suggests that the flow rate is unimportant.
Since, experimentally, the flow rate had a great

134

effect, the back reaction was involved significantly
in the kinetic behavior of Mo®*, b is the order of

df or larger, and k,/k, <f,/f , consistent with the
experimental values

k, 1.2x 1073 hr= ! ppm~! 10
k, " 1.2x 1074 hr~ ! ppm~!

for

f2 12 liters/hr 16
£ 0.75 liter/hr

1

The first two points (at 0 and 0.75 hr) in Fig.
11.2 are associated with a lower temperature,
since the heatup from 500 to 700°C was involved.
However, since the rates were faster at the lower
temperatures and since the rates are known to in-
crease with temperature, the short-term effect
must have another explanation. Apparently this
was the period necessary to establish the steady-
state condition. These kinetic experiments will
be continued with the following objectives:

1. To determine effects of UF , UF3, ZtF ,, and
ThF, additions as well as INOR-8 and graphite
on the rate and extent of removal of molybdenum
from the melt.

2. To identify and determine the fate of the
volatile species.

3. To obtain the corresponding information for
niobium, technetium, and ruthenium.

11.2 MASS SPECTROMETRY OF THE
MOLYBDENUM FLUORIDES

R. A. Strehlow J. D. Redman

The mass spectrometric study of molybdenum
fluoride vaporization behavior was continued.
Previous work!? was conducted with a conventional
Knudsen cell suitable for the study of low-vapor-
pressure solids. This cell was not suitable for
study of gaseous species at elevated temperature.
Accordingly a cell (made of nickel) was con-
structed which permitted addition of gases directly
into the cell region. Using the new cell, the mass
spectrum due to admitted MoF ;. was obtained at

12vsr Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191.
135

Table 11.8. Mass Spectrometric Cracking Patterns for Molybdenum Fluorides and Oxyfluorides

a
MOF6 MOF5 MOF4 MOOF4 M002F2
Ion Intensity Ion Intensity Ion Intensity Ion Intensity Ion Intensity
MoF 100 MoF " <7 MoF 100 MoOF," 100 MoO,F " 100
MoF , * 32 MoF ¥ 100 MoF " 15 MoF , 6 MoO,F 65
MoF , * 19 MoF ¥ 56 MoF 10 MoOF, " 10+ MoOF," 6
MoF " 15 MoF " 18 Mo 10 MoF " 11 MoF > " 2
MoF * 11 MoF © 8 MoOF © 10 MoOF * 23
Mo 8 Mo ™ 8 MoF 9 MoF * 5
MoO 3.5 MoO 7
Mo 5 Mo 16
2This assignment is not certain; MoOF , is possible.
various temperatures from 30 to 850°C. The ORNL-DWG 6875543
. of s . SOURCE MATERIAL
cracking pattern at 30°C differed only slightly SPECIES  MoF,  MoF.
from that obtained eatlier and is reported in MoFe .
Table 11.8. With a steady rate of MoF ; admission miij .
to the cell containing molybdenum wire, an in- 00
crease in temperature resulted first in production
of MoF5 and, at higher temperatures, increasing 90
amounts of MoF ,. This behavior paralleled that
of MoF , (which was studied using the earlier 80
cell design) except for the presence of dimeric
impurity in the MoF , sample. The dimer, which 70
has a principal peak family corresponding to &
M02F9+, may have as a neutral precursor M02F9O 5 60
=
or MOQFIOO. The mass spectroscopic evidence S o
alone is not adequate to distinguish between §
these possibilities. & 40
The rather striking similarity of results from the <
MoF , and the MoF , studies is shown in Fig. 11.3, 30
which displays the derived percentages of the
species MoF _, MoF, and MoF , as a function of 20
temperature for both the tri- and hexafluoride. The
{0

data for MoF exclude the dimer. The difference
in free energy of formation between MoF ; and MoF,
is clearly very slight, since their relative amount
varies only slowly with temperature. The earlier
conclusion that kinetic factors dominate the
descriptive chemistry of these materials is still
valid, although we have obtained data indicating

OL’.
o i

00

300

400

500

600

700

KNUDSEN CELL TEMPERATURE {(°C)

Fig. 11.3. Monomeric Yapor Species over MoF6 vs

Cell Temperature,
that some thermochemical relations can be derived
for the Mo-F system.

The dimer M°2F101 was not the only dinuclear
fluoride observed. During admission of MoF . an
erosion of the tantalum heat shields in the Knudsen
cell assembly occurred and was evidenced by the
presence of MoTaF *ions in the spectrum as well
as by subsequent examination of the tantalum
elements.

Continued work on the molybdenum fluoride
system will include a more complete treatment of
the data obtained so far and consideration of the
approach needed to obtain the desired thermo-
chemical and kinetic data for this system. Some
other fission product fluoride and oxyfluoride
systems will also be examined.

11.3 SPECTROSCOPIC STUDIES OF FISSION
PRODUCT FLUORIDES

L. M. Toth J. P. Young
G. P. Smith

Because spectrophotometry offers a direct meas-
urement of fission products such as the lower-
valent niobium and molybdenum fluorides, a study
of their spectra has been undertaken in the window-
less cell developed by J. P. Young.'® This cell
has the advantage that the melt need only be ex-
posed to a container which has been selected on
the basis of the melt’s reactivity with it. For
this reason graphite and copper windowless cells
have been selected for use with M0F314 and NbF
solutes in LiF-BeF , (66-34 mole %). *°

The original intent was to compare the spectra
of NbF, and MoF ; with the ligand-field spectra
of d! and d3 configurations, respectively, in
octahedral symmetry.

Initial attempts to dissolve 0.1 wt % concen-
trations of MoF , in L.,B'° failed due to the forma-
tion of a finely divided precipitate which required
16 to 24 hr to settle out. After it had settled, no
spectrum was evident in the L,B. The cause of

13_]. P. Young, Anal. Chem. 36, 390 (1964).

14Supplied by C. F. Weaver and H. A. Friedman.

15I-Iem:efcn-th abbreviated as “LZB’”

18 This L2B was from a standard HF-H2 treated batch
of J. H. Shaffer’s, Reactor Chemistry Division.

17Reactor Chemistry Division.

136

the precipitate has been attributed to the forma-
tion of insoluble molybdenum oxides or oxy-
fluorides.

Successful dissolution of MoF, in L B has
been achieved by first treating the solvent with
SiF, and filtering it through a quartz frit. The
spectra of such solutions display peaks at 350 and
470 mp. which are not inconsistent with those ex-
pected for a d* cation in octahedral symmetry.
To verify that this spectrum is not due to contri-
butions from SiF ,, it was compared with one
produced by a sample taken from the molybdenum
fluoride kinetic studies of C. F. Weaver and
H. A. Friedman.!? Although the latter sample
yielded only a very weak spectrum, similar ab-
sorbancies at 350 and 470 my could be identified.
The poor intensity of this spectrum has been
tentatively attributed to the fact that much MoF
had precipitated out due to oxide contamination
of the sample. Experiments are currently under
way to demonstrate this further as well as to
establish a quantitative value for the absorption
coefficients at 350 and 470 my. With these values,
limits of spectroscopic detectability will be
available. Further experiments will be made in
an attempt to demonstrate conclusively that the
spectrum is due to dissolved MoF, in octahedral
symmetry.

Niobium tetrafluoride is apparently sparingly
soluble in L_B. Presently, only a very weak
spectrum has been obtained which displays a
single absorbance at 350 mu. This spectrum is,
however, similar to that produced by NbF, in
molten NbF_.'® Further work shall endeavor to
circumvent the solvation problems by generation
of NbF  in solution.

Additional information available from the
spectroscopic studies includes the compatibility
of the solutions with the container materials. In
reactor-grade graphite, the spectrum attributed to
MoF , disappears completely within 4 hr at 650°C
according to first-order kinetics. No intermediate
spectra have been seen during this reaction. On
the basis that the same solution shows no percep-
tible change under similar conditions in a hydrogen-
fired copper container, reaction with the graphite
is suggested. Further experiments will describe
the stability of the solutions in more detail.

181, wm. Toth and G. P. Smith, Reactor Chem. Div.
Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 64.
11.4 PREPARATION OF NIOBIUM
PENTAFLUORIDE

L. M. Toth H. A. Friedman

C. F. Weaver

Because of the lack of information concerning
the behavior of fission product fluorides in the
MSRE, their chemistry in molten fluoride solutions
is being actively pursued. One such study is con-
cerned with niobium and its fluorides.!?

Niobium pentafluoride has been the most con-
venient starting point for the synthesis of lower
niobium fluorides as well as a low-temperature
solvent for the lower niobium fluorides.?® It had
been previously prepared in a flow system at
250°C by passing fluorine over niobium powder in
a Monel reaction vessel,?° but this procedure
would have taken several days reaction time for
the hundred-gram quantities of product required.

19L. M. Toth and G. P. Smith, Reactor Chem. Div.
Ann. Progr. Rept. Dec. 31, 1967, ORNL-4220.

0%, Fairbrother and W. C. Frith, J. Chem. Soc. 1951,
p. 3051,

137

It was found in this work that decreasing the
reaction temperature accelerated the reaction rate
to the point that it could be completed within 8 hr.
This is to be expected for an exothermic reaction
which involves a decrease in the entropy of the
system:

Nb + 5/2 F, - NbF, .

The effect also provides a built-in safety device
in which the reaction rate decreases abruptly if
the temperature goes too high.

The synthesis was performed in a two-chamber
Monel reaction tube which was entirely sealed
except for the F, inlet valve. The F, was admitted
at a constant rate into the reaction chamber, con-
taining powdered niobium metal at 70 to 100°C,
and the NbF | sublimed into the other chamber,
which was at room temperature.

Niobium oxyfluorides present as impurities were
then removed from the product by resublimation of
the NbF . at 100°.
12. Physical Chemistry of Molten Salts

12.1 THERMODYNAMICS OF LiF-BeF , MELTS
BY EMF MEASUREMENTS

B. F. Hitch  C. F. Baes, ]Jr.

A potentiometric study! has been completed of
the cell

Bé’lBer,Lfl?|H2,HF,Pt,

for which the cell reaction
Be?(c) + 2HF(g) = BeF ,(d) + H,(g)

was assumed. These measurements extended over
the composition range 0.30 to 0.90 mole fraction
BeF, and over temperatures in the range 500 to
900°C (Fig. 12.1).

Since the cell potential E should be related to the
activity of BeF , in the melt by

a
. EO RTI I—l2 BeF2
= ——=1n
2 b
23 P2

activity coefficients could be derived for BeF, (and
by a Gibbs-Duhem integration for LiF). Usefully
accurate measurements could not be made with pure
BeF, in the cell because of its very high viscosity;
hence values of E° were calculated using values

for agep derived from the phase diagram and a re-

ported heat of fusion? of 1.13 kcal/mole, This
gave, in volts,

E° = 2.4430 — 0.0007952T .

1B, F. Hitch and C. F. Baes, Jr., An E.M.F. Study of
LiF-BeF2 Solutions, ORNL-4257.

ZA. R. Taylor and T. E. Gardner, Some Thermal Prop-
erties of Beryllium Fluoride from 8° to 1200°K, U.S.
Bureau of Mines Report RI-6644 (1965).

This comparison of the emf data with the LiF-BeF,
phase diagram also indicated that the heat of fusion
for BeF, is <2.0 kcal/mole.

A power series in x ;. was assumed for log

yBer , and the coefficients were determined by a

least-squares fit to the data. This gave (Fig. 12.2)

2
LiF

2353.5

log yBeF2 :<3.8780 -

36292.8\
+{ -40.7375 + )53
84870.9\
+< 94.3997 — ————)xLiF
52923.5\ |
- (—67.4178 " x5

A Gibbs-Duhem integration of this expression gives

fory ;g

232.08
log y, ;p =0.9384 —

2
Xper

14652.7
+{ —36.9734 + ——
2

3
XBEF

+ (126.0947 -
2

74588.5 )

4

X
BeF2

113592.3
+ (—158.4173 +——-——)

5
XBeF

52923.5
+ (67.4178 —- .
2
(The integration constant was determined from the
LiF liquidus in the LiF-BeF ) system.) Formation
free energies and heats for BeF and BeO were also

138
139

ORN|.—DWG 67-13723R

!
|
|
|
|

195 T —’— T

-

=
LUu \\
070
1.60 ( \
1.55
0.30, 0.33, 0‘40,‘ 050, 060 Ber 550 600 650
0.70 BeF2 600 650 700
0.80, 0.90 Ber 650 700 750

TEMPERATURE (°C}

Fig. 12.1. Measured Cell Potentials, Corrected to a Unit Value of the Pressure Quotient Py /PE’F, as a Function
of Temperature at Various LiF-BeF2 Melt Compositions: 2

RT 9
E =E+—= In(Py /P5)
140

ORNL-DWG 68-231

wl |

]

/

/

//fl

/0

/
i
0 oA 0.2 0.3 04 0.5 10
“BeF,

Fig. 12.2. Activity Coefficients of BeF, and LiF Derived from the EMF Measurements.

Table 12.1. Formation Heats and Free Energies of Bel"2 and BeO

Temperature AH[(kcal/mole) AGf(kcal/mole)
Compound State
(°K) Present Work JANAF? Present Work JANAF?®
BeF, 298 Crystalline  —246.01 -242.30(1£2.0) —234.39 —230.98(12.0)
900 Liquid —243.12(%1.1) —211.90(%1.1)
1000 Liquid —242.54 —208.47
BeO 298 Crystalline  —145.85 —143.10(£0.1) —138.36 —136.12($0.1)
900 Crystalline  —145.57(%1.5) —123.20(%1.5)
1000 Crystalline  —145.46 —120.73

4Reference 3.

calculated (Table 12.1) by combining the results of
the present study with available thermochemical
data® for HF and H,O and with the results of a pre-
vious study* of the equilibrium

2HF(g) + BeO(c) = BeF (d) + H,0(g) .

3JANAF Thermochemical Tables, Clearing House for
Federal Scientific and Technical Information, U.S. De-
partment of Commerce, August 19635.

4A. L. Mathews and C. F. Baes, Jr., J. Inorg. Chem.
7, 373 (1968).

The Be2*|Be® and the HF,H,|F~ electrodes used
in this study performed acceptably for use as refer-
ence electrodes, both being stable and reproducible.

12.2 ELECTROLYSIS OF LiF-BeF, MIXTURES
WITH A BISMUTH CATHODE

K. A. Romberger  J. Braunstein

Experiments have been initiated on the electrol-
ysis of LiF-BeF, mixtures with a bismuth cathode
141

in a silica cell. The purpose of these experiments
is (1) to develop lithium-bismuth alloy electrodes
containing about 0.1 at. % lithium as reference
electrodes for equilibrium and nonequilibrium elec-
trochemical measurements in molten fluorides, (2)
to ascertain the compatibility of lithium-bismuth
alloys with silica containers, and (3) to determine
whether reduced bismuth species are present in
melts contacted with lithium-bismuth alloys under
conditions proposed for extraction processes for the
single-region MSBR.® (At very much higher lithium
concentrations in bismuth, that is, with saturated
alloys, there have been indications of the dissolu-
tion of reduced bismuth species in LiCI-LiF mix-
tures. ©)

In a preliminary experiment, 41 g (1.2 moles) of
LiF-BeF2 (66-34 mole %) which had previously
been purified for use with the MSRE’ was elec-
trolyzed, under a helium blanket, between a bis-
muth cathode (99.999% Bi, hydrogen fired at 600°C
to reduce any oxide) and a graphite anode isolated
from the main portion of the melt in a silica J-tube.
Forty-one grams of bismuth was used as the cath-
ode, and the temperature was 473°C. Approxi-
mately 160 coulombs were passed (~1.7 x 1073
equivalent), enough to produce about 0.8 at. %
lithium-bismuth alloy if the current efficiency were
100%. The current density was 2 x 10~ 3 amp/cm?.
The silica cell permitted visual observation of the
melt and electrodes during the electrolysis. From
the inception of electrolysis, gas bubbles were
formed at the cathode (possibly hydrogen from the
reduction of residual HF, which had been used in
the purification of the salt,’ or traces of OH™).

A black deposit was formed which partially

covered the surface of the bismuth cathode. Chem-
ical analysis of the phases has not been completed,
but it is believed that the black deposit was metal-
lic beryllium. Because of the low rate of diffusion
of lithium from the surface into the bulk of the bis-
muth, the local activity of lithium at the surface
may have been high enough to permit beryllium,
which is insoluble in bismuth, to deposit on the
surface.

5_]. H. Shaffer et al., MSR Program Semiann. Progr.
Rept. Aug. 31, 1967, ORNL-4191, p. 148.

®M. S. Foster et al., J. Phys. Chem. 68, 980 (1964);
M. S. Foster, Advan. Chem. Ser. 64, 136 (1967).

7_]. H. Shaffer et al., Reactor Chem. Div. Ann. Progr.
Rept. Jan. 31, 1965, ORNL-3789, p. 99.

Samples of salt were withdrawn for analysis after
passing about 40 coulombs and again after about
160 coulombs (equivalent to about 0.2 and 0.8 at %,
respectively, lithium in bismuth). Preliminary ana-
Iytical results showed that the bismuth content of
the salt samples was below the limit of detection
of 10 ppm. There was no visible attack of the
silica by the metal phase. Experiments are con-
tinuing, to evaluate the behavior of the lithium-
bismuth electrodes and to elucidate the electro-
chemical behavior of bismuth in molten fluorides.

12.3 REVIEW OF ELECTRICAL
CONDUCTIVITIES IN MOLTEN FLUORIDE
SYSTEMS

G. D. Robbins

A review of electrical conductivity measurements
in molten fluoride systems covering the period 1927
to 1967 has been made.® The results may be sum-
marized as follows:

1. Because of the high specific conductance of
most molten salts (1 to 6 ohms~! cm™!),° one of
two experimental approaches!? is usually employed:
the use of capillary-containing cells, which results
in a measured resistance of several hundred ohms,
or the use of metallic cells in which the container
is one electrode with a second electrode positioned
in the melt. Measured resistances in the latter type
of cells are less than 1 ohm. Boron nitride encased
in graphite has been successfully employed in cap-
illary construction,’!~13 while platinum, platinum-
thodium (20%), Inconel, molybdenum, and graphite
have been used for container and electrode mate-
rials.

2. The common practice of measuring resistance
with a Wheatstone bridge having a parallel re-

8G. D. Robbins, Electrical Conductivity of Molten
Fluorides, a Review, being reviewed for publication.

1. S. Yaffe and E. R. Van Artsdalen, J. Phys. Chem.
60, 1125 (1956).

10G. J- Janz, C. Solomons, and H. J. Gardner, Chem.
Rev. 58, 461 (1958).

11E. A. Brown and B. Porter, Electrical Conductivity
and Density of Molten Systems of Uranium Tetrafluoride
and Thorium Fluoride with Alkali Fluorides, U.S. Dept.
of Interior, Bureau of Mines, 128.23:6500 (1964).

128, W. Yim and M. Feinleib, J. Electrochem. Soc.
104, 622 (1957).

L31bid., 626 (1957).
142

sistance and capacitance in the balancing arm can
result in considerable error if the relation for re-
sistive and phase balance,

_ 2 2 2
Rp = RS[I + RPCp (27hH?*],

is not employed in determining the solution resist-
ance.® In this expression R, and C are the par-
allel balancing resistance and capacitance, RS is
the solution resistance, and f is the measuring fre-
quency. Use of this correction can be avoided by
employing a bridge with the balancing resistance
and capacitance in series.!*

3. The practice of measuring resistance at a
series of frequencies and extrapolating to infinite
frequency as a function of {~!/? is examined in
terms of electrode process concepts. Some authors
followed this procedure,!®:1® while others found no
appreciable variation of resistance with frequency
(refs. 11-13, 17, 18). Unfortunately, the majority
of investigators did not address themselves to this
question. (Research on the frequency dispersion of
electrical conductivity in molten salts is continuing
in this laboratory.)

4. Table 12.2 lists some values of the specific
conductance x and the equivalent conductance A®4
of those alkali-metal and alkaline-earth fluorides
which have been reported.!1—13:15:17=23 These
quantities are defined as

1
K=—(l/a) ,
R( )
Aed _ Keq wt
p b

14G. D. Robbins and J. Braunstein, Reactor Chem.
Div. Ann. Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 57.

151 D. Edwards et al., J. Electrochem. Soc. 100, 508
(1953).

161 D. Edwards et al., ibid. 99, 527 (1952).
17y, E. Mackenzie, J. Phys. Chem. 32, 1150 (1960).
18_]. A. Mackenzie, Rev. Sci. Instr. 27, 297 (1956).

191 s. Yaffe and E. R. Van Artsdalen, Chem. Div.
Semiann. Progr. Rept. June 20, 1956, ORNL-2159, p. 77.

201—{. R. Bronstein and M. A. Bredig, Chem. Div. Ann.
Progr. Rept. June 20, 1959, ORNL-2782, p. 59.

21y R. Bronstein, A. S. Dworkin, and M. A. Bredig,
Chem. Div. Ann. Progr. Rept. June 20, 1960, ORNL-
2983, p. 65.

22, Baak, Acta Chem. Scand. 8, 1727 (1954).

231—1. Winterhager and L. Werner, Forschungsber.
Wirtsch. Verkehrsministeriums Nordrhein-Westfalen,
No. 438 (1957).

ORNL-DWG 68-5544

12 ,
:
10 /)
I
|
23
el 1 NaF-NaBFZY S NoF -NaBFg?®Y ———
(10’90W'a/o) (20'80 wt ‘70)
e
S KF ~KBF 2>
E6 (10-90 wt%)
K =y
o
¥
*
al //
' LiF-The™
/(78~22mole%)
(t1%) LiF-ThE!"
- le To
2 o G 1%) (72-28 mole %) |
L
(24}
KBF4
|
o I
400 500 600 700 800
TEMPERATURE (°C)
Fig. 12.3. Specific Conductances of Some MSRE-

Related Systems.

where R is the resistance, [/a is the cell constant,
and p is the density of the molten fluoride at the
measuring temperature.

Specific conductance as a function of temperature
is shown in Fig. 12.3 for several LiF-ThF , mix-
tures and alkali fluoride—alkali fluoroborate mix-
tures of relevance to the Molten-Salt Reactor Pro-
gram. (Insufficient experimental details are
contained in ref. 24 to permit a critical evaluation
of the data from this source.)

24y. G. Selivanov and V. V. Stender, Russian J.
Inorg. Chem. 4, 934 (1959).
143

Table 12.2.
System Reference T (OC) 62 K(ohms_1 cm_l) A®¢ (cm2 ohms ™! eq—l)
LiF 19 847-1027 3.805 + 1.004 X 102 T(°C) —~
3.516 x 10~0712 P
12,13 900 1.05 8.43 (t1%) 128
23 875 8.663
958 9.058
1037 9.306
11 870—1010 1.2 9.06 + 5.83 x 10~ 3(T ~ 160.8
870°C) (£1%)
NaF 15 1000 5.52 118
1040 5.74 » (fseveral %)
1080 5.95
12,13 1020 1.05 5.15 (£1%) 113
23 1003 4.960
1086 5.179
1138 5.335
11 1030—-1090 1.2 5.29 + 5.64 X 10~ 3(T — 156.6
1030°C) (+1%)
KF 19 869—1040 —3.493 + 1.480 x 10~2 T(°C) ~
6.608 x 107012 °¢
12,13 900 1.05 3.80 (£1%) 124
23 859 3.573
938 3.793
1012 4.021
20 905 3.77 (£2%)
CSF 19 725-921 —4.511 + 1.642 x 1072 1(°C) —
7.632 x 1078712 ¢
21 737 2.51
784 2.73
852 3.03
BeF, 17,18 700 0.71 x 1077
800 15.3 x 107° } (+i0%)
950 236 x 1075
CaF, 22 1020 1.5,

P T measured (OK)

T melting (°K)
bStandard deviation = 0.008 ohm_1 cm
°Standard deviation = 0.009 ohm_1 cm

-1
—~1
12.4 MEASUREMENT OF SPECIFIC
CONDUCTANCE IN LiF-BeF , (66-34 MOLE %)

G. D. Robbins J. Braunstein

The silica cell shown in Fig. 12.4 was designed
for use in the determination of electrical conduct-
ance in molten LiF-BeF, (66-34 mole %). The

ORNL -DWG 68-5545
ROD, 1 x 30 ¢m

——— Pt WIRE, 1 mm

GLASS TAPE
jK
—]
——
I
> 4

AF/TUBING,O.3X2O cm
| —

THERMOCOUPLE WELL

Fig. 12.4, Silica Conductivity Cell.

144

small-diameter tubing increased the conduction
path and resulted in measured resistances of the
order of 100 ohms. A Wheatstone bridge incor-
porating a variable resistance and capacitance
connected in series was constructed, and its use
eliminated errors resulting from approximations
often employed in conductivity measurements.?>
The platinized platinum electrodes (d = 1 mm)
were held in fixed position relative to the cell
assembly by glass tape.

The cell constant, I/a, of the silica cell was de-
termined at 340 and 380°C in molten potassium
nitrate by measuring the resistance in the frequency
range 600 to 4000 hertz and extrapolating to in-
finite frequency vs f~!/2. The observed frequency
dependence is shown in Fig. 12.5. From the
infinite-frequency resistance R_ and literature
values of the specific conductance « vs tempera-
ture for KNO,,?¢ the cell constant was determined
to be 129 cm—! at both temperatures. These quan-
tities are related by the expression

K :El- (I/a) . (1)

e o]

The frequency dependence of measured resistance
of LiF-BeF  (66-34 mole %) is shown in Fig. 12.6
at several temperatures. The specific conductivity,
calculated from the values of resistance at infinite
frequency and Eq. (1), is presented in Fig. 12.7 as
a function of temperature. From the slope of this
line, the temperature coefficient at 500°C is cal-
culated to be 3.9 x 1073 (°C)~!. Employing the
density data of Cantor?” and the relation

mol wt
A= ———— (2)
density

the molar conductance A™ at 500°C is 25.5 cm?
ohms~! moles~!. The measurements were com-
pleted during the first hour the cell was in con-
tact with the molten fluoride, and we believe the
accuracy to be better than £10%.

It was observed that in the molten fluoride sys-
tem the measured resistance increased slowly as
a function of time. In a subsequent experiment

25G. D. Robbins, Electrical Conductivity of Molten
Fluorides, a review, being reviewed for publication.

264, Klemm, p. 567 in Molten Salt Chemistry (ed. by
M. Blander), Interscience, New York, 1964.

27y, Cantor, internal memorandum, 1967.
145

ORNL-DWG 68-5546 ORNL -DWG 68-5548

f (Hz}

4000 2000 1200 800 600

| !
210 —_205.4 — -"—.—'—.7‘#’3_8’5: —
- —'r !

200 —T— T — AIJ S

« (ohms/cm)

A (ohms)

190 - —— _L . 47 . %_
Wl

171.8 o—*T"340 |
| ."'.T | . |
| ‘ ‘ 450 470 490 510 530
170 :

o) oo 0.02 0.03 0.04 0.05 TEMPERATURE (°C)

Fig. 12.7. Specific Conductance vs Temperature for
LiF-BeF2 (66-34 Mole %).

Fig. 12,5. Measured Resistance vs Frequency in

Molten KN03.
ORNL- DWG 68-5549

180 T T L °
F o%- ’
ORNL-DWG 68-5547 ‘ ‘ \
f (Hz) ‘

4000 2000 1200 800 600 160

100

80

Fig. 12,6. Measured Resistance vs Frequency in

Molten LiF-BeF2 (66-34 Mole %). f—]/2 at

Fig. 12.8. Slope of Measured Resistance vs
485°C in LiF-Ber (66-34 Mole %).

employing a different cell [(I/a) = 164 (1) cm™1 ]

and platinized platinum electrodes which had been value of resistance extrapolated to infinite fre-
previously exposed to molten LiF-BeF, (66-34 quency also varied (Fig. 12.9). This cannot be
mole %), the slopes of the measured resistance vs explained in terms of changes in cell constant,
f~1/2 plots were investigated as a function of since (I/a) was redetermined at the completion of
time. The results are shown in Fig. 12.8. The the experiment to be 171 cm~!. Investigation of
frequency range over which R vs f~1/2? was linear the frequency dependence of measured resistance

increased at longer exposure times; however, the is proceeding.
146

ORNL-DWG 68-5550

140 ——————— ,
130 / —_ = —

0 20 40 60 80 100 120
TIME (hr)

A (obhms)}

S
-
A S
|
!
|
|
|
J
—
|
|
|
|

Fig. 12.9. Values of Resistance Extrapolated to
Infinite Frequency vs Time for LiF-BeF2 (66-34 Mole %)
at 485°C,

12.5 A STIRRED REACTION VESSEL
FOR MOLTEN FLUORIDES

C. E. Bamberger
T. J. Golson

C. F. Baes, ]Jr.
J. Nicholson

In the past, gas sparging, stirring, and rocking
have been used to provide agitation in studies of
heterogeneous equilibria involving molten fluorides.
The stirred vessels, while providing the most vig-
orous mixing, tended to leak at the seal of the
stirrer shaft. A rocking furnace or gas sparging
did not provide very vigorous agitation.

More recently, the need for an improved reaction
vessel became urgent in connection with the
planned study of the distribution of U**, Th*™,
and Pa* " between fluoride and oxide phases.??
The attractiveness of such an extraction system as
a protactinium recovery process obviously depends
as much on favorable equilibration rates as on favor-
able equilibrium behavior. But to achieve good ex-
traction rates, one needs good suspension and mix-
ing of the heavy oxide solids with the fluoride salt;
this requires a suitably designed mixer.

An improved stirred vessel has therefore been de-
signed and constructed. It includes the following
features (Fig. 12.10):

1. The shaft is sealed by means of Teflon chevron
rings which are stacked around the shaft and
compressed by a nut. This seal is water cooled.

28p R, Hitch, C. S. Bamberger, and C. F. Baes, Jr.,
MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 136.

2. The upper (cool) shaft bearing is a sleeve of
oil-impregnated metal. The lower (hot) bearing
is a sleeve of graphite.

3. The stirrer is driven by a flexible cable con-
nected to a variable-speed dc motor. It allows
controlled stirrer speeds in the range 60 to
1750 rpm.

4. The shaft and stirrer blade (of INOR-8) can
easily be removed and replaced. The vessel,
flanged to the stirrer assembly, is cheap to
fabricate and is easily replaced.

The apparatus currently is being used to com-
plete a study of the reactions of SiO, with LiF-
BeF, melts (see Sect. 12.6). The helium cover
gas, maintained at ~ 0.4 atm (gage), is circulated
continuously through the melt and then around a
closed circuit which includes an infrared gas ab-
sorption cell. Measurement of the infrared ab-
sorption bands of SiF, is used to monitor the
course of the reaction. The growth of the CO,
absorption peak also provides a sensitive mea-
sure of the rate of oxygen inleakage, since CO,
is produced by reaction of oxygen with the traces
of graphite present in the melt.

With the stirrer-shaft seal lubricated by a small
amount of oil, the stirrer has been operated con-
tinuously at >300 rpm for several days at a time.
While wear of the Teflon rings requires an occa-
sional tightening of the packing nut, the out-
leakage of helium is easily maintained below
~1 cm Hg of pressure per day, while the rate of
air inleakage, as indicated by the CO, content of
the cover gas, has been nearly undetectable.

This reaction vessel should prove a versatile
apparatus for a variety of future studies of
heterogeneous equilibria involving molten salts.

12.6 THE CHEMISTRY OF SILICA
IN MOLTEN LiF-BeF,

C. E. Bamberger C. F. Baes, Jr.

As reported previously,??-3? silica has been

found to be a satisfactory container for melts in

29¢. E. Bamberger, C. F. Baes, Jr., and J. P. Young,
J. Inorg. Nucl. Chem., to be published.

30C. E. Bamberger, R. B. Allen, and C. F. Baes, ]Jr.,
Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1967,
ORNL-4229, p. 60.
147

|
! FLEXIBLE DRIVE

THERMOCOUPLE WELL
__/ ] /—
SPARGE GAS

TUBE

SAMPLING TUBE

OFF GAS
OFF GAS

HZO COOLING COlLS

"0"RING SEAL

GRAPHITE BEARING

NI VESSEL

STIRRER
INOR-8

CENTERING GUIDE

Fig. 12.10. Improved Stirred Reaction Vessel for Molten Fluorides.

OILITE BEARING

TEFLON CHEVRON
PACKING SEAL

ORNL-DWG 68-3432

THRUST BEARING

PACKING NUT
148

which Li,BeF  is the solvent salt. The partial
pressures of SiF generated by the reaction

$i0,(c) + 4F~ = 20%~ + SiF () (1)

generally are low, and the oxide ion produced by
this reaction can easily be held to low levels if
SiF is introduced into the cover gas. In visual
experiments and in spectrophotometric studies,
silica containers were found to remain intact and
did not contaminate the melt provided the SiF4
generated was retained in the cover gas or if SiF,
was already present in the cover gas. The chem-
istry of Si0, in the presence of MSR fluoride mix-
tures is being studied further with the objectives
of (1) establishing the equilibrium solids which
can coexist with such melts and (2) examining
their ion exchange properties.

If the above reaction is allowed to proceed to the
right by sweeping away the SiF 4 produced, the
oxide ion concentration should increase until a
new oxide-containing solid phase precipitates.
Beryllium oxide is normally the oxide phase which
first precipitates in molten Li,BeF , but according
to the Si0,-BeO phase diagram, SiO2 should react
with BeO to produce Be ,5i0, (phenacite),

2BeO(c) + Si0,(c) = Be,SiO, . 2)

Hence, unless some more stable oxyfluoride phase
can form in the presence of the molten fluoride, it
is expected that as reaction (1) proceeds far
enough to the right, phenacite should appear as a
stable solid phase, establishing the equilibrium

25i0,(c) + 2BeF ,(d) = Be,Si0,(c) + SiF ,(¢) , .

2
SiF4/aBeF

K=P .
2

This equilibrium has been investigated directly
by the recirculation of helium carrier gas through
an Li,BeF, melt to which both Si0, and Be,SiO,
had been added and then through an infrared ab-
sorption cell, where the intensity of the SiF, peak
at 1033 cm~—! was determined. The gas was pumped
around this circuit by means of a finger pump.

The following expression reproduces the partial
pressure measurements within their estimated un-

certainty (+10%, Fig. 12.11):

log P

SiF

(mm) = 8.745 — 7576/T . @
4

ORNL-DWG 68-1617R

7 (°C)
800 700 600 500 400
200 1 I W 1
a
100 - —]
Ak) )
\
N\
10 \ b — —
[ N\ \\
S \.__ \
N\ \ -
\\ \ |
z N ‘ —
£ - NN
~— N\
Lgl‘ \ \
g N e
o L A W——
° - R
SN
\ o %
AAY
\ ®y
N\
\
0.1 e
] N ]
- N |
I | ]
\‘
\
0.04 A
0.9 1.0 1.4 1.2 1.3 1.4 1.5
1000/7 (ek)

Fig. 12,11, Partial Pressures of Sil"'4 in Equilibrium
with 5502, Be2$i04, and Molten 2LiF-BeF2. The point
(O) at upper left is from Novoselova et al.32 The solid
line is based on Eq. (4); the dashed line is based on

available thermochemical dafa.30'31

The observed partial pressures are in general
slightly higher than may be predicted from avail-
able thermochemical data3°~32 (cf. Fig. 12.11).
These partial pressures are seen to be quite low,
however, confirming that quite modest overpres-
sures of SiF", should prevent the precipitation of
an oxide-containing phase in the presence of SiO,.
At present, a stirred vessel (see Sect. 12.5) is
being used in continuing these measurements. The

31JANAF Thermochemical Tables, Clearing House for
Federal Scientific and Technical Information, U.S. Depart-
ment of Commerce, August 1965.

324, V. Novoselova and Yu. V. Ashikina, Inorg. Ma-
terials USSR (English Transl.) 2(9), 1375 (1966).
149

ratio of BeO to 5i0O, in the added solids is being
varied to ascertain whether or not any stable oxide-
containing phases other than Be 3iO_, SiO,, and
BeO are formed in the presence of the molten fluo-
rides.

12.7 A SILICA CELL AND FURNACE FOR
ELECTROCHEMICAL MEASUREMENTS WITH
FUSED FLUORIDES

K. A. Romberger

Two distinct advantages are gained if an electro-
chemical cell can be constructed from fused silica
rather than from graphite or corrosion-resistant
metals. First, fused silica provides excellent
electrical insulation for the electrodes, and second,
the cell components are visible so that formation
of gas bubbles, solids, etc., may be detected
rapidly. Even so, fused silica has seldom been
used with fused fluorides in the past because of
the corrosive attack of the fluoride salt on the
silica. Recently, however, Bamberger et al.33
suggested that the presence of an SiF ,-containing
atmosphere would reduce the rate of attack of some
molten fluorides on fused silica. This protection of
the silica should be the greatest with those fluoride
salts which have a relatively low basicity or alkali
fluoride activity, for example, LiF-BeF, mixtures
that contain 0.33 or higher mole fraction of BeF .

In order to utilize the advantages of silica, a
silica cell and furnace has been constructed for
electrochemical studies in LiF-BeF2 mixtures.

The design is such that it will permit, in a single
system with a controlled atmosphere, both equi-
librium and nonequilibrium electrochemical meas-
urements including emf, chronopotentiometry, linear
sweep voltammetry at fixed solid or liquid elec-
trodes, rotating disk voltammetry, and polarography
at a dropping (bismuth) electrode. The purpose of
these measurements is to yield information about
the diffusion coefficients, decomposition potentials,
and oxidation states of the species dissolved in
the molten salt.

The heating element of the furance is a 20-gage
asbestos-covered Nichrome wire helix supported by
a vertical 4-in.-OD silica tube. A 21/2-in.-ID, 6-in.-
high silica cup, centered within the heated zone,

33c. E. L. Bamberger et al., MSR Program Semiann.
Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 137.

holds the fused salt. Atmospheric containment (and
secondary salt containment) is provided by an 11-
in.-long, 3-in.-ID closed-end silica tube. This
latter tube extends upward to the top of the furnace,
where it is joined, via a silicone rubber seal, to a
water-cooled nickel collar.

A nickel head having eight ports which terminate
in Swagelok fittings provides entry to the cell.
Five ports are vertically aligned risers. Three of
these risers are used for entry by electrodes. A
fourth riser is used either for solid sample addi-
tions or for a fourth electrode. The fifth riser
admits either a stitrer or a rotating disk electrode
through a Swagelok tee fitted with Teflon seals.
Helium pressure is maintained through the tee to
provide an inert-gas seal. The remaining three
ports are for a thermocouple well and inlet and
outlet gas lines.

The usable working depth in the cell of approxi-
mately 4 in. is centered on a line between two 6-
in.-square windows. The heating element is sur-
rounded by a platinum-covered nickel reflector.
Insulation is of Lavite. The furnace requires a
208-v ac source and can be safely operated at
800°C, at which temperature the required power
is 2700 w. A detailed drawing of this cell and
furnace is shown in Fig. 12.12.

12.8 STATUS OF THE MOLTEN-SALT
CHEMISTRY INFORMATION CENTER

H. F. McDuffie

In the summer of 1965, we decided to establish
a Molten-Salt Chemistry Information Center. The
primary purpose was to provide a detailed subject
index to all the chemical material representing
ORNL work in the field since 1950. A secondary
purpose was to provide a framework into which
other technical reports and open-literature publi-
cations could be inserted. The center has never
been formally funded but has required approximately
half the time of one technical man and half the time
of one secretary. Because of the low level of
funding available and the purposes of the center,
we designed it so that it could be used by tech-
nical staff members without assistance and so
that new entries could be created with the minimum
investment of technical time.

An optical coordination indexing system (Terma-
trex) was chosen as being the most appropriate
150

ORNL—DWG 68-4387

ELECTRODES

, SWAGELOK FITTING
TEFLON LINERS

— —’ COVER OR

HER P
THERMOCOUPLE D— —0 ¢mem SPARGING GAS

HEAD ( NICKEL) 5 e B

v TEFLON O-RING GASKET
COOLING COLLAR (NICKEL)
‘Oq

— I WATER

TOP PLATE (STAINLESS STEEL)

&)
[ [\
L AN I
FRAME (STAINLESS STEEL) —
NICHROME —| - LAVITE
@ N /
N A /
REFLECTOR (PLATINUM) SN y
IRl ¢ y g
T e || |
GRAPHITE —F—F—] ’q s N _
: / @I -

L l I !
INCHES

OVERFLOW CuP
(STAINLESS STEEL)

SUPPORT AND SALT e
\

BOTTOM PLATE
(STAINLESS STEEL) % /208 volt AC
— <

Fig. 12,12, Silica Furnace and Cell for Electrochemical Measurements.
(simplest and cheapest) for the purpose. This is
the same system used by the ORNL Isotopes In-
formation Center; it was possible to use some of
their equipment in setting up the Molten-Salt
Chemistry Information Center.

The two most important functions, from the tech-
nical point of view, in establishing an information
center for the retrieval of the information contained
within are the selection of an adequate list of key
words for indexing purposes and the assignment of

these key words to particular documents or portions

of documents. We have tried to use care in the
selection of our key-word list and have, so far,
lirrited it to 480 terms. The list is broken into
two parts, general terms and materials. General
terms include words and phrases such as ABSORP-

TION SPECTRA, BIBLIOGRAPHY, BLANKET RE-

PROCESSING, CALORIMETRY, DENSITY OF
LIQUID, HYDROLYSIS, KINETICS, etc. Materials
words include more specific items such as ACTI-
NIDES, ALUMINUM, ARGON, GLASS, GRAPHITE,
HF, KF, LiF-BeF2 (the system), MERCURY,
MOLYBDENUM, MSRE FUEL, etc. Technical
people selected these key words, and technical
people have made the assignments of key words
to articles.

Clerical and secretarial people have handled all
the other aspects of entering the information into
the system. A 5 by 8 card is typed containing the
assigned serial number, the identification of the
document, the abstract (if available), and the as-
signed key words. A single cross-reference card
is typed, showing the serial number and the iden-
tification of the document; it is filed by journal or
report number. Then the Termatrex cards for the
several selected key words are assembled and
punched with the serial number of the document.
(The arrangement of the serial number punches is
on a 100 by 100 grid, so that any number up to
10,000 can be located by a separate hole.)

In addition to the indexing of relevant reports
which are brought to our attention, we maintain a
subscription to the ASCA (automatic subject cita-
tion alert) function of the Institute for Scientific

151

Information. Each week, their current scientific
literature is scanned for articles written by any of
106 source authors whom we have chosen as pub-
lishing wholly or significantly in the molten-salt
field. The titles of all the articles thus brought to
our attention are screened, and the relevant articles
are indexed, thus previding some coverage of cur-
rent scientific literature as well as of our own
reports.

At the present time the system has accumulated
a total of 6600 entries; the rate of accumulation
has been somewhat more rapid than we had ex-
pected, particularly in view of the low level of
effort which we have assigned to the work. The
system has been most useful to new members of
our staff in permitting thein ready access to the
earlier project work in the Molten-Salt Reactor
Program. Additional support, which would be ex-
pected to materialize with an increase in the
overall program, would permit us to expand our
coverage of both the report literature and the
open literature through systematic surveying of
Chemical Abstracts and Nuclear Science Ab-
stracts for relevant documents. When the storage
capacity, 10,000, of the present system is reached,
we can set up a second deck of Termatrex cards
and continue with another 10,000 entries, or we
can consider whether we have arrived at the stage
of justifying the staff and complexity required to
put the information into a digital computer data
processing system. This operation would be es-
sentially nontechnical, since the system was de-
signed with this ultimate conversion in mind.

The flexibility of the system permits continuous
improvement of the indexing and easy correction
of any errors which are found. For the present,
it seems preferable to maintain the center as a
device for retrieval of requested information rather
than as a place for continuing analytical reviews
with a regular output to customers. We can answer
some telephone queries quickly and can handle
some requests from outside writers for specific in-
formation, but we do not have the staff to do
elaborate state-of-the-art studies at this time.
13. Chemistry of Molten-Salt Reactor

Fuel Reprocessing Technology

13.1 MSBR FUEL REPROCESSING BY
REDUCTIVE EXTRACTION INTO
MOLTEN BISMUTH

D. M. Moulton W. R. Grimes
J. H. Shaffer

Studies of the reductive extraction of protactinium,
uranium, and rare earths from molten fluoride mix-
tures into molten bismuth have continued to yield
results which favor its application for reprocessing
fissile and fertile mixtures of the two-region molten-
salt breeder reactor concept. As previously re-
ported in this series, the experimental data relate
the equilibrium distribution of extracted metals to
concentrations of reducing agent dissolved in the
metal phase! and permit the construction of an
electromotive series where the relative positions of
metals indicate their order of extraction.? These
values have demonstrated the chemical feasibility
of reductive extraction for reprocessing a two-region
breeder reactor within a reasonable degree of confi-
dence and have provided a basis for initiating an
engineering evaluation of the processing concept.

More recent efforts have been primarily oriented
toward reprocessing requirements of the conceptual
single-region breeder reactor. This concept com-
bines both fissile and fertile material in a single
fluoride mixture which will nominally contain 12
mole % ThF4 and 0.3 mole % UF, dissolved in an
LiF-BeF solvent mixture, The reprocessing re-
quirements remain the same, but the chemistry be-
comes more difficult since there are now two com-

I1MSR Program Semiann, Progr, Rept, Feb, 28, 1966,
ORNL-3936, p. 141,

2MSR Program Semiann, Progr, Rept, Aug. 31, 1967,
ORNIL.-4191, p. 148.

152

binations of elements with similar reduction
potentials, Earlier data for a somewhat different
solvent, Lil"‘-Be13‘2~T11F4 (73-2-25 mole %), indicate
that uranium will have to be removed as the first
step, with protactinium following under slightly
stronger reduction conditions. If the separation
factor (~10) obtained from these data is valid for
the single-region system, the process of separating
uranium from protactinium should be feasible but
will require more careful process control, Further
studies of the extractability of these elements in
the single-region fuel mixture are in progress.

Fission product removal will logically be per-
formed on the salt stream effluent from the uranium-
protactinium removal unit ahead of the fuel re-
constitution section of the reprocessing plant. Con-
sequently, laboratory studies are concerned with
the extraction of rare earths from a mixture of LiF,
BeF " and ThF where the concentration of thorium
will be about 12 mole %, lithium may be varied from
58 to 72 mole %, and beryllium will make up the
balance., In a preliminary experiment a small
quantity of ThF4 was added to LiF-BeF, (66-34
mole %) containing about 100 ppm europium spiked
with 152—154FEy activity, Extraction from this salt
mixture (salt A) into bismuth was carried out as in
previous experiments by making metallic lithium
additions to the metal phase and by monitoring the
progress with a beryllium electrode. Reduction po-
tentials calculated from the measured distributions
are shown as part of Table 13.1, which contains
current values for several elements obtained from
the same salt mixture, If the thorium activity re-
mains equal to its concentration, then at 12 mole %
ThF , in the single-region fuel mixture, thorium bis-
muthide will precipitate at —1.49 v, The distribu-
tions of various important elements between salt A
and bismuth at 600°C are illustrated in Fig. 13.1 as
153

Table 13.1. Extroction Potentials of Fission Products from LiF-BeF, (66-34 Mole %) into Bismuth

Element Valence 80/
500°C 600°C 700°C 800°C
Li 1 2.00 1.93 1.88 1.83
Be? 2 1.93 1.85 1.78 1.71
Ba 2 1.80
Eu b 1.69(2.0) 1.61 (1.8) 1.53(1.8) 1.45(1.8)
Th 4 1.56 1.50
Nd 2.5 1.52
Ce b 1.58(3.0) 1.48(2.6) 1.41(2.4) 1.33(2.2)
Sm 1.6 1.50
La 2.7 1.48
U 4 1.39

880 (as referred to pure metal); Be is insoluble in Bi.

bValence in parentheses at each temperature,

functions of reduction potentials, The vertical lines
mean that a solid phase of constant composition is
being formed and will limit the reduction potential
of the system until its concentration in the salt
phase is exhausted. For thorium, the virtual posi-
tion of 80 is taken from dilute measurements in
salt A, but the onset of solid phase formation is
calculated for ThF =12 mole % in the salt. The
important considerations for the single-region MSBR
reprocessing scheme are the voltage difference be-
tween uranium and protactinium and the rare-earth
positions relative to the vertical thorium line If
these values are valid, the extraction process
should be feasible but nol a> favorable as its ap-
plication with the two-region reactor.

Because of the uncertainty of thorium activity at
higher thorium concentrations in the salt mixture

and its effect on the extraction process, a systematic

investigation of rare-earth extraction from salt mix-
tures of practical interest as a fuel solvent has
been initiated. Since increasing the LiF concen-
tration of the mixture (hence increasing its free

fluoride content) might increase the stability of
Th** by fluoride complex formation,? initi~l

studies have held the thorium concentratio.. of the
salt at 12 mole % and varied the ratio of lithium to
beryllium. Three experiments with europium,
samarium, and neodymium, each dissolved in LiF-
BeF -ThF , (58-30-12 mole %), were conducted by
adding thorium metal to the bismuth pool. Thorium
saturation in bismuth was inferred when the po-
tential of a beryllium electrode inserted in the salt
mixture stopped falling upon further addition of
thorium metal. The amount of extracted rare earth
in each case was too low for quantitative evaluation
under analytical conditions imposed on the systir:,
Radiochemical analyses were further complicated v
the presence of gamma-emitting thorium daughters,
some of which extracted readily into bismuth.
Neodymium was seen in the bismuth at 700 and

33, Cantor and C. T, Baes, private communication,
December 1967,
ORNL-DWG €8-3119

4 .
[ [ [ | !
NOTE: FORMATION OF ThBi, CALCULATED
— FOR ThF, =12 mole % IN SALT
o |
HE :
Ele ‘
5|8
ol
ge /
" /
1 /
7
Cr——71 7

LOG DISTRIBUTION COEFFICIENT (D

- ——t

| |

1.5 1.6 1.7 1.8 1.9
£. REDUCTION POTENTIAL (volts)

Fig. 13.1. Distribution of Uranium, Protactinium, and
Rare Earths Between LiF-BeF2 (66-34 Mole %) and
Bismuth at 600°C,

800°C, but samarium, though it should extract more
easily, was not definitely detected, probably be-
cause it has only a weak and short-lived radioiso-
tope. Potentiometric readings in this salt mixture
did indicate changes in the activities of the major
salt constituents from values reported for salt A.
Values for beryllium and thorium were found about
0.1 and 0.05 v, respectively, more noble than in
salt A if lithium is assumed unchanged.

Since the feasibility of the extraction process de-
pends on the relative extractability of rare earths as
compared with thorium, the experimental procedures
are being refined to yield significant values for
very low extraction efficiencies. If we define this
separation coefficient a as

XRE(BI) /XRE(salt)
X

o =

Th(Bi) /XTh(salt)

154

values for o appreciably above 1 will denote feasi-
ble separations, and the magnitude of a will indi-
cate extraction efficiency. Thus, if the concentra-
tion of rare earth in the salt mixture is 100 ppm by
weight, the thorium concentration in the salt is 12
mole %, and the limiting potential corresponds to
3500 ppm thorium (i. e., XTh(Bi) =3.14x 1073) in
the bismuth, then the experimental procedure should
detect rare-earth concentrations as low as 0.8 ppm
in the bismuth phase. Experiments are now in
progress to examine the effects of salt composition
on the extractability of rare earths, These experi-
ments will be conducted with cerium because of the
convenient radiochemical properties of the 144Ce
isotope for analytical application. When the salt
composition has been optimized, extractions of the
various important rare earths will be studied. The
results of one experiment (still in progress) with
the salt mixture LiF-Ber-ThF4 (72-16-12 mole %)
have shown the favorable extraction of cerium into
bismuth at 600°C. Figure 13.2 illustrates the de-
pendence of rare-earth extraction on the thorium
concentration in bismuth up to its saturation value.
The separation coefficient o is also related to the
thorium concentration in bismuth, as shown in Fig.
13.3. The decrease in a with increasing thorium
results from the higher valence of thorium than
cerium,

ORNL-DWG 68-1692

T T | |
INITIAL CONC OF Ce IN SALT =107 ppm
WEIGHT OF SALT =3.26 kg
WEIGHT OF Bi=3.00kg
e =POINTS AT 700°C
4 0 =POINTS AT 600°C
Ce CONTENT OF SALT REMAINED AT

107 * 2 ppm DURING ENTIRE EXPERIMENT

e

2 o
e

/

CERIUM FOUND IN BISMUTH (ppm BY wt)

0 0.5 1.0 1.5 2.0 2.5 3.0
THORIUM FOUND IN BISMUTH (mole fraction x103)

Fig. 13.2. Extraction of Cerium from LiF-Ber-Th F4
(72-16-12 Mole %) into Molten Bismuth by Reduction
with Thorium at 600°C.
o ORNL—-DWG 68-1690
1

CERIUM CONCENTRATION IN SALT = 107 ppm BY WEIGHT

= |
D= NMETAL /NSALT !
WEIGHT OF SALT = 3.26 kg |

|
|
7 WEIGHT OF Bi=3.0 kg T—’*F{{[“‘A
‘ | | | |
L]

|
0
0 0.5 1.0 15 2.0 2.5 3.0

THORIUM FOUND IN BISMUTH (mole fraction x 103 )

DISTRIBUTION RATIO OF Ce AND Th (Dge/D1p)

Fig. 13.3. Relative Extraction of Cerium and Thorium
from LiF-BeF2-ThF4 (72-16-12 Mole %) into Bismuth at
600°C.

13.2 REMOVAL OF STRUCTURAL METAL
FLUORIDES FROM A SIMULATED MSRE
FUEL SOLVENT

F. A. Doss L. E. McNeese
J. H. Shaffer

Current plans for the MSRE call for substituting
233y for 235U in the reactor fuel mixture. The fuel
solvent mixture, LiF-BeF2—ZrF4 (65-30-5 mole %),
together with its fission product inventory, will be
retained; this will be accomplished by fluorinating
the fuel salt to remove its present uranium content
and by adding 233U as the concentrate mixture LiF-
UF, (73-27 mole %). During fluorination the salt
will become contaminated with fluorides of nickel,
iron, and chromium as a result of corrosion of the
Hastelloy N fluorinator. Although the expected con-
centrations of these fluorides will not exceed their
solubilities in the fuel salt, their return to the re-
Therefore, the reduction of
structural metals from solution in the fluorinated
fuel solvent and their removal by filtration from the
solvent during its transfer to the reactor system are

actor is undesirable.

requirements of the refueling operation. An experi-
mental simulation of these steps has been conducted
to define better the feasibility of the refueling pro-
cedure,

155

The experimental program was carried out with a

104-kg batch of the fluoride mixture LiF-BeF -ZrF
2 4

(65-30-5 mole %), prepared from the component
fluoride salts by routine production procedures. In
order to represent conditions expected after fluori-
nation of the MSRE fuel salt, approximately 80 g of
CiF,, 50 ¢ of FeF , and 800 g of NiF, were added
to the salt mixture, which was hydrofluorinated for
6 hr with 10 to 20 mole % HF in helium to ensure
complete dissolution of the corrosion product fluo-
rides, The analysis of a filtered salt sample taken
after the hydrofluorination period is compared with
expected concentrations in Table 13.2.

Table 13.2. Concentrations of Corrosion Product

Fluorides after Hydrofluorination

Estimated Concentration? Analysisa
Material (PPm) (ppm)
CrF2 445 470
Fer 286 620
NiF 4680 3700

4Concentrations reported on a metal basis,

13.2.1 Reduction of Structural Metal Fluorides

Equilibrium data* for reduction of the corrosion
product fluorides indicate that if equilibrium is
achieved, a hydrogen sparge will reduce NiF
easily, FeF2 with difficulty, and CrF2 to a negli-
gible extent in a practical time period. The aim of
the first part of the experiment was to demonstrate
NiF2 reduction with a hydrogen sparge and to com-
pare hydrogen utilization with that predicted from
equilibrium data. After the initial hydrofluorination,
free HF was stripped from the salt mixture by helium
sparging, and the molten salt was sparged for 9 hr
with hydrogen at a rate of 10 standard liters per
minute at 650°C. The 48.2-liter salt mixture was
contained in a 12-in.-diam vessel, and the sparge
line terminated 26 in, below the melt surface, The
gas effluent from the treatment vessel was analyzed

‘c. M. Blood, Solubility and Stability of Structural
Metal Difluorides in Molten Fluoride Mixtures, ORNL-
CF-61-5-4 (Sept. 21, 1961).
ORNL-DWG €8-5551

T =T
H, FLOW: 10 liters/min NTP
SALT VOL: 48.2 liters
SALT wt: 104.4 kg
INJ 7 3700 ppm

o

(¢}

[N

-

HF FOUND IN GAS EFFLUENT (meg/titer H,)

REACTION TIME {hr)

Fig. 13.4. Reduction of NiF2 from Simulated MSRE
Fuel Solvent by Hydrogen Sparging at 650°C.

periodically for HF (Fig. 13.4), and sparging was
discontinued when the HF concentration decreased
to a low but near constant value, The quantity of
HF produced during this period was equivalent to
about 93% of the NiF2 initially present in the salt
mixture. Hydrogen utilization was considerably
lower than would have resulted if equilibrium had
been achieved; under equilibrium conditions a
sparge time of 20.4 min would have resulted in re-
duction of 99% of the NiF2 present initially.

Since equilibrium data predict very low hydrogen
utilization during reduction of FeF_ and CtF _, data
were obtained on reduction of these fluorides with
zirconium metal, The zirconium was first machined
in order to increase the surface area and thereby
reduce passivation of the surface by deposition of
reduced iron and chromium. The turnings were
pressed into 344-in.-diam cylinders (p ~4.6 g/cm? )
having lengths of 0.3 to 1.0 in. to facilitate addition
to the treatment vessel, The pressed zirconium
slugs were added to the salt mixture after reduction
of NiF | with hydrogen sparging. After each addi-
tion the approximate extent of reduction was noted
by sparging the melt with hydrogen and analyzing
the effluent gas for HF. As shown in Fig. 13.5, the

156

ORNL-DWG 68-5552

T |
|
|

_+__

SALT COMP:
LiF 65mole T
BeF2 30mole %
ZrF4 5 mole %
SALT wt: 104 4 kg

0.6 — ]—

T

ND- PCINT BY EMPIRICAL RELATION TO GAS ANALYSIS

jfl
—

:

HF FOUND IN GAS EFFLUENT (meq /liter Hy)
o
o~
’:L

STOICHIOMETRIC POINT BY CHEMICAL ANALYSIS

|

CALCULATED STOICHIOMETRIC POINT

]

|

|
"
T
|

i

FIRST SALT SAMPLE TAKEN
SECOND SALT SAMPLE TAKEN

ol

o] 100 200 300 400 500
ZIRCONIUM METAL ADDRED (g)

Fig. 13.5. Pseudoequilibrium of Hydrogen with
Structural Metal Fiuorides During Their Reduction from
Simulated MSRE Fuel Solvent by Zirconium Metal at
650°C.

HF concentration decreased steadily, and an arbi-
trary end point for the reduction reaction was
established at a concentration of 0.02 meq of HF
per liter of H, as a result of experience during
preparation of the initial MSRE fuel carrier salt.
The calculated quantity of zirconium required to
complete reduction of the corrosion product fluo-
rides was 115 g; this was based on the analysis of
the salt mixture shown in Table 13.2 and assumed
reduction of 12.2 equivalents of these metal fluo-
rides during the hydrogen reduction period.
Analyses of filtered samples taken near the con-
clusion of the zirconium addition period are shown
below.

Zirconium Added Metals Found (ppm by weight)

() Cr Fe Ni
253 9 60 <30
327 26 44 36
157

These values are below limits established for the
initial MSRE fuel carrier salt. On the basis of these
data, zirconium utilization was at least 46 to 65%.

13.2.2 Salt Filtration Studies

The reduction of structural metal corrosion products

from solution in the MSRE fuel solvent is antici-
pated to produce approximately 330 g (0.73 1b) of
solid particles per cubic foot of salt mixture. The
suspension or entrainment of a major fraction of
these solids during the transfer of the fuel solvent
back to the reactor drain tank assembly could have
adverse effects on filter performance. Therefore,
the primary purpose of this program was an exami-
nation of salt filtration characteristics under con-
ditions which would reasonably simulate those
anticipated in the reactor application. Filtration
tests were separately conducted on a sintered nickel
product of the Micro Metallics Corporation, having
pores about 40 ;4 in diameter, and two grades of
Feltmetal, a product of Huyck Metals Company.
Each filter was fabricated as a 25/8—in.-diam plate

so that geometric surface areas of all filters would
be identical. One grade was of Monel with pores

20 u in diameter and a second of 347 stainless steel
having 41-y-diam pores. The latter, designated
FM250 by the manufacturer, had a porosity of 78%
and a rated capability for removing 98% of all
particles over 10 y in diameter. This material was
chosen for application in the reactor system because
of its adaptability to fabrication requirements and
its satisfactory performance during these filtration
tests,

The credibility of the multiple filtration tests that
were performed on a single salt preparation is based
on the anticipation that the primary process for re-
moving solids from the MSRE fuel solvent will be
that of decantation. This conclusion is supported
by considerable experience in filtering fluoride mix-
tures through sintered nickel during the production
of fluoride mixtures for the MSRP. Examinations of
filters used in this process have consistently shown
the collection of very small quantities of solids on
the filter plates. This condition has persisted
despite the accumulation of considerable quantities
of solids in the salt treatment vessels during their
repeated use in the production process. Thus, the
novel feature of these current filtration tests is,
perhaps, the indirect estimate of particle size dis-
tribution of solids formed by the reduction process
proposed for the reactor application.

The experimental assembly used for the filtration
tests was essentially that used for the routine pro-
duction of fluoride mixtures. The salt preparation
vessel was connected by a 1/Z-irx.—diam Inconel tube
to a heated salt receiver vessel, This transfer tube
extended near the bottom of the salt preparation
vessel and only penetrated the top of the receiver,
The porous metal filter was mounted vertically in
the transfer line above the receiver vessel, A
second salt return line extended near the bottom of
the receiver vessel and connected to the salt prepa-
ration vessel. These lines were used alternately
for forcing the 1.7 ft3 of salt mixture from one ves-
sel to the other and were capped when not in use.
This arrangement permitted repeated filtration tests
to be conducted on the salt mixture. Eight filtra-
tions were conducted during this investigation.

Although conditions which would promote the
decantation process are desired for the reactor
application, attempts were made also to observe
filter loading capacities. Tests under static con-
ditions prevailed when the melt had remained
quiescent for a minimum time of 4 hr prior to trans-
fer. Agitated melt conditions were established by
rapidly sparging the melt with helium just prior to
salt transfer. The temperature of the melt was con-
trolled at 650°C (1200°F) for all tests. The indi-
cated temperatures of the salt transfer line fell to
about 565°C (1050°F ) during the extended transfer
operations. Lower temperatures were erroneously
recorded from thermocouples that became separated
from the transfer line., In no instances did the salt
freeze in the transfer line or filter during the tests.

In each test, salt transfer was induced by in-
creasing to 11 psig the pressure of helium above
the salt mixture in the treatment vessel and evacu-
ating all gases from the receiver vessel. These
conditions were maintained throughout each test.
The pressure drop across the filter varied as the
level of salt in the treatment vessel decreased. A
summary of the filter performance tests is shown in
Table 13.3. The 20-y Feltmetal was tested first,
and it was disqualified after 2 hr operation, Es-
sentially no salt was passed through the filter
during this period. Inspection of the filter also
showed that the predominant part of the 22-g holdup
was salt, and only a thin layer of metallic material
was on the filter plate. The second test, with
sintered nickel, showed remarkable performance but
collected only 7 g of material on the filter plate,
Since no visible failure of the filter plate was ap-
158

Table 13.3. Summary of Filtration Tests

Salt composition: LiF~BeF2-ZrF4 (65-30-5 mole %)

Weight of salt mixture: 104.1 kg
Volume of salt at 650°C: 1.7 ft3

Indicated pressure differential:

11 psig forepressure vs vacuum

Pore Transfer Weight
Test Filter Material Diameter Salt Time Gain Remarks
No. (W Conditions (hr) ()
1 Monel Feltmetal 20 Static 22 Test terminated after 2 hr; es-
sentially no salt transfer
2 Sintered nickel 40 Static 0.5 7 No visible material on filter or
evidence of failure
3 Sintered nickel 40 Agitated 1.75 44
4 347 SS Feltmetal 41 Static 2.0 77 Test stopped after 90-kg
transfer
5 Sintered nickel 40 Static 2.17 63 Filter plugged after 40-kg trans-
fer; material on filter pre-
dominantly salt
6 Sintered nickel 40 Static 1.84 19 Balance of salt transferred; filter
ruptured
7 347 SS Feltmetal 41 Agitated 2.0 67
8 Sintered nickel 40 Static 1.75 22 80-kg back transfer of salt from re-

ceiver to treatment vessel; filter

plugged.

parent, we presume that either the decantation
process was extremely efficient or the suspended
particles were too small to impinge on the filter
plate. The balance of the filtration tests showed
comparable results which were essentially inde-
pendent of the static or agitated condition of the
melt. Under the test conditions, filtration times
for FM250 Feltmetal were about 1.19 and 1.36 hr
per cubic foot of salt mixture. The longer transfer
time probably reflected a lower salt temperature at
the filter plate. The occasional plugging of the
filter in tests 4 and 5 suggests that the loading
capacity of the 40-y filters was about 50 to 75 g of
metal particles,

Samples of the salt mixture were withdrawn by
filter stick prior to the first filtration experiment
and by a composite sampler downstream from the

filter plate after tests 4, 6, and 7. The results of
chemical analyses of these samples are shown in
Table 13.4. Although test 6 resulted in a filter
rupture, corresponding chemical analyses do not
reflect excessively large concentrations of structural
metals in the filtered salt. Variations in the other
results are probably within the combined reproduc-
ibility of the analytical methods and the test proce-
dure. If these concentrations are indicative of
solids passing through the filter plates rather than
unreduced metals, then the data reflect the particle
size distribution of the reduced material. We con-
clude that only 1.8 to 3.4% of metals reduced from
solution in the MSRE fuel solvent will pass through
the filter material proposed for use in the reactor
application,
159

Table 13.4., Summary of Analytical Results During Filtration Tests

Impurity Concentration (ppm)

Sample Interval

Cr Ni Fe Total
Before test 19 26 36 44 106
After test 4 15 84 66 165
After test 6 16 256 132 404
After test 7 17 19 49 85

4Rilter stick sample.

13.3 PROTACTINIUM STUDIES IN THE
HIGH-ALPHA MOLTEN-SALT LABORATORY

C. J. Barton
H. H. Stone

J. C. Mailen
W. R. Grimes

Technical feasibility of the ‘“Brillo process,’’ a
method for removal of protactinium from 7LiF—BeF2—
ThF4 melts by reduction from the salt phase and
adsorption on steel wool, was demonstrated pre-
viously and has been described.® Because applica-
tion of this method involves separations of liquids
and solids in highly radioactive environments, its
adaptation at the engineering scale does not seem
to be attractive, For this reason, we have begun
studies in which protactinium is reduced by liquid
bismuth-thorium alloys. The results of similar
studies® conducted at the tracer level indicate that
removal of protactinium from molten fluorides by
liquid metal extraction is a potentially successful
fuel reprocessing method. This investigation is
designed to confirm the chemical validity of the
process and to provide basic information needed to
establish optimal processing conditions.

Before beginning liquid-metal extraction studies,
we performed one solid thorium reduction experiment
that was planned to elucidate the role of iron in
protactinium precipitation, The results of this ex-
periment are summarized below,

5C. J. Barton and H. H. Stone, MSR Program Semiann.
Progr, Rept, Aug, 31, 1967, ORNL-4101, p. 153.

6_]. S. Watson and M. E. Whatley, Protactinium Removal

from Molten-Salt Reactor Fertile Salt, internal memorandum,

Jan. 24, 1968.

13.3.1 Protactinium Reduction by Solid Thorium
in the Near Absence of Iron

Previously, we found that in LiF-ThF4 melts,
iron is reduced rapidly at 600°C by thorium and at
a much slower rate by hydrogen. We chose to ex-
amine the reduction of iron at 700°C in an LiF-ThF4
melt which also contained protactinium. Treatment
of the melt with hydrogen reduced the iron concen-
tration from an original concentration of 500 ppm to
3 ppm (as indicated by °°Fe counts) in about 18 hr,
in contrast to a 40-hr period required at 600°C.7
The quantitative significance of the results obtained
was somewhat impaired by partial loss of protac-
tinium during hydrogen treatment and during transfer
from the nickel pot to a graphite-lined pot. How-
ever, it appears qualitatively that there is little
difference in the behavior of protactinium metal
during reduction with solid thorium except that in
the presence of iron a slightly greater fraction of
protactinium remains in the unfiltered salt sample
after reduction than is found in using the bismuth-
thorium alloy.

13.3.2 Reduction of Protactinium by
Bismuth-Uranium Alloy

Two equilibration experiments were performed
with bismuth-uranium alloys which initially con-
tained 1 at. % uranium in order to simulate condi-

c. J. Barton and H. H. Stone, Reduction of Iron Dis-
solved in Molten LiF-ThF4, ORNL-TM-2036 (Nov, 2,
1967).
160

tions under single-fluid reactor conditions, The
first of these experiments is described elsewhere,?
In summary, analyses of both filtered and unfiltered
samples of bismuth and salt phases removed after
16 and 20 hr contact accounted for only about 42%
of the uranium added initially to the bismuth, This
probably indicated that part of the uranium was
precipitated as UO2 by oxygen or water vapor in-
advertently introduced into the system. The
presence of UO, means that uranium might have
been introduced into the salt by the reaction UO )t
ThF4 > UF4 + ThOz, as well as by the reaction
U®+ ThF, - UF, + Th. Both reactions have
positive free energy for the direction indicated, but
since ThF‘4 is present in large excess over the
amount of U° or UO, and since complexing of the
UF, by LiF occurs in the molten-salt phase, either
reaction might be expected to take place. Since the
relative contribution of the two possible reactions
cannot be evaluated, the data obtained have only
qualitative significance. The uranium content of
the frozen unfiltered salt was much higher than that
of the unfiltered samples removed from the molten
salt. This indicates that the UO_ either dissolved
or dispersed in the salt as it froze, and since the
salt was quite green, it seems most likely that it
dissolved.

Repetition of the experiment resulted in a signifi-
cant improvement, Uranium losses were found to be
negligible, and the protactinium removal rate was
encouragingly rapid. Protactinium distribution data
displayed in Fig, 13.6 showed a marked drop in
concentration in the bismuth during the first S hr
with relatively little change during the last 11 hr.
Although the concentrations of protactinium in the
salt phase appear to vary somewhat, about 90% of
the initial amount of the element was removed from
the salt after 5 hr of contact. The initial low value
in the salt phase coincides with the initial high
uranium concentration in the filtered bismuth (Fig.
13.6).

13.3.3 Reduction of Protactinium by
Bismuth-Thorium Alloys

We have examined the transfer of protactinium from
molten fluoride mixtures to bismuth-thorium alloys

8¢c. J. Barton and H. H. Stone, Reactor Chem. Div, Ann,

Progr. Rept. Dec. 31, 1967, ORNL-4229, p. 47.

ORNL-DWG 68-5553

|

¢ UNFILTERED SALT
— - 99

A

7 _ PROTACTINIUM CONCENTRATION —{ 94
o N “_ |INFILTERED SALT

PROTACTINIUM CONCENTRATION IN SALT (% of total)

PROTACTINIUM CONCENTRATION IN BISMUTH-URANIUM ALLOY (% of total)

6 %— 89
5 H»— R 87
g bt PROTACTINIUM CONCENTRATIONJ 85
IN FILTERED BISMUTH
3 - 83
o 2 4 & 8 10 12 14 6

CONTACT TIME (hr)

Fig. 13.6. Protactinium Distribution Between LiF-
Ber-ThF4 (73-2-25 Mole %) and Bismuth-Uranium
Alloy. Run 1-16.

using the two-region breeder blanket composition

LiF-Ber-ThF . (73-2-25 mole %) and the single-

region fuel composition LiF—Ber—ThF4 (72-16-12
mole %). They will be discussed in that order.

13.3.4 Two-Region Breeder Blanket Composition

Two experiments were performed to measure the
extent of transfer of protactinium to bismuth-thorium
alloy and the distribution of uranium. In both ex-
periments the amount of thorium added initially to
the bismuth was sufficient to give a concentration
of 2500 ppm if all the thorium dissolved. Analysis
of a sample of bismuth-thorium alloy removed 2 hr
after addition of the thorium showed 2220 ppm, but
five days later, after extended treatment of the salt
phase occasioned by difficulty in keeping the pro-
tactinium in solution during hydrogen treatment, the
thorium concentration had dropped to 220 ppm. We
assume that water vapor or oxygen in the system
was causing the precipitation of protactinium be-
161

cause 90% of the thorium in the bismuth-thorium
alloys was converted to ThOz. Despite the presence
of ThOz, about 1.0 to 1.5% of the protactinium was
transferred to the bismuth with low concentration of
thorium metal in the bismuth. The protactinium
transfer rate was increased sharply to about 20% of
the total within 1 hr when an additional amount of
thorium (totaling 2500 ppm) was added. The pro-
tactinium content of the bismuth dropped to about
12% of the total in the system after standing over-
night under flowing helium without agitation, and a
further drop, to about 3%, occurred when the phases
were mixed by sparging with helium for 11/4 hr, The
abovementioned existence of solid ThO_ in the sys-
tem probably accounts for the removal of the pro-
tactinium from bismuth. A large increase in the
protactinium content of unfiltered salt was noted
after the gas sparging.

An additional experiment resulted in the most
complete transfer of protactinium from salt to bis-
muth that has been experienced to date. The re-
sults, displayed in Fig. 13.7, show about 80% of
the protactinium in the filtered bismuth after only
30 min contact of the phases, Since the first set of
samples accounted for 125% of the protactinium, it
seems probable that the protactinium was not
homogenously dispersed in the bismuth at that time.

ORNL-DWG 68-2747

[T ]
°© FILTERED SALT
o FILTERED BISMUTH
--- Ao FILTERED
FILTER

100 ¢ }

BISML‘JTH PLUS —

: A | :
S S, | R — .
I —FILTERED BISMUTH

\__fl

|
8 | SN R
i
i
i
|
i

»

ADDED THORIUM

Pa CONTENT ( % of total}
(621
o

30 - .___]!__. S

FILTERlED SALT '
% \\_J_L/PAI
’ ‘ T

3 B § ———— - e

]

0 2 4 6

Fig. 13.7.

8 10 12 14 16
TIME OF CONTACT { hr)

18

20 22

Distribution of Protactinium Between

Bismuth-Thorium Alloy (2500 ppm Th) and LiF-BeF2-

ThF4 (73-2-25 Mole

%) at 625°C.

If we assume that the calculated protactinium con-
tent of the bismuth in the first sample was too high,
then it appears that there was relatively little
change in the protactinium content of the bismuth
over an 18-hr period. The minimum protactinium
balance observed in this experiment was 80% (18-hr
samples). The uranium distribution data are dis-
played in Fig. 13.8. The uranium was present in
the salt at an initial concentration of about 50 ppm.
About 80% was found in the filtered bismuth, and
there was very little left in the filtered salt after
18 hr contact time. Furthermore, the data indicate
that the transfer of uranium to the bismuth occurred
rapidly, as did the transfer of protactinium,

13.3.5 Single-Region Fuel Composition

A series of four experiments was performed with
the single-region reactor fuel composition LiF—BeFZ-
ThF4 (72-16-12 mole %). The principal variables
were the concentration of uranium fluoride added to
the salt phase and the thorium concentration in the
bismuth. Lanthanum was added to the salt in some

ORNL-DWG 68-5554

100 o
o
90 F— = ‘
URANIUM CONTENT
. OF FILTERED BISMUTH
H -~ ‘-—"—"
(o]
S
Y— 70 ] -
(o]
8
= 60 ) i
o o FILTERED SALT
7 e UNFILTERED SALT
x50 4 FILTERED BISMUTH
< A FILTERED BISMUTH
o PLUS FILTER
g 40
(&)
2
Z 30 ‘
< 1
m !
=) ;
20 .
[ ]
10 I . - ;
URANIUM CONTENT
°© OF FILTERED SALT
0 1 © | T —0
0 2 4 6 8 10 12 14 16 18

CONTACT TIME (hr)

Fig. 13.8. Distribution of Uranium Between LiF-
Ber-ThF4 (73-2-25 Mole %) and Bismuth-Thorium
Alloy. Run 1-8.
162

Table 13.5. Experimental Conditions for Extraction Experiments with Single-Region Fuel,
LiF-BeF,-ThF, (72-16-12 Mole %)

Uranium Lanthanum - _ Weight of Thorium
Run Concentration Concentration Pa Concentration Added to Bismuth
No. (mole %) (ppm) (ppm) ©
1-30 0.00 Tracer 98 3.0
2-5 0.25 100 97 3.0
2-15 0.125 100 82 1.93
2-26 0.010 None 140 0.50 + 0.50

of the experiments in an effort to compare the trans-
fer of a rare earth with that of protactinium under
the same conditions. Experimental conditions in
the four experiments are included in Table 13.5.

Protactinium distribution data in run 1-30, shown
in Fig, 13.9, indicate that protactinium reduction
was nearly complete after 7Y, hr contact of the salt
and bismuth phases, with gas (helium) mixing.
However, protactinium transfer to the bismuth was
much less complete than in the case of the two-
region blanket composition (Fig. 13.7), and the
protactinium balance after the first (30-min) set of
samples was poor. This shows that a large fraction
of the protactinium was associated with some solid
phase and was inaccessible to sampling. The
amount of thorium metal present was more than
enough to saturate the bismuth at the temperature
of the experiment (625°C), so that insoluble inter-
metallic complexes may have been present in addi-
tion to solid thorium. Lanthanum reduction was
quite incomplete, and the !4%La balance was not
very good except for the 30-min samples, where
98% was found in the filtered salt and 2% in the
filtered bismuth. Subsequent filtered bismuth sam-
ples showed only 0.2 to 0.5% of the initial amount
of 14%La (not shown in Fig, 13.9).

The amount of thorium added in run 2-5 (3.0 g of
thorium to 300 g of bismuth) was calculated to pro-
vide an excess of 2500 ppm Th®. The protactinium
distribution data are shown in Fig. 13.10. Protactin-
ium in filtered salt samples went through a minimum
at about 8 hr contact time and then increased, while
that in the filtered bismuth decreased after reaching
a maximum at the same contact time, The uranium
distribution (Fig. 13.11) shows a minimum in the
uranium concentration of filtered salt samples, but

the uranium concentration in the bismuth did not
drop off during the same period. The thorium con-
centration in bismuth, also shown in Fig. 13.11,
dropped off steadily during the course of the experi-
ment, from an initial concentration of 2500 ppm to
160 ppm after 19 hr contact time, The final low
concentration of thorium in the bismuth provides a
possible explanation for the protactinium distribu-
tion data observed in this experiment (Fig. 13.10),
but there is no obvious explanation for thorium loss.
Except for short exposure of stainless steel sam-
plers during removal of bismuth samples, graphite
was the only structural material in contact with the
two phases. The copper salt samplers only contact
the salt phase during sampling. Further investigation
will be required to provide an explanation for tho-
rium loss and the poor protactinium balance observed
in the later part of the experiment. In this experi-
ment we demonstrated that the reduced protactinium
could be returned to solution in the molten salt in
the presence of bismuth by treating the mix with
anhydrous H2-HF (H2 : HF volume ratio varied from
10:1 to 5:1). Approximately 28 hr were required to
redissolve all the protactinium because it was nec-
essary to hydrofluorinate all the metallic thorium
and uranium in the system as well as the reduced
protactinium, We had previously demonstrated
reversibility of the protactinium precipitation in
some experiments with solid thorium, but this was
the first time that we showed that liquid bismuth
does not interfere with the hydrofluorination process.
(Previously reported?® tracer-level experiments indi-
cated that this result would be expected.)

9_]. H. Shaffer et al., MSR Program Semiann, Progr,
Rept, Aug. 31, 1967, ORNL-4191, p. 148.
163

ORNL-DWG 68-5555
T | | - L ]
LANTHANUM CONTENT OF FILTERED SALT

100

a LANTHANUM CONTENT OF FILTERED SALT T ’
LANTHANUM CONTENT OF UNFILTERED SALT
PROTACTINIUM CONTENT OF FILTERED SALT _|
PROTACTINIUM CONTENT OF UNFILTERED SALT
PROTACTINIUM CONTENT OF FILTERED BISMUTH
PROTACTINIUM CONTENT OF FILTERED BISMUTH PLUS FILTER _|

Pa CONTENT OF UNFILTERED SALT -

|

LANTHANUM OR PROTACTINIUM CONCENTRATION (% of total)

P T
_~—Pa_CONTENT OF FILTERED BISMUTH

~1_[ [T

10 12
CONTACT TIME (hr)

6

Fig. 13.9. Reduction of Protactinium and Lanthanum in LiF-Ber»ThF4 (72-16-12 Mole %) by Bismuth-Thorium
Alloy at 625°C,

Uranium distribution data in run 2-15 (Fig. 13.12) tactinium balance was poor, and the amount found

show that uranium reduction was slow and incom-
plete with a poor uranium balance, especially during
the later part of the experiment. The thorium con-
centration in the metal phase (corrected for salt
contamination by assuming that all beryllium
present was due to salt) is also shown in the same
figure. The decrease in thorium concentration
proceeded at a linear rate (177 ppm/hr) during the
first 9 hr and much more slowly during the last 131/2
hr., This compares with the uniform rate of decrease
of 125 ppm /hr over an 18-hr period in run 2-5. The
protactinium distribution data (Fig. 13.13), like the
uranium data, show slow and incomplete reduction,
probably because of the thorium loss from the metal
phase. The protactinium concentration in filtered
bismuth reached a maximum (3% of total) at 9 hr
contact time and then decreased to 0.6%. The pro-

on the stainless steel filter through which the
filtered bismuth passed was relatively large, show-
ing that there was a great deal of insoluble pro-
tactinium in the metal phase, The lanthanum data
were much like those in run 2-5. The 14%La con-
centration in filtered salt samples decreased to 82%
of the initial value after 221/2 hr contact, but the
maximum concentration observed in filtered bismuth
was 0.5% (3 hr contact time).

The last experiment in the series (run 2-26)
represented an effort to avoid insoluble phases in
the bismuth. The initial addition of thorium (504 mg)
was calculated to leave 1000 ppm excess thorium in
the bismuth after all uranium and protactinium were
completely reduced. After 177 hr, a second addition
of 506 mg of thorium was made. Only 233Pa data
are available at present, On the basis of these data,
164

ORNL-DWG 68-5556

[T ng -

THORIUM CONCENTRATION IN FILTERED BISMUTH| 180C

1600
j 140G
1200

60 - \URANIUM CONCENTRATION IN FILTERED SALT 1000
j 800
800

s— 400

ORNL-DWG 68-—-2748 1020

R
. T ‘\L S ‘{_T 1010

" e Pa CONTENT OF FILTERED ~| 1000
Bi PLUS FILTER

: — P CONTENT OF UNFILTERED

100fi

| & Pa (_erNTENT OF F!LTERED

PROTACTINIUM CONTENT (% of total)
o
@]

URANIUM CONCENTRATION (% of total)

THORIUM CONCENTRATION IN BISMUTH {(ppm)

0 2 4 & 8 10 42 14 16 18 20

CONTACT TIME (hr} 20 | 200

A
_ 10 "3URANIUM CONCENTRATION IN FILTERED BISMUTH| O
Fig. 13.10. Distribution of Protactinium Between LiF- ] ) ‘

BeF_ .ThF , (72-16-12 Mole %) and Bismuth-Thorium 0 ¢
2 4 0 3 6 9 12 45 18 24 24
Alloy at 625°C, CONTACT TIME (hr}

Fig. 13.12. Distribution of Uranium Between LiF-
Ber-ThF4 (72-16-12 Mole %) and Bismuth-Thorium

ORNL— DWG 68 -2746 Alloy. Run 2-15.

the protactinium distribution does not appear to be
drastically different from that observed in the three
previous experiments, The protactinium content of
the filtered salt appeared to be leveling off at about
46% of initial concentration after 16 hr contact, and
the second thorium addition produced only a slight
further decrease (to 34%). Only about 4% of the
protactinium was found in the filtered bismuth prior
to the second addition of the thorium, but a sharp
increase, to a maximum of 11%, was noted after the
addition. However, after another 18-hr period, this
value had dropped to 3.1%. This experiment will be
discussed in more detail when the analytical data
are complete,

Transfer of protactinium from two-region breeder

THORIUM CONCENTRATION IN BISMUTH (ppm)

SALT PHASE URANIUM CONCENTRATION (% of total)

CONTACT TIME (hr) blanket compositions to bismuth-thorium alloys oc-

curred rapidly and very nearly completely. Reduction

Fig. 13.11. Uranium Distribution Between LiF-BeF .. of both uranium and protactinium in the single-region
ThF, (72-16-12 Mole %) and Bismuth-Thorium Alloy at reactor fuel composition LiF-—BeFQ-ThF4 (72-16-12

625°C. mole %) occurred much more slowly in contact with
165

ORNL—DWG 68~5557

Th CONCENTRATION IN BISMUTH (ppm)

PROTACTINIUM CONTENT (% of ‘total)

ISMUTH

‘ 1
0 2 4 6 8 40 12 14 16 18 20 22 24
CONTACT TIME (hr)

Fig. 13.13. Distribution of Protactinium Between LiF-Ber-ThF4 (72-16-12 Mole %) and Bismuth-Thorium Alloy.
Run 2-15.

bismuth-thorium alloys, and the fraction of pro- periments with single-region fuel may be attributed
tactinium found in filtered bismuth samples was to reaction between thorium and Ber. The reaction
much smaller than for the breeder blanket composi- product could be Be?, BeZC, or ThBe13 , and the
tions tested. Since the principal difference is in available thermodynamic data seem to favor the

the BeF2 content (16 mole % for single-region vs compounds rather than pure metallic beryllium,

0 or 2% for two-region compositions), it seems likely Further investigations are planned in an effort to
that the disappearance of thorium noted in two ex- elucidate the thorium loss phenomenon.
14. Behavior of BF; and Fluoroborate Mixtures

14.1 PHASE RELATIONS IN
FLUOROBORATE SYSTEMS

14.1.1 The System NaF-NaBF -KBF ,-KF

Investigation of phase relations in the system
NaF-KF-BF | at atmospheric pressure is limited
to the portion of the system bounded by the com-
pounds NaF—NaBF4-KBF4-KF, because fluoro-
borate preparations containing more than 50 mole
% of BF , exist only in equilibrium with high
pressures of BF,. This system is of interest be-
cause it would be desirable to have a coolant
with an even lower liquidus temperature than is
provided by the NaF-NaBF , system (380°C).

Phase diagrams for two pseudobinary systems
that form part of the ternary system have been
published in another report.> The NaF-KBF
system has a eutectic containing about 96 mole
% of KBF , melting at 540 * 5°C. The high-
temperature forms of the compounds NaBF , and
KBF4 make a continuous series of solid solutions
with a minimum-~melting composition containing
approximately 85 mole % NaBF .. A tentative
diagram for the temary system is shown in Fig.
14.1. There is a temary eutectic composition in
the triangle NaF-KBF -KF close to the KF-KBF
binary eutectic. The NaF primary phase field
covers a large area of the diagram. There ap-
pears to be a valley forming the boundary between
the NaF and NaBF -KBF, solid solution primary
phase fields, approximately parallel to the NaBF "
KBF, join and quite close to it. The minimum-
melting composition along this valley, containing
approximately 47 mole % NaF, 5 mole % KF, and
48 mole % BF ,, is the minimum-melting composi-
tion in the ternary system, with a liquidus tem-
perature of 386 + 5°C,

C. J. Barton
L. O. Gilpatrick

H. Ins]Ley1
T. N. McVay!

We mentioned in out previous report? that there
were indications that NaBF , recrystallized from
dilute hydrofluoric acid solutions was less pure
than KBF  prepared by the same procedure and
that alternative methods of purifying NaBF4 were
being considered. One process that has worked
well with some materials, slow recrystallization
from the melt by passing a tube filled with the
material through a temperature gradient, failed to
produce any significant improvement in the sharp-
ness of the differential thermal analysis (DTA)
melting curve, the most sensitive measure of
compound purity that we have available. We found
that improved purity of NaBF , was achieved by
treating molten NaBF | at 425°C with a mixture of
anhydrous HF, BF ., and helium (2 volumes of
BF | per volume of HF). The sharpness of the
DTA melting curve for NaBF , resulting from this
treatment approached that produced by our best
KBF,.

We believe that this treatment removes oxygen,
which is probably present as NaBFsOH formed
by hydrolysis of NaBF4. The presence of a small
quantity of this impurity could account for at
least a portion of the corrosive action of earlier
fluoroborate preparations.

A sample of material from a large batch of the
NaBF4 purchased by a special order from a com-
mercial producer® had a melting point only about
1°C lower than that of our best laboratory prepa-

ration. 3
Harshaw Chemical Co. — Division of Kewanee (il
Co., Cleveland, Ohio. 8pecifications supplied by the
manufacturer were: NaBF,, 99.08%; oxygen, 0,025%;
Pb, 0.004%; Si, 0.01%; Ca, 0,.01%; Fe, 0.023%; water
Iconsultant. insoluble <0.01%; H20, 0.01%. These specifications

2C. J. Barton et al., MSR Program Semiann. Progr.
Rept. Aug. 31, 1967, ORNL-4191, pp. 158-59,

166

were confirmed by ORNL analyses, and a total of
2400 1b of this material was received., Major portions
are available for study.
167

ORNL—DWG 68-5560

542¢°

KBF,

5558 (570°)

455°

5000 - 440°

4 £50°

600°

650°

700°

750°

KF
800°

y/ \/ KE

(995°)

720 (858°)

Fig. 14.1. The System NaF-KF-NoBF4-KBF4.

14.2 NONIDEALITY OF MIXING IN POTASSIUM
FLUOROBORATE-SODIUM (OR POTASSIUM)
FLUORIDE SYSTEMS

M. A. Bredig

Experimental phase diagram data available from
the preceding semiannual report? were examined
for deviations from ideality of mixing. Slightly
revised liquidus curves for KF and NaF in the
binary salt system KF-KBF , and the reciprocal
salt system NaF-KBF ,, respectively, are shown
in Fig. 14.2. The revision is based upon an
appropriate application of the known enthalpies
of fusion of NaF and KF (AHm = 7.90 and 6.75
kcal/mole) to the liquidus near 100% NaF and
KF respectively. It is apparent that the KF-KBF4
system deviates little from ideality, whereas the
reciprocal salt mixtures NaF-KBF ,, containing
cations and anions of widely differing sizes, show
large positive deviations from ideality. The latter
were estimated semiquantitatively in terms of the
partial molar excess free energy of NaF, G%, by
means of the equation

GFE=RTlny  .=AH [(T/T)-1]
+ AC (T, -~ )~ T x AC,
xIn(T_/T)~RTIn N} __,

ORNL-DWG 68-5558

1000 o e T T
\\\\ \-{ | | !
900;:\<\:L"Af-<*<~*<‘*4*;¢'—>
N ‘ ’ .
800 F— — — ]'lr\\. — _\ -
%) ~
= 900 P' — =t — = l\DEAL'
% .\**\ \ 1\\
5 800 - 7\-5c\-ifx: — b
& BN
= 700 ~~’T—!DEAL>T‘\\: -
600 — — —— — — L
| \
500 #,J]_-,{ -
400 S U S
0 20 a0

male % KBF,

Fig. 14.2. Revised Liquidus Interpolations for the

Systems KF-KBF4 and NoF-KBF4. Experimental points

by Barton et 31.2
ORNL-DWG 68-5559

(kcal/mole)

= E
GNuF

i
02 03 04 05 0.7
(1=Nyor 2= Ear (= agF, IDEAL,TEMKIN)

Fig. 14.3. Excess Chemical Potential of NaF in NaF-

KBF4 from Phase Diagram. Experimental points by

Barton et a1.2

with AC_ = 1.0 cal deg™! mole~'. Figure
14.3 gives GF as a function of, and shows it to
be not simply proportional to, the ideal ¢‘Temkin’’
activity of KBF ,, (1 — N,.o)? (tmustbe
noted that this is not an isothermal plot.) The
shape of the NaF liquidus in Fig. 14.2 and the
high value (12,000 cal/mole) of the (interpolated)
slope of Gy __ below (1 — NNaF)2 = 0.1 in Fig.
14.3 suggest strongly that replacement of NaF by
LiF, with the much greater difference in size of
Li* from K*, would lead to separation into two
liquids.

14.3 HEAT CONTENT OF NoBF -NaF
(92.5-7.5 MOLE %)

A. S. Dworkin

The heat content of the eutectic NaBF ,-NaF
has been measured using a copper block drop
calorimeter.* The salt mixture was supplied by
L. O. Gilpatrick and was from the same batch as
that used for the phase diagram studies.” The

*A. S. Dworkin and M. A. Bredig, J. Phys. Chem. 64,
269 (1960).

SMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 158=-59.

168

ORNL-DWG 68-5561

200

C)
< 150
3 LA Heygion =31 cal/
8
T
% 100 S
AHrransiTion =14-7 col/g
50
e}
500

Fig. 14.4. Heat Content of NuBF4-NuF (92.5-7.5
Mole %).

mixture was contained in a sealed platinum cap-
sule, which in turn was sealed into an Inconel
capsule especially designed for our heat content
apparatus.

The results are shown in Fig. 14.4. The heat
of fusion is 31 cal/g, and the heat of transition
is 14.7 cal/g. The heat capacity of the liquid
from 380 to 600°C is 0.36 cal g~! °C~! and the
heat capacity of the high-temperature solid from
243 to 380°C is 0.34 cal g~ ! °C~ 1.

14.4 SOLUBILITY OF THORIUM METAL IN
LITHIUM FLUORIDE-THORIUM
TETRAFLUORIDE MIXTURES

H. R. Bronstein M. A. Bredig

A knowledge of the free energy of formation of
’I‘hF4 in various molten mixtures of LiF-ThF
would be of value to the MSR fuel reprocessing
study. An emf study of a cell of the type

Th®| ThF (x), LiF (1 -x)|H,, HF|Pt

is the most obvious method for determining this
quantity. The cell container must of necessity
be nickel because of the use of the H,, HF | Pt
electrode.

600
However, a study by C. J. Barton and H. H.
Stone® has cast some doubt about the feasibility
of such measurements in a nickel container. A
thorium rod suspended in a molten mixture of the
LiF-ThF , (73-27 mole %) contained in a nickel
crucible disappeared after 16 hr at a temperature
of 600°C. Black magnetic material removed from
the cooled melt analyzed 45% Ni and 30% Th,
while nonmagnetic material contained 22% Ni and
49.5% Th. An explanation was based on the
assumption that the reaction

3ThF, + Th? — 4ThF

occurs and that the Tth, when it diffuses to the
nickel wall, disproportionates because of the
formation of the Th-Ni intermetallic compound.
However, the speculative nature of this explana-
tion was emphasized, as very little is known
about the existence of a ThF .. 7

If such a reaction with the nickel containet
actually occurs, severe complications would
arise in an attempt to measure the free energy of
formation of ThF4 in the above cell. Before
abandoning this much-preferred cell arrangement,
a reinvestigation of the reaction of Th® with the
LiF-ThF , melt was deemed desirable.

A carefully dehydrated mixture of LiF-ThF
(73 and 27 mole % respectively) was melted in a
thorium crucible and held in the molten state
under an atmosphere of dry argon at a temperature
of 620°C. The melt was frequently stirred with
a thorium stirrer. After a heating period of 16 hr,
a sample of the molten liquid was withdrawn in a
tantalum cup. The apparatus utilized for this
experiment has been previously described® and
illustrated.®

The analysis of the sample indicated that ap-
proximately 2.0 mole % thorium metal had dis-
solved in the melt,.

Next, a 1/g-in. nickel rod was suspended in the
melt for a period of 5 hr at 620°C. At the termi-
nation of the experiment the solidified melt was
removed from the thorium crucible. A blackish

®¢. J. Barton and H. H. Stone, ORNL-TM-2036
(November 1967).

7y, C. Warf, J. Am. Chem. Soc. 74, 1864 (1952),

8y, R. Bronstein, A. S. Dworkin, and M. A. Bredig,
J. Phys. Chem. 66, 44 (1962).

9A. S. Dworkin, H. R. Bronstein, and M. A. Bredig,
Discussions Faraday Soc. 32, 188 (1962).

169

layer of material was found at the bottom of the
melt. Apparently segregation had occurred upon
slow cooling. This black material reacted vig-
orously with hydrochloric acid. A hydrogen
evolution analysis yielded a value of 12.5% by
weight calculated as free thorium metal. Spectro-
graphic analysis showed the presence of only
very minot quantities of other metals as impurities.
Of some significance is the absence of detectable
quantities of nickel, possibly attributable to the
very small surface ratio of nickel rod to thorium
crucible. X-ray examination of this blackish
material showed lines only of the compounds
existing in this melt composition, 1 © that is, mainly
Li,ThF and a small amount of LiThF .. In agree-
ment with this, microscopic examination showed
the blackish material to be comprised mainly of
the salt mixture,

Further experimental work will be performed to
corroborate the above findings.

14.5 DISSOCIATION PRESSURE OF BF, FOR
THE MSRE SUBSTITUTE COOLANT

Stanley Cantor

Melts composed mainly of NaBF ,, because they
possess attractive thermal properties, are under
consideration as coolants in molten-salt reactors.
An unattractive property of fluoroborates is the
relatively high vapor pressure of BF , caused by
dissociation, for example,

NaBF (1) = NaF(!) + BF ,(g).

To determine the equilibrium constant of the
above reaction and to derive other thermodynamic
data (see next section), BF , pressures in equi-
librium with melts of the system NaBF -NaF have
been measured. Some of the results are shown in
Fig. 14.5. (Preliminary measurements reported in
ORNL-4191, pp. 159-61, are superseded by more
accurate results shown in Fig. 14.5.)

For the mixture proposed as the MSRE substi-
tute coolant, 92-8 mole % NaBF4-NaF, vapor
pressures of BF , in equilibrium with the melt of
this fixed composition can be represented by the
equation

1fl'R. E. Thoma (ed.), Phase Diagrams of Nuclear
Reactor Materials, ORNL-2548, p. 72 (1959),
ORNL-DWG 68-5562

10,000 e e S — D — ”44f557

oo - — 1 — = = ——F - -
__—=—— EXTRAPOLATED , __,!_fi
T SUPERCOOLED LIQUID —_—

T i T
—— —— 1000° K (1340° F)
500 [——f — 4 —

PRESSURE OF BF

~
10 “::% 750° K (890° F)

mole % NCBF4

Fig. 14.5. Vapor Pressure of BF3 in Equilibrium with
NuBF4-NuF Melts.

920

log P (mm) = 9.024 — .
g P (mm) TCK

At normal power operating temperature (1000°F)
in the MSRE coolant pump bowl, the pressure of
BF , will be 53 mm; at zero-power temperature
(1200°F) the pressure will be 401 mm.

14.6 CHEMICAL THERMODYNAMICS OF THE
SYSTEM NaBF ,-NaF

Stanley Cantor

In a closed apparatus of known volume, BF,
pressures in equilibrium with the melt were meas-
ured manometrically in the composition range 2 to
~~100 mole % NaBF ,. For each composition,
pressures were measured over at least a 150°
temperature interval within overall limits of 425
to 1200°C.

From the observables of the experiment (pres-
sure and volume of BF3 vapor, temperature, num-
ber of moles of NaBF , and NaF initially charged
into the apparatus), an equation was derived for
the thermodynamic activity of NaBF,. The
equation is:

f(v )
lnaN BF =f 2
28N, f(N,)~ 0

f¥) din Py, (1)

where Pg is the pressure of BF ; in equilibrium
with a melt whose composition is denoted by
f(N ), defined as the number of moles of NaF
initially loaded minus the number of moles of BF
in the vapor divided by the number of moles of
NaBF, and NaF initially loaded. The composi-
tion variable, f(NQ), is very close to, but not
exactly equal to, the equilibrium mole fraction of
NaF in the melt. Equation (1) was derived from
the Gibbs-Duhem equation for the molten phase at
constant temperature.

The integral on the right-hand side of Eq. (1)
was solved graphically, using Simpson’s rule. By
integration of the Gibbs-Duhem equation, activi-
ties of NaF were derived from the activities of
NaBF 4+ For the most part the activities ex-
hibited slightly positive deviations from ideality.
In other words, activity coefficients (defined in
terms of the pure liquid components) were greater
than unity. Figure 14.6 shows the activity co~
efficients of NaBF , and NaF at 1000°K.

By combining BF ; pressures and activities the
equilibrium constant K p Was obtained for the
reaction

ORNL-DWG 68-2994

N

32

mole fraction NaBF,

Fig. 14,6, Activity Coefficients of NuBF4 and NaF at
1000°K.
NaBF (1) = NaF(I) + BF (g) ,

where

NaF

3 aNaBF
4

At 1000°K, Kp equals 0.182 atm, while at 1100°K,
Kp is 0.670 atm.

14.7 CORROSION OF HASTELLOY N AND
ITS CONSTITUENTS IN FLUORO-
BORATE MELTS

Stanley Cantor

A preliminary corrosion test!! showed that
metallic chromium (the component of Hastelloy N
most readily oxidized in molten fluoride media)
reacted with the molten mixture proposed as the
MSRE substitute coolant, NaF-NaBF , (92-8 mole
%). Two products of the reaction were crystal-
line Na,CrF ( and gaseous BF,. Most likely, the
removal of NaF from the melt in forming Na CrF
caused a shift in the equilibrium

NaBF (1) = NaF(!) + BF ,(8) ,

thereby establishing a higher vapor pressure of
BF .
A program of tests is in progress to
1. determine if Hastelloy N itself reacts un-
favorably with NaBF , melts,
2. verify the initial results with pure chromium,
3. observe whether nickel and iron also react.

In each test a salt sample of about 100 g par-
tially decomposes (according to the above re-
action) in an evacuated container fabricated from
or containing samples of the test metal; the gas
pressure and temperature are continuously moni-
tored. On the basis of the preliminaty study, we
anticipated that for a constant temperature, the
gas pressure would increase at a rate proportional
to the chromium corrosion rate. After runs are
completed, the containers are cut open and the
contents examined. A summary of six tests is
given in Table 14.1.

g, Cantor, MSR Program Semiann. Progr. Rept.
Aug. 31, 1967, ORNL-4191, p. 161.

171

The condition of the Hastelloy N containers
and specimens following tests 1 and 2 (see
Table 14.1) indicated rather sluggish reaction
with NaBF . On the other hand, increases of
nickel and chromium in the salt indicated that
moderate corrosion occurred; taking into account
the surface area of metal exposed to the melt, the
concentrations of nickel and chromium in the
salt are equivalent to a corrosion rate of 1.5 to
3 mils/year.

Test 3, however, provided more encouraging
results with regard to both nickel and chromium
corrosion. The lack of nickel corrosion in test 3
was essentially verified in test 5. It is possible
that the vessels in tests 1 and 2 had undergone
some corrosion prior to being loaded with fluoro-
borate and that the corrosion product was sub-
sequently dissolved by the salt.

Test 4 indicates that chromium reacts with
either fluoroborate or some impurity contained in
NaBF ,. Asyetwe have not found reduced boron,
either in an elemental form or as a metallic
boride. Hence, it cannot be stated with any cer-
tainty that NaBF , oxidizes chromium. The
blackish material noted after completion of test 4
is being carefully examined for a possible boron
content.

14.8 COMPATIBILITY AND IMMISCIBILITY
OF MOLTEN FLUORIDES

C. E. Bamberger
C. F. Baes, ]Jr.

J. P. Young
C. S. Sherer!?

In connection with the investigation of fluoro-
borates as a coolant for molten-salt reactors, the
following tests of compatibility and solubility of
KBF4 and NaBF, with various fluoride mixtures
and with B O, were carried out. It was found
that silica and, at lower temperatures, even Pyrex
containers were suitable for these tests, thus
providing the great advantage of direct visual
examination. As a result, the occurrence of
liquid-liquid immiscibility was quickly detected,
a phenomenon which was not detected by less-
direct DTA and quenching techniques. In all
tests the samples were evacuated several times
and flushed with helium prior to melting,

12Research participant, Alabama College, Montevallo,
Alabama.
172

Table 14.1. Summary of Corrosion Tests with 92-8 Mole % NaB F4-N0F

Test

Time at

Metal Exposed Temperature

Vapor Pressure

Observations at Termination of Test

Hastelloy N 284 hr at 560°C

Hastelloy N, salt 449 hr at 580°C
loaded was pure

NaBF4

603 hr at 605°C,
then 253 hr at
657°C

Hastelloy N

50 g chromium 837 hr at 600°C
chips; nickel

container

Nickel coupons; 1428 hr at 606°C

nickel container

Iron coupons; 936 hr at 600°C

nickel container

Rose from 100 to 210 mm

in 96 hr; remained at

210 mm thereafter

585 mm continuously

150 mm continuously,
then 370 mm con-

tinuously

Siowly rose from 170 to
280 mm

170 mm unchanged for
384 hr; slow rise to
460 mm thereafter

220 mm continuously

Metal exposed to salt and BF ; vapor
had blackened appearance; Hastel-
loy N specimens immersed in melt
had not changed weight. Salt cake
was white with black scum at inter-
faces. Concentrations of chromium

and nickel in salt increased by 100

and 650 ppm respectively.

Slight blackening of exposed metal
surfaces; no change in weight of
Hastelloy N specimens. Chromium
and nickel in salt phase increased
by 230 and 1590 ppm respectively.

Slight blackening of exposed metal
surfaces; specimens unchanged in
weight. Salt had slight greenish
cast, but Na3CrF6 was not detected.
Chromium had increased in salt phase
by 100 ppm but nickel remained
<20 ppm.

Chromium chips lost 0.5 g. Much
green Na 3CrF‘6 formed and located
on or near chromium chips within the
salt cake. Water-insoluble blackish

material mixed with Na3CrF6.

Pressure remained at 310 mm when
sample cooled to room temperature.
It may therefore be inferred that the
rise in pressure during the run was
caused by a very slow inleakage of
air, Nickel in salt phase increased by

by about 25 ppm. Nickel metal cou-

pons were unchanged in weight and

showed no evidence of attack.

Test still continuing.

In the following solubility tests, except one,
the temperature was 480°C and the containers
were Pyrex. Samples remained molten for periods

up to 12 hr.

1. NaBF, + CtF, (0.3 wt %): no solubility de-
tected; after 1 to 2 hr the insoluble CtF,
changed to a bright green color, possibly

3NaF.CrF ,.
2. NaBF, + CtF, (V1 wt %): no solubility de-
tected; change of color indicating possible
3NaF.CrF .

3. NaBF, + UF,: neither solubility of UF, nor
formation of U0, was detected, even after
adding B,0,.

4. NaBF, + HoF ;: this rare earth was readily
available, and its spectrum in molten fluorides
is known; no measurable solubility was de-
tected.

5. KBF, + B (amorphous) at 650°C in fused silica:

no significant solubility was detected; the
finely divided powder coagulated with time.

The following list summarizes the tests in
which liquid immiscibility was observed; unless
otherwise stated the container material was fused
silica. Since UF, did not show spectrophoto-
metrically any solubility in NaBF , or in KBF ,,
it was used for spiking other fluorides as a visual
aid in the detection of phase separations. In some
instances KBF , was used rather than NaBF | be-
cause its higher melting point was closer to the
melting point of some fluoride mixtures.

1. NaBF + blanket salt (0.71 LiF, 0.02 BeF ,
0.27 ThF ), phase ratio ~1:10, temperature
585°C.

2. KBF, + blanket salt (0.71 LiF, 0.02 BeF ,,
0.27 ThF ), temperature 625°C.

3. NaBF, + BULT-4 (0.65 LiF, 0.30 BeF ,, 0.01
UF,, 0.04 ThF ), phase ratio ~1:10, tempera-
ture 585°C.,

4. NaBF4 + 3LiFoUF4, phase ratio ~1:30, tem-
perature 480°C, Pyrex container.

13y. Hellriegel, Ber. 70B, 689--90 (1937).

173

5. NaBF, + 2LiF-BeF ,, phase ratio ~10:1, tem-
perature 480°C, Pyrex container.

6. NaBF, + MSRE salt (0.64 LiF, 0.29 BeF,,
0.015 Z:F,, 0.009 UF,), phase ratio ~1:10
and 4:1, temperature 480°C, Pyrex container.

It has been reported’® that mixtures of NaBF
and B,O, react vigorously on heating to produce
BF

NaBF, + 2B,0 ,=<=BF, + NaF-B,0, . (1)
Indeed this is a commonly used method to gener-
ate BF ,. Nonetheless, we decided to explore the
effect of B,0, additions to immiscible liquids of
NaBF, plus LiF-BeF 2-ThF4-UF4. Gas evolution
was detected in all tests where B,0, was added
directly, and no extraction of UF, or precipitation
of UO, was noted. The resulting borate phases
were very viscous.

Finally, in another series of tests 2LiF-BeF
was equilibrated with NaBF | in sealed nickel
containers at 600°C. The capsules were quenched
and the two phases sampled and analyzed. The
results (Table 14.2) show appreciable distribution
of all the components (NaF, LiF, BFS, and Ber)
between the two phases. In each phase

suggesting that the phases may be represented
approximately as reciprocal mixtures of the ions
Li*, Na+, BF4“, BeF42". In terms of these ions
there is a tendency for the smaller ions, Li* and
BeF 2, to favor one phase and the larger ions,
Na® and BF 7, to favor the other phase.

Table 14.2. Liquid-Liquid Distribution Behavior for LizBeF4—NuBF4 Mixtures at 600°C

Initial Phase Ratio

Mole Fraction

Phase
(g NaBF4/g leBeF4) NaF LiF BF3 BeF,
3 Top 0.391 0.139 0.419 0.052
Bottom 0.165 0.435 0.129 0.271
2 Top 0.365 0.174 0.406 0.055
Bottom 0.156 0.448 0.138 0.257
1 Top 0.373 0.164 0.432 0.031
Bottom 0.110 0.545 0.044 0.300

Part 4. Molten-Salt Irradiation Experiments

E. G. Bohlmann

Molten-salt breeder reactors will operate at fuel
salt power densities of 300 to 700 w/cc in com-
parison with the peak level of 24 w/cc in the
MSRE. Knowledge of the effects of such ex-
posures on fuel salt stability and materials com-
patibility for both design and foreseeable off-
design conditions is essential to the success of
the breeder program. The fates of the fission
products are also of interest in terms of separa-
tions processing and possible poisoning due to
accumulation on or in the graphite moderator.
Molten-salt convection loop experiments are being
operated in beam hole HN-1 of the ORR to develop
data pertinent to the breeder program.

Examination of the second loop experiment was
completed during this period. No unexpected ma-
terials problems were encountered except for the
fact that the salt had wetted the graphite, pre-
sumably because of small amounts of moisture
present in the gas used in loading, sampling, and
draining manipulations. Fission product behavior

was well defined (good material balances for most
isotopes of interest) and generally consistent with
results of analyses on MSRE materials.

A prototype of a third loop, modified to provide
for better graphite examinations both pre and post
irradiation, has been constructed, and components
for the in-pile version have been fabricated of
Hastelloy N modified to improve high-temperature
ductility after irradiation. Work on the experiment
has been discontinued because of budget limita-
tions and investigations of effects of moisture
levels in the cover and manipulatory gas.

In the interim a program of capsule irradiations
of fluoroborate coolant salt in spent HFIR fuel
elements is in progress. Gamma fluxes comparable
with those present in the MSRE heat exchanger
are available in the center of such elements. Ex-
amination of the first capsule experiment showed
no gross effects of accumulation of a dose of
2.5%x 1010 1.

15. Molten-Salt Convection Loop in the ORR

E. L. Compere

We have previously described the operation and
postirradiation examination of the second in-pile
molten-salt convection loop experiment operated
in beam hole HN-1 of the ORR.! During this re-
port period, we completed additional analyses of

lg. L. Compere et al., MSR Program Semiann. Progr.
Rept. Aug. 31, 1967, ORNL-4191, pp. 176-93.

H. C. Savage

174

J. M. Baker

the fuel salt, graphite, and metal in contact with
fissioning fuel salt and cover gas and thereby de-
termined the distribution of fission product activ-
ities in the loop components. A prototype of a
third loop assembly has been designed, and com-
ponent parts have been fabricated.

Results of these additional postirradiation
analyses and present design features of the third
loop assembly are described below.
15.1 ISOTOPE ACTIVITY BALANCE (LOOP 2)
As reported previously,! the activity of a given
isotope to be expected in the system at a partic-
ular time was estimated by a detailed application
of standard equations? to the individual irradiation
and inventory periods with adjustment for decay
to final ORR shutdown for the experiment. Data
received during the current period permit comple-
tion of the isotope activity balance.

The activities of the fission products !37Cs,
144Ce, and °5Zr were used as internal standards
to estimate the average flux received by the salt
under the assumption that they were not appreci-
ably lost from the salt. A mean flux to the salt
of 0.88 x 10! % was thus estimated. From this
value, total activities of the various isotopes
produced in the ORR loop experiment were cal-
culated.

The calculations described above provided an
estimate of the amount of each isotope to be
accounted for. We determined the total amount
of isotope actually found in the system by dividing
the loop into regions, analyzing a specimen from
each region, and allocating a proportionate
amount of activity to the region.

For the various samples obtained from the loop,
activity determinations for the 13 isotopes shown
in Table 15.1 were made, as well as sensitive de-
terminations of 235U. The activities have been
totaled for each isotope under the categories of
graphite, loop metal, salt sample lines, gas
sample and inlet lines, and salt. These values,
plus the estimated total activities calculated from
inventory and irradiation history, are also shown
in Table 15.1.

It may be seen that over half (but generally less
than all) the expected activity was accounted for
in the cases of 2°Mo, '32Te, 25Zr, 89St, 137Cs,
141Ce, 144Ce, 91Y, and 1*7Nd.

A substantial proportion, although less than
half, was accounted for in the cases of °>Nb,
149Ba, and !3'1. Inasmuch as iodine readily
volatilizes from all solid samples (without doubt
from powdered graphite in particular), it is to be
expected that iodine determinations would be low.

2J. M. West, pp. 7—14 in Nuclear Engineering Hand-
book, ed. by H. Etherington, McGraw-Hill, New York,
1953.

175

Only about 11% of the !23Ru was found, although
proper traps were used to recover any ruthenium
compounds volatilized during the preparation of
radiochemical samples.

Molybdenum, tellurium, ruthenium, and niobium
are almost entirely departed from the salt. These
elements showed no dominant preference for
graphite or metal but seemed to deposit on what-
ever surface was available. Short-lived noble
gases appeared to have diffused appreciably into
the graphite, as shown by the presence of daughter
isotopes such as 8°Sr, 1*%Ba, and others. How-
ever, the major proportions of these, and almost
all of other alkali, alkaline-earth, and rare-earth
isotopes (all of which form relatively stable,
nonvolatile fluorides), were found in the salt.

15.2 PENETRATION OF FISSION PRODUCTS
INTO GRAPHITE AND DEPOSITION
ONTO SURFACES

The amounts of the respective fission product
isctopes which penetrated the graphite to given
depths were obtained from the samples shaved
from the fuel channels. Data for each isotope
and each common cut depth were summed over
all the fuel channels. The value was corrected
for the amount of the isotope in salt in the sample
based on the amount of 23U which was found and
on the concentration of the isotope in regular salt
samples. This correction was appreciable only
for those isotopes found principally in salt and
was never dominant.

The amount of an isotope in the graphite up to
a given depth was then divided by the total
amount of the isotope found in the loop, yielding
the percentage of the particular fission product
that penetrated to that depth in the graphite.
Thus, about 43% of the fission product *3%Te
that was found was deposited within 1.3 mils of
the graphite surface, and about 58% within the
first 35 mils.

Values so obtained are shown in Fig. 15.1,
where the percentage of each fission product
which penetrated to given depths is shown as a
function of depth.

The fission products ?3Zr, 141Ce, 144Ce, °'Y,
and '*7Nd are found in low amounts in the graphite
(between 0.4 and 3%) with nearly all that observed
being found close to the surface.
176

Table 15.1. Comparison of Fission Product Activity Found in Various Loop Regions with

Activity Produced in Loop

Percentage of loop inventory (calculated from power history) found in given region

Fission Inventory, ? Salt Loop Salt Sj::le Gas Inlet Total
Isotope Yield Calculated Samples Graphite© Metal Sample Line Line Foundd
(%) (10'° dis/min) Line  (Heateqy (COld)
Isotopes That Leave the Salt
66-hr ° Mo 6.1 940 0.2 41.0 23.4 9.9 0.02 <0.01 74
39.7-day 1%3Ru 3.0 8,100 0.07;0.09 6.0 4.5  0.64 0.0005 0.0001 11
78-hr 1327Te 4.3 1,140 0.8;1.8 31.6 14.0 7.2 0.04 <0.01 54
35-day °°Nb® 6.2 7,400 0.07 13.4 17.6 3.3 <0.01 <0.01 35
Isotopes That Remain in the Salt
65-day 95Zr 6.2 12,800 69;107 0.4 0.6 0.8 <0.01 <0.01 71;106
8.05-day 311 2.93 5,000 30;31 0.3 1.6 1.1 0.1 0.02 33
12.8-day !*%Ba®  6.35 16,500 29;30 2.3 1.3 0.9 <0.01 <0.01 34
50.5-day 2%sr® 4.79 11,500 66;82 8.7 1.1 0.7 0.1 0.01 76;92
58.3-day 2ly® 5.8 12,900 121;61 3.0 0.6 0.6 0.003  <0.00001 125;65
30-year 137Cs® 6.0 108 87,95 1.8 2.2 0.02 0.15 0.02 91;99
32.8-day !4lce 6.0 17,300 70;78 1.3 0.4 0.7 <0.01 <0.01 72; 80
284-day !*4Ce 5.6 3,500 109;124 1.3 0.5 0.7 0.0003 <0.0003 112;126
11.1-day '47Nd 2.6 6,100 72,64 2.0 0.7 1.9 <0.01 <0.01 77;69

“Assuming a mean flux to salt of 0.88 X 1013 based on average of values from 95Z'r, 137CS, and '*%Ce in final

salt samples.

bEstimated for total salt based on each of two final samples. Niobium-95 based on earlier sample because pro-

duction from °3Zr in salt during final period tended to remain in the salt, which was frozen most of the final two

weeks.

€Corrected for salt content.

(Seventeen and twenty-one percent were found in the two final samples.)

dBoth totals are shown if the two salt samples differed appreciably.

®Isotopes with noble-gas precursors.

On the other extreme in quantity are 32Te,
?9Mo, and '°3Ru. These isotopes showed 35 to
43% in the first 1.3 mils, 52 to 56% within 3.1
mils, and not much more at greater depths. These
isotopes are thus indicated to deposit strongly
from the salt onto the graphite surface but migrate
only weakly if at all after deposition.

Noble-gas isotopes would be expected to dif-
fuse into graphite,® so that a more gradual con-
centration gradient of daughter isotopes resulting
from their decay should be encountered. Thus,

3R. J. Kedl, A Model for Computing the Migration of
Very Short-Lived Noble Gases into MSRE Graphite,
ORNL-TM-1810 (July 1967).

the total amount of daughter isotope within a
given depth would continue to increase to
appreciable depths. This pattern is evident for
895y 140Ba, 137Cs, and possibly °'Y.

Iodine-131 also exhibited such a pattern, al-
though it was present in low quantity and amounts
varied from sample to sample. Since iodine is
not expected to be volatile from molten salt,
either a volatile precursor is implied or migration
could have occurred after the salt was frozen or
during subsequent handling.

Niobium-95 is present at concentrations far in
excess of its parent ?5Zr, and its action does
not appear coupled to that of °5Zr. Accumulated
amounts increase rapidly for the first 5 mils from
177

ORNL—~DWG 68-1994R

ITE
£
Q
)
©
g
o]
]

[ 13214 (54%,)
03R4 (1.3 %)
L 990(74%%)

S B—

89, (84%,)

405, (34%)
f

147Nd (73 %)

+,

Moo (76 %)

m

PERCENTAGE OF TOTAL FISSION ISOTOPE PENETRATING TO GIVEN DEPTH IN GRAPH

02 I DR

FIGURES IN PARENTHESES
SHOW PERCENTAGE ACCOUNTED
FOR IN ISOTOPE BALANCE

ok

15 20 25

DEPTH (mils)

Fig. 15.1. Penetration of Various Fission Product Isotopes into Graphite as a Percentage of Total Isotope

137C 5 (95 To) 1
¥ (34%)
_ !

- 987, (89%)

in Loop.
178

11 to 32% and then increase regularly but more
slowly up to about 37% at 35 mils. Apparently
this isotope was deposited relatively strongly on
the graphite but also tended to penetrate inward,
in this way exhibiting the pattern attributed above
to gases.

15.3 STUDIES OF SURFACE WETTING
OF GRAPHITE BY MOLTEN SALT

During postirradiation examination of the second
in-pile molten-salt loop, evidence of surface wet-
ting of the core graphite by salt was noted. We
wish to establish procedures to prevent this in
the next loop experiment. We have recently
studied in a vacuum box the wetting characteristics
of droplets of MSRE solvent salt on a platelet of
MSRE gtaphite. We confirm generally the observa-
tion of Kreyger, Kirslis, and Blankenship* that
wetting is due to three-phase contact of gas,
graphite, and molten salt at moisture levels as
low as 10 ppm or lower.

In an atmosphere of tank helium (below 4 ppm
water), the droplets on melting did not wet the
graphite. However, in a few minutes they de-
veloped a translucent crust and in 1 to 2 hr after
melting had slumped and spread spasmodically
over the graphite surface. Another droplet was
melted on graphite under good vacuum (less than
10— 4 torr, the present gage limit) and remained
clear and round for about 20 hr. However, when
tank helium was admitted to a pressure of 0.2
torr, the molten droplet promptly slumped and
wetted the graphite surface.

The surface wetting behavior of the salt thus
appears to be very sensitive to very slight im-
purities, presumably water, in the helium. This
could have been accentuated in these studies
because of the small amount of salt and the
large amount of gas. Further, the lack of wet-
ting of the graphite by salt in the experiment
under vacuum indicates that the prior condition
of graphite or salt did not control the wetting
behavior. Consequently, gas cleanup procedures
are being tested as the next step in studying the
wetting phenomenon.

“p. J. Kreyger et al., MSR Program Semiann. Progr.
Rept. July 31, 1963, ORNL-3539, p. 125.

154 DESIGN OF A THIRD IN-PILE
MOLTEN-SALT LOOP

In-pile operation of the second molten-salt loop
experiment was terminated because of a crack in
the core outlet pipe’ with resultant fission product
leakage. From evidence of postirradiation exami-
nation and evaluation of the effects of neutron
irradiation on properties of Hastelloy N, it was
concluded that this failure was caused by de-
terioration of stress rupture properties due to
neutron irradiation.® Therefore, the third loop
assembly will be constructed of a modified
Hastelloy N containing titanium as an additive
for improving resistance to radiation-induced
high-temperature embrittlement.” Required shapes
for construction of the loop components have been
obtained from an ingot of such (~0.5% Ti, 2% W)
Hastelloy N.

We have also redesigned the graphite core sec-
tion of the next loop to provide additional graphite
surveillance specimens. The redesigned core has
four holes of % -in.-square cross section instead
of the eight ¥ -in.-diam round holes for salt flow
used for the two loops previously operated in-
pile.® Graphite specimens with a 3‘/8-in.—square
cross section are inserted coaxially in each hole,
resulting in a % _-in. annular salt flow channel,
as shown in Fig. 15.2. The flat surfaces of these
graphite specimens will permit improved preirradia-
tion evaluation of the graphite quality and post-
irradiation examination to determine interaction of
graphite and salt. Metal locating brackets will be
used to position the graphite specimens and will
also serve as surveillance specimens for metals
of interest.

In addition, we plan to install an adsorption
trap immediately adjacent to the vapor space in
the loop gas separation tank in an effort to trap
and identify the gas-borne fission product species.

SE. L. Compere ef al., op. cit., p. 182.

6H. C. Savage et al., Operation of Molten-Salt Con-
vection Loop in the ORR, ORNL-TM-1960 (December
1967).

H. E. McCoy, Jr., et al., MSR Program Semiann.
Progr. Rept. Aug. 31, 1967, ORNL-4191, p. 18.

SE. L. Compere et al., op. cit,, p. 190.
179

ORNL-DWG 68-5563

Yo x Y, _—HASTELLOY CAN
FLOW CHANNEL

_— GRAPHITE CORE

THERMOCOUPLE
WELL
SALT FLOW

~ ¥g x % GRAPHITE
SPECIMEN

Fig. 15.2. Graphite Core and Specimens, Molten-Salt Y6 SALT FLOW
In-Pile Loop. ANNULUS

SALT FLOW E E
—————

———HASTELLOY END CAP

— GRAPHITE SPECIMEN

HASTELLOY
SUPPORT PIN

SALT FLOW
—

DIMENSIONS IN INCHES
16. Gamma Irradiation of Fluoroborate

E. L. Compere

H. C. Savage

J. M. Baker

The use of sodium fluoroborate as a coolant in
the Molten-Salt Breeder Reactor will involve the
possibility of irradiation damage due to gamma
rays and delayed neutrons from the fuel salt in
the heat exchanger.

Gamma rays might ionize and decompose sodium
fluoroborate to produce boron, fluorine, and sodium
fluoride. Neutrons in a '°B(n,a) reaction would
be expected to yield an alpha particle, lithium
fluoride, fluorine, and sodium fluoride. The high-
energy alpha particle could also ionize and de-
compose the sodium fluoroborate. The fluorine
generated by either process would probably react
with structural metals of the system to form their
fluorides. Thus, a major effect of radiation damage
to sodium fluoroborate by either mechanism would
be a corresponding amount of corrosion. In the
projected test of the 8% NaF-92% NaBF, eutectic
mixture in MSRE, the neutron flux is negligible
(~6 x 10® nv); therefore, the present studies are
emphasizing the effects of gamma radiation. The
gamma irradiation experiments are conducted in
the central channel of spent HFIR fuel elements
in the storage pool of the HFIR, where very high
gamma fluxes are available (™8 x 107 t/hr for
elements four days old); this is comparable with
the gamma flux in the MSRE heat exchanger.

For the first gamma irradiation experiment 34 g
of an NaF-NaBF, eutectic mixture with a melting
point of ~380°C was placed in a Hastelloy N
capsule (0.93 in. OD x 0.78 in. ID x 3.5 in. long)
containing a Hastelloy N corrosion test specimen.
The salt was melted in the capsule in an argon
atmosphere box, and then the capsule was welded
shut. A capillary tube connected the gas space
above the salt to a pressure transducer. A heater
and three thermocouples were used to control and

180

monitor temperature. The capsule assembly was
then placed in an aluminum container to isolate it
from the pool water. !

Prior to the start of irradiation, the capsule
containing the sodium fluoroborate was vacuum
pumped (5 x 1073 torr) for 16 hr at 150°C to re-
move any moisture and residual gas. The capsule
was then sealed by closing the valve on the gas
line andheld at 450°C for 3 hr. On cooling to
20°C a residual pressure of ~5 torrs was observed.
Analysis of the residual gas by mass spectrograph
showed it to contain N, (73%), Ar (13%), CO, (8%),
H, (4%), O, (2%), and H,0 (0.2%). The capsule
was then vacuum pumped at 150°C for 1 hr, closed
off, and operated for 68 hr at 600°C. The observed
pressure at 600°C was 145 mm Hg — close to the
value anticipated for BF, vapor pressure accord-
ing to Cantor’s data.? When cooled to 20°C,
residual gas pressure was below the limit of de-
tection ("5 torrs).

The capsule assembly was then placed in the
center of HFIR fuel element 34 on February 2,
1968. This element had been removed from the re-
actor on January 29, 1968, after operating for 23
days at 100 Mw. The sodium fluoroborate salt
temperature reached 465°C from gamma heat alone
(estimated to be 0.25 w/g). Electrical heat was
then added to bring the temperature to 600°C.

On February 4, 1968, it was observed that the
electrical power required to maintain the capsule
at 600°C was higher than anticipated and that the
indicated wall temperature at the center of the
capsule (below the salt-gas interface) was below

'H. c. Savage et al., MSR Program Semiann. Progr.
Rept. Aug. 31, 1967, ORNL-4191, pp. 194-95.

Zs. Cantor, MSR Program Semiann. Progr. Rept. Aug.
31, 1967, ORNL-4191, pp. 159—61.
that of the upper section (gas space). These
observations indicated the possibility of water

in the container surrounding the capsule, and the
experiment was removed from the fuel element.
No evidence of water in the container was found
by vacuum pumping while heating the capsule to
~100°C. However, only the thermocouple located
on the wall by the gas space of the capsule was
still operating. The experiment was again placed
in the center of the fuel element, and operation at
600°C, as indicated by the above thermocouple,
was continued.

The experiment was removed from the fuel ele-
ment on February 26, 1968, after 533 hr of irradia-
tion. Observed pressure during most of the ir-
radiation period ranged between 100 and 150 mm.
For the 533 hr of irradiation the NaF-NaBF, salt
accumulated an estimated absorbed dose of ~1 x
1024 ev/g.

Residual gas pressure in the capsule was ~ 10
torrs at 50°C and was found to contain a large
percentage of hydrogen by mass spectrographic
analysis (H, 85%; N, 12%; H,0, 2%; CO,, 1%;
Ar, 0.3%; 02, 0.1%), which was confirmed by gas
chromatography. The source of the hydrogen in
the residual cover gas is not known; incompletely
fluorinated fluoroborate (NaBFSOH) and in-diffu-
sion of radiolytic or corrosion hydrogen from the
water in the containment are possibilities.

Upon opening the aluminum container can, it
was found to contain sufficient water to soak the
magnesia insulating pad under the capsule. Since
the capsule bottom was in contact with a water-
soaked insulation pad, a severe temperature gra-
dient undoubtedly existed from the top to bottom
of the capsule for most of the irradiation period.
The two thermocouples at the center and near the
bottom of the capsule were lost early in the run,
thus the temperature at the bottom of the capsule
is unknown. However, it is estimated that at
least part of the NaF-NaBF | salt in the capsule
remained frozen after the first two days of opera-
Since no leaks were found in the aluminum
can by pressure testing, it is assumed that water
was inadvertently spilled into the can through the

tion.

container access tube.

The capsule was cut open, and the corrosion
test specimen and salt were removed. No visual
evidence of corrosion was seen. The salt, how-
ever, was discolored. Several sharply defined
layers of salt of varying shades of pink from the
bottom to about the center were observed with a

181

thin dark-brown layer on the bottom. The upper
half of the salt was a mottled gray white. The
layers are believed to be associated with solidifi-
cation progressing upward from the bottom as the
internal gamma heating diminished, with the moist
insulation permitting significant heat removal.

An unirradiated capsule assembly identical to
the irradiated capsule was operated molten for
840 hr (mostly at 600°C) as a control for the ir-
radiation experiment. Salt removed from this ex-
periment was entirely white and rather frangible
with none of the pink-brown discoloration ob-
served in the irradiated salt. However, small
brilliant-green crystals identified microscopically
as Na CrF were found near metal surfaces —
particularly at the bottom. For the unirradiated
capsule, there was very little temperature gradient
from top to bottom as was initially the case for
the irradiated capsule.

Samples of salt removed at various levels from
the respective irradiated and unirradiated experi-
ments were examined by mass spectrographic
analyses. In each case, the bottom layer con-
tained 10,000 ppm Cr or more and about 2000 ppm
Ni, Fe, and Ca, and 500 ppm of Mo, Ti, Mg, and
Al, as well as smaller but appreciable amounts of
a number of other impurities. The Na/B atom
ratio ranged between 0.96 and 1.12 for most regions
as might be expected for the sodium fluoroborate—
sodium fluoride mixture (nominally, Na/B ~1.09).
The region of the unirradiated material with green
crystals showed an Na/B ratio of 1.26, whereas
the corresponding dark-brown bottom region of the
irradiated salt had a ratio of 1.09.

The upper surfaces of both salts had a dark
scum, and samples from this region in each case
showed an Na/B ratio of ~0.9.

It is believed that a mild corrosion of Hastelloy
N occurred in both cases, with a dark boron scum
being produced. The corrosion products collected
near the bottom.

Hastelloy N coupons exposed in the irradiated
and unirradiated tests exhibited negligible attack
other than a few surface stains.
amination revealed a general slight dulling — more
evident at the liquid surface and above and some-
what more pronounced on the unirradiated coupon.
The weight change of the unirradiated specimen

Microscopic ex-

corresponded to less than 0.01 mil/year; for the

irradiated coupon, the weight change was zero.
The above findings are not indicative of any

noteworthy effect of gamma irradiation on the
sodium fluoroborate salt or its compatibility with
Hastelloy N.

A second gamma irradiation capsule assembly
has been fabricated, and it is planned to irradiate
this experiment to a higher total dose than the

182

first experiment. Additional insulation at the top
and bottom of the capsule has been added to en-
sure that the entire capsule is at a nearly uniform
temperature.
Part 5. Materials Development

H. E. McCoy, ]Jr.

The primary structural materials in molten-salt
reactors are Hastelloy N and graphite. Our ex-
perience with the MSRE has demonstrated the ex-
cellent compatibility of these materials with fluoride
salts in a nuclear environment. We have a surveil-
lance facility in the MSRE that allows us to period-
ically remove graphite and Hastelloy N specimens
for examination. Both materials have been used
in our fission product studies, and we have run
mechanical property tests on the Hastelloy N to
follow the changes in the properties of the reactor
vessel,

Future MSBR’s will probably utilize similar
structural materials, but some advances in tech-
nology are desirable. The requirements placed on
the graphite are somewhat more severe in an MSBR

J. R. Weir

than in the MSRE. The graphite will be exposed
to higher fluences and must have low permeability
to gaseous fission products. We presently have a
facility for irradiating graphite to a fluence of
about 3 x 1022 nvt (>50 kev) per year, and we are
investigating the dimensional changes that occur
in various grades of graphite. We plan to reduce
the permeability of graphite by surface sealing
with metals or pyrocarbon. We have developed a
modified Hastelloy N with superior resistance to
irradiation damage and are studying its properties
in detail. Some new salt compositions are of
interest for MSBR’s, and we are expanding our cot-
rosion program to investigate their compatibility
with Hastelloy N.

17. MSRE Surveillance Program

17.1 GENERAL COMMENTS

H. E. McCoy, ]Jr. W. H. Cook

We maintain a graphite and Hastelloy N surveil-
lance program to follow the changes in properties
with irradiation of the graphite and Hastelloy N
used in the MSRE. A special fixture has been de-
signed! so that these materials can be exposed to
the MSRE environment and removed periodically
for examination. The graphite has been used
principally for fission product deposition studies,
since we presently do not have hot-cell techniques

TMsR Program Semiann. Progr. Rept. Aug. 31, 19685,
ORNL-3872, pp. 87—-92.

183

for measuring the very small changes in physical
properties that have likely taken place. The
Hastelloy N is in a form that can be converted to
mechanical test specimens quite easily, and we
have evaluated its postirradiation properties by
tensile and creep-rupture tests.

The space in the surveillance fixture is some-
what larger than required to follow the property
changes in the same materials used to construct
the MSRE. Hence, we have inserted some special
types of graphite and Hastelloy N (1) to gain a
better understanding of fission product behavior,
(2) to obtain compatibility information, and (3) to
determine irradiation damage information.

We plan to remove another group of surveillance
specimens from the MSRE about April 1, 1968.
184

17.2 EXAMINATION OF HASTELLOY N
SPECIMENS FROM MSRE
SURVEILLANCE FACILITY

H. E. McCoy, ]Jr.

We removed a second set of Hastelloy N speci-
mens in June 1967. The specimens removed from
the core had been at temperature for 5500 hr and
had accumulated a peak thermal dose of approxi-
mately 4 x 102°% neutrons/cm?. The core speci-
mens were modified alloys containing approxi-
mately 0.5% Ti (heat 21545) and 0.5% Zr (heat
21554). A stringer of standard Hastelloy N speci-
mens from outside the reactor vessel was also re-
moved. These specimens had been exposed to the
MSRE cell environment for about 11,000 hr and had
accumulated a peak thermal dose of about 3 x 101°
neutrons/cm?. The vessel specimens were made
of the same heats used in constructing the MSRE.

The creep-rupture properties of the heats of
standard Hastelloy N are shown in Fig. 17.1. Data
are shown for (1) the first set of specimens re-
moved from the core after 4800 hr with a thermal
fluence of 1.3 x 102° neutrons/cm? and (2) the
second set of specimens located outside the
vessel and removed after 11,000 hr with a thermal
fluence of 3 x 1019 neutrons/cm?. The specimens
irradiated to the lower fluence generally had a
larger rupture life at a given stress, although the
differences become quite small at low stresses
(compare points for heat 5085). We have no ex-
planation for the superior rupture life of heat 5081.

The minimum creep rates are compared in Fig.
17.2 for several heats of Hastelloy N in their-
radiated and unirradiated conditions. As observed
previously,? this parameter is not affected.

2. E. McCoy, Jr., and J. R. Weir, Nucl. Appl. 4(2),
96 (1968).

ORNL-DWG 68~-6576

70 -
T F
60
AN
\\ | —UNIRRADIATED
50 ‘ — N a
\J A \
T
% a0 |- rr\4‘~\;_F:i L Lk
a ~ ] —fi
2 ] ] §
: e T T \
A ° Ton ol & | N
W L J N~ N
=
wn | To \W\ \A
N\ \

20 L

MSRE SURVEILLANCE

VESSEL  CORE HEAT
o 5065
o n 5085
. A 508 B
19 20
3X10 1.3%102° THERMAL FLUENCE
11,000 4800  TIME AT 650°C (hr)
R R e
T 5  10° 2 5 10 2 5 10° ) 5 103 2 5 0%

RUPTURE TIME (hr)

Fig. 17.1. Postirradiation Creep-Rupture Properties of MSRE Surveillance Specimens at 650°C.
185

The fracture strains of the vessel heats are
shown in Fig. 17.3. The scatter band was de-
termined from a large number of data points, with
heat 5065 being on the lower side and heat 5085
being on the upper side. The minimum fracture
strain at a strain rate of about 0.1%/hr clearly
exists for materials exposed to irradiation for
about 1000 hr at 650°C, whereas the data for
the surveillance specimens do not suggest the
existence of this minimum. The fracture strains
of most of the surveillance specimens fall be-
tween 1.5 and 2.5%; however, heat 5081 is some-
what anomalous and exhibits higher fracture
strains with the trend of increasing fracture strain
with decreasing creep rate. The results on heat
5085 indicate that the fracture strain may be de-
creased only slightly by increasing the fluence

from 3 x 109 to 1.3 % 102°% neutrons/cm?. How-
ever, the data at higher strain rates indicate a
large reduction in the fracture strain as a result
of the increased fluence.

The tensile properties of the air-melted surveil-
lance materials are shown in Table 17.1 for test
conditions of 25 and 650°C. The fracture strains
at 650°C are considerably lower at the higher
fluence. Thereduction in ductility at 25°C is
thought to be associated with carbide precipita-
tion in the alloy and is probably a function of both
the thermal history and the fluence.

The property changes that have been noted in
Hastelloy N as a result of service in the MSRE
are similar to those noted for Hastelloy N ir-
radiated in an inert-gas environment, thus reflect-
ing the excellent compatibility of the alloy with

ORNL—-DWG 68—~ 6577

70 T T 7{

il

o - _J_.J,’_J

60 F——1 [ HH——— -

- 1

50 hk“r“—”—**—l—?*‘

%‘
4 A

S‘ 40 — I~ J
S
= /M
) y
4 PP
o |
'f_
w

20 Tfl“‘

T AT
f
; !
20 J/" | B I
on | ] i

MSRE SURVEILLANCE

VESSEL CORE HEAT
O 5065
0 ] 5085 |

A 5081
3x10"°  13x102° THERMAL FLUENCE
1,000 4800  TIME AT 650°C (hr)

10 f _J

1 Rl

0% 2 5 1073 2 5 072 2

5 40" 2 5 10° o 5 i0

MINIMUM GREEP RATE (%/hr)

Fig. 17.2. Variation of Minimum Creep Rate of Hastelloy
Tests at 650°C.

N MSRE Surveillance Samples in Postirradiation Creep
186

ORNL-DWG 68-6578

r Tzs.s
22.4
l @ “]
W 1 [ l (e
P i \ Smunas
2 4, MSRE SURVEILLANCE \r ' !
VESSEL CORE HEAT
i o 5065 N mul
o . 5085
10 A 5084 1| H
331019 1.3x102°  THERMAL FLUENCE j
14,000 4800 TIME AT 650°C (hr) / L
9 —\f H T
8 TFLL—“_T**_I*

FRACTURE STRAIN (%)

L DAL T T
il ALY s wesonreo

A I I

13}“3‘; M Bl 1l

o i I LD

1073 10! 102 10°

STRAIN RATE (%/hr)

Fig. 17.3. Variation of Fracture Strain with Strain Rate for MSRE Surveillance Specimens at 650°C.

fluoride salts. The fracture strain and the rupture
life at a given stress are both reduced, but not to
levels that should prevent the continued success-
ful operation of the MSRE. The property changes
noted in increasing the fluence from 3 x 109 to
1.3 x 102° neutrons/cm? were quite small. The
vessel should not reach a thermal fluence of 1.3 x
102° neutrons/cm? until after 150,000 Mwhr of
operation. Thus, if the effects of thermal aging
are not adverse, the properties of the vessel
should not deteriorate much below those observed

for the material irradiated to 1.3 x 102° neutrons/
cm?.

The titanium- and zirconium-modified alloys that
were removed from the core were heat treated to
obtain a very fine grain size. At the time these
were inserted, it was thought that the fracture
strain increased with decreasing grain size. How-
ever, subsequent tests showed that the reverse
was true. The postirradiation properties of these
materials were quite comparable with those noted
for the standard Hastelloy N.
187

Table 17.1. Tensile Elongation (%) for MSRE Surveillance Specimens

Test Heat 5065 Heat 5085 Heat 5081
Temperature 1 1 ] 1 ) 1
©c) 0.05 min™ " ° 0.002 min~ 0.05 min— 0.002 min~ 0.05 min— 0.002 min—
19 2
Vessel (3 x 10" 7 neutrons/em*)
25 34.6 42.5
650 25.8 11.4 22.4 12.0
Core (1.3 x 1020 neufrons/cmz)
25 27.3 38.7
42.6
650 13.7 9.4 14.3 9.0

4Strain rate.
18. Graphite Studies

18.1 PROCUREMENT OF SPECIAL GRADES
OF GRAPHITE

W. H. Cook

The current studies on graphite for molten-salt
breeder reactors include (1) the determination of
the physical and mechanical properties before and
after irradiation, (2) sealing research with pyro-
lytic carbon and chemical-vapor-deposited niobium

and molybdenum, (3) graphite-to-metal joining
studies, and (4) fabrication of test assemblies.
Graphite for the initial phases of items 1 through
3 has been obtained in reasonable quantities.
Some graphite for item 4 is on order.

The graphite manufactured by Poco Graphite,
Inc., continues to be our reference graphite for the
sealing and joining work because it (1) has reason-
able uniformity, (2) has pore entrance diameters

Table 18.1. Receipt and Utilization of Special Grades of Graphite Received Since January 1, 1968

Nominal
Grade Source Type Bulk Dex;s1ty Dimensions Pieces Utilization
(g/cm”) .
(in.)
9950 Speer®  Near isotropic 1.71 21/2 X 4 x 8 1
9949 Speer? Near isotropic 1.71 21/2 X4 X8 1
99048 Speer? Near isotropic 1.92 21/2 X4 %8 1 Grade 9948 will be evaluated in
HFIR irradiation studies
AXF Poco? Isotropic ~1.82 7/16 diam X 6 200 For sealing studies
AXF-5Q Poco Isotropic 1.81 4 diam X 18 1 Graphite-to-metal joining studies
AXF-50BG Poco Isotropic 1.89 11/2 X4 xX6 4 Graphite-to-metal joining studies
AT]J-S CPD®  Near isotropic ~1.83 8 diam x 12 1 Made by a special process which
has potential for fabrication
of MSBR shapes; evaluation
to include irradiation in the
HFIR
AT]J-SG CPD Near isotropic 1.81 8 diam X 12 1 Made by the AT]J-S process

using a more isotropic filler
material, Gilsocarbon flour;
evaluation to include irradia-
tion in the HFIR

4Gratis material from Speer Carbon Company.

bpoco Graphite, Inc.

°Carbon Products Division of Union Carbide Corporation.

188
189
that are in the desired range (approx 1 p), and (3) ment was used to remove fingerprints or other
is readily obtainable in small but useful sizes for contaminants from graphite test pieces before
these studies. they were sealed by the vapor deposition of
The grades and disposition of graphite obtained metals. Specimens given this pretreatment were
recently are summarized in Table 18.1. unusually permeable and were not sealed by rea-

sonable metal deposits. Microstructural examina-
tion showed that the normal, relatively uniform
18.2 POROSITY CREATED IN GRADE AXF pore structure with entrance diameters approxi-
GRAPHITE BY OXIDATION PRETREATMENT mately 1 p now had a secondary set of pores with
some more than a hundred times as large as the
basic pores.?
This was a new lot of grade AXF graphite in
stock sizes of 7/1 -in.-diam, 6-in.-long rods; so

W. H. Cook

The porosity of grade AXF graphite! used in
coating and sealing studies was increased to an
unacceptable extent when the specimens were
treated for 1/2 hr in air at 600°C. This pretreat-

6

2The original pore entrance diameters were measured
1 with a mercury intrusion porosimeter, and the secondary
Manufactured by Poco Graphite, Inc., Garland, Tex. pores were measured with a microscope.

PHOTO 91877

oD

~ 7/32 In.
R—— S

Fig. 18.1. Microstructures of Grade AXF Isotropic Graphite Specimens from 7¢16oin.-diam Rods. (a) As received,
(b) after exposure for 1/2 hr in air at 600°C, and (c) machined to 0.400 in. OD x 0.100 in. ID and then given the same
treatment given the specimen shown in b. The normal pores are not resolved in these microstructures; they are usu-
ally associated with the dark-gray phase of the structure, but assuming that they are greater than 1 p (the approximate
diameter of their entrances), they would be only approximately 0.004 in. on the photomicrographs. However, the

oxidized voids (pores) are clearly visible as the large black circular spots. 100x. No etchant.

a series of tests were made to determine if the
secondary set of pores was the result of the
pretreatment or was characteristic of the new

lot of material. Control and test specimens were
taken from three random rods. Test specimens
from each rod were given the standard pretreat-
ment, and duplicate specimens were given the
same anneal in an argon environment. The normal
grade AXF graphite microstructure, as shown in
Fig. 18.1a, was observed for all three controls
and the test specimens treated in argon. All
those given the standard oxidation pretreatment
developed the large secondary set of pores shown
in Fig. 1856 and c. The voids were at irregular
depths ranging from 20 mils to completely through
the 7 _-in.-diam specimens. The oxidized void
(pore) sizes graded from large to small from the
external surfaces to inside the specimens.

The samples used in the sealing studies were
0.4 in. OD x 0.1 in. ID x 1.45 in. long, and the
large voids introduced by oxidation generally
penetrated the entire wall. The pretreatment in
air will not be used in future work. Surface con-
tamination will be reduced by more careful han-
dling, followed by anneals in argon or vacuum
environments.

18.3 X-RAY STUDIES ON GRAPHITE
0. B. Cavin

X-ray diffraction is a useful technique for ex-
amining many of the properties of graphite that

190

are closely related to its resistance to radiation
damage. We are concemed with the determination
of lattice parameters (in both the ¢ and a direc-
tions), preferred orientation, and crystallite size
(L ). Quantitative data are difficult to obtain and
are strongly affected by equipment and analytical
procedures.

We have begun measurements of the crystallite
sizes and lattice parameters of nearly isotropic
polycrystalline graphite to develop an accurate
approach for data gathering and analysis and to
characterize these particular graphites. Pre-
liminary results are given in Table 18.2. The
lattice parameter measurements are reasonably
consistent and agree well with those obtained by
other investigators. The values for the a param-
eter are reasonably constant, but the values for
¢ vary. These variations probably reflect dif-
ferences in crystal perfection.

The crystallite size measurements were de-
termined independently from the broadening of
the (002) and (004) diffraction peaks. These de-
terminations should give identical values for L,
but the data in Table 18.2 show that this was not
the case. The basic Scherrer?® equation used in
analyzing these results involves an instrumental
correction factor. Again different results were
obtained using the correction factor suggested by
Scherrer® and the one later proposed by Warren.?

3H. P. King and L. E. Alexander, X-Ray Diffraction
Procedures, pp. 491, 500, Wiley, New York, 1954,

Table 18.2. X-Ray Measurements on Several Graphites

Sample 2 (A) ¢ (A) L_(&)° L, (a)°
(110) (004) (002) (004) (002) (004)

AXF 2.462 6.769 189 116 250 156
AXF-5QBG 2.464 6.763 196 130 262 183
AXF-5QBG-3 2.463 6.749 224 142 313 205
BY-12 2.463 6.737 272 204 406 342
RY-12-00029-32 2.463 6.732 250 198 361 330
RY-12-00031-34 2.464 6.736 272 191 406 312
CGB base stock 2.464 6.731 277 235 417 424

8Using Warren’s instrumental correction.

bUsing Scherrer's instrumental correction.
These apparently contradictory results are due to
the fact that the Scherrer equation assumes that
the only source of line broadening is that due to
the crystallite size.

In addition to crystallite size, there are at least
six other factors which may contribute to line
broadening:* (1) compositional variations, (2)
small crystallite size, (3) inhomogeneous strains,
(4) stacking faults, (5) range of wavelengths in
incident beams, (6) instrumental factors, and (7)
low absorption in the sample. Thus the Scherrer
equation leaves us with a rather meaningless
summation of these factors. Therefore, in the
future we plan to use a more accurate Fourier
approach for the analysis of peak shapes rather
than using only peak breadths. In this manner
we can separate more of the factors and obtain
a more realistic measurement of the crystallite
size Lc.

18.4 GAS IMPREGNATION OF GRAPHITE
WITH CARBON

R. L. Beatty D. V. Kiplinger

One of the requirements for graphite to be used
in a molten-salt breeder reactor is a surface with
low permeability to prevent xenon absorption.
Calculations suggest that a helium permeability
of less than 107 cm?/sec at the graphite surface
will be required to keep the xenon concentration

in the core to the desired level. Since commercially

available graphites usually have helium permeabil-
ities some five or six orders of magnitude higher
than this required level, it is necessary to con-
sider coating or sealing the graphite surface by
some means.

Present considerations are to seal the graphite
surface either with pyrolytic carbon or with a metal
such as molybdenum or niobium. If a metal is em-
ployed, the amount must be strictly limited to
avoid excessive neutron absorption penalties.
Though no such volume limitation per se exists
in the case of pyrolytic carbon, there is a limita-
tion on the means of employing it. Since the
crystalline character of the pyrolytic carbon
deposit is likely to differ markedly from that of

‘c. s. Barrett and T. B. Massalski, Structure of
Metals, pp. 155, 454, McGraw-Hill, New York, 1966.

191

the base graphite, the neutron-induced dimensional
changes of the two materials will probably be dif-
ferent also. Therefore, if the pyrolytic carbon is
applied simply as a coating, it may be subject to
spalling during irradiation. This problem may be
circumvented, however, if the carbon is deposited
in the pores near the surface rather than on the
surface.

We are exploring methods of gas impregnating
graphite with carbon. A technique for providing
a very low-permeability surface seal by immersing
a graphite specimen in a fluidized bed has been
demonstrated,® but this method may be impractical
for large bodies such as are required in a molten-
salt reactor. Another method we are studying in-
volves forcing the carbon-bearing gas radially
through a heated specimen, up a thermal gradient.
The thermal gradient is achieved radially in the
specimen by heating it with a 1.2-kw, 450-kc in-
duction generator while cooling the inside by
means of an axial %  -in. water line. We selected
the specimen geometry to meet the requirements
of the HFIR irradiation facility (i.e., a hollow
right circular cylinder of nominal dimensions
0.400 in. OD, 0.125 in. ID, and 0.500 in. long).
The base stock is Poco graphite grade AXF.
During impregnation the specimen is clamped
between spring-loaded mullite tubes which seal
against the ends with Grafoil gaskets. The
carbon-bearing gas and a diluent are forced into
one of the mullite rams and can exit only by pass-
ing radially through the graphite. Sources of
carbon being considered are methane, propylene,
and butadiene.

We checked 30 as-machined AXF specimens for
leak tightness and found great variation. We ob-
tained vacuum levels ranging from 4 to 120 ; Hg
when the specimens were pumped with the rough-
ing pump of a Veeco leak detector.

Preliminary impregnation experiments indicate
that this range of as-received permeabilities may
not be important for carbon impregnation, but we
will examine this point in detail. With the above
techniques we have sealed two specimens so that
the helium leak rates measured by the Veeco de-
tector are less than half that of our 4.5 x 10— 8
std cm?/sec standard leak. Though sealing at
the ends of the specimen is a problem with this

SMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 212.
technique, we believe it is promising and can be
used either alone or in conjunction with surface
diffusion impregnation.

We are also studying a system which cycles be-
tween vacuum and hydrocarbon atmosphere while
the graphite substrate is heated to temperatures
of 800 to 1300°C. The variables being considered
in this system are temperature, cycle time, con-
centration of hydrocarbon gas, system pressure,
and original substrate porosity.

The experimental apparatus includes a silica
tube with the test specimen suspended inside by
a graphite thread. The silica tube is closed at
each end by a solenoid valve, the bottom one
being connected to a vacuum pump and the top
one to a source of hydrocarbon gas. The two
solenoids are controlled by a pulse timer that
automatically switches the system between
vacuum and hydrocarbon atmosphere at preset
intervals.

The first carbon source we used with this
system was a mixture of 1,3-butadiene in argon
(10% C,H, +90% Ar). We found that by using
argon as the diluent, we could operate from 150
to 200°C hotter with the same power setting than
with helium as the diluent. We later found that
the time required to seal a sample can be reduced
by using undiluted butadiene. We selected buta-
diene as the carbon source because it decomposes
at a lower temperature than hydrocarbons such as
methane and propylene and has a high carbon to
volume of gas ratio.

The impregnation process appears to be quite
sensitive to temperature. At temperatures above
1000°C the processes studied tend to coat the
samples rather than impregnate them. At tem-
peratures in the range of 800 to 950°C, undiluted
butadiene appears to penetrate the pores, but the
deposit is probably one or more of a family of
highly viscous hydrocarbons. These tars, when
baked out, release hydrogen, which tends to re-
open the pores to some degree,

Several samples were prepared by pulsing be-
tween 5 sec of vacuum (<25 in. Hg) and a pres-
sure of 10 psig of undiluted butadiene in the
temperature range 800 to 900°C. We found that
samples could be made leak-tight after a total
hydrocarbon contact time of 1 to 2 hr. The ex-
pression ‘‘leak-tight’’ here means a leak rate less
than the standard used, which was 4.5 x 10~ ¢ std
cm?®/sec of helium. These samples were then
heated to 1300°C, cooled, and rechecked for

192

leakage. In all samples studied, except one
sample imptegnated at 900°C, the heat treatment
was found to increase the leak rate.

The sample impregnated at 900°C that remained
impermeable to helium after heating at 1300°C was
heated to 2000°C for 5 min. The dimensions re-
mained the same with a very slight loss in weight,
and the helium leak rate was 200 x 10~ 8 std cm?3/
sec after 20 min of helium purge.

This specimen was then reimpregnated at 900°C
and heated at 2000°C. This time no leak was de-
tected by the Veeco instrument, which means that
the helium permeability was less than 10—° cm?/
sec. There was no measurable thickness of coat-
ing on the outside of the specimen as a result of
the two impregnations. Thus we feel that the
pulsing technique for gas impregnation of graphite
with carbon is the most promising technique we
have tried. Further, this technique should be
relatively easy to scale up and is equally appli-
cable to either hollow tubes or solid rods.

18.5 GRAPHITE SURFACE SEALING
WITH METALS

W. C. Robinson

Detailed analysis of the molybdenum-coated
R-0025 graphite samples reported previously®
indicated that there was no direct connection be-
tween the thickness of metal coating and the
helium permeability of the coated sample. The
first experiment, M-Mo-1, which had a coating of
approximately 0.05 mil, had a permeability too
small to be detected by a Veeco leak detector.
The other eight samples had coatings up to 0.275
mil and would not pump down low enough to
measure the permeability. We decided that this
lack of correlation must be due to the inhomo-
geneity of the R-0025 starting material.

Twenty samples of AXF graphite were obtained.
Prior to coating, each sample was tested by at-
tempting to pump a vacuum on it with the pumps
of the Veeco leak detector. Pressures down to
0.001 torr can be measured. After coating, the
sample was again pumped and the pressure was
recorded. If the pressure was £0.001 torr, an
attempt was made to measure the permeability.

SMSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 211.
193

Table 18.3. Surface Sealing of Graphite by Metal Coatings

Gas
Experiment I;t;‘g/:i:; Tem%erature Pressure Time T:iii::fs I;::Cuou: Vz:za:m 1'-’ermesabilitya
No. o) (torrs) (min) (mils) (torrs) (torrs) {cm” /sec)
MoF _ H,
M-Mo-10 50 800 700 5 0.033 0.120 0.110
M-Mo-11 50 800 700 10 0.073 0.115 0.082
M-Mo-12° 50 800 700 10 5 0.049 0.010 0.017
M-Mo-13 50 800 700 0.046 0.075 0.065
M-Mo-14 50 800 700 10 0.091 0.120 0.020
M-Mo-15 50 800 700 10 5 0.102 0.135 0.022
M-Mo-16 50 800 700 10 30 0.520 0.155 0.01
M-Mo-17 50 800 700 10 10 0.066 0.082 0.05
M-Mo-18 50 800 800 5 5 0.080 0.001  <0.001 <1.5x 10~1°
M-Mo-20 Oxidized
M-Mo-21 Oxidized
M-Mo-22 50 800 800 5 0.081 0.050 0.010
M-Mo-23 50 800 800 10 0.252 0.075 0.001 3.15x107°
M-Mo-24 50 800 800 10 5 0.136 0.100 0.010
M-Mo-25 50 800 800 10 0.161 0.090 0.001 >7 x 107 %¢
M-Mo-26 50 800 800 20 0.374 0.095 0.001 >7x107%°¢
M-Mo-27 50 800 800 10 10 0.055 0.090 0.075
M-Mo-28 50 800 800 10 20 0.485 0.100 0.001 >7x1076¢

%A helium leak rate of <108 cm3/sec corresponds to a surface diffusion coefficient of about 10~ 2 cmz/sec.

b ingerprint.

“Would pump down sufficiently to leak check, but leak was too large to measure.

The experimental conditions and results are
shown in Table 18.3. The coatings reduced the
permeability of all samples except the third,
M-Mo-12, which was more permeable after coat-
ing. On this sample the coating outlined a
fingerprint very clearly, and this contaminant
probably contributed to the poor coating character-
istics. All subsequent samples were annealed for
% to 1 hr at 600°C in air before coating to remove
surface contaminants.

For all the remaining samples (M-Mo-13 through
M-Mo-28) the vacuum achieved on the coated
samples seems to be a function of both the precoat
vacuum and deposit thickness. Only five samples
had permeabilities low enough to be pumped to a
pressure less than 0.001 torr. Permeability within
the range of 7 x 107°% t0 1.5 x 10~ 1° cm3/sec can
be measured with the leak detector. In M-Mo-23
the measured helium permeability was 3.15 x 10—

cm®/sec. This required a deposit of 0.252 mil on
a sample that pumped to 0.075 torr originally. In
M-Mo-18 the permeability of the control sample
was too low to measure on the Veeco leak de-
tector (i.e., less than 1.5 x 1071% cm3/sec).
This sample had only 0.080 mil of molybdenum,
but the sample pumped to a vacuum of 0.001 torr
before coating. For M-Mo-25, M-Mo-26, and M-
Mo-28, the coated samples could be pumped to
0.001 torr, which is sufficiently low to leak test,
but they leaked too much to be measured (i.e.,
the permeability was greater than 7 x 10—9
cm3/sec).

The test results are plotted in Fig. 18.2 to de-
pict the pressure drop obtained before and after
coating as a function of deposition thickness.
This correlation indicates that thinner deposits
are required at 700 than at 800°C, particularly on
samples that will not originally pump below 0.080
0.6

| |
PRECOAT GRAPHITE HEAT TREATMENT
&4 700°C DEPQOSITION TEMPERATURE

B 1 l |

MINIMUM DEPQOSIT THICKNESS (mils)

o 0.02 6.04 .08

05 ™= "= e0g00°C DEPOSITION TEMPERATURE
CLOSED SYMBOLS: AS RECEIVED
04 OPEN SYMBOLS: { hr AT 600°C IN AIR

(PRECOAT VACUUM GIVEN IN PARENTHESES)

S S S

MAXIMUM THICKNESS OF COATING ALLOWA\BLE
BECAUSE OF NEUTRON ABSORPTION
-{0A15 !

194

ORNL-DWG 68-3986

o (0.075)

0.08 0.0 0.2 0.4
(PorG=PrnaL) REDUCTION PRESSURE (forr)

Fig. 18.2. Change in Pressure Drop Acrass o Graphite Specimen with Coating Thickness.

to 0.090 torr. On samples that will pump below
0.075 torr originally, both coatings are equally
sound, and less than 0.1 mil Mo is sufficient to
reduce the final vacuum to the limit of the Veeco
pressure gage. On samples that will not pump
below 0.095 torr, over 0.3 mil of molybdenum is
necessary for an 800°C deposit. A 700°C deposit
may be capable of sealing the samples with 0.1
mil of molybdenum if the sample will originally
pump below 0.110 torr. If the original samples
will not pump below this range, then our data
indicate that they cannot be sealed with a rea-
sonable amount of molybdenum. The validity of
the plot is being checked with 20 more samples
of the air-fired graphite, which are on hand.

When one compares M-Mo-18 with the other
samples, it is apparent that the original perme-
ability is very important. A recent study by
Cook (Sect. 18.2) has revealed that the 600°C
air fire can open large pores in the graphite
samples. These large pores may be more dif-
ficult to seal than the pores in the as-received
sample. The 600°C air fire will therefore be
discontinued in the hope that the argon-fired
samples will have better permeability prior to
coating and will thus require thinner coatings.
New coating experiments on AXF graphite with
an argon heat treatment will be compared with the
air-treated samples to test this hypothesis.

Examination of all the coatings revealed that
the metal was principally bridging the external
pores of the graphite, with only the most open

and shallow voids being filled. An attempt was
made to exert a vacuum on the inside of the
graphite cylinder during deposition in an effort
to pull the reacting gases into the graphite pores
and possibly achieve the desired reduction in
gas permeability with less molybdenum. This
technique is called the differential pressure CVD
coating process. The first attempts to do this
were not successful because we were not able
to obtain a vacuum seal around the graphite
during heating and therefore pulled only a par-
tial vacuum of 1 to 2 torrs inside the graphite
sleeve. However, as shown in Fig. 18.3, some
penetration was achieved. The Welding and
Brazing Laboratory will make a braze joint be-
tween the graphite cylinder and a molybdenum
tube, which should solve this problem. Thus we
will obtain a lower vacuum inside the graphite
cylinder and perhaps obtain deeper penetration
of coatings within the pores in subsequent coat-
ing experiments,

A series of 700°C argon anneals were performed
on selected samples of molybdenum-coated
graphite to evaluate the tendency of spalling dur-
ing thermal cycling and to see if the coating
totally converted to a carbide. The samples were
heated to 700°C for 17 hr, furnace cooled, and
then given three more 48-hr anneals at 700°C
with cooling to room temperature between anneals.
No tendency to spall was observed on any sample.
After the final anneal the samples examined still
had 10% molybdenum carbide or less, as esti-
mated using x-ray diffraction techniques.
Mo COATING |

GRAPHITE

195

PHOTO 91878

® 8

Fig. 18.3. Penetration of Molybdenum Deposit into Graphite, M-Mo-30. 500x.

18.6 GRAPHITE IRRADIATION PROGRAM
C. R. Kennedy

We have initiated graphite irradiation experi-
ments in the target rod positions in the core of
the HFIR. This facility is being used to obtain
high fluences in a relatively short time. The
first two experiments, described previously,’
have completed a single cycle in the HFIR and
have been examined. The purpose of these first
two experiments was primarily to confirm the
heating rates and the design of the experiment.

The design utilized tungsten susceptors to
compensate for the lower nuclear heating near
the ends and thus obtain an axial temperature
distribution of 690 to 730°C. The reactor itself
is quite stable, and temperature variations over
a cycle or from cycle to cycle should be less
than 10°C. The results of the first series show
that the peak nuclear heating was 33.6 w/g,
compared with an estimated 35 w/g. The axial

"MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 215.

gradient did not follow the estimated cosine
function but was slightly distorted, with a ratio
of end heating to peak heating of 0.56. Calcu-
lations based on flux monitors of stainless steel
doped with cobalt indicate a flux of 1.3 x 10°
neutrons cm~?2 sec™! (E > 50 kev).

The results of the first two experiments sug-
gested a few minor design changes. These
changes have been made, and two long-term ex-
periments have been constructed and inserted
in the HFIR. These experiments will be removed
from the reactor after five cycles with an expo-
sure of about 1 x 1022 neutrons/cm? (E > 50 kev).
The specimens will be evaluated by measurements
of physical dimensions, permeability, elastic
modulus, and other physical parameters. Selected
specimens will be subjected to low-angle x-ray
scattering and electron microscope studies. The
materials included in the two experiments are
listed in Table 18.4.

The program of evaluating candidate materials
is still in its beginning stages, and many of the
more promising materials have not been delivered.
This affords a chance to include grades AGOT,
Table 18.4. Materials Used in Graphite Irradiation Experiments

Source Grade Remarks

Poco AXF Base stock

Poco AXF-3000 Base stock heated to 3000°C

Poco AXF-50B Impregnated base stock

Poco AXF-50QB-3000 Impregnated base stock heated to 3000°C

Y-12 BY-12 85% graphite flour—15% fines—pitch, 3000°C
Y-12 RY-12-31 85% graphite flour—15% fines—Varcum, 3000°C
Y-12 RY-12-29 85% graphite flour—15% Thermax—Varcum, 3000°C
Y-12 Natural flake Natural flake—pitch, 3000°C

GLCC H315-A Pipe material similar to GA materials, isotropic
GLCC H364 Fine grained, high fracture elongation, isotropic
UcCcC 1425 Pipe material similar to grade CGB

UucCcC AT]J-S An improved AT]J grade

UCC AGOT Material used in PNL irradiations

UcCcC Hot-pressed pyrographite To be used for x-ray scattering experiments
UKAEA UK isotropic Material used in UK irradiations

ORNL Pyrocarbon-coated Poco Surface sealed with pyrocarbon

UK isotropic, and H315-A in the HFIR irradia-
tions for direct comparison with PNL, UK, and
GA irradiation results. Those grades demon-
strating poor irradiation resistance will be re-
moved and replaced during the recycle periods.
The Y-12 grades are a controlled fabrication
series to isolate material variables affecting the
growth rates and ultimate irradiation resistance.
These four graphites are unique in that although
several different starting materials were used, the
physical properties and their isotropics are about
the same. This will allow us to assess the effects
of starting materials with a minimal change in
other parameters. The Poco series is included to
gain a better understanding of the exceptional
dimensional stability demonstrated by this ma-
terial.® The coated materials were included to
obtain some indication of the ability of these
materials to adhere to a graphite body. The
coatings were not applied by an optimized process,
since our coating work is in its early stages.

8Y osh Kawa, BNWL, private communication.

18.7 NONDESTRUCTIVE TESTING STUDIES

H. L. Whaley R. W. McClung

18.7.1 Graphite Ultrasonic Velocity Measurements

H. L. Whaley

Progress has been made toward the measurement
of ultrasonic longitudinal (VL) and shear (VS) ve-
locities for determining the elastic properties of
proposed MSBR graphites. Longitudinal measure-
ments have been performed with both commercial
transducers with oil coupling and with thin ceramic
disks bonded to the samples with Salol (phenyl
salicylate). Shear crystals require a solid cou-
plant and have been bonded with Salol. One major
problem arises from the configuration of the speci-
mens (0.400-in.-diam cylinders, 0.500 in. long,
with a 0.120-in.-diam axial hole). The concentric
smooth surfaces result in mode conversion of the
ultrasound and spurious reflections that make
identification of the desired signals difficult.

To help overcome this situation, bulk pieces of
one kind of the graphite (Poco AXF) have been
obtained to establish approximate time intervals
required for the ultrasonic pulses to travel known
distances. Unfortunately, bulk pieces are not
available for all types of the graphite to be
measured.

To avoid activation in the reactor due to prior
contamination by a couplant, ultrasonic measure-
ments will be performed initially on pieces taken
adjacent to the irradiation samples and then, after
exposure, on the irradiation samples themselves.
About 25 of the cylinders (machined from Poco
AXF and Great Lakes H315-A graphite) have been
obtained for studying the uniformity of the mate-
rials and the reproducibility of bonding and mea-
suring techniques. The anisotropy of grade
H315-A was detected and may pose some problems.

18.7.2 Low-Voltage Radiography

R. W. McClung

Low-voltage radiographs were made for the Re-
actor Chemistry Division on several sections of
graphite which had undergone salt impregnation
testing in an irradiation field (Fig. 18.4). The
radiographs showed the presence of high-density

material (compared with graphite) in obvious cracks

and in other areas which may have been cracks
oriented so that a characteristic shape could not
be observed. The inspection was a further demon-
stration of the low-voltage radiographic technique:

1. to achieve high-sensitivity radiographs on thin

sections of low-density material,

197

R-40933

Fig. 18.4. Low-Voltage Radiograph of a Graphite
Slice Taken from ORR Natural Circulation Loop No. 2.
Uranium-bearing fluoride salt passed through the eight

holes. The dark areas indicate the presence of the

high-density salt. The diameter of the piece is 2 in.

2. to detect minute amounts of higher-density ma-
terial in the low-density matrix,
3. to perform these evaluations despite the speci-

men being radioactive due to irradiation testing.
19. Hastelloy N

19.1 INFLUENCE OF STRAIN RATE ON THE
FRACTURE STRAIN OF HASTELLOY N

H. E. McCoy, ]Jr.

Previous results have indicated that the fracture
strain of irradiated Hastelloy N is a function of
strain rate.!*? We have now obtained sufficient
test results to see the influence of this variable
more clearly.

The variation of fracture strain at the strain
rates normally encountered in tensile tests, 2 x
10~3 to 2 min~!, is shown in Fig. 19.1 for a
typical standard air-melted alloy in the unirra-

IMSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, p. 103.

2MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 200.

diated condition. In the unirradiated condition
the fracture strain at temperatures <500°C is

only weakly influenced by the strain rate. At
500°C the fracture strain decreases at very low
strain rates, due to the transition from transgran-
ular to intergranular fracture. At 650°C the frac-
ture strain is very dependent upon strain rate,
with the lower ductilities being characterized by
intergranular fracture. At 760 and 850°C the frac-
ture strains are again quite high, with only a slight
dependence on the strain rate.

The results from specimens that were irradiated
in the MSRE are shown in Fig. 19.2. At 400°C the
fracture strains are almost independent of strain
rate, although the strains are slightly lower than
those in Fig. 19.1 for unirradiated materials. At
500°C a distinct effect of strain rate is obvious,
the decreased ductility again being a result of the

ORNL-DWG 67-2455A

s

l l
Ry i A i i
§ 40 T‘ T ?|60°c pm—— T ‘ ’| ‘1
i
% 30 % Tl | | 1 i
o | T o Wl
L ’ A500°C ¢ 760°C J
0 850°C J
10 ' H ‘
LD T
0.004 00! 0 * 10

STRAIN RATE (min~1)

Fig. 19.1. Influence of Strain Rate on the Ductility of MSRE Surveillance Control Specimens. Heat 5081.

198
60

50

199

ORNL-DWG 67-3961R

TOTAL STRAIN (%)
o N
o o

n
o

0
0001 00!

ol} { 10

STRAIN RATE (min™")

Fig. 19.2.

Irradiated to thermal fluence of 1.3 x 1020 neutrons/cmz.

transition from transgranular to intergranular frac-
ture. However, the transition occurs at a higher
strain rate for the irradiated than for the unirra-
diated material. At 650°C and higher temperatures
the ductility decreases progressively as the test
temperature increases. The variation in strain
over the temperature range from 650 to 850°C be-
comes smaller as the strain rate decreases, thus
suggesting that the fracture strains at very low
strain rates (creep tests) may be independent of
test temperature (above 650°C). We shall later
present data that indicate that this may indeed be
the case.

Most nuclear applications expose the materials
to creep conditions rather than the high strain
rates encountered in tensile tests. The fracture
strains for several heats of itrradiated air-melted
material are given in Fig. 19.3. The data are rep-
resented by a scatter band, but there is an obvious
trend for the data from heat 5065 to fall on the

lower side and that for heat 5085 on the upper side.

However, the spread is much larger at high strain
rates. The fracture strain is a minimum at a
strain rate of about 0.1%/ht. The irradiation tem-
perature does not seem to be a significant factor
in the fracture strain at lower strain rates, but
can cause a factor of 2 variation at high strain
rates, the higher values being obtained for the
lower irradiation temperature.

Influence of Strain Rate on the Ductility of Hastelloy N (Heat 5081) MSRE Surveillance Specimens.

The fracture strains are shown in Fig. 19.4 for
several vacuum melts of standard Hastelloy N.
These materials seem to show a larger influence
of irradiation temperature, although the scatter
in the data prevents this from being an unequiv-
ocal conclusion. Generally, the material irra-
diated at <150°C has the higher ductility.
Specimens irradiated at 650°C have fracture
strains quite close to those shown in Fig. 19.3
for the air-melted heats. This observation is
quite interesting since the vacuum-melted alloys
have boron levels of 0.5 to 10 ppm, whereas the
boron levels of the air-melted materials are as
high as 50 ppm. Although there are many other
differences between air- and vacuum-melted ma-
terials, our tests do not indicate that reducing the
boron level alone is an effective solution to the
problem of irradiation damage in this material.

We have found that modifications in the chemical
composition of Hastelloy N are very effective in
reducing the radiation damage. The microstructure
is generally characterized by large stringers of
carbide. These carbides are of the M.C type and
are very high in molybdenum and contain from 2
to 5% Si. We found that the stringers were elimi-
nated if the molybdenum content were reduced to
12 to 13% rather than the 16% normally used. The
strength penalty for the reduction in molybdenum

content seems rather modest. However, this
200

ORNL-DWG 68-4199A

13—

o

I

l

/1
1 |———+ PRETEST ANNEAL—1 hr AT 1177:cr} {
REE: et SRR |
L AR
g | | / i
z _ 1y [
T AT
& | -
: ] .7 f
w 6 1| | . _} A1l # |
5 F————+H — / ; /V —1 — #L
| I/
‘ ' / / ; ’ =~ TENSIILE TESTS l
SR Vil
RN /1 fiili. _
SO
FENA Nl A i
T
NI i AR AR

0.001 0.04 od 1 10 100 1000
MINIMUM CREEP RATE (% /hr)

Fig. 19.3. Influence of Strain Rate on the Fracture Strain of Several Air-Melted Heats of Hastelloy N. Irradiated
to a thermal fluence of 2 to 8 x 1020 neutrons/cmZ and tested at 650°C. Solid points indicate specimens irradiated

at less than 150°C; open points indicate an irradiation temperature of 650°C.
p

change alone does not seem to improve the re- ductility is not as low for the titanium-modified
sistance to irradiation damage. material and rises very sharply with strain rate
The further modification of adding about 0.5 above or below the value of about 0.1%/hr as-
wt % Ti to the basic composition of Ni—7% Cr— sociated with the minimum ductility. The varia-
0.2% Mn—~0.05% C was found to give additional tion in the properties of the two titanium-modified
improvement. The creep-rupture properties of two heats is presently unexplained, although it is
small commercial melts are shown in Fig. 19.5. quite interesting that heat 21545, which has the
The fracture strains are shown in Fig. 19.6. higher boron level, has the better ductility.
Again, the ductility minimum as a function of The influence of test and irradiation tempera-

strain rate is apparent. However, the minimum tures is also shown in Fig. 19.6. The standard
201

ORNL—-DWG 68-4198A

° e 7] r282 7]
[||lw15.8 7es.3
| 14.5 8.7
14 ] : 14.3 | {]
HEAT I
3 a-5444 I
©—2477 |
o—7304 |
12 06252 5
v —65-552 |
. L iz
N I
PRETEST ANNEAL-1 hr AT 1477°C |
10 SOLID POINTS-IRRADIATED 150° C |
OPEN POINTS—IRRADIATED 500-650°C |
_ 9 | /
& I
z I
< 8 ) A
= ! /
) I/ /
g, / /
= I
g /
3 | /
“ 6 /[ I //
| o
il
5 /
7 TR
Wi
4 pa
/
~r_ / /| |
LLor T /1
&
v M TN / I TENSILE TESTS
~N
~ 4 g 'I"'I"/I | !
2 S F
. .“‘4\ (& // . !
, 3 % |
A P I
|
0 I
0.001 004 oY 1 10 100 1000

MINIMUM CREEP RATE (% /hr)

Fig. 19.4. Influence of Strain Rate on the Fracture Strain of Several Yacuum-Melted Heats of Hastelloy N.
lIrradiated to a thermal fluence of 2 to 8 X 1029 neutrons/cm? and tested at 650°C. Solid points indicate speci-

mens irradiated at less than 150°C; open points indicate an irradiation temperature of 650°C.

heats included are only the vacuum-melted ma-
terials. They were irradiated and tested at the
same temperature — either 650 or 760°C. Although
the fracture strains observed in a tensile test were
quite different, the strains under creep conditions
were similar for 650 and 760°C. The titanium-
modified specimens (21545 and 66-548) were all
irradiated at 650°C and then tested at 650 or’
760°C. Again, the fracture strain in a tensile test
varies with test temperature, but the strains under
creep conditions do not vary significantly with test
temperature.

19.2 STATUS OF DEVELOPMENT OF THE
MODIFIED ALLOY

H. E. McCoy, ]Jr.

Alloying additions to Hastelloy N of titanium,
zirconium, and hafnium have been found to improve
the resistance of the alloy to irradiation damage.?
Several small laboratory heats have been produced

3MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 217.
202

ORNL—DWG 68—4204A

70
Nereicas | 11
N
NTYPICAL
UNIRRADIATED
60
N
p \
A / B \L
- 50 9 \
a DN \
: W
S 40 %’*
= % ? N
0 < %
@ 30 HEAT v ;
= a 59U
—~ 4
2 o 2477 \
20 o 7304 0N
o 6252 ]
v 65-552
10 PRETEST ANNEAL—1hr AT #477°C
SOLID POINTS—IRRADIATED 150°C
OPEN POINTS— [RRADIATED 500-650°C
o RET Rl
10

01 4

100 41000 10,000

RUPTURE TIME (hr)

Fig. 19.5. Stress-Rupture Properties of Irradiated Titanium-Modified Hastelloy N. Irradiated and tested at

650°C. Thermal fluence was 2 to 5 X ]020 neutrons/cm2.

that contain up to 1% of each alloying addition.
Two 100-1b commercial melts containing 0.05 and
0.5% Hf have been procured and found to have
excellent properties. The zirconium addition re-
duced the fabricability of the alloy, so that the
first small commercial melts containing 0.5 wt %
Zr broke during fabrication at 1177°C. A subse-
quent melt was fabricated successfully by lower-
ing the working temperature to about 1038°C.
Thus, two 100-1b commercial melts containing 0.05
and 0.5% Zr were obtained. Both of these mate-
rials exhibited extensive hot cracking during
welding.* Twenty-five 100-1b commercial melts
have been procured with titanium levels up to
1.2% and carbon levels ranging from 0.003 to
0.1%. These alloys have fabricated well, have
exhibited good welding characteristics, and have
shown good pre- and postirradiation properties.
We have decided to continue further development
of only the titanium-modified alloy. This decision
was based on: (1) the poor weldability of the
zirconium-bearing alloys, (2) the excellent proper-
ties of the titanium-modified alloy, and (3) the
comparatively large amount of experience that we

4MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, p. 221.

have with titanium-bearing materials compared
with the hafnium-modified alloys.

The first large-scale melt has been delivered.
This 5000-1b melt was produced as a double-
vacuum-melted ingot and subjected to standard
mill practice. The yield will be about 2000 1b,
which is quite reasonable for the various shapes
of products that were made. The composition of
this melt is given below. This material will be
included in all phases of our testing program.

Chromium 7.84%
Molybdenum 12.36%
Tungsten 0.12%
Iron 0.25%
Carbon 0.066%
Silicon 0.02%
Cobalt 0.07%
Manganese 0.14%
Vanadium 0.04%
Phosphorus 0.002%
Sulfur 0.002%
Aluminum 0.12%
Titanium 0.56%
Copper 0.01%
Nitrogen 0.004%
Boron 0.2 ppm
203

ORNL-DWG 68-4203A

IRRADIATED TEST
TEMPERATURE  TEMPERATURE
HEAT (°C) (°C)
o 6252 650 650
® 6252 760 760
o 5914 650 650
® 5914 760 760
15 - 4 24545 500-650 650 :
1841 A 21545 500-650 760 019.8
. 1 o 66-548 500-650 650
@ 16.45 " 66-548 500-650 760
ANNEAL 1 hr AT 1477° C BEFORE TESTING {
4
13

g /
< A
= o
z 8 \21545
(- a
n \ /
B A \ L 1
=3 \ /
5 \§6-548 \ /
g:i N\
w . A \ / ]
o \ / {760 C
. o)
\ \ 650°c/ /
/ /
5
\\ \ , / / /6
4 \\&3 / / /
’ / /650°C
3 o =1 //
. STANDARD /
2 ™ /
> /
® \\ q g o 0
* N o [ ) g
TN 760°C 4~
1 \0\ = < ) - A
* \\. PuRR 4 /:___//
~—& ot
0
0.004 0.01 04 { 10 100

MINIMUM CREEP RATE (% / hr)

Fig. 19.6. Variation of Fracture Strain with Strain Rate for Standard and Titanium-Modified Hastelloy N.
20
0

Thermal fluence was 2 to 8 X 1 neufrons/cmz.
Further large melts are necessary before we or
the vendors will consider this alloy a standard
production item. The next large melt to be pro-
cured will have approximately the same composi-
tion as that shown previously, with relaxed
specifications on iron and manganese to allow
more liberal use of scrap by the vendor.

19.3. EFFECT OF CARBON AND TITANIUM
ON THE UNIRRADIATED CREEP-RUPTURE
PROPERTIES OF Ni-Mo-Cr ALLOYS

C. E. Sessions

As part of our efforts to optimize the mechanical

properties of titanium-modified Hastelloy N, we
are studying the effects of carbon concentration

Five small laboratory heats of modified Hastel-
loy N (Ni—12% Mo—7% Cr—0.5% Ti) were made with
carbon levels of 0.003, 0.007, 0.037, 0.053, and
0.27%. A sixth heat containing 0.04% C and no
titanium was also included. After fabricating into
l/4--in.~rod stock, miniature mechanical property
specimens were machined and given one of three
different heat treatments in argon. Samples were
then creep tested in air at 650°C at stresses of

47,000, 40,000, and 32,350 psi.

An evaluation of the creep rupture results indi-

cated the following:

1.

on the alloy. The interaction of carbon with strong

carbide formers such as molybdenum, chromium,
and, more significantly, titanium is expected to
strengthen the alloy through precipitation reac-
tions. In addition, however, we suspect that the
beneficial effects of small titanium additions on
the postirradiation ductility of these alloys is in
part due to titanium-(boron, carbon) interactions
which reduce the concentration of helium at the
grain boundaries following neutron irradiation.
Since boron and carbon should act similarly in
this alloy, we have attempted to define the ti-
tanium-carbon interactions in order to better de-
duce the role of titanium in reducing the irradia-
tion damage of nickel-base alloys.

The rupture life at a given stress increases
with increasing carbon content, as shown in
Fig. 19.7.

Heat treating for 1 hr at 1177°C prior to test-
ing resulted in longer rupture lives than the
other two heat treatments (i.e., 1 hr at 1260°C,
or 1 hr at 1177°C followed by aging 100 hr at
870°C).

The creep ductility at 40,000 psi showed little
variation (14—20%) for carbon levels up to
0.05%, but at the highest carbon level, 0.27%,
the creep elongation increased significantly to
30—-40% (Fig. 19.8).

The minimum creep rate at all stress levels de-
creased continuously with increasing carbon
content.

The addition of 0.5% Ti to thebasic Ni—12%
Mo—7% Cr—0.04% C alloy increased the rupture

ORNL-DWG 68-1993

60 —
48 (NOT) \l\ ‘ H
™~ "'\L.QS N
50 o \\ NN "
Y| Iy N N

% 194 ™ N \\
240 \Nfi> | \‘ TR o ™~
o 133 RS TR ISR
Q T N Nliss
o [ NS
= - '__‘._ N A O
® 30 ——
& HEAT CARBON CONTENT
'_
@ 20 |l Y193 0.003% B
o v 194 0007 %
v ® 195 0037 %
© A 196 0.05%

10 11 o0 197 027 %

A 198 004% (WITHOUT TITANIUM)
L
Sult I |
0.4 1 10 100 1000

RUPTURE LIFE (hr)

Fig. 19.7. Effect of Carbon on the Rupture Life of Ni-12% Mo~7% Cr-0.5% Ti Alloy at 650°C.
205

ORNL-DWG 68-6084

42 T T T 1117717 w HHHI
BASIC ALLOY: Ni-12% Mo-7% Cr )
NO Ti 0.5%Ti || |
% —— . a ANNEALED fhr AT #177°C ’ %
. ° ANNEALED ihr AT 1260°C ) .
/ }

W
O

n
D

CREEP ELONGATION (%)

N

a ANNEALED H77°C, AGED
100 hr AT 870°C

|

®

- |
el
1”2 I

AN
RN
AN

|

|

T

|

I“ |
ij__

0.001 0.002 0.005 001 0.02

0.05 041 0.2

CARBON CONTENT (wt %)

Fig. 19.8. Effect of Carbon and Titanium Content on the Total Creep Strain at 40,000 psi and 650°C.

life by a factor of from 3 to 10, depending on
the stress level.

6. The 0.5% Ti addition at the 0.04% C level did
not change the creep ductility of samples tested
in the solution-annealed condition; however, the
ductility of samples that were aged 100 hr at
870°C before testing was increased by a factor
of 2 at the lower stress level.

These observations indicate that there is a pro-
nounced strengthening effect due to titanium-carbon
interactions in this alloy; however, no significant
increase in creep ductility occurs at the 0.5% Ti
level when compared with samples containing no
titanium addition.

Changes in the microstructure with increases in
carbon concentration are shown in Fig. 19.9 for
samples solution annealed at 1177°C and aged
100 hr at 870°C. The lower carbon levels are
virtually single-phase alloys, but the amount of
second phase increases significantly with in-
creasing carbon. Heat 198, which contains 0.04% C
and no titanium, shows a large amount of finely
divided precipitates that are probably M +C type.
Each sample, except the 0.27% C level, was
single phase after a 1260°C heat treatment.

To summarize, the beneficial effect of titanium
additions on the out-of-reactor properties of this
alloy is primarily a strengthening advantage.
There appears to be no effect of titanium on the

creep ductility for samples tested in the solution-
annealed condition; however, a small increase in
ductility was found at lower stresses for samples
heat treated to precipitate the carbon prior to
testing. Carbon increases the creep strength
significantly, especially over the range from
0.007 to 0.037 wt %.

19.4 ELECTRICAL RESISTIVITY OF TITANIUM-
MODIFIED HASTELLOY N

H. E. McCoy, ]Jr.

The resistivity of Hastelloy N varies with tem-
perature in a somewhat anomalous fashion.® Upon
heating, the resistivity increases with temperature
up to about 600°C, decreases with increasing tem-
perature up to 1000°C, and then increases with
further increases in temperature. The resistivity
is decreased by cold working rather than in-
creased, as normally noted for metals. This be-
havior is generally attributed to short-range order,
with the ordering reaction causing an increase in
resistivity.

We have investigated the resistivity of the
titanium-modified alloy and found that the same

H. E. McCoy, Jr., Resistivity Anomaly in Nickel-
Base Alloys (in preparation).
0.003 % C 193 0.007%C

0.05%C

196 0.27%C

206

PHOTO 94879

194 0.037%C 195

197 0.04 %C—-NO Ti . 198

f=——0.005in—»]

[

500X 1

Fig. 19.9. Effect of Carbon on the Microstructure of Titanium-Modified Hastelloy N (Ni-12% Mo-7% Cr-0.5% Ti).
Samples solution annealed 1 hr at 1177°C and aged 100 hr at 870°C.

general behavior prevails. However, the resis-

tivity curve is lower by about 12% for the modified

alloy. The chemical changes involved for the

modified alloy include a decrease in the molybdenum

content from 16 to 12%, the addition of 0.5% Ti, a
decrease in the iron content from 4 to 0%, and a
general decrease in residual elements such as
silicon and manganese. These changes lead to a
decrease in the total alloy content and hence
would be expected to yield an alloy with lower
resistivity. A composite set of curves is shown
in Fig. 19.10 for the titanium-modified alloy.
Cold working decreases the resistivity by about
15%. After a complete heating and cooling cycle,
an ‘‘equilibrium’’ curve is established that is fol-
lowed during subsequent cycles. Annealing the
material for 1 hr at 1177°C reduces the resistivity
slightly, but the equilibrium curve is followed
after the first heating. The resistivity changes
over a normal operating range of 25 to 700°C are
quite small, only 4%.

19.5 ELECTRON MICROSCOPE STUDIES
OF HASTELLOY N

R. E. Gehlbach

19.5.1 Phase ldentification Studies in Hastelloy N

We are continuing our investigation into the char-
acterization of precipitated phases occurring in
Hastelloy N. As we reported previously® the mi-
crostructure of standard Hastelloy N is character-
ized by stringers of blocky M_C carbides after
annealing in the temperature range 1177 to 1260°C.
After aging and/or testing at temperatures between
427 and 871°C, a fine dendritic grain-boundary
precipitate is generated, which coarsens with in-
creasing aging temperature and with increasing
times at the lower temperatures.

6R. E. Gehlbach, MSR Program Semiann. Progr. Rept.
Aug. 31, 1967, ORNL-4191, pp. 219-21.

207

ORNL—DWG 68-—6085

] ]
=
2
(O]
w
2
16 — ——— e 3 } {
Z O
w o
-
oo
Wz
0 <
Ha —— uD> -
M2 l
Mo | —— ‘ —
o | | |
L9 . B
- N~ S IO I . ]
g 108 = g ' '7 ‘ |
| —
£ < 3 ‘ ‘
S = 9
4 < 5 (=) ’ |
> 106 |—— Yoo I e e
= ‘ a Z ) ‘ |
> w = ’ ‘
_ = (@)
= Z o O
%) < O =
@ 3
iy s bu
x 404 (— - -F & R R S
< <
i b
T <«
w
—~ T
0 |
i e !
102 b € — o — 4 — ]
i
100 S o I
Ti MODIFIED HASTELLOY N ‘
HEAT 21545 ‘ |
! i
|
98— —_— e Jy—
96 - —— — — — — — =
94 / :
0 100 200 300 400 500 600 700 800 900 1000 1100
TEMPERATURE (°C)

Fig. 19.10. Variation of Resistivity with Temperature for Titanium-Modified Hastelloy N.
208

Two tensile specimens were examined that had
been annealed at 1177 and 1260°C and tested at
082°C. They exhibited elongations of 58 and
4.6% respectively.” Transmission electron mi-
croscopy of electrochemically thinned slices from
the gage lengths showed grain boundaries nearly
completely free of precipitates and completely re-
crystallized for the specimens annealed at 1177°C,
whereas the grain boundaries of samples taken from
the specimen with the higher temperature anneal
contained a very extensive amount of precipitate
(Fig. 19.11). Extraction replication revealed a
cellular precipitate morphology (Fig. 19.12) in the
grain boundaries of the latter specimen and con-
firmed the relative absence of precipitation in the
grain boundaries of the former. The cellular
morphology has the same lattice parameter (11A)
as the blocky and dendritic forms of the M_C
carbides.

Chemical analysis of the various M _C-type car-
bides in standard Hastelloy N has been completed,
employing both conventional electron microprobe Fig. 19.12. Morphology of M,C-Type Grain-Boundary
techniques and also the electron microprobe ana- Precipitate in Fig. 19.115.
lyzer accessory on one of our electron microscopes.

With the latter instrument, we are able to analyze

} YE-93537

precipitates on extraction replicas, which permits
] the study of fine grain-boundary precipitates that .
H. E. McCoy, Influence of Several Metallurgical . &
Variables on the Tensile Properties of Hastelloy N, cannot be analnyad m.the bulk sample. The re
ORNL-3661, p. 8 (August 1964). sults are summarized in Table 19.1. Several

PHOTO 91880

Fig. 19.11. Effect of Annealing at (a) 1177°C and (b) 1260°C on Microstructure After Tensile Fracture at 982°C.

Fracture elongations were 58 and 4.6% respectively.
209

Table 19.1. Analysis of Precipitates in Standard Hostelloy N in %

Large Stringer

Major Nominal® Matrix? Precipitate® Bofijisy a4t High-Tempjrature

Element Phase?'#
b d,e Precipitate
Ni 72 74.5 33.8 37.4 15-18 64.6
Mo 16 12.1 55.8 49.2 7075 22.2
Cr 7 7.1 4.8 5.7 2-4 7.4
Fe 4 4.2 0.8 2.7
Si 0.6 0.3 2.7 3.7 2.5«4.5 1.1
C 0.06 21.7

#Nominal alloy composition from bulk chemical analysis.

bAnalysis employing conventional microprobe techniques on bulk specimen; normalized to 98%.

CM6C-type carbides.
d

Analysis employing microprobe analyzer accessory on electron microscope.

®Corrected for absorption, atomic number, fluorescence; normalized to 96%.

fThin precipitates, no corrections required; normalized to 96%.

€Noncarbide.

errors are inherent in the techniques involved, and
the results are only semiquantitative. The blocky,
or stringer, morphology approximately fits the
stoichiometric NisMOSC carbide, whereas the
grain-boundary morphology corresponds closely to
the Ni,Mo,C composition.

We have carried out additional work on Hastel-
loy N which contained '4C introduced into the
molten alloy.® Autoradiography performed on
these specimens indicates that the M _C-type pre-
cipitates actually are carbides, as can be seen
in Fig. 19.13. .

The high-temperature phase present after anneal-
ing the '*C at 1316°C and cooling quickly was re-
ported previously® as being a noncarbide phase,
probably resulting from a transformation from the
carbide precipitates. Electron microscope micro-
probe analyses performed on extractions of this
phase indicate that it is primarily molybdenum and
chromium in the ratio of 9:1 by weight. No other
elements of atomic number greater than 11 were
detected.

The carbides that are also present in this speci-
men have been identified by electron diffraction as

8H. E. McCoy, Studies of the Carbon Distribution in
Hastelloy N, ORNL-TM-1353, p. 2 (February 1966).

the Mo, C type, and analysis of individual par-
ticles and films indicates that the actual com-
position is close to (Mo €1y ,0,C.

It is possible that the high-temperature phase
in the 14C material is not the same phase as that
formed in the commercial heats, because conven-
tional microprobe analysis of the high-temperature
phase in standard Hastelloy N differs markedly
from the above results (see Table 19.1) and be-
cause Mo, C-type carbides have not been observed
in specimens from standard alloys which were an-
nealed at high temperature. We are now attempting
to identify the high-temperature phase after ex-
traction from standard Hastelloy N.

19.5.2 Effect of Silicon on Precipitation
in Hastelloy N

R. E. Gehlbach  H. E. McCoy, ]Jr.

In air-melted heats of Hastelloy N, the large car-
bides do not go into solid solution, even at high
annealing temperatures, but appear to melt and
transform to a lamellar phase between 1260 and
1316°C. However, in vacuum-melted heats the
precipitates do go into solution, and the high-
temperature phase is not formed (Fig. 19.14). The
PHOTO 91881

Fig. 19.13. Carbon-14 Specimen Annealed at 1177°C
and Aged 100 hr at 649°C, Showing Activity at Precipi-
tates and in Grain Boundaries. (a) Lightly etched. (b)

Specimen with NTB-2 autoradiographic emulsion, exposed
100 hr. 500x.

major compositional difference between the air-

and vacuum-melted alloys is the silicon concentra-

tion.?

To evaluate the effect of silicon on precipitation,

several alloys of nominal composition (16 Mo, 7

Cr, 4 Fe, etc.) with silicon concentrations of 0.02,
0.12, 0.48, and 1.05 wt % were prepared. The fol-
lowing observations were made regarding the ef-

fect of increasing silicon concentration on pre-
cipitation:

1. The amount of M C-type precipitates increases
noticeably; however, many are present in the

low-silicon heats after annealing at 1177°C.

2. The as-cast structure is retained somewhat in

the 0.48% Si heat, greatly in the 1% alloy.

°H. E. McCoy, Influence of Several Metallurgical

Variables on the Tensile Properties of Hastelloy N,
ORNL-3661 (August 1964).

3. The precipitates go into solution after anneal-

ing the low-silicon heats at temperatures of
1260 to 1371°C, whereas those in the high-
silicon heats melt and transform to the high-
temperature phase seen in Fig. 19.14.

The temperature for the melting and transforma-
tion was lower for the 1% than the 0.48% alloy.

The concentration of silicon in the precipi-
tates, determined by conventional microprobe
analysis, increases with increasing silicon in
the specimens and is generally two to five
times the matrix silicon concentration. The
matrix composition is approximately half the
nominal silicon concentration in the alloy.
These data are shown in Table 19.2.

We extracted precipitates from a specimen of 1%
Si which had been annealed at 1260°C, using a
direct carbon extraction replica (Fig. 19.15). The
AIR MELTED Ji
(0.6% Si)

VACUUM MELTED @
(0.015 % Si)

2150°F

211

PHOTO 914882

- L

4100 X

fe—————0.035 in

Fig. 19.14. High-Temperature Phase Transformation Occurring in Air-Melted Hastelloy N. Note that the M6C-

type carbides go into solution in the low-silicon vacuum-melted heat and do not melt or form the high-temperature

phase.

grain boundaries contain much precipitate in the
form of both particles and thin films associated

with these particles. In contrast, very little pre-
cipitate is found in grain boundaries of standard

Table 19.2. Semiquantitative Microprobe Analysis
of Silicon in Hastelloy N

Nominal Matrix Precipitates?
et (wt %) (wt %) (wt %)
42 0.02 0.02 0.03
46 0.12 0.1 0.6
47 0.48 0.3 5
48 1.05 0.5 1.7

@Precipitates are very small; silicon concentration is
actually greater due to matrix dilution resulting from
bulk metal surrounding particles.

Hastelloy N. Selected area electron diffraction
indicates that the film precipitate has essentially
the same lattice parameter as the M_C-type car-
bides found in the standard alloy. X-ray diffrac-
tion powder patterns confirm this observation, and
long exposures fail to show any diffraction lines
other than those associated with the M C-type
carbide. Microprobe analysis of these extracted
particles using the electron microscope micro-
probe accessory resulted in a composition similar
to the blocky carbides present in the commercial
material with the exception of silicon, which exists
in concentrations up to about 5%.

We know that the M_C-type precipitates in
standard Hastelloy N are rich in silicon (Sect.
19.5.1). It is quite obvious that silicon is playing
an important role both in the high-temperature non-
carbide phase and in the behavior of the M C-type
precipitates. We have not been able to correlate
the M_C-type particles with carbon, but their be-
212

Y-85738

Fig. 19.15. Direct Carbon Extraction Replica Showing Blocky Precipitates and Grain-Boundary Films Extracted

from Hastelloy N with 1.05% Si. Specimen was annealed at 1260°C before extraction.

havior is sensitive to silicon concentration in the in Fig. 19.16, as well as some larger precipitates,
alloy. Both the blocky and grain boundary pre- is actually of the Mo ,C type, the major phase
cipitates are rich in silicon and increasingly so present. The lattice constants are slightly con-

with increases in the silicon concentration of the
bulk material. The available information suggests
that the silicon may occupy carbon positions with
the lattice parameter not being changed signifi-
cantly. If silicon were to occupy all carbon sites,
about 5.2% Si could be accommodated. Microprobe
analysis of individual particles and precipitate
films has not suggested that the silicon concen-
tration exceeds this value.

19.5.3 Titanium-Modified Hastelloy N

We have identified fine grain-boundary pre-
cipitates which occur in titanium-modified Has-
telloy N after annealing at 1177 and 871°C. It
was previously reported!? that the thin-film
morphology was probably face-centered cubic
TiC and the particles a noncubic precipitate.
However, we have found by electron and x-ray
diffraction that the thin-film morphology shown

Fig. 19.16. Grain-Boundary Precipitate in Titanium-
Rept. Aug. 31, 1967, ORNL-4191, pp. 219-21. Modified Hastelloy N. See text for discussion.

s 8. Gehlbach, MSR Program Semiann. Progr.

tracted from those for pure Mo, C, and we have
found that the metallic portion of the carbides
contains approximately 10 wt % chromium.

Electron diffraction also shows that the small
particles associated with this film, as well as
the needle-like morphology, have a face-centered
cubic lattice parameter of 4.24 A, similar to the
values reported for TiN. X-ray diffraction shows
a slightly expanded (4.26—4.27 A) face-centered
cubic parameter, suggesting a complex TiC-type
phase which might contain nitrogen, oxygen, and
possibly another metallic atom besides titanium.
The amount of this phase is quite small compared
with the amount of Mo, C present.

19.5.4 Summary

These studies have shown that standard Hastel-
loy N annealed at 1177°C contains large M_C car-
bides that are rich in molybdenum. Aging causes
the precipitation of fine M_C particles along the
boundaries. All the M,C particles studied are en-
riched in silicon. A series of special alloys re-
vealed that the concentration of silicon in the
precipitate increased as the silicon content of the
alloy increased. This segregation of silicon to
these precipitates is probably responsible for the
localized melting that occurs when this alloy is
heated to about 1300°C. The low-temperature
M,C transforms to a lamellar product when the
alloy is heated to about 1260°C. Carbon-14
studies show that this product is not a carbide,
but an identification has not been made. The
titanium-modified Hastelloy N was found to have
small amounts of grain-boundary precipitates of
the MozC type and also a complex face-centered
cubic phase.

19.6 DIFFUSION OF TITANIUM IN MODIFIED
HASTELLOY N

C. E. Sessions  T. S. Lundy
Experiments and service in the MSRE have dem-
onstrated the excellent corrosion resistance of
Hastelloy N in fluoride salt systems. Our radia-
tion damage studies have shown that the resistance
of this alloy to radiation damage can be improved
markedly by the addition of about 0.5% Ti. Ti-
tanium forms a very stable fluoride, and we are

213

concerned with whether this element would in-
crease the corrosion rate. The corrosion of
Hastelloy N in fluoride salt systems is controlled
by the diffusion of chromium, and we have carried
out diffusion measurements to determine whether
titanium diffuses rapidly enough to contribute
significantly to the corrosion rate.

We have completed measurements of the dif-
fusion coefficient for the radioactive tracer 44Ti
in modified Hastelloy N over the temperature range
800 to 1250°C. The data obtained can be described
by an Arrhenius expression,

73,000 + 3000
RT

D = (15 = 2) exp cm? /sec , (1)

as illustrated in Fig. 19.17. Technique limitations
prevented meaningful experiments at lower tem-
peratures.

ORNL-DWG 68- 3734

7(°C)
_, 1200 1000 800 700 600 500
0" 17— 7T —
N
\ | | |
l l |

| | ~ |
161 \» — _D=(15.31 2.2) exp(~72,000% 3300/RT) - —
\(/ (cm?/sec)

DIFFUSION COEFFICIENT (cm?/sec)

10,000/ o)

Fig. 19.17. Temperature Dependence of the Diffusion
Coefficient of Titanium in Modified Hastelloy N,
The data were used to predict the maximum rate
of titanium diffusion out of Hastelloy N under
steady-state corrosion conditions. This occurs
for the case where the titanium concentration at
the surface in contact with the salt is maintained
constant at zero. The appropriate solution of
Fick’s second law of diffusion is then

X

cofx, t) = C0 erf , (2)

2Dt

where

c(x, t) = concentration at a distance x from the
surface,

t = time at temperature,
D = diffusion coefficient,

C0 = initial uniform concentration throughout
the specimen.

Using Eq. (2) and experimental values of the dif-
fusion coefficient D, we have calculated titanium
concentration distributions that would result after
various times at 700 and 800°C. At 700°C the ex-
trapolated value of D is approximately 5 x 10~18
cm?/sec. However, since this value is for volume
diffusion, one might argue that the effective dif-
fusivity, which includes the effect of short-circuit
diffusion paths, might be as much as an order of
magnitude larger than the extrapolated value. Thus
a value for D of 5 x 10~ '% cm?/sec has been used
in our calculations for 700°C. The measured value
at 800°C was 2 x 10~ 1% cm?/sec. The latter value
corresponds roughly to the effective chromium dif-
fusion rate in Hastelloy N at 650°C.!!

The depth of the diffusion zone from which ti-
tanium is depleted from the alloy increases with
both increasing time and increasing temperature.
Table 19.3 lists the depth at which the titanium
concentration is about 95% of its original value.
For the maximum probable coefficient at 700°C
(i.e., 5 x 10~ !%), the depleted zone after 30
years (™ 250,000 hr) extends only over a distance
approximately 60 u from the surface of the alloy.

The total amount of material M, which diffuses
from the alloy held under isothermal conditions
with a zero surface concentration is given by!?

llj. H. DeVan, Effect of Alloying Additions on Cor~
rosion Behavior of Nickel-Molybdenum Alloys in Fused
Fluoride Mixtures, M.S, thesis, The University of Ten-
nessee, August 1960.

Table 19.3. Depth at Which Titanium Concentration
Is Within Approximately 95% of Its Ilnitial Concentration

in the Alloy
D Time Depth
(cm2 /sec) (hr) (cm)
5% 10196 4,000 0.00025
8,000 0.00037
16,000 0.0005
250,000 0.002
5x10715 4,000 0.0007
8,000 0.0010
16,000 0.0015
250,000 0.0060
2 x 10”14 4,000 0.0015
8,000 0.002
16,000 0.003
250,000 0.0120
Di\!/?
M,=2C, (_._> : 3)
I

Using this equation and various values of D and
t, we calculated the expected buildup in solute
concentration in the salt of a typical system such
as the MSRE. (The total surface area was as-
sumed to be 852 ft?, and the total amount of
molten salt 4.92 x 10° g.) The results are shown
in Table 19.4. At 800°C the amount of titanium
removed would be approximately 169 g after 30
years, which is the equivalent of only 30 ppm in-
crease in the salt concentration; at 700°C the
comparable values would be 84 g of titanium and
17 ppm.

We conclude from these studies that the addi-
tion of titanium to Hastelloy N has not affected
the corrosion resistance of the alloy. Because
of the lower rate of titanium diffusion and the
lower concentration of titanium present, the se-
lective removal of chromium will continue to be
the primary corrosion process in this alloy in
molten fluoride salts.

12]. Crank, The Mathematics of Diffusion, pp. 11-13,
Clarendon, 1956.
215

Table 19.4. Calculated Titanium and Chromium Loss from Modified Hastelloy N&

Under Zero Surface Concentration Steady-State Corrosion Conditions

Hastelloy N surface area used was 852 ft2. Amount of fluoride salt considered at 4.92 X 10° g.

Salt

Alloy Temperature Diffusion Exposure Flux Out Total Amount Concentration
Constituent (°O) Coefficient Time (g/cm2) Removed (g) Increase

(cm2 /sec) (hr) (ppm)

Titanium 700 5x 1016 1,000 2.13 x 107° 1.7 0.3

4,000 4.26 x 107° 3.4 0.7

9,000 6.39 x 10~° 5.0 1.0

16,000 8.52 x 10~° 6.7 1.3

250,000 3.37 x 1077 26.5 5.1

5% 10715 1,000 6.74 x 107° 5.3 1.1

4,000 1.35 x 105 10.7 2.2

9,000 2.02 x 1073 15.8 3.2

16,000 2.69 x 1077 21.2 4.3

250,000 1.06 x 10~4 84.0 17.0

800 2 x10"14 1,000 1.35 x 1075 10.7 2.2

4,000 2.69 x 1077 15.8 3.2

9,000 4.04 x 1077 31.9 6.5

16,000 5.39 x 1073 42.6 8.7

250,000 2.13 x 1074 169 30.5

Chromium ~650 2 x 10714 1,000 1.89 x 10~* 149 30.0

4,000 3.77 x 1074 298 60.6

9,000 5.65 x 1074 446 90.8

16,000 7.55 x 1074 596 121.2

250,000 2.99 x 1073 2360 480
@ The chromium and titanium concentrations were 7 and 0.5 wt % respectively.

19.7 MEASUREMENT OF RESIDUAL STRESSES The technique used for machining the specimens
IN HASTELLOY N WELDS has been modified slightly from that previously re-

ported.!3 A lathe was substituted for the milling
machine previously used. The cutting tool is

fixed more rigidly, resulting in an improved cut
with more uniform quality and reduced work hard-
ening. Large volumes of cutting fluid are now

used for both lubricating and cooling. The tem-
perature of the welded specimen is monitored, and
a thermocouple placed near the strain gages showed

A. G. Cepolina  D. A. Canonico

We are continuing to study the effects of welding
conditions and postweld heat treatment on the dis-
tribution and level of the residual stresses in Has-
telloy N. The method of measuring the residual
stresses is a modification of the Boring-Sachs
technique,!® which permits the continuous in- a maximum temperature rise during machining of
vestigation of the distribution of the planar about 3°C. The normal lathe chucking technique
stresses in and near the weld. has been revised. The 12-in.-diam, 15-in.-thick

welded specimen is bolted to a Micarta bed.

13MSR Program Semiann. Progr. Rept. Aug. 31, 1967, The specimen is then held by clamping to the
ORNL-4191, pp. 223-26. plate. This system eliminates the stresses in-

duced in the plane of the specimen by the con-
ventional holding system. The machining op-
eration on the lathe requires that the strain

gages and their associated wires be handled in

a way that will prevent them from being damaged.
To accomplish this, the wires are fed through the
hollow lathe shaft and connected to silver-coated
plugs. During machining, these plugs are held in
a manner that permits their rotating with the
shaft. After a cut is completed, the lathe is
stopped and the plugs are connected to the strain-
measuring unit. After the reading is taken, the
plugs are disconnected, and the next cut is made.
The error due to contact resistance in the plugs
is negligible.

The tangential residual stress (with respect to
the weld axis) is a function of the change of the
tangential strain with the radius, that is, dET/dR,
where €,. is the gage reading in microinches per
inch and where R is the value of the radius of the
cut in inches. In order to evaluate deT/dR, it is
necessary to fit the experimental readings of
strain and radius with a continuous function. The
curve fitting and evaluation of the slope at various
values of R are done by a computer. The tangen-
tial stresses are evaluated as a function of loca-
tion across the weldment. The radial stresses are
determined mathematically from the longitudinal
strain data. These calculations have also been
computerized.

19.7.1 Experimental Results

Bead-on-plate welds (fusion pass with no addi-
tion of filler metal) have been examined. Two

6-in.-diam circular welds with a weld bead 1/4 in.

216

wide were made simultaneously on each face of a
1/2 -in.-thick Hastelloy N disk. Because of this
welding procedure, the stress distribution induced
in the plate is assumed to be planar.

Welds involving variations in shielding gas and
postweld heat treatment have been investigated.
Prior to welding, all plates were annealed for 1
hr at 1177°C in hydrogen. Table 19.5 lists the
variations studied to date.

The computer printout curves showing the dis-
tribution of radial and tangential stresses across
the welds of specimens 1, 5, and 6 are shown in
Fig. 19.18. From an analysis of the welds made
without postweld heat treatment, it appears that
the maximum tangential stress occurs about 3/1 6
in. from the weld axis toward the disk center.14
Using helium as the arc shielding gas, the peak
stress was found to be approximately 61,000 psi;
for argon shielding gas, the value was about
56,000 psi. It appears that the stress gradient
across the weld is steeper for argon than for
helium. This probably results from the hotter
and wider weld puddle which characterizes welds
made under helium.

Two postweld heat treatments were also con-
sidered (specimens 5 and 6). Sample 5 was given
a full anneal at 1177°C for 1 hr, while sample 6
was heated 6 hr at 871°C, a standard stress relief
for Hastelloy N. Both treatments reduced the re-
sidual welding stress to extremely low values.
Whereas the full anneal eliminated any peaking of

14 phis asymmetry might be explained by considering
the differential heat sink and restraint conditions that
exist between portions of the sample. This is attribut-
able to the difference in the amount of material present
(ratio of approx 1/3).

Table 19.5. Conditions Used for Making Weld Test Specimens

Heat Input
Specimen for Each of Two Shielding Postweld
No. Simultaneous Welds Gas Heat Treatment
(3/in.)

1 15,000 Argon None

5 15,000 Argon 6 hr at 871°C in H,
6 15,000 Argon 1 hr at 1177°C in H,
7 15,000 Helium None

CRNL-DWG 68-60886

SPECIMEN NOA
{AS WELDED)

SCALE OF 10,000 (psi)

SPECIMEN NO.6

RE | SoMR
STRESS RELIEVED o) SIGMA T

6 hr AT 1600°F 1
-1

1
SPECIMEN NO.5 o B~ : { : ~~ jSIGMAR
ANNEALED SIGMA T
thr AT 2450°F L
25 .

Fig. 19.18. Residual Stresses in Various Hasteliloy N

Weldments. Sigma R = radial stress, and sigma T =

tangential stress.

the stress, the 871°C treatment left a maximum of
about 5000 psi in the weld deposit. From these
results, it appears that the heat treatment at
871°C for 6 hr is indeed beneficial. The differ-
ence in stress distribution between argon and
helium welds is also noteworthy.

19.8 APPLICATION OF THE NARROW-GAP
WELDING PROCESS TO THE JOINING
OF HASTELLOY N

D. A. Canonico

The final report on the application of the Narrow-
Gap Welding Process (NGW)'® to the joining

217

of thick Hastelloy N plate has been received from
Battelle Memorial Institute (BMI). Battelle was
supplied with a 1-in.-thick plate of MSRE grade
Hastelloy N and a 25-1b coil of 0.125-in.-diam
Hastelloy N weld wire. Because of the needs of
the NGW process, it was necessary to reduce the
wire diameter to 0.045 in. Problems were en-
countered by BMI during processing, and only 7
Ib of usable wire was finally obtained.

The NGW process is a proprietary process de-
veloped by BMI. It is basically a metal-arc inert-
gas process wherein welding is done at the bottom
of a narrow gap formed by the root faces of the two
materials being joined. The process has the ad-
vantages that it requires no edge preparation and
the amount of filler metal needed to fill the joint
is minimal.

Operationally, the two plates to be joined are
placed about ’ in. apart. Two torches slightly
out of alignment (within the allowable 7, in.
dimension) are employed. The second torch trails
the first by about 2 in. The lay of the weld wires
is oriented so that each tends to deposit its bead
toward the wall nearest it. The process permits
low-heat-input welds, which, owing to the narrow
joint, results in the fabrication being completed
with a minimum number of passes.

Three welds were made during the course of the
investigation. Only the first weldment was made
under the normal twin-wire—narrow-gap technique.
This weldment was made using an argon-rich at-
mosphere (60 Ar—40 He). Center-line cracking
occurred in the trailing weld bead after about
half the plate thickness had been welded.

The second weld was made using a single-wire
technique and a helium-rich atmosphere (70 He—-30
Ar). Cracking again occurred after slightly less
than half the plate thickness had been welded.

The third weld was preceded by a bead-on-plate
study to further optimize the weld bead contour.

It was found that an 80 He—20 Ar atmosphere pro-
duced the best weld bead, one with a reduced
capillary and no long columnar grains. Special
precautions were taken with the filler wire prior
to welding, with the interpass temperature, and
with the condition of the surface. This weld was
also unsatisfactory, since it contained large blow-
holes and a number of cracks and fissures.

15R. P. Meister and D. C. Martin, Brit. Welding J. 13,
252 (May 1966).
This study concerning the applicability of the
NGW process was inconclusive. In all instances
the welds were unsatisfactory. However, a strong
possibility exists that the filler wire was not of
the best quality.

19.9 NATURAL CIRCULATION LOOPS
AND TEST CAPSULES

J. W. Koger A. P. Litman

Six loops and four capsules are presently in op-
eration. Tables 19.6 and 19.7 detail the service
parameters of these test units. Since last year we
have utilized a newly designed natural circulation
loop. The device is essentially a test bed system
wherein metal specimens suspended in the salt
and salt samples can be removed and/or replaced
at operating temperature without disturbing flow or
introducing air contamination. This design makes
it possible to obtain kinetic data on corrosion and
to study corrosion mechanisms prior to dismantling
and subjecting an entire loop to metallurgical
analysis. Figure 19.19 shows a schematic of the
new loop configuration, while Fig. 19.20 displays
two loops, NCL-15 and -16, in a protective hood
prior to operation.

We are continuing to concentrate on the com-
patibility of Hastelloy N with fuel, blanket, and
coolant salts. The compositions of standard Has-
telloy N and the titanium-modified Hastelloy N are
shown in Table 19.8.

19.9.1 Fuel Salts

Loop 1255, constructed of Hastelloy N and con-
taining a simulated MSRE fuel salt plus 1 mole %
ThF ,, continues to operate without difficulty after
almost six years.

Loop 1258, constructed of type 304L stainless
steel and containing the same salt as loop 1255,
has operated about 4.6 years with only minor
changes in flow characteristics. In January 1967,
ten new stainless steel specimens were placed in
the hot leg of loop 1258. The specimen in the
hottest position was replaced after 3700 hr. A
plot of the weight change of all the specimens as
a function of {ime and temperature is given in Fig.
19.21. The curves are rather typical of corrosion
processes that are diffusion controlled. As ex-

218

pected, the weight loss is highest at the hottest
point in the loop and decreases with decreasing
temperature. Salt analyses show that only chromium
is leached from the metal, and a diffusion coeffi-
cient can be calculated from Eq. (1) by assuming
that the rate-controlling process is the diffusion

of chromium to the solid surface:

2
AW =fi(c0 - C), VDT, 1)

where
AW = weight loss, g/cm?,

C o = bulk chromium concentration in material,
g/cm?,
C = surface chromium concentration in material,
g/cm?,
t = time, sec,

D = diffusion coefficient, cm?/sec.

Assuming a surface chromium concentration of
zero, the calculated values for D at 649 and
677°C were 4.0 x 10~12 and 1.4 x 10~ !! cm?/sec
(ref. 16). The diffusion coefficient for chromium
in austenitic stainless steel at 677°C was ex-
trapolated from existing data as 4 x 10~!3 cm?/sec.
Thus our corrosion data indicate an apparent dif-
fusion coefficient two orders of magnitude higher
than that extrapolated from diffusion measurements.

There are several possible reasons why these
values do not agree., (1) The measured values
were obtained over the temperature range from
849 to 1398°C and extrapolated to 677°C. Short-
circuit diffusion mechanisms have been shown to
be effective at lower temperatures and lead to a
deviation in the standard Arrhenius plot used for
extrapolation. (2) A corrosion mechanism more
complicated than leaching of chromium from stain-
less steel is involved. (3) The solid-state dif-
fusion of chromium in stainless steel is not the
rate-controlling step in the process. We are con-
tinuing to investigate these factors in an effort
to explain the differences above.

Loop NCL 16, the first two-fluid MSBR fuel salt
natural circulation loop incorporating the new loop
configuration, is now operating. The weight change
obtained after 250 hr on the titanium-modified Has-
telloy N specimen at 704°C was 0.35 mg/cm?.

16R. A. Wolfe and H. W. Paxton, Trans. Met. Soc.
AIME 230, 1426 (1964).
Table 19.6. MSRP

Natural Circulation Loop Operation Through February 29, 1968

Maximum

Time (hr)
Loop Loop Specimens Salt Salt Composition (mole %) Temperature OAT
No. Material Type (OC) e Scheduled Operated
1255 Hastelloy N Hastelloy N + 2% Nb&:P Fuel LiF—BeF2—ZrF4-UF4»ThF4 704 90 Indefinite 51,810
(70-23-5-1-1)
1258 Type 304L Type 304L stainless Fuel LiF-BeF2-ZrF4-UF4-ThF4 677 100 Indefinite 40,510
stainless steel?: € (70-23-5-1-1)
steel

NCL 13 Hastelloy N  Hastelloy Ne-d Coolant NaBF4-NaF (92-8) 607 150 5000 2,930

NCL 14  Hastelloy N Titanium-modified Coolant NaBF4-NaF (92-8) 607 150 Indefinite 2,840
Hastelloy N©&d

NCL 15 Hastelloy N  Titanium-modified Blanket LiF-BeF2-ThF‘4 (73-2-25) 677 55 Indefinite 530
Hastelloy N;
Hastelloy N controls®’9

NCL 16 Hastelloy N  Titanium-modified Fuel LiF-Ber-UF‘4 704 170 Indefinite 340
Hastelloy N; (65.5-34.0-0.5)
Hastelloy N controls®d

NCL 17 Hastelloy N Titanium-modified Coolant NaBF4-NaF (92-8) plus 607 150 Indefinite Estimated startup
Hastelloy N; water vapor additions 5-1-68
Hastelloy N controls * 9

NCL 18 Hastelloy N Titanium-modified Fuel LiF-BeF2-UF4 (65.5- 704 170 Indefinite Estimated startup
Hastelloy N; 34.0-0.5) plus FeF2 5-1-68
Hastelloy N controls®’ 9 additions

NCL 19 Hastelloy N  Titanium-modified Fuel LiF-BeF2-UF4 (65.5- 704 170 5000 Estimated startup

Hastelloy N;
Hastelloy N controls®’9

34.0-0.5) plus bismuth

plus lithium in molybdenum

hot finger

5-1-68

a .
Permanent specimens.

bHot leg only.

[+ .
Removable specimens.

9Hot and cold leg.

61¢C
Table 19.7. MSRP Capsule Program

Container Specimens Test Fluid Temperature Time (hr) Purpose
Material (mole %) (°C) Scheduled Operated®
Hastelloy N Hastelloy N in vapor, NaBF4-NaF (92-8) plus 607 2000 840 (No. 1) Support coolant salt project and
(four containers) liquid, and interface BFs at 120 mm Hg 822 (No. 2) determine effect of BF3 pres-
(No. 1), 50 psig (No. 815 (No. 3) sure on compatibility of so-
2), 100 psig (No. 3), 805 (No. 4) dium fluoroborate salts with
and 400 psig (No.4) Hastelloy N
TZM? TZM LiF-BeF2 (88-12) 1093 500 Estimated Support MSRP fuel processing
startup program; potential vacuum
4-1-68 still material

@Through February.
5M0—0.5% Ti—0.08% Z1—0.02% C.

0ZC
221

ORNL-DWG 68~ 3987 There is little doubt that an unsteady-state con-

dition still prevails in the system, and any data
obtained at this time must be considered tentative.

In support of the MSRP fuel processing scheme
using vacuum distillation, we have fabricated a
TZM alloy capsule that will contain LiF-BeF,
(88-12 mole %). This composition is typical of the
salt distilland expected during processing. The
test will be run at 1093°C, the approximate dis-
tillation temperature. This temperature precludes
use of nickel-base alloys, and thus the molyb-
denum-base alloy TZM has been selected as a
candidate retort material.

STANDPIPE

TWIN
BALL VALVES

19.9.2 Blanket Salts

Loop NCL 15 is now operating with a typical
two-fluid MSBR blanket salt. Considerable dif-
ficulty was experienced in filling this loop. The
salt seemed to have an exaggerated tendency
toward incongruent melting and segregation during
3 solidification. The weight change obtained after
i 400 hr on the titanium-modified Hastelloy N speci-
men at 677°C was 0.05 mg/cm?. This figure is
- cLAMSHELL lower than that for any other loop we have studied

HEATERS recently. It is again emphasized that these data
are preliminary, since the loop has not reached
a steady state.

v |

\

]
(S
I fi\

INSULATION w

CORROSION
SPECIMENS

19.9.3 Coolant Salts

Loops NCL 13 and NCL 14, containing NaBF -
NaF (92-8 mole %), continue to operate under
identical conditions. Loop NCL 13 contains
standard Hastelloy N specimens, while loop
NCL 14 has titanium-modified Hastelloy N
specimens suspended in the salt stream. These
differences make possible an excellent comparison
of the effect of fluoroborate salt on these alloys.
Utilizing weight changes on the specimens, changes
in fluoroborate salt chemistry, and assuming loop
behavior and specimen behavior are the same at a
given temperature, a mass balance of the system
was obtained within 10%:

r SAMPLER

INCHES

AW = AW + AC

system loss system gain

salt ’ (2)

where

Fig. 19.19. Schematic of New Natural Circulation Awsystem loss = weight loss for specimens and

Loop. loop components,
222

Fig. 19.20. Natural Circulation Loops NCL-15 and -16 Prior to Operation.
223

Table 19.8. Composition of Hasteiloy N Alloys
Chemical Content (wt %)
Alloy
Ni Mo Cr Fe Si Mn Ti

Standard

Hastelloy N 70 17.2 7.4 4.5 0.6 0.54 G.02
Titanium-modified

Hastelloy N 78 13.6 7.3 <0.1 <0.01 0.14 0.5

ORNL—DWG 68-6087

WEIGHT 0SS {mg/cm?)

bl

" B EQUIVALENT TO { mil/year — 1 —
O EQUIVALENT TO 1.5 mil/year

(

SPECIMEN TIME IN SYSTEM (hr)

Fig. 19.21. Specimen Weight Loss as Functions of Time and Temperature for Loop 1258 Heated Section. Loop

and specimens fabricated from type 304L stainiess steel.

AW

= weight gain for specimens and
components,

system gain

AC

salt = Chemistry change in salt.

A profile of the weight change of loop NCL 13 at
various times is given in Fig. 19.22. It can be
seen that there are two positions (balance points)
at 520 + 15°C which have not changed weight as
a function of time. One of these points is located
in the upper crossover section and the other in the
lower portion of the hot leg. Considering these

balance points, we estimate that about 55% of the
loop is gaining weight and the remainder is losing
weight. The change in chromium and iron content
in the salt with time is plotted in Fig. 19.23. At
the end of this reporting period, 2200 hr exposure
time, about 1500 mg of material has been lost from
the metal. Approximately half of this stayed in
the salt, and the other half deposited in the cold
portions of the loop.

We believe the container material, standard
Hastelloy N, in loop NCL 14 is behaving like
ORNL~-DWG 68-3385

o S538hr
& 867 hr
A {462 hr
a 2249 hr

—— e s g

WEIGHT CHANGE (mg/cm?)

TEMPERATURE (°C)

80 20 100
DISTANCE (in.)
UPPER CROSS OVER COLD LEG BEND LOWER BEND HOT LEG
(VERTICAL) CROSS OVER (VERTICAL)

Fig. 19.22. Weight Change as Function of Position (Temperature) for Standard Hastelloy N Specimens in Loop NCL 13.

loop NCL 13 tubing despite the results on speci-
mens from loop NCL 14, which are titanium-
modified Hastelloy N. A weight change vs
distance (temperature) profile for the specimens
in loop NCL 14, Fig. 19.24, illustrates the
lower weight changes for the modified alloy com-
pared with the standard Hastelloy. The value
of AC__,, calculated using Eq. (2) was 150 ppm
Cr, in excellent agreement with experiment. This
suggests that a loop fabricated from modified
Hastelloy N should behave like our test specimens
of the same material. Proof of this thesis will be
studied as modified material becomes available
for construction of a loop.

Weight change—diffusion (rate constant) calcu-
lations similar to those done on other loops were
performed on loops NCL 13 and NCL 14 using Eq.

3):

2
AW, - c Dt

loss \/;
= C1 vDt,

T>520 £ 15°C, (3)

where

C = concentration of Fe + Cr in Hastelloy N,

g/cm3,

D = diffusion coefficient, cm?/sec,
t = time, sec.

It was found that by using the weight changes
and the combined iron and chromium concentration
for the loop specimens, almost identical D values
were obtained for the standard and modified Has-
telloy N. This is quite significant, since the
iron content of titanium-modified Hastelloy N
is negligible. Thus D, or in reality a rate con-
stant K, is independent of the concentration of
chromium and iron and other diffusing species.
The weight change, however, is quite sensitive
to the total concentration of the migrating ele-
ments. Confirmation of this hypothesis is evi-
dent when comparing K, from our loops with a
value obtained for D in Hastelloy N;!7 K, was
found to be ='10% exp (~57,500/RT), and specifi-
cally K, for loops NCL 13 and 14 at 607°C is

171, H. DeVan, Effect of Alloying Additions on Cor-
rosion Behavior of Nickel-Molybdenum Alloys in Fused
Fluoride Mixtures, ORNL-TM-2021 (to be published).
225

ORNL-DWG 6€8-6088

400

350

300

250

200

CONCENTRATION (ppm)

150

100

50

TIME (hr)

Fig. 19.23. Iron and Chromium Levels in the NaBF ,-
NaF (92-8 Mole %) Salt in Loop NCL 13.

5 x 10— 12 cm?/sec, while the literature value for
D_ is 5x 10~ !4 cm?/sec.

We believe the sodium fluoroborate salts may
be less selective in their dissolution tendency
of the major elements in nickel- and iron-base
alloys. However, it is emphasized that the cor-
rosion rates of Hastelloy N in fluoroborate salts
are relatively small and now appear to be leveling
off. This behavior increases our confidence in the
compatibility of Hastelloy N in candidate salts for
molten-salt reactors.

Capsule tests to evaluate the compatibility of
Hastelloy N with fluoroborate salts at 607°C as a
function of BF, pressure have operated 800 hr
through February. There are four capsules with
three titanium-modified Hastelloy N specimens in
each; one specimen is in the salt, one in the BF
vapor, and one in the liquid-vapor interface. The
BF3 pressures used are approximately 130 torrs
(the normal BF, vapor pressure at 607°C), 50, 100,
and 400 psig. The time, originally scheduled for
1000 hr, has been extended to 2000 hr to permit
gathering of more conclusive results. We have ob-
served that the specimens in the vapor phase of
loops NCL 13 and NCL 14 continue to lose weight
and retain a blackened appearance. These capsule
tests should shed light on this phenomenon.

ORNL - DWG 68-6089

3 | ] l 1 T ]
o 513 hr J L { ™ HEATED AND INSULATED |
o 827 h
24 taz2 hrt——— R N A R —
A 2191 hr ‘ ‘ f l ’
Sy A T S S S
S e
g i\?!"’ e,
A7 'Y
e | | T *~<;| 659
NS,
I | o
= | | | | | | | \A 600 €
o | | I P | Y E
R R R R S S e R
\ l | | ]
| , ‘ | 7 _J‘__ _____ - o
| | 1 5 J =
-4 JI J L I L | l J
0 10 20 30 40 50 60 70 80 90 100
DISTANCE (in.)
R | U [ -]
UPPER COLD LEG BEND LOWER BEND HOT LEG l
CROSS OVER (VERTICAL) CROSS OVER (VERTICAL)

Fig. 19.24. Weight Change as Function of Position (Temperature) for Titanium-Modified Hastelloy N in Loop NCL 14,
19.10 FORCED CIRCULATION LOOP

P. A. Gnadt W. R. Huntley

Work is in progress on a loop facility that will
enable us to study the compatibility of alloys and
salts under forced-circulation conditions of interest
to the MSRP. At present, we are constructing a
test facility for studying the compatibility of Has-
telloy N with NaBF -NaF (92-8 mole %). The loop,
MSR-FCL-1, will operate at a bulk maximum tem-
perature of 607°C with a temperature difference of
150°C. This salt is very attractive as a coolant
because of its low cost and low melting point. If
found to be compatible and satisfactory in other
respects, it will be used in the MSRE in place of
the present coolant, 'LiF-BeF, (66-34 mole %).
The loop and facility designs from earlier forced

226

circulation loops were used, with minor modi-
fications, in an attempt to reduce costs.

The loop MSR-FCL-1 has been designed to ap-
proximate conditions in the MSRE coolant circuit.
Liquid velocities will be limited to 7 fps because
an available model LFB pump is being used. Other
salient features of the loop include a BF . purging
and addition system over the free liquid surface in
the pump tank, means for sampling molten salt dur-
ing operation, and permanent metallurgical speci-
mens installed in the bulk stream. The salt in the
loop will be heated by electrical resistance heat-
ing of the tubing wall and will be cooled by a
finned air-cooled heat exchanger. Special con-
trols are provided to reduce the possibilities of
freezing the salt in the loop piping during elec-
trical power interruptions or other equipment
failures. The loop is shown in simplified sche-

ORNL-DWG 68-2797

SALT
SAMPLE
LINE AJUSTO SPEDE
5 hp MOTOR
———— e —————
|
|
l | OIL LINES
—— TO PUMP
GAS - -—
LINES
TO — 850 °F
PUMP
METALLURGICAL
SPECIMEN HEATER
LUG
LFB PUMP RESISTANCE HEATED SECTION
3 gpm
850 °F 1120 °F
HEATER METALLURGICAL
LUG SPECIMEN
FINNED
COOLER
BLOWER
METALLURGICAL
SPECIMEN HEATER LUG
r—{/ 1
/‘-‘ RESISTANCE HEATED SECT|0N/f
FREEZE 850 °F HEATER LUG

VALVE
_—“\i:::¥b~t—AlR
‘ SumP l
C

OPERATING PRESSURE
18 psia
REYNOLD'S NUMBER
~3400
VELOCITY ~7 f1/sec

Fig. 19.25. Schematic of Molten-Salt Forced-Convection Corrosion Loop.
227

ORNL-DWG 68-2793

FIRST SECOND FINNED
HEATER HEATER COOLING COIL
A
@T T T
| S |
: | (M) METAL COUPON LOCATION
| — LIQUID TEMPERATURE
| | —— WALL TEMPERATURE
| | |
| { | |
1400 | l % |
r ) | /F
1300 — | | -
s
SR
1200 — | ; I
[T
o |
; 1100 :f/ : } | |
x | | | / | |
£ 1000 1 - |
5 2 HEERN
B_J 900 | / | | II SN
= N | | \fi'
800 —i , | ik_____""_
|
700 ; : | ;
10 20 30 40 50
LENGTH (1)

— @
PUMP

60

Fig. 19.26. Temperature Profile of Molten-Salt Forced-Convection Corrosion Loop.

matic form in Fig. 19.25. Figure 19.26 shows the
expected temperature profile of the liquid and the
metal wall. Selected engineering data for the
loop design are given in Table 19.9. The data
given are the design conditions, but uncertainties
in the physical properties of the salt may cause
the actual operating conditions to be different.

Table 19.9. Engineering Data for Loop MSR-FCL-1

Table 19.9 (Cont.)

Heat transfer

Heat load at finned cooler

Liquid Reynolds number

Liquid film heat transfer
coefficient

Length of finned 1/2-in.-OD

cooler coil

Materials, temperature, velocities

Tubing and specimens

Nominal tubing size

Total tubing length

Bulk fluid temperature
(max)

Bulk fluid AT

Flow rate

Liquid velocity

Standard Hastelloy N
1/2 in. OD by 0.045-in.

wall

Coolant air flow
Coolant air AT
Pumping requirements

System AP at 3 gpm

54 ft Required pump speed
o

607°C Circulating salt volume
o Tubing

150°C Pump bowl

3 gpm

Approximately 7 1/4 fps

Total volume

322,000 Btu/hr (approx
94 kw)

Approximately 3450

Approximately 775 Btu
hrt o2 (OF) !

26 ft

3000 cfm
50°C

53 psi (60 ft)
4700 rpm

85 in.3
47 in.3
132 in.3

Approximately 98% of the design modifications
have been completed during the reporting period.
Ninety-five percent of the loop components have
been procured and fabricated. Assembly of com-
ponents and installation of electrical equipment
are in progress.

19.11 OXIDATION OF HASTELLOY N

B. McNabb, Jr.

Although Hastelloy N has generally performed
quite well, its susceptibility to radiation dam-
age! 819 has caused us to consider small modi-
fications of its chemical composition. The oxida-

228

tion resistance of this alloy is quite good,%° and
we have been concerned with whether these chem-
ical modifications would reduce the oxidation
resistance.

The rate of oxidation at a given peak tempera-
ture is usually higher when the material is being
cycled periodically to lower temperatures. Al-
though a very protective and tenacious oxide may

!8y. R. Martin and J. R. Weir, Nucl. Appl. 1(2), 160—67
(1965).

19%. R. Martin and J. R. Weir, Nucl. Appl. 3, 167
(1967).

20y, E. McCoy, Jr., and J. R. Weir, Jr., Materials De-
velopment for Molten-Salt Breeder Reactors, ORNL-TM-
1854 (June 1967), pp. 31-34.

Table 19.10. Compositions of Test Alloys

Chemical Content (wt %)

Heat
No. Mo Cr Fe Mn C Si Al Ti Other
2477 16.32 7.05 4.25 0.04 0.057 0.015 0.055
X2497 17.1 7.15 <0.02 0.048 0.016
5065 16.48 7.25 3.90 0.55 0.065 0.595 0.01
5067 17.30 7.40 4.0 0.48 0.06 0.43 0.01
5320 16.85 6.92 0.05 0.016 0.056 <0.006
5911 17.01 6.14 0.03 0.21 0.056 0.05 0.15
7304 16.1 7.16 2.36 0.92 0.104 0.037 0.26
7305 16.3 7.27 2.40 0.90 0.059 0.21 0.41
Y8487 16.78 7.32 4.11 0.30 0.05 0.17 0.16
6252-2 16.53 7.26 0.12 0.20 0.051 0.05 0.20
65552 15.85 7.43 4.45 0.43 0.045 0.15 0.25
61 12.14 7.77 0.21 0.048 0.01
71 11.69 7.93 0.23 0.059 <0.005 0.05
72 11.74 7.90 0.23 0.057 <0.005 0.08
73 11.87 7.90 0.27 0.058 0.005 0.15
74 11.81 7.94 0.25 0.062 <0.005 0.43
75 11.89 7.89 0.24 0.062 <0.005 0.95
76 11.93 7.84 0.28 0.127 0.01 1.0
102 11.2 6.92 0.089 0.20 0.045 <0.005 0.39
104 11.74 7.14 0.10 0.21 0.05 <0.005 0.55
105 12.28 6.72 0.089 0.20 0.049 <0.005 0.66
66535 12.66 7.2 0.07 0.12 0.072 0.03 0.06 0.15
66536 12.36 6.9 0.07 0.10 0.028 0.025 0.08 0.32
66541 13.17 6.81 0.03 0.066 0.05 <0.03 0.27
66548 12.44 7.66 0.03 0.039 0.05 <0.03 0.45
67502 12.74 7.24 0.08 0.15 0.04 <0.01 0.12 0.53 2.15w
67503 12.54 7.75 0.15 0.08 0.06 <0.01 0.12 0.06 0.08 Hf
67504 12.68 7.49 0.12 0.12 0.07 0.02 0.03 <0.01 0.43 Hf

229

be formed at the peak temperature, this oxide may alloys were made by air and vacuum melting and
spall during cooling due to a difference between contain various amounts of Fe, Al, Mn, and Si.
the thermal expansion of the oxide and the metal However, the variation in silicon appears to be
substrate. When reheated to the peak temperature, most important, with the weight loss at 982°C
a period of rapid oxidation occurs until another varving by a factor of about 6 over the range of
protective layer of oxide is formed. We used a 0 to 0.6 wt %. At 760°C a similar trend is
cycle period of 25 hr in our tests (i.e., 25 hr at noted, although about 0.05% Si seems sufficient
temperature), after which the specimens were re- to improve the oxidation resistance appreciably.
moved from the furnace, allowed to cool to room Several 2-1b laboratory melts of the modified base
temperature, weighed, and reinserted in the fur- composition (12% Mo~7% Cr—0.2% Mn~0.05% C)
nace. The test times were generally 1000 hr, with were prepared with various amounts of titanium.
the samples being weighed periodically. Test The results of tests on this material are shown in
temperatures of 760 and 982°C were studied. The Fig. 19.28. These melts are all quite low in sili-
test specimens were Y, in. in diameter by 7, in. con, and the weight losses are quite similar to
long; the materials studied are given in Table those shown in Fig. 19.27 for the low-silicon
19.10. alloys. The titanium additions seem to decrease
The results from several melts of standard the weight loss slightly, the effect being greater
Hastelloy N are shown in Fig. 19.27. These at 760°C.

Several commercial heats of titanium-modified
materials have been tested, and the results are
shown in Fig. 19.29. With the exception of the
one point at 760°C, the oxidation rates are rea-
sonably independent of titanium level. The rates

ORNL - DWG 68— 3990

600

ORNL—DWG 68—3991

500

n
o
(@]

W
o
o

WEIGHT LOSS IN 1000 hr (mg/cm?)

200

WEIGHT LOSS IN 4000 hr (mg/cm?)

100

0 0.2 0.4 0.6 0.8 1.0
PERCENT Ti

PERCENT SILICON

Fig. 19.28. Effects of Titanium Additions on the

Fig. 19.27. Effects of Silicon on the Scaling Re- Scaling Resistance of Laboratory Melts of Ni-12%
sistance of Hastelloy N. Mo-7% Cr-0.2% Mn-0.05%C Alloys.
are slightly lower than those shown in Fig. 19.28,
probably because of the higher concentration of
beneficial impurities in the commercial alloys.
These results indicate that the removal of
silicon and other residuals from the alloy is more
detrimental to the oxidation resistance than the
addition of small amounts of titanium. The weight
losses experienced at 760°C are still relatively
low for such stringent conditions (e.g., a weight
loss of 100 mg/in.? in 1000 hr corresponds to the
loss of about 50 mils of metal in 10,000 hr). Iso-
thermal oxidation tests are being run on these ma-
terials, and the rates are much lower; however, the
trends noted here under cyclic conditions seem to

apply.

230

ORNL—-DWG 68—6090

700

600 ss.| 982°C
« 500 S
5
~
o
E
£ |
g 400
(@]
<
w
g 300t 1 —
-
—
T
©
ES

200 +— 0_

\K
L ]
760°C
./
100 .\ |
0 0.2 0.4 06 08 4
7, Ti

Fig. 19.29. Variation of Scaling Resistance with Ti-

tanium Content for Several Commercial Melts of Ni—-12%

Mo—-7% Cr-0.2% Mn-0.05% C Alloys.
20. Graphite-to-Metal Joining

20.1 GRAPHITE BRAZING DEVELOPMENT hancing brazability. Thus, more conventional
ductile, corrosion-resistant brazing alloys such
Holl. Wetiet as pure copper and 60 Pd—40 Ni (wt %) could be
In addition to developing brazing filler metals used. Both molybdenum and tungsten are being
which directly wet and flow on graphite,! we are deposited on high-density, low-permeability
investigating the use of thin (0.0002 to 0.0005 in.) graphite, and we have found that brazing filler
pyrolytically deposited metallic coatings for en- metals readily wet and flow on the coated graph-

ite. Figure 20.1 shows the morphology of a thin

1MSR Program Semiann. Progr. Rept. Aug. 31, 1967, Coatlr?g of pyrolyt1c rflolybdenum o M.SRE-g.rade
ORNL-4191, p. 236. graphite. X-ray studies showed that immediately

Y-83253

P e

1000X

0.0035 INCHES

o

Jou

Fig. 20.1. Microstructure and Morphology of 0.0002-in. Coating of Pyrolytic Molybdenum on CGB Graphite. As
polished.

231
232

adjacent to the graphite the molybdenum had re-
acted with the graphite to form molybdenum car-
bide. Also, the morphology of the coating sug-
gests a high degree of mechanical bonding be-
tween the materials.

The test joint shown schematically in Fig. 20.2
was devised to check the strength of graphite—
Hastelloy N joints made by the above method.

The graphite used in these tests was high-density,
low-permeability isotropic material made by Poco. ?
The joints were brazed with copper and pulled at
room temperature. Figure 20.3 shows one of the
broken test specimens. The position of fracture
varied from specimen to specimen. This was
taken as an indication of the extent of cracking
due to differential thermal expansion in the plane
of the taper. In addition, one specimen fractured
at the metal-graphite junction during the brazing
cycle. Thus more severe stresses than antici-
pated were encountered with this particular test
specimen design. Currently, a series of coated
graphite-to-molybdenum brazed specimens is

being prepared.

Experimental results on full-size joints (31/2 in.
oD, 1/2-in. wall) are preliminary. Most of the
tests were direct joints of Hastelloy N to pyrolytic-
coated graphite using copper as the brazing filler
metal. Both induction and furnace brazing tech-
niques were used. The induction-brazing tech-
nique is quite attractive in view of the large
number of joints needed and the length and uniform
configuration of the reactor components. Induction-
brazing experiments have resulted in fracture of
the graphite at the graphite—Hastelloy N junction

’Manufactured by Poco Graphite, Inc., Garland, Tex.

ORNL-DWG 68-6091

%/

227727

v 2

[/

GRAPHITE

HASTELLOY N OR
MOLYBDENUM

Fig. 20.2. Schematic lllustration of Graphite—Hastelloy
N or Graphite-Molybdenum Test Joint.

PHOTO 89770

Fig. 20.3. Graphite—Hastelloy N Brazed Specimen

Tested at Room Temperature. The location of the
fracture was approximately in the middle of the taper.

Joint design shown in Fig. 20.2.

during a relatively fast cooldown from the brazing
temperature. Uniform heating has been difficult
to obtain, and, as a result, we found portions of
these joints containing unmelted brazing filler
metal. We are presently upgrading the induction-
brazing unit to minimize these problems. Several
direct joints of various designs have also been
furnace brazed. Again the direct joints between
graphite and Hastelloy N cracked in all cases,
although not to the extent that the induction-brazed
joints did. This experience indicates the poten-
tial advantages of the low-expansion transition
metal concept. For example, a brazed graphite-
to-molybdenum joint showed no visual or radio-
graphic evidence of cracking. Radiographically,
good flow of the brazing filler metal was noted
throughout the joint area. Figure 20.4 shows

the microstructure of such a joint (note the ex-
cellent flow of the copper filler metal). Next, a
piece of Hastelloy N will be brazed to the molyb-
denum to form the complete transition joint, and
then the entire joint will be thermally cycled.
233

GRAPHITE

0.018 INCI'ES

Fig. 20.4. Microstructure of Pyrolytic-Molybdenum-Coated ATJ Graphite Brazed to Molybdenum Transition Piece
with Copper. Etchant: H202 and NH4OH.
20.2 RADIATION STABILITY OF BRAZING
ALLOYS OF INTEREST FOR
BRAZING GRAPHITE

W. J. Wemer H. E. McCoy, Jr.

We have irradiated a series of brazing filler

metals which find application for joining Hastelloy

N and/or molybdenum to high-density graphite.
Hastelloy N test specimens of the Miller-Peaslee
design? were used in the study. The specimens
were 0,125 in. thick x 0.375 in. x 1.875 in. The
joint spacing before brazing was maintained at
0.002 in.

The specimens were irradiated in a single-test
capsule in the poolside position in the ORR. The
peak thermal flux was 6 x 10?3 neutrons cm™?
sec™!, and the peak fast (>2.9 Mev) flux was
5 x 1012 neutrons cm™?2 sec™!. The duration of
the experiment was 1128 hr (time at temperature
and full power), so the thermal and fast doses
were 2.43 x 1020 and 2.03 x 10'? neutrons/cm?
respectively. The compositions of the brazing
filler metals used in the study are shown in
Table 20.1.

The pure copper and nickel-palladium-chromium
brazes were made in a hydrogen atmosphere at
1125 and 1250°C respectively. The copper-base
filler alloys were brazed in vacuum at 1200°C.

After irradiation the specimens were tested on
an Instron tensile machine. As-brazed control
specimens were also tested. In addition, a set
of specimens having the same thermal and en-

3F. M. Miller and R. L. Peaslee, Welding J. Res.
Suppl. 34(4), 144s~150s (1958).

234

vironmental history as the reactor specimens is
being prepared.

Strength data for the reactor and control speci-
mens are shown in Table 20.2. It can be seen
that all the brazing filler metals seem to have
adequate mechanical strength for nuclear system
joints. We cannot obtain quantitative values for
the shear strains until the width of the sheared
region is determined. Complete analysis of the
data must await the test results from the speci-
mens having the same themal history as the
irradiated specimens.

Table 20.2. Properties of Several Brazing Alloys

Maximum Shear
Strength (ksi)

Material
25°C 700°C

Copper

Unirradiated 29 9

Irradiated 15 7
35 Ni~60 Pd~-5 Cr

Unirradiated 66 28

Irradiated 66 12.5
68 Cu~17 Ni~10 Cr-~5 Be

Unirradiated 64 32

Irradiated 54 17
60 Cu~15 Ni~20 Ta—5 Be?

Unirradiated 78 31

Irradiated 49 18

“Specimens failed in Hastelloy N due to the high
strength of the brazing alloy.

Table 20.1. Composition of Brazing Filler Metals Used in Irradiation Test

Alloy Composition
(wt %)

Brazing Application

Pure copper

Pyrolytic-molybdenum-coated graphite to molybdenum

or Hastelloy N and molybdenum to Hastelloy N

35 Ni-60 Pd-=5 Cr
68 Cu—~17 Ni—-10 Cr—5 Be
60 Cu—15 Ni~20 Ta~5 Be

Graphite to molybdenum or Hastelloy N
Graphite to molybdenum or Hastelloy N

Graphite to molybdenum or Hastelloy N

20.3 GRAPHITE-TO-HASTELLOY N
TRANSITION JOINT

J. P. Hammond

Work was initiated to develop a transition joint
for joining graphite to Hastelloy N. The transi-
tion joint concept is an attractive solution to the
problem of joining materials of widely differing
thermal coefficients of expansion and consists
in inserting a series of connected segments whose
properties change gradually from one component
of the juncture to the other.* The segments are
usually fabricated by powder metallurgy and are
joined by diffusion bonding or brazing. The con-
stituents of the joint should be metallurgically
compatible with one another under the fabrication
and service conditions.

There are two widely differing properties of
graphite and Hastelloy N which give concern:
the coefficient of thermal expansion and the di-
mensional stability under irradiation. Both can
produce damaging strain at the joint interface,
the former being of concern while cooling from the
brazing temperature and during thermal cycling
and the latter being of concern during prolonged
irradiation of the graphite member.

20.3.1 Conceptual Design

Two families of materials exist which offer
unique characteristics for designing the type of
link we require. One is the familiar class of
heavy-metal alloys® of tungsten or molybdenum
with a nickel alloy as the binder constituent,
which is fabricated by liquid-phase sintering,.

The structure consists of round heavy-metal

grains enveloped by a nickel alloy matrix. The
other class of materials is the graphite—transi-
tional metal carbide composites recently developed
for rocket nozzles.® These are fabricated by hot
isostatic compaction.

As noted in Table 20.3 the coefficient of thermal
expansion of the heavy-metal alloy should be

‘r. Zimmer, Metal Progr. 83, 101 (January 1963).

5F‘. C. Holtz, Development and Evaluation of High-
temperature Tungsten Alloys, Final Report, ARF-2209-
7, LAR 59 (September 1961).

by, Harada, Graphite—Metal Carbide Composites,
NASA-Cr-507 (June 1966).

235

Table 20.3. Expected Coefficients of Thermal

Expansion for Thermal Joint Materials

Coefficient of

Material Crystal  ppermal Expansion®
Structure (ptin./oC)
Isotropic graphite Hex 4.3
Hastelloy N Fcc 12.3
Tungsten Bcc 4.7
Molybdenum Bcc 5.3
Graphite-carbide Hex-~fcc 4.3 to ™5.3
composites
Heavy-metal alloys Bcce-fcc ~5 to ™11

fMean value between 20 and 600°C.

alterable over a wide range by varying the con-
centration of heavy-metal phase in the structure.
Since tungsten and molybdenum are near graphite
in expansion coefficient (Table 20.3), the seg-
ment of highest heavy-metal concentration would
be located adjacent to the graphite member, while
the segment of highest nickel matrix would be
next to Hastelloy N. To mollify undesirable
effects stemming from placing dimensionally un-
stable graphite against stable heavy-metal alloy,
a transition with respect to irradiation-induced
dimensional instability is also introduced. This
is done by incorporating one or more segments of
a graphite-carbide composite between the graphite
and the heavy-metal members. Since the graphite-
carbide composite would be comprised of a highly
dense isotropic-type graphite without any graphi-
tizable binder,” it should be fairly resistant to
dimensional change during irradiation. Observe
in Table 20.3 that the combination of graphite-~
carbide composites and heavy-metal alloys spans
the wide difference in coefficient of thermal ex-
pansion between the graphite and the Hastelloy N.
Schematic drawings illustrating the principles of
this design are shown in Fig. 20.5.

Fabrication of suitable graphite-carbide com-
posites for this joint appears straightforward using

7D. C. Carmichael, W. C. Chard, and M. C. Brockway,
Dense Isotropic Graphite Fabricated by Hot Isostatic
Compaction, BMI-1796 (March 1967).
GRAPHITE-CARBIDE

ORNL-DWG 67-10338

HEAVY METAL

COMPOSITES { ALLOYS
LN iy -
TRANSITION N
ETMASS'ST,(')O,\TALN COEFFICIENT OF
INSTABILITY THERMAL EXPANSION
- /
.
Afl/g 3
2

t {

EXPOSURE

N\
HASTELLOY N - &70{
E = éflb’\’X
Wwoo 4 V
cug &
w o, ‘\(’ =X X
S “\90‘9\/ -
X
<
GRAPHITE 4-  x
L_L__L___L | | I 1L
® @ & @ & @

COMPOSITION OF SEGMENTS

Fig. 20.5. Transition Joint, Graphite to Hastelloy N,
Top: array of segments comprising the joint, showing
Middle:

diation-induced dimensional instability of graphite-

transitions in properties. transition in irra-
carbide composite segments. Bottom: transition in co-
efficient of thermal expansion of graphite-carbide

composite and heavy-metal segments,

the high-temperature gas-pressure bonding tech-
nique. The joining of the components of the joint
is believed to be feasible with current methods
and knowledge relative to brazing and diffusion
bonding.

20.3.2 Heavy-Metal Alloy Development

Preliminary fabrication studies were conducted
on tungsten heavy-metal alloys to confirm the good
sinterability and ductility experienced with this
class of materials in other quarters.® By achiev-
ing microstructures wherein a ductile nickel phase
surrounds the otherwise fragile tungsten grains,
exceptionally ductile composites have been ob-
tained. As indicated in Table 20.4, good fabri-

236

cation results were achieved with a nickel-iron
mixture (4/1 nickel-to-iron ratio) as the binder
constituent. The tungsten concentrations studied
were 90, 75, and 60 wt %; these levels appear
appropriate for the individual segments of the
transition joint. The high toughness of these
particular alloys is attested to by the high degree
to which they are cold rollable without edge
cracking (Fig. 20.6a and b).

We extended our studies to include alloys based
on molybdenum because they are expected to show
superior resistance to embrittiement by irradia-
tion. The important consideration here is that at
the service temperature (around 700°C), fast dis-
placement damage would anneal out of the molyb-
denum but possibly not out of tungsten. Early
attempts to fabricate alloys based on molybdenum
either gave rather brittle alloys or they were dif-
ficult to fabricate (Table 20.4, specimens 4 to 11).

The observation that these molybdenum heavy-
metal alloys were prone to be brittle whereas the
tungsten alloys were not is explained by compar-
ing the nickel-molybdenum and nickel-tungsten
phase diagrams. On cooling mixtures of molyb-
denum and nickel which have been heated at the
liquid-phase sintering temperature (just above
1453°C), a eutectic containing the brittle inter-
metallic compound NiMo forms., This happened
for the 90% Mo—7% Ni—3% Fe alloy (specimen 4
of Table 20.4) and gave a very brittle alloy.
Nickel-tungsten alloys do not form an intermetallic
compound during cooling from the liquid phase,
and tungsten which was liquid-phase sintered with
nickel~iron as the binder was free of the nickel-
molybdenum intermetallic and was quite ductile.

The problem of developing suitable molybdenum
heavy-metal alloys thus resolves into one of find-
ing additives to molybdenum-nickel that suppress
the intermetallic-forming reaction sufficiently to
preclude brittle alloys while at the same time
ensuring good fabricability. Palladium, platinum,
coppet, and tungsten were selected for examina-
tion — palladium, platinum, and copper as soluble
matrix phase additions and tungsten as a molyb-
denum displacement additive.

The tungsten additive greatly reduced the
amount of intermetallic in the alloys, but some of
this phase still remained (specimens 5 to 8 of
Table 20.4). Also, these compacts, when con-
taining the larger amounts of binder constituent,
had very narrow sintering temperature ranges,
237

Table 20.4. Preliminary Fabrication Results on Tungsten and Molybdenum Heavy-Metal Alloys

Optimum Sintering

Specimen Composition Temperature® Remarks?

1 90 W—-10 (7 Ni/3 Fe) 1450 Good structure with one-phase matrix;
ductile

2 75 W—25 (7 Ni/3 Fe) 1435 Good structure with one-phase matrix;
very ductile

3 60 W—40 (7 Ni/3 Fe) 1415 Good structure with one-phase matrix;
very ductile

4 90 Mo~10 (7 Ni/3 Fe) 1400 Good structure, but matrix was a
eutectic mixture; brittle

5 48 Mo—35 W—~15 Ni—2 Fe 1410 Fair structure, but matrix contained
some eutectic; not ductile

6 85 (7 Mo/10 W)—15 (4 Ni/1 Fe) 1475 Fair structure, but some intermetallic
in matrix; not ductile

7 70 (7 Mo/10 W)-30 (4 Ni/1 Fe) ~1310 Similar to 6 and difficult to sinter
without forming *“glob?®*‘

8 55 (7 Mo/10 W)—45 (4 Ni/1 Fe) Similar to 6 and very difficult to sinter
without forming *“glob?*?*®

9 75 Mo—20 Ni-5 Pt 1370 Poor structure, with some eutectic and
porosity in the matrix; not ductile

10 87 Mo—~6.5 Ni—6.5 Pd 1475 Fair structure, with one-phase matrix;
not ductile
11 90 Mo—7 Ni-3 Cu 1375 Excellent structure with one-phase

matrix but poor surface character
and difficult-to-control sintering;

ductile

aFirings were conducted in a dry hydrogen atmosphere.

bComments relative to quality of microstructure refer to degree of success in achieving the desired dispersion of
round heavy-metal grains in a nickel alloy matrix; ductility was assessed by determining amenability to cold rolling.

°Glob refers to a slumping rounded mass.

showing a marked tendency to ‘“glob.”” Platinum flaws. The difficulties experienced with the

also gave unsatisfactory results (specimen 9), latter alloy were partially attributed to a tempera-
but palladium and copper (specimens 10 and 11) ture-lowering endothermic reaction which occurred
gave the desired continuous-matrix-type structures just as sintering began.

free of any discermible intermetallic. However, These preliminary results on heavy-metal alloys
the palladium-containing alloy, 87 Mo—6.5 Ni—6.5 are encouraging, and we feel that the transition-
Pd (wt %), was unexpectedly brittle. Since palla- joint approach to joining graphite to Hastelloy N
dium has a rather high affinity for hydrogen, this will prove practicable. A joint involving tungsten-
may have been caused by having sintered in hy- base alloys is presently possible, and efforts to
drogen as an atmosphere., The copper-containing develop suitable molybdenum heavy-metal alloys
material, 90% Mo—7% Ni—-3% Cu, was ductile but will continue. Also, cursory evaluations will be

proved difficult to sinter without severe surface made of graphite-carbide composites.
238

0.007 INCHES

B s00x -

Fig. 20.6. Microstructures of a Number of Experimental Heavy-Metal Alloys. (a) 90% W=7% Ni-3% Fe, cold rolled
60% in reduction of area without cracking. (b) 60% W-28% Ni-12% Fe, cold rolled 95% without cracking. (c) 87%
Mo~-6.5% Ni-6.5% Pd, cracked with first rolling pass. (d) 90% Mo~7% Ni—-3% Cu, as sintered, ductile. Etchant:
equal parts of hydrogen peroxide (30%) and ammonium hydroxide.

20.4 NONDESTRUCTIVE TESTING EVALUATION
OF GRAPHITE-TO-METAL JOINTS

K. V. Cook

We are developing nondestructive testing tech-
niques to detect nonbond in various designs of
graphite-to-metal joints. Our preliminary studies
indicate that an acceptable pulse-echo ultrasonic
method can be developed for certain joint con-
figurations. Mechanical scanning is a problem
for the joint designs with which we have been
working. We assembled a turntable device to use

in conjunction with our large ultrasonic tank for
X-Y scanning; however, modifications will be
necessary before we can economically scan a
number of joints of different configurations.

We were able to demonstrate that the ultrasonic
method (using the turntable scanning device) could
distinguish bond and nonbond on one joint design.
The ultrasonic data indicated that bonding was
accomplished on only one area of the total joint.
The destructive test of this joint resulted in the
separation of the graphite material in the bonded
area. Figure 20.7 shows that result. As is evi-

Fig. 20.7. Results of a Destructive Test on a

Graphite-to-Metal Brazed Joint. The dark areas, where
the graphite material separated due to the presence of
good bond, coincide with the area which the ultrasonic

test had predicted to be bonded.

~ PHOTO 89768

dent, only one portion of the joint, the dark area,
was bonded, coinciding with the prediction based
on the ultrasonic examination. These results
demonstrate the feasibility of the ultrasonic
pulse-echo method for this particular joint con-
figuration. No attempt has been made to determine
the optimum sensitivity to nonbonding.

A second joint design we have been working
with is a tensile specimen with the graphite-to-
metal joint in the gage length. The maximum
diameter of the specimen is 1.00 in., and the
joint is a 30° cone, so that the joint diameter
varies from 1 in. to a point. The graphite com-
prises the male part of the joint and the Hastelloy
N the female part. Since both the radius of curva-
ture of the joint and the area that reflects the
ultrasonic energy change with the diameter, cali~
bration is a problem because we encounter varia-
tions in reflected signal as the diameter changes
even for complete nonbond. A further complica-
tion is the changing metal thickness from the
outer surface of the specimen to the joint inter-
face as the joint diameter changes. Attempts to
evaluate this joint with the pulse-echo ultrasonic
method have been unsuccessful thus far; however,
we feel that it is possible to tolerate some varia-
tion in our test, and we plan to pursue this tech-~
nique when modifications on our mechanical
system are completed. Other joint designs to be
fabricated will also require investigation by NDT
evaluation techniques. These needs are being
coordinated closely with the Welding and Brazing
Group.
21.

21.1 REMOTE WELDING

R. W. Gunkel

The Metals and Ceramics Division and the Re-
actor Division are engaged in a joint program for
investigating and developing a welding process for
automated remote maintenance of the MSBR. Most
work so far has been in the form of literature sur-
veys and contacts with automatic equipment manu-
facturers and users. Automated remote cutting and
welding have been done at several facilities in the
United States, and information from these programs
is beneficial to us. Most of the applicable work
has been conducted using the gas-tungsten arc
(cold-wire feed) and gas-metal arc processes.

Hastelloy N will be the metallic structural ma-
terial to be joined, and both automatic and manual

240

Support for Components Development Program

gas-tungsten arc welding processes have been dem-
onstrated adequately for this alloy. The primary
problem lies in adapting the techniques to remote
operation. This is not true for the gas-metal arc
process, which has been used very little in the
welding of Hastelloy N. Hence, we have in prog-
ress a modest effort to adapt this potentially de-
sirable process to Hastelloy N. Test plates have
been machined, and filler wire is being fabricated
and spooled for use in these studies. Approximate
parameters will be obtained for depositing sound
weld metal.

Since the maintenance may involve the welding
of highly irradiated (with consequent low ductility)
Hastelloy N, we plan to conduct welding studies
on some irradiated creep specimens. If weld crack-
ing appears to be serious, various approaches will
be investigated to obtain a suitable joint.
Part 6. Molten-Salt Processing and Preparation

M. E. Whatley

The development work on processes for molten-
salt breeder reactors has undergone a significant
change in direction during this reporting period.
The change resulted from the recognition and ac-
ceptance of reductive extraction as a feasible and
attractive basis for processes to isolate protactin-
ium from the reactor and, possibly, to remove fis-
sion products. A protactinium isolation flowsheet
based on reductive extraction now appears appli-
cable to a one-fluid breeder reactor. This process-
ing development, along with the recognition that
good breeding performance can be achieved with a
single-fluid system, has changed the direction of
the reactor development and hence has redirected
the development of chemical processing methods
away from a two-fluid concept.

While fluorination and distillation for a two-fluid
reactor will probably find application as steps in
the processing of a single-fluid machine, they will
be secondary to reductive extraction, which will

comprise the heart of the processing flowsheet.

Our work on protecting exposed surfaces from cor-
rosion by a dynamically cast layer of frozen salt,
originally developed for the fluorination step, will
find application in many steps of the new process.

This section includes the presentation of some
of the data on reductive extraction which lend
credit to its feasibility, and the calculational anal-
ysis of some proposed flowsheets with alternatives.
Although hardware for engineering experiments with
reductive extraction is being assembled, this work
is not yet ready to report.

Our participation in the activities of the Molten-
Salt Reactor Experiment during its shutdown in
March 1968 will include recovery of the 235U from
the present fuel salt, processing this salt to re-
move chromium and corrosion products, delivery of
a prepared charge of 233UF ,-LiF eutectic for re-
fueling the reactor, and distillation of 48 liters of
MSRE carrier salt in an experimental vacuum still.

22. Measurement of Distribution Coefficients

in Molten-Salt—Metal Systems

L. M. Ferris J. F. Land

Evaluation of reductive extraction methods for
the processing of molten-salt breeder reactor fuels
is being continued. The process now under study
involves the selective extraction of uranium, prot-
actinium, and rare earths from the molten salt
using liquid lithium-bismuth solutions. Originally !
the main effort was devoted to the processing of

J. J. Lawrence

241

C. E. Schilling

two-fluid MSBR salts: specifically, the removal of
protactinium from the blanket salt and the removal
of the rare earths from the fuel salt. More recent

1M. W. Rosenthal, MSR Program Semiann. Progr. Rept.
Aug. 31, 1967, ORNL-4191, pp. 148 and 248.
work 2’3 has indicated that uranium and protactin-
ium can probably be separated and recovered from
a single-fluid MSBR salt; furthermore, a prelimi-
nary evaluation? showed that it should be pos-
sible to engineer such a process. Consequently,
most of the current effort is directed toward devel-
opment of processing methods for single-fluid
MSBR fuels. In order to obtain a more detailed
evaluation of these methods, accurate distribution
coefficients for uranium, thorium, and the rare
earths must be determined. Our program is, there-
fore, designed to provide these data.

22.1 EXTRACTION OF URANIUM AND
RARE EARTHS FROM FUEL SALT
OF TWO-FLUID MSBR'S

The equilibtium distribution of a component M
between a molten salt and a liquid metal phase can
be expressed as a distribution coefficient D that
is defined as

mole fraction of component M in metal phase

M

mole fraction of component M in salt phase

Thermodynamic treatment® (supported by experi-
mental data) of the equilibria involved with salts
containing LiF indicates that, at a fixed tempera-
ture, each distribution coefficient should vary with
the lithiwn concentration in the metal phase accord-
ing to the equation

log D,=nlogX . . +logl 1)

(m)
if the concentration of M in each phase is low. In
Eq. (1), n is the valence of the component (as its
fluoride, MFn) in the salt, XLi my 18 the equilib-
rium lithium concentration in the metal phase (atom
fraction), and I is a constant.

Presentation of equilibrium distribution data in
the manner indicated by Eq. (1) is desirable; how-
ever, because of the preliminary nature of our ini-
tial experiments, this is not possible. In our ex-

?J. H. Shaffer, memo to W. R. Grimes, Nov. 27, 1967.
3C. J. Barton, personal communication, December 1967,

4M. E. Whatley, L. E. McNeese, and J. S. Watson,
personal communication, Feb. 6, 1968.

5W. R. Grimes, Reactor Chem. Div. Ann. Progr. Rept.
Dec. 31, 1966, ORNL-4076, p. 34.

242

periments, usually only a few data points were
obtained at a given temperature. Our data can be
compared with those obtained by others either in
terms of the quantity Q, defined as

Oy = DMXI—:?(m) ’ (2)

or as the difference in standard reduction poten-
tials (as defined by Moulton %):

E{ - Eg i =AE;=(RT/nF)In D,

~RT/F)InD_,. (3)

Values for Q and AE(; can be calculated from a sin-
gle distribution coefficient and the corresponding
equilibrium lithium concentration. From Eqs. (2)
and (3), the relationship between Q and AE; is
found to be

nF AE’

log 0= ————9 _nlog X . 4
°8 0= 3mrT " o8 XLir )

in which X, ,p is the mole fraction of LiF in the
salt.

Data obtained on the distribution of several sol-
utes between two-fluid MSBR fuel salt, LiF-BeF,
(66-34 mole %), and lithium-bismuth solutions are
given in Tables 22.1 and 22.2. In calculating val-
ues of Q and AE, n was assumed to be unity for
lithium and sodium, 2 for europium, 3 for lanthanum,
and 4 for uranium. Average AE_ values (Table
22.1) are given at those temperatures where more
than one distribution coefficient was obtained.

The values obtained for lanthanum and europium
are in reasonable agreement with those reported by
Moulton and Shaffer.” 8 However, our value for
uranium at 600°C (based on three data points,
Table 22.2) is 0.14 v higher than the value re-
ported by Moulton and Shaffer.” The reason for
this difference is inexplicable at present. It is
interesting to note that the AEO' values appear to
be practically constant over the temperature range
from 500 to 700°C. If AE_ is constant, plots of
log Q vs 1/T should be linear [see Eq. (4)]. Plots

M. w. Rosenthal, MSR Program Semiann. Progr. Rept.
Feb. 28, 1967, ORNL-4119, p. 150.

7'D. M. Moulton and J. H. Shaffer, unpublished results,
Dec. 14, 1967.

8M. W. Rosenthal, op. cit., pp. 155-56.
243

Table 22.1. AE(; Values? for the Distribution of Various Solutes Between Different Fluoride Salts

and Lithium-Bismuth Solutions

Salt Composition (mole %) Temperature AE; v)
LiF BeF ThF CC) U La Eu Na Th
66 34 0 500 0.27
66 34 0 600° 0.32
66 34 0 600° 0.40 0.30
66 34 0 600° 0.54 0.45 0.32 0.36
66 34 0 600 0.68
70 19 11 600 0.66 0.33 0.41
66 34 0 602 0.31
66 34 0 605 0.44 0.27 0.20
66 34 0 675 0.60 0.43
66 34 0 700 0.30
66 34 0 700? 0.34
aAE(; = E(;,M — E(;,Li' where M is the other metal involved; see text.

5D. M. Moulton and J. H. Shaffer, ref 7.

°D. M. Moulton and J. H. Shaffer, ORNL-4191, p. 155; ref 8.

of the data for europium and lanthanum are shown
in Fig. 22.1; the lines were placed using average
values of AE_ of 0.43 and 0.30 for lanthanum and
europium respectively (Table 22.1). (Moulton’s
data for cerium® give a very good straight line over
the temperature range from 500 to 700°C.) Although
the data points are highly scattered, it does appear
that Q decreases with increasing temperature. This
dependence of Q on temperature shows that the rare
earths are less easily extracted at the higher tem-
peratures.

The data given in Tables 22.1 and 22.2 show
that uranium can easily be removed from the fuel
salt of a two-fluid MSBR by reductive extraction.
Furthermore, separation from the rare earths is
very good; at 600°C the separation factors
(D,/D ) are at least 1000. The data also show
that, once the uranium has been extracted, the rare
earths can be removed from the salt by increasing
the lithium concentration in the metal phase.

22.2 EXTRACTION OF URANIUM FROM
SINGLE-FLUID MSBR FUEL

Two experiments involving equilibration of sim-
ulated single-fluid MSBR fuel salt with lithium-
bismuth solutions at 600°C have been completed.
In each experiment the salt, initially LiF-BeF .
ThF ,-UF, (about 70-19-11-0.3 mole %), and about
an equal volume of bismuth were heated in a mild
steel crucible under argon to 600°C; then lithium
(as Li-Be alloy) was added to the bismuth in small
increments, and the equilibrium distribution of the
various components between the two phases was
determined. The salt in one experiment contained
250 ppm LaF
other system.

The distribution coefficients obtained are given
in Table 22.3. Those for uranium can be repre-
sented by the relationship

5 while EuF , was present in the

log D, =41log C, . +7.874,
244

Table 22.2. Effect of Temperature on the Distribution of Europium and Lanthanum Between LiF-BeF2
(66-36 Mole %) and Li-Bi Solutions

4 Li Concentration Qa
Temperature 10°/T, in Metal Phase D D D

Cc) CK) at. %) . La v Eu La
x 103 x 10%

500 12.94 4.02 14.3 8.83

583 11.68 4.95 16.4 6.67

583 11.68 3.46 17.1 14.3

600 11.45 0.0133 1.74

600 11.45 0.0232 5.41

600 11.45 0.0313 53.6

600° 11.45 6.68 0.29

600° 11.45 11.4

600° 11.45 11.4 2.16

602 11.43 2.56 5.41 8.26

605 11.39 0.119 0.092 0.55

605 11.39 0.119 0.90 5.34

605 11.39 0.208 0.705 0.78

605 11.39 0.298 5.12 1.93

605 11.39 4.84 6.36 2.72

608 11.35 0.85 0.82 11.3

675 10.55 0.238 0.487 1021 0.36

675 10.55 0.298 0.672 1788 0.25

675 10.55 0.298 0.926 0.35

675 10.55 0.416 1.09 0.15

675 10.55 0.506 2.01 0.16

700 10.28 5.11 8.66 3.32

700° 10.28 7.64

800° 9.32 5.53

aQ = DLnX;?(Bi); Ln = Eu or La; the values of n used for Eu and La were 2 and 3 respectively.

bCalculated from data given by D. M. Moulton and J. H. Shaffer, ref. 8.
°D. M. Moulton and J. H. Shaffer, ref. 7.
245

Table 22.3. Distribution of Lithium, Uranium, Thorium, and Europium Between LiF-Ber-ThF4
(70-19-11 Mole %) and Bismuth Solutions at 600°C

Experiment Lithium Concentratiog DU DTh DEu
No. in Metal Phase (at. %)
72 0.000672 0.00021
72 0.00119 0.00013
2 0.00321 0.0638
2 0.00579 0.0267
72 0.00584 0.311
2 0.00726 0.132
2 0.00747 0.238
72 0.00873 0.134
2 0.00988 2.47
72 0.00997 0.827
2 0.0112 1.12
72 0.0128 1.47
72 0.0133 2.28
72 0.0136 4.08
72 0.0176 5.07
2 0.0182 15.1
72 0.0285 15.2
72 0.0376 26.1
2 0.0570 102 0.00131 0.0041
72 0.0623 129 0.00229
72 0.0786 168 0.00327
72 0.0816 155 0.00573
72 0.0855 0.00802
2 0.0939 0.00998 0.011
2 0.0997 333 0.00712 0.014
72 0.114 0.0131

2 0.175 759 0.0152

ORNL -DWG 68-6095
TEMPERATURE (°C}
800 700 600 500

0.1 A DATA OF D.M. MOULTON, ORNL-4191, p. 155.
® D.M, MOULTON, PERSONAL COMMUNICATION.

9 10 {1 12 13

10,000/, )

Fig. 22.1. Dependence of log QO on Temperature,

in which €, . is the lithium concentration in the
metal phase (at. %). This equation was obtained
by visually fitting what appeared to be the best
line of slope 4 through the points in the lithium
concentration range of 0.002 to 0.03 at. %, where
the analyses were the most accurate. These data
are shown in Fig. 22.2; the distribution coeffi-
cients obtained using LiF-BeF, (66-34 mole %)
as the salt phase are included for comparison.

The distribution coefficients for thorium (Table
22.3) can be represented by the equation

log D, =4logC ., +2.00,

which was also obtained by a visual fit of the data
(Fig. 22.2). The line through the three points ob-
tained for europium (Fig. 22.2) was drawn with a
slope of 2.

In these experiments, as lithium-bismuth alloy
was added to the system, the thorium concentration
in the metal phase increased to what appeared to
be a limiting (steady-state) value of 1700 £ 100
ppm. The corresponding uranium and lithium con-

246

5 ORNL-DWG 68-6096
S ETEREE — T
[ o i 1 A F — -
I ,J I N
| SoRil | L
2 1o ( ]
g 10 b =
= o _— N I N
S S * HiES
- i
L|c
516 - Hd
TG v
|5
ol 100 — — o
ole —— —
EiE T T
R - T
s L,W, o 7+fl[f — ¥| s {
m ) | e, EXPT. 72 L
u B Z o U, EXPT. 2
o | -t v U, EXPT. WiTH Lif — BeF.
z 0 = a (66-34 mole %) -
P - Ny AEAN 4 Th, EXPT. 72
@ - I aTh EXPT 2
'?7:) _ - . v Fu, EXPT. 2 -
& I Y AR AN jt T
I0° b—at—+t ] }a_fi::
,,_k:“l: 511 o— S 4 7 A N N
e e e
‘ i -
L - S 1l
/ A
% B, S
50—3 [ Jq (L
0,00 00! oY 10

LITHIUM CONCENTRATION IN METAL PHASE fat. %)

Fig. 22,2, Effect of Lithium Concentration ir Metal
Phase on the Distribution of Uranium, Thorium, und
Evuropium Between LiF-BeFZ-ThF4 (70-19-11 Mole %)
and Bismuth at 600°C.

centrations were 2500 = 300 and 50 % 10 ppm re-
spectively. The distribution coefficients at steady
state were about 0.014 for both thorium and euro-
pium and less than 0.07 for lanthanum (the lantha-
num concentration in the metal phase was always
less than 4 ppm). The data indicate that the sol-
ubility of thorium in bismuth is markedly depressed
by the presence of uranium and/or lithium, since
the solubility of thorium in pure bismuth at 600°C
is 3000 to 3900 ppm. /10

9M. Hansen and K. Anderko, Constitution of Binary
Alloys, 2d ed., p. 341, McGraw-Hill, New York, 1958,

IOR. P. Elliott, Constifution of Binary Alloys, First
Supplement, p. 202, McGraw-Hill, New York, 1965.
247

The data presented above show that uranium can
be easily extracted from single-fluid MSBR salt,
leaving thorium and rare earths in the salt; the
separation factors (D,/D ., and D /D _ ) are at
least 10*. These data also show by extrapolation
to low lithium concentrations that the rare earths
can probably be separated from thorium; however,
in order to achieve reasonable separation factors
in this system, the thorium concentration in the
metal phase will probably have to be somewhat
lower than its maximum solubility.

22.3 EXPERIMENTAL PROCEDURE AND
LITHIUM *‘LOSS"’

It has been noted previously !'!! that during re-
ductive extraction experiments up to 75% of the
lithium added to the system was consumed in an
inexplicable manner. In most of these experiments,
the lithium was added as the pure metal. The re-
sults of our latest experiments show that this prob-
lem can be largely eliminated when the lithium is
added to the system as pieces of frozen lithium-
bismuth alloy. An example of the equivalent bal-
ances that can be achieved by this method (Fig.
22.3) was derived from data obtained in an experi-
ment in which LiF-BeF ,-UF, (initially 66-34-0.3
mole %) was equilibrated with liquid bismuth to
which lithium-bismuth alloy (7 at. % Li) was added
periodically. The figure is a plot of the milliequiv-
alents of Li + U found in the metal phase [4(mg-
atoms U) + mg-atoms Li] vs the milliequivalents of
lithium added to the system as lithium-bismuth alloy.
The first 10 meq of lithium added was probably
consumed in reactions with FeF ,» OHT, and other

! IR. B. Briggs, MSR Program Semiann. Progr. Rept.
Aug. 31, 1966, ORNL-4037, p. 142.

easily reduced species that were present in the
salt. Then, as more lithium-bismuth alloy as
added, the lithium was consumed by reduction of
uranium fluoride from the salts. The consumption
of lithium was nearly stoichiometric for the reac-
tion

4Li+UF4 - 4LiF +U .

This behavior indicates that the uranium in the
salt phase was still in the tetravalent state (un-
doubtedly as UF ,). After the uranium was ex-
tracted into the metal phase, the increase in milli-
equivalents of Li + U was due entirely to the addi-
tion of more lithium. As may be seen in Fig. 22.3,
the equivalent balance was quite acceptable (i.e.,
the slope of the line was unity) throughout most of
the experiment.

ORNL-DWG 68-6097

U + Li FOUND IN METAL PHASE (meq)

0 10 20 30 40 50 60 70
Li ADDED TO SYSTEM AS Li—Bi ALLOY (meq)

Fig. 22.3. Reductant Balance in Experiment Involving
Equilibration of Lithium-Bismuth Solutions with LiF-
Ber-Ul""'4 (Initially 66-34-0.3 Mole %) at 600°C.
23. Protactinium Removal from a Single-Fluid MSBR

L. E. McNeese

Laboratory experiments have shown that pro-
tactinium and-uranium can be extracted into bis-~
muth from molten salt which contains fluorides
of lithium, beryllium, and thorium by using metal-
lic thorium as the reductant. In the proposed
flowsheet (Fig. 23.1), a salt stream from the re-
actor enters the bottom of the extraction column
and flows countercurrently to a stream of bismuth
containing reduced metals. Ideally, the metal
stream entering the top of the column contains
sufficient thorium and lithium to extract only the
uranium entering the column. The system ex-
ploits the fact that protactinium is less noble
than uranium but more noble than thorium. Hence,

M. E. Whatley

in the lower part of the column, uranium is prefer-
entially extracted from the incoming salt, while
the protactinium progresses farther up the column
to where it is reduced by thorium. In this manner,
protactinium refluxes in the center of the column,
and relatively high protactinium concentrations
result. A retention tank is provided at the center
of the column, where the maximum protactinium
concentration occurs in the salt, to retain the
protactinium until it decays to uranium.

An essential part of the flowsheet is an elec-
trolytic oxidizer-reducer which serves the dual
purpose of recovering the extracted uranium from
the metal stream leaving the extraction column

ORNL-DWG 67-12861A

F2

REACTOR

OXIDIZER
(ANODE)

REDUCER
(CATHODE)

Fa4
F3

Fig. 23.1. Single-Fluid MSBR Processing by Reductive Extraction.

248
and of preparing the lithium-thorium-bismuth
stream which is fed to the extraction column.

The metal phase containing the uranium extracted
in the column can serve as the cell anode, where
uranium and lithium will be oxidized to UF, and
LiF. Salt from the top of the extraction column
serves as the cell electrolyte and first passes
over a pool of liquid bismuth which serves as the
cathode into which thorium and lithium are
reduced.

Typical performance of the protactinium isola-
tion system is shown in Fig. 23.2 for a case
which will be taken as the reference case. The
minimum reactor protactinium concentration is
obtained when the bismuth flow rate is just suffi-
cient to extract the uranium entering the system.
At slightly higher bismuth rates, protactinium
will also be extracted, since it is the next com-
ponent in order of decreasing nobility. At bis-

249

muth rates slightly lower than the optimum rate,
some uranium will not be extracted; this uranium
and most of the protactinium will flow out the top
of the column. In either case, some protactinium
is allowed to return to the reactor, and the effec-
tiveness of the system is diminished. The pro-
tactinium isolation system becomes ineffective
almost immediately for bismuth flow rates lower
than the optimum rate; for bismuth flow rates
higher than the optimum, the reactor protactinium
concentration increases from the minimum value
of 22 ppm at the rate of 28 ppm for each percent
increase in metal flow rate for the conditions of
this particular case. Similar effects would be
produced by variations in salt flow rate or in
total concentration of reduced metals (equivalents
of reduced metals per mole of bismuth) fed to
the column. Calculated concentration profiles in
the extraction column are shown in Fig., 23.3 for

ORNL~DWG 68-12907A

1400 [
CONDITIONS
REACTOR VOLUME 1.5x10°% gmoles
1200 - DECAY TANK VOLUME 0.3x40% gmoles - |
\ SALT FLOW RATE: 0.5x10% gmoles /day
\ TOWER ABOVE TANK: 6 STAGES
21000 DECAY TANK TOWER BELOW TANK: 6 STAGES
s
S
QL
g
— 800 r
‘e
»
2
O
=
< 600
-
P
w
Q
=
(@)
[
© 400
REACTOR AND
/ DECAY TANK REACTOR
200 — /(
////l/y
0
48 4.9 5.0 5.1 5.2 5.3 5.4 55 5.6 5.7

5.8

BISMUTH FLOW RATE 1075 (gmoles /day)

Fig. 23.2. Variation of Protactinium Concentration in Reactor and Protactinium Decay Tank with Bismuth Flow Rate.
ORNL.— DWG 67 -12893A

-0
10
S LiF
Q BEFE
1072 -
1ThFa Th N METAL ]
— — —0—1552. C O O O O
X
v ‘F
N
1074 / A AN ]
—n A Po IN METAL
108 ‘\
1085
i
=
" m\
T
2 1©0'°g %
s al
[ray
w 5 g ‘
o O alin METAL
g g2 by d
1 I
I SALT

~14
) | |

U=0000063 | 0.000083

0.00130 ®

Pa = 0.00132

O 2 4 © 8 10 12
STAGE NUMBER

Fig. 23.3. Calculated Concentration Profiles in Re-

ductive Extraction Tower.

steady-state operation under optimum conditions.
The concentration of uranium in the salt decreases
from the reactor concentration of 3 x 1072 mole
fraction to 6.3 x 10~5 mole fraction at the inlet
to the protactinium decay tank, whereas the pro-
tactinium concentration increases from the reactor
concentration of 2.2 x 10~ ° mole fraction to
1.32 x 10~ mole fraction at the inlet to the de-
cay tank. The concentrations of protactinium
and uranium in the decay tank are 1.3 x 10~* and
8.3 x 10~ ° mole fraction respectively. Abov=
the decay tank the uranium and protactinium con-
centrations decrease steadily to negligible values.
The flowsheet has several very desirable char~
acteristics, which include a neglibible holdup of
fissile 2*?( in the processing plant; an almost
immediate return of newly produced 233U to the
reactor system; and a closed system which pre-

cludes loss of protactinium, 2331y, or other com~
ponents of the reactor fuel salt. However, the
protactinium: removal efficiency is undesirably
sensitive to riinor variations in operating condi-
tions such as salt or bismuth flow rate and con-
centration of reduced metals in the bismuth stream
fed to the extraction column. Methods for making
system performance less sensitive to minor varia-
tions in operating conditions have been considered.
Removal of uranium from the center of the column
by fluorination of molten salt was found to make
steady-state system performance insensitive to
small changes in operating conditions; the con-
centration of UF _in the fluorinator off-gas was
found to be extremely sensitive to changes in
operating conditions and can be used as the
primary process control point.

The effectiveness of a system stabilized by
uranium removal was compared with that of a
nonstabilized system by assuming the error in
control of bismuth flow rate to be distributed
normally around a best mean flow rate FMBopt
with standard deviation o, so that an average
reactor protactinium concentration € could be
calculated by

1

f:’o c(x) e“"2/2 dx ,
V2r

where

x = FMB — FMBopt/O',
FMB = bismuth flow rate,

c(x) = steady-state reactor protactinium con-
centration at bismuth flow rate corre-
sponding to x.

The average reactor protactinium concentration
is shown in Fig. 23.4 as a function of the stand-
ard deviation in bismuth flow rate for cases with
stabilization by uranium removal as well as for
two cases with no uranium removal which have
protactinium processing cycle times of three days
(reference case) and one day. For a standard
deviation of 0.35% of FMB_ (optimum bismuth
flow rate), the minimum reactor protactinium con-
centration (22 ppm) is obtained if the fraction of
uranium removed from the inlet to the decay tank
is 2% or greater; with no uranium removal the
average concentration is 38.5 ppm. With a pro-
tactinium processing cycle time of one day and
a standard deviation of 0.35%, an average con-
251

280 ORNL-DWG 68-2090A
f T — T ]
T | | T |
’ £=FRACTION OF URANIUM REMOVED ’ |
FROM STREAM ENTERING DECAY

240 — —_—

TANK ;

Pa PROCESSING CYCLE TIME =3 days J
200 |——— — — — Pa PROCESSING CYCLE TIME=1 day — E—

160

NO REMOVAL
{3-day CYCLE)

e}
O

CONCENTRATION x 10% (mole fraction)
)
O
|

AVERAGE REACTOR PROTACTINIUM

40

0
0.0707 04 0.2 0.5 1.0 2.0 50
(o / FMB__,) x 100 (%)

opt

Fig. 23.4. Variation of Average Reactor Protactinium Concentration with Standard Deviation of Bismuth Flow

Rate for Cases with Stabilization by Fluorination and Without Stabilization.

centration of 24.5 ppm is obtained. A protactinium In choosing the optimum protactinium isolation
isolation system having a one-day processing system, the relative costs associated with higher
cycle time with no uranium removal produces processing rates, with closer process control, and
roughly the same average reactor protactinium with increased uranium removal must be con-
concentration as a system having a three-day sidered as well as the value of obtaining a given
cycle time with 1% removal of uranium from the average reactor protactinium concentration.

salt stream entering the decay tank for standard
deviations larger than about 0.35%.
24. Continuvous Fluorination of Molten Salt

B. A. Hannaford

Uranium present in the fuel stream of an MSBR
must be removed prior to the distillation step
since UF , would not be completely volatilized
during this operation. Equipment is being de-
veloped for the continuous removal of UF, from
a salt stream by countercurrently contacting the
salt with F in a salt-phase-continuous system.
The equipment will be protected from corrosion
by freezing a layer of salt on the vessel wall,
the heat necessary for maintaining molten salt
adjacent to frozen salt will be provided by the
decay of fission products in the salt stream.
Recent development work has been directed toward
demonstrating operability of a frozen-wall fluori-
nator using countercurrent flow of molten salt and
an inert gas.

The experimental equipment consisted of a
S-in.-diam, 8-ft-high column fabricated from sched
40 nickel pipe (Fig. 24.1). An internal heat source
consisting of three Calrod heaters contained in a
J-in.-diam sched 40 Inconel pipe is used to simu-
late a volume heat source in the molten salt. Two
sets of internal thermocouples located near the
center of each of two test sections indicated the
radial temperature gradient, from which one could
infer the location of the interface between molten
and frozen salt. Each test section was inde-
pendently cooled by air flowing through spirally
wound 3/8--in.-- diam nickel tubing. Additional Cal-
rod heaters were wound on the external surface of
the fluorinator to provide auxiliary heat during
heatup and to provide control of temperatures at
the ends of the column. A 66-34 mole % LiF-ZrF
mixture, which has a liquidus of 595°C and a
phase diagram similar to the LiF-BeF  system,
was metered from the feed tank for periods as
long as 5 hr, which allowed collection of data for
a 1~ to 2-hr period of steady-state operation.

252

L. E. McNeese

The principal objective of the experiments,
demonstrating that a layer of frozen salt could be
formed and maintained under approximate operating
conditions, was achieved. Experimental condi-
tions are compared in Table 24.1 with reference
conditions for a fluorinator having a salt through-
put of 15 ft3/day and an inlet uranium concentra-
tion of 0.8 kg per cubic foot of salt with a 50% F,
utilization.

In general the frozen wall thickness and tem-
perature profiles in the frozen salt were in good
agreement with the values predicted by relations
obtained by assuming purely radial heat flow from
a volume heat source. Frozen wall thickness
ranged from 0.3 to 0.8 in. The effect of heat
generation in the layer of frozen salt was not
simulated in these experiments.

The thermal conductivity of the frozen salt was
calculated for each run from the experimentally
determined temperature gradient and the measured
heat flux; the relative agreement of calculated
values was taken to be indicative of consistency
of the experimental data. Thermal conductivities
calculated from the upper test section data were
closely grouped around 0.75 Btu hr~?! ft—! (°F)~!;
however, values from the lower section were more
widely scattered and were generally about 100%
higher.

The heating-cooling system used on the column
produced some variation in extemal wall tempera-
ture (and hence frozen wall thickness); in a typi-
cal run the difference between salt liquidus and
wall temperature ranged from 85 to 140°C.

Protection of the fluorine inlet nozzle from cor-
rosion is an anticipated problem associated with
operation of a frozen-wall fluorinator. A possible
solution incorporated in the present system con-
sists in introducing the fluorine through a short
253

ORNL- DWG 68-3340

 SALT

Y

}lNTERNAL
THERMOCOUPLES,
——) TEST SECTION A

5-in. SCHED 40
NICKEL PIPE

COOLING
AIR

VENT ARGON

INTERNAL
___}THERMOCOUPLES,
TEST SECTION B

ARGON

ufi"CA'—ROD HEATERS IN |_ ] —
3/,-in, INCONEL PIPE

A b 8 -

SALT RECEIVER FROZEN WALL FLUORINATOR FEED TANK
T = 625°C T=600°C T £ 625°C

Fig. 24.1. Experimental Equipment for Studying Formation of Frozen Salt Layer for Corrosion Protection.

Height of column, 8 ft; height of each test section, 1.75 ft.,
Table 24.1. Comparison of Experimental and
Reference Conditions for Fluorination of

15 £13 of Molten Salt per Day®

254

Experimental Reference
Salt flow rate, liters/hr ™ 3.3 17.7
Gas flow rate, standard 0.5~2.0 (Ar) 2.0 (Fz)
liters/min
Heat flux, w per foot of 600--1600 ~2000

column height

415 fts/day corresponds to processing fuel stream of
a 1000 Mw (electrical) two-fluid MSBR.

section of 3-in.-diam pipe which intersects the
fluorinator at a 45° angle, as shown in Fig. 24.1.

The inlet section would be protected from corro-
sion by a layer of frozen salt as in the fluorinator.
Tests with the present system indicate satis-
factory operation when the surfaces of the inlet
section are covered by a layer of frozen salt pro-
duced by maintaining wall temperatures below the
salt liquidus. Heat is supplied to this section in
the present case as a result of turbulence in the
molten salt caused by bubbles; in an actual case,
heat would be generated in the salt as a result of
fission product decay. This method of gas intro-
duction appeats to be feasible, although it will
not produce small-diameter gas bubbles.

In future experiments, the heat flux will be in-
creased to the reference value, and salt flow rate
will be increased to approximately 50% of the
reference rate.
25. Relative Volatility Measurements

by the Transpiration Method

F. J. Smith

Liquid-vapor equilibrium data necessary for the
evaluation and development of the possible dis-
tillation step in the processing of two-fluid MSBR
fuel salt, LiF-BeF, (66-34 mole %), are being
determined by the transpiration method. In the
absence of any information regarding complex
molecules in the vapor phase, the partial pres-
sures of LiF, BeF,, and solute fluorides were cal-
culated on the assumption that only monomers ex-
isted in the vapor. In each experiment, the ap-
parent partial pressures P could be described
adequately by linear expressions,

log P (mm Hg) = A — B/T (°K) ,

in which A and B were constants over the tem-
perature range investigated, 900 to 1050°C. The
data are summarized in Table 25.1. Relative
volatilities are also included, as well as effective
activity coefficients. For a multicomponent sys-
tem, the activity coefficients of component 4 at
each temperature are given by

Ya= T
A ?
NA Pf;
where NA is the mole fraction of A in the melt and

Pi is the vapor pressure of pure A. Relative
volatility is defined by

Ya/Vp
AB XA/XB ’
where a , _ is the relative volatility of A with re-

AB
spect to B, y is the mole fraction of the designated

component in the vapor phase, and x is the mole
fraction in the liquid phase.

C. T. Thompson

L. M. Ferris

Data for three LiF-BeF2 solutions are given in
Table 25.1. The composition LiF-BeF  (90-10
mole %) is approximately the one expected in the
still pot during vacuum distillation and gives LiF-
BeF, (66-34 mole %) as the condensate.! The
relative volatilities of BeF with respect to LiF
obtained in experiments with LiF-BeF , binary
systems are in reasonable agreement with those
reported by Cantor,? who also used the transpira-
tion method. For example, Cantor obtained values
of 4.28 and 3.75 at 1000°C for LiF-BeF (85-15
mole %) and LiF-BeF, (90-10 mole %) respectively;
the corresponding values from our work were about
3.8 and 3.77 (Table 25.1). The value obtained with
LiF-BeF, (90-10 mole %) is somewhat lower than
the average value of 4.71 obtained by Hightower and
McNeese,! who used an equilibrium still method,
and is higher than the values obtained when the
salt contained small amounts of RbF, CsF, ZrF4,
or UF, (Table 25.1). This scatter in values is
not surprising when one considers that small
variations in the composition of the liquid and/or
vapor lead to large changes in the relative vola-
tility. For example, Hightower and McNeese! re-
ported that the vapor in equilibrium with LiF-BeF
(90-10 mole %) at 1000°C was LiF-BeF, (66-34
mole %). This gives a value for

34/66
- 10/90

a

lj. R. Hightower, Jr., and L. E. McNeese, Measure-
ment of the Relative Volatilities of Fluorides of Ce,
La, Pr, Nd, Sm, Eu, Ba, Sr, Y, and Zr in Mixtures of
LiF and BeFQ, ORNL-TM-2058 (January 1968).

’R. B. Briggs, MSR Program Semiann. Progr. Rept.
Feb. 28, 1966, ORNL-3936, p. 128.
256

Table 25.1. Apparent Partial Pressures, Relative Volatilities, and Effective Activity Coefficients
in LiF—Ber—Meml Fluoride Systems

Apparent Partial

Effective Activity Relative Volatility

Salt Composition (mole %) a
Species Pressure Coefficient at with Respect to
LiF BeF2 3d Component A B 1000°C LiF at 1000°C
86 14 LiF 8.497 11,055 1.60
BeF, 7.983 10,665 4.42 x 10~2 3.82
90 10 LiF 7.604 10,070 1.30
BeF 8.707 11,884 3.55 x 102 3.77
95 5 LiF 8.804 11,505 1.30
BeF 11.510 15,303 4.33 x 102 4.40
90 10 0.02 UF, LiF 9.481 12,386 1.33
BeF, 9.339 12,411 5.96 x 102 6.19
UF, 4.361 12,481 7.36 x 103 2.9 x10™2
89.6 9.9 0.5 UF, LiF 8.384 10,987 1.34
BeF, 7.421 10,112 4.65 x 102 4.78
UF 6.686 13,443 1.09 x 102 4.2 x 102
86.4 9.6 4.0 UF, LiF 10.790 13,992 1.55
BeF, 10.177 13,726 3.84 x 102 3.42
UF, 10.272 16,786 1.25 x 10~2 4.2 X 1072
90 10 0.09 RbF LiF 8.286 10,811 1.47
BeF, 6.596 10,552 3.11 x 102 2.93
RbF 5.187 8,907 2.19 24.7
89.9 10 0.03 CsF LiF 9.654 13,459 1.99
BeF, 8.310 11,313 4.07 x 102 2.82
CsF 0.819 3,375 1.17 95.1
90 10 0.083 Z1F, LiF 7.915 10,358 1.41
BeF 7.167 10,070 2.83 x 1072 2.77
ZiF, 13.095 20,382 5.39 x 10~*4 2.19

“Log p(mm) = A — B/T (°K). Temperature range: 900 to 1050°C.

fluorides existed only as monomers in the vapor.

On the other hand, Cantor? reported that the
vapor in equilibrium with LiF-BeF  (88-12 mole
has the composition LiF-BeF , (67-33 mole %),
which leads to a value for

33/67

a= =3.6
12/88

The partial pressure data obtained in our work
are incompatible with the total pressure data of
Cantor.® He reports the total pressure of LiF-

%)

BeF, (90-10 mole %) to be 1.8 mm Hg at 1000°C.

It was assumed that LiF, BeF2, and the solute

For the same system at 1000°C, we obtained
P =0.55and P = 0.23 mm Hg, cor-

LiF BeF,
responding to a total pressure of 0.78 mm (as-
suming that no association or dissociation oc-
curred in the vapor phase). Association (known
to occur* in the vapor phase above pure LiF) or

3W. R. Grimes, Reactor Chem. Div. Ann. Progr. Rept.
Dec. 31, 1965, ORNL-3913, p. 24.

4R.S. Scheffee and J. L. Margrave, J. Chem. Phys.
31, 1682 (1959).
complexation (observed mass spectrometrically in
the vapors above LiF-BeF | solutions®) would
make the total pressure even lower than that pre-
dicted by our transpiration experiments. Our re-
sults are also in disagreement with those of
Cantor at several other LiF—BeF2 compositions.
The reason for this disparity is not obvious.

The activity coefficients for BeF, are in good
agreement with the values obtained by Kelly?
using distillation data and assuming unit activity
for LiF.

Experiments performed with an LiF-BeF, (90-
10 mole %) solvent containing UF4 at concentra-

257

tions of 0.02, 0.5, and 4.0 mole % (by analysis),
respectively, gave relative volatilities of UF
with respect to LiF (Table 25.1) which confirmed
the earlier assumption that adequate uranium re-
covery cannot be achieved in the distillation step.
The experimental relative volatilities of RbF and
CsF with respect to LiF are within a factor of 2
of the theoretically predicted values (assuming
ideal behavior of the system). Hence, as pre-
viously assumed, these two fission products will
probably set the discard rate. The relative
volatility of ZrF ,, present at a concentration of
0.083 mole %, was 2.19.
26. Distillation of MSRE Fuel Carrier Salt

L. E. McNeese

We have described previously® equipment for
study of low-pressure distillation of MSRE fuel
carrier salt which includes a 48-liter feed tank,

a 12-liter still pot, a condenser, and a 48-liter
condensate receiver. The equipment has been in-
stalled and checked out in a test facility to per-
form nonradioactive experiments prior to opera-
tion with irradiated MSRE fuel carrier salt. Two
experiments have been made to date. During non-
radioactive operation, four 48-liter batches of
MSRE fuel carrier salt (65-30-5 mole % LiF-BeF -
ZtF ) will be distilled at 1000°C and at a pressure
of about 1 mm Hg. The thermal insulation and
heaters will then be removed from the still, and a
thorough examination of the equipment will be
made. After examination the still will be moved
to a cell at the MSRE site for distillation of a
48-liter batch of fluorinated fuel salt from the
reactor.

The first distillation run was completed in 83
hr using approximately 48 liters of MSRE fuel
carrier salt. At the beginning of the run, 9.4
liters of salt was transferred to the still pot, and
the condenser pressure was lowered to 2.0 mm Hg.
Startup of the equipment. was hampered by un-
satisfactory operation of the liquid level probes
in the still pot. It is believed that argon dis-
solved in the molten salt formed bubbles on the
liquid level probes initially as the pressure was
reduced. The system was held at 900°C and 2 mm
Hg for approximately 4 hr, after which time the
probes performed adequately. The system was
then operated with a still pot temperature of
900 to 950°C and a condenser pressure of 0.65 to
2.0 mm Hg for approximately 40 hr. Salt was trans-

IMSR Program Semiann. Progr. Rept. Feb. 28, 1967,
ORNL-4119, p. 211.

J. R. Hightower

ferred to the still pot periodically in order to main-
tain a volume of about 9 liters. This operating
period reduced the concentration of BeF in the
still pot from the initial concentration of 30 mole
%. The still pot temperature was raised to 990°C,
and distillation rates were measured during a 40-hr
period at condenser pressures of 0.5, 0.3, and
0.055 mm Hg. Results at these conditions are
summarized in Table 26.1. Distillation rates were
calculated from the rate of change of salt level in
the feed tank and condensate receiver. During the
last 40 hr of operation the still liquid level and
salt feed rate were controlled automatically.
Equipment operation was smooth. Four condensate
samples were taken and have been submitted for
analysis. At the end of the run, approximately 8
liters of the initial salt mixture was used to flush
the high-melting salt from the still pot in order to
produce salt having a liquidus of less than 700°C.
For the second run, approximately 45 liters of
MSRE fuel carrier salt (65-30-5 mole % LiF-BeF -
ZiF ) was transferred to the feed tank. Salt was
transferred to the still pot to yield a volume of
about 9 liters. A still pot temperature of 900°C

Table 26.1. Distillation Rates and Operating
Conditions for Distillation of MSRE Fuel
Carrier Salt

Still Pot Run

Temperature Time

Condenser Distillation Rate

1

Pressure

kg/nr ft° day ™! £t~

(mm Hg) °c) (hr)
0.5 1.54 1.15 990 7.2
0.3 1.60 1.20 990 9.5
0.055 1.67 1.20 a90 16.5
0.07 2.02 1.51 1005 29.8

and a condenser pressure of 2 mm Hg were held
for 1.5 hr, after which (over a 13-hr period) the
still pot temperature was slowly increased to
1005°C and the condenser pressure was decreased
to 0.07 mm Hg. These conditions were maintained
for about 30 hr. The distillation rate during this
period is given in Table 26.1. Still pot liquid
level and salt feed rate were controlled auto-
matically during the entire run. Minor difficulty
was experienced in taking condensate samples
when small amounts of salt vapor condensed in
the 1.5-in. line through which samples are taken.
Two condensate samples were obtained.

Under the operating conditions in the still, the
distillation rate is controlled by the condition that
pressure drop in the passage connecting the vapor-

259

ization and condensation surfaces equals the dif-
ference between the vapor pressure of salt in the
still pot and the pressure at the lower end of the
condenser. For this reason, distillation rate should
be essentially independent of condenser pressure
for condenser pressures much lower than the salt
vapor pressure (1.0 to 1.5 mm Hg), as was ob-
served. With condenser pressures much lower than
the salt vapor pressure, distillation rate should be
proportional to salt vapor pressure and hence
quite dependent on still pot temperature. A 21%
increase in distillation rate was observed as the
still pot temperature was increased from 990 to
1005°C; the corresponding increase in salt vapor
pressure is 28%.
27. Protactinium Removal

J. S. Watson

A higher breeding ratio can be achieved in
molten-salt reactors if a low protactinium concen-
tration is maintained in regions of high neutron
flux so as to avoid parasitic capture by the prot-
actinium before it decays to 233U. One con-
ceptual MSBR design is based on a two-fluid
concept, with the fertile stream circulating in
separate channels through the high-flux core
region as well as through the blanket surrounding
the core. The protactinium concentration in the
fertile stream can be kept low by processing the
stream rapidly for removal of protactinium soon
after it is formed, leaving little time for neutron
capture. Promising protactinium removal processes
based on reductive extraction using liquid bismuth
have been analyzed to evaluate their feasibility.

F, (SALT)

F3 (METAL)

Ho

EXTRACTOR 1

F, (SALT) }T F, (METAL)
FERTILE BLANKET
SALT REDUCER

BLANKET N1 Fr_2(SALT)

\

from a Two-Fluid MSBR

M. E. Whatley

The flowsheet chosen for study is shown in Fig,
27.1. The modification shown in Fig. 27.2 could
be used if development of a reliable reducer
proves more difficult than expected. In both flow-
sheets, fertile salt from the reactor (stream 1) is
contacted with liquid bismuth (stream 3) contain-
ing thorium (0.003 mole fraction). The bismuth
contains a lithium concentration such that no
thorium or lithium transfers between the metal and
the fertile salt. The metal stream from the ex-
tractor containing protactinium flows to an oxidizer
which converts the thorium, protactinium, and
lithium to fluorides. The bismuth is not oxidized
and is recycled to the extractor after the proper
amounts of thorium and lithium reductant are
added by electrolytically reducing all or part of a

ORNL-~DWG 68-787A

SALT

F, (SALT)

EXTRACTOR II

FLUORINATOR

—

T Fa_ o (METAL)

Fig. 27.1. Reference Flowsheet for Removing Protactinium from a Two-Fluid MSBR.

260
261

F, (SALT)
F, (METAL)

H

BLANKET Fy (SALT)
SALT  FERTILE BLANKE
SALT REDUCTANT
ADDITION
TO WASTE =t
7-2

F (SALT)

ORNL—-DWG 68~ 788A

NATURAL LiF OR
NaF ADDITION

Fs (SALT)

Li® Th®

| ko,
EXTRACTOR 11 EXTRACTOR 1

Fa_op (METAL)

Fig. 27.2. *'Throwaway’’ Flowsheet for Removing Protactinium from a Two-Fluid MSBR Without Use of an

Electrolytic Oxidizer-Reducer.

recycle salt stream from the decay tank (stream 7).

The salt mixture formed by oxidation of stream 4
would have an undesirably high liquidus; the
liquidus is lowered in the preferred flowsheet
(Fig. 27.1) by recycling salt from the decay tank
through the reducer and into the oxidizer. To
minimize the protactinium return to the extractor,
salt recycled from the decay tank is contacted
with the metal stream leaving the first, or main,
extractor. This transfers most of the protactinium
to the metal stream, which then flows to the
oxidizer; there the protactinium is oxidized and
returns to the decay tank without passing through
the main extractor.

In the modification shown in Fig. 27.2, fresh "Li
and thorium metal are added to bismuth to form
stream 3; no reducer is used. The metal stream
from the first extractor is contacted with a recycle
from the decay tank, as in Fig. 27.1, before enter-
ing the oxidizer, but the recycle salt is sent to
waste. Lithium fluoride is added to the oxidizer
to reduce the liquidus of the resulting salt, and
since this material will not return to the reactor
blanket, natural lithium can be used. Another
alkali metal fluoride (e.g., sodium fluoride) could
be used if it does not interfere with protactinium
transfer in the second extractor.

To evaluate the relative merits of these flow-
sheets, flow rates and compositions of all streams
were calculated over a wide range of conditions.
The lithium and thorium concentrations in the
blanket were fixed at 0.72 and 0.28 mole fraction,
respectively, and the protactinium generation
rate was fixed at 10.6 g-moles/day. [This cor-
responds to a 2200 Mw (thermal) reactor.] All
calculations were based on a processing rate of
5.81 x 10° g-moles/day of blanket salt, which
is 4750 ft°/day (25 gpm, approximately two blanket
volumes per day). Results are shown in Figs. 27.3
and 27.4.

Figure 27.3 shows the required metal flow rate
in gram-moles per day (multiply by 3.9 x 1075 to
get gallons per minute) as a function of the number
of stages in the extractor for three protactinium
concentrations in the blanket (5 x 10~ ¢, 10 x
107, and 20 x 10~ ° mole fraction). These blanket
concentrations may be compared with the value of
approximately 80 x 10~ % mole fraction which re-
sulted in the blanket of a ‘‘no-Pa-removal’’ re-
actor system which has been considered.’ The

1R. B. Briggs, Summary of the Objectives, the Design,
and a Program of Development of Moiten-Salt Breeder
Reactors, ORNL-TM-1851 (June 12, 1967).
262

ORNL-DWG 68-6098

fl
b
I
[

=

o

~

~l -

o

E BLANKET CONCENTRATION

= K L mole fraction
[

" | 6

: Eesniees s

= 1 o

% -

& 14072

@ l

LE

R WL \ |
-
|
2

5 A0 20 50
N, NUMBER OF STAGES

Fig. 27.3. Metal Rate Required as a Function of

Number of Stages in Extractor.

metal rate is particularly important for two reasons:
it affects the extractor design and bismuth inventory
and it sets the reducer capacity (or "Li and

thorium consumption in the ‘“‘no-reducer’’ flow-
sheet). A low metal rate is desirable because the
reducer anode surface may be expensive to main-
tain.

Raising the protactinium concentration in the
blanket raises the equilibrium (or maximum) metal
loading and lowers the required bismuth rate. In-
creasing the number of stages in the contactor
also improves the performance of the extractor and
lowers the required metal rate; however, Fig. 27.3
shows that the gain will be relatively small for
more than two or three stages. Required metal flow
rates will be about 0.3 x 10° g-moles/day (0.1
gpm).

In the reference flowsheet the salt recycle rate
to the oxidizer is determined by the desired liquidus
temperature. Figure 27.4 shows the decay tank
composition as a function of the ratio of the salt
recycle rate to the metal rate F . With an infinite

ORNL —DWG 68— 793A

0.4

0.2

0.02

Q.04
0.008

0.006
0.58

0.68
LiF CONCENTRATION [N DECAY TANK (mole fraction )

0.60 0.62 064 0.66 0.70 0.72

Fig. 27.4. Ratio of Salt Rate from Oxidizer to Metal

Rate as a Function of LiF Concentration in Decay Tank.

recycle rate, the lithium composition only ap-
proaches the blanket composition, so that it is
necessary to operate with a slightly lower lithium
composition in the decay salt than in the blanket.
The no-reducer flowsheet has the option of a
lower reductant composition, which increases the
lithium-to-thorium ratio in the metal. This allows
a lower salt recycle but requires higher metal
circulation rates,

The volume of the decay tank is set by the
concentration of protactinium desired in the tank
but is subject to limitations. The tank must hold
sufficient protactinium that 10.6 g-moles decay in
the tank per day (neglecting decay outside the
tank). Under the conditions of interest, the tank
volume will be limited by heat-removal considetra-
tions rather than by the maximum concentration
which could be achieved. There will be 6.7 Mw
of heat generated in the decay tank from protactinium
decay, which suggests a minimum tank volume of
a few hundred cubic feet.
Important considerations unique to the no-reducer
flowsheet are Li and thorium consumption and
protactinium losses. The “Li and thorium con-
sumptions reflect both material transferred to
fertile salt in extractor I (0.5 to 5% of material
fed) and the material remaining in the effluent
metal stream from this extractor, which is ulti-
mately sent to waste. The protactinium loss re-
sults from incomplete removal of protactinium from
the salt stream flowing through extractor II. The
"Li consumptions are 5.7 and 1.4 1b/day, re-
spectively, for blanket protactinium concentra-
tions of 5 x 10~ % and 20 x 10~ ® mole fraction. The
corresponding thorium consumptions would be 150

263

and 38 Ib/day. Consumption of these quantities
of "Li and thorium would be tolerated, but the
costs are significant. With a 400-ft*® decay tank,
the protactinium loss would be less than 0.02% of
that produced in the reactor.

The reference flowsheet shown in Fig. 27.1 is
the preferred flowsheet, and further studies and
development work for two-region processing should
be aimed at this system. This flowsheet provides
a minimum-size reducer and bismuth rate while
placing little restriction on the decay tank volume;
the decay tank volume is governed largely by heat-
removal considerations.
28. Recovery of Uranium from

MSRE

Fuel Salt by Fluorination

G. 1. Cathers

M. R. Bennett

C. J. Shipman

The future schedule of MSRE operation includes
removal of the uranium (about 30% eariched in
235U) from the carrier salt, LiF-BeF ,-Z1F, (65-
30-5 mole %), by fluorination. The purified carrier
salt will be reused for the subsequent operation
of the MSRE with ?*°UF,. Fluorination of UF,
dissolved in a molten fluoride salt, with the at-
tendant evolution of volatile UF, is quite corro-
sive to all known metallic containers. It is anti-
cipated that significant amounts of chromium, iron,
and nickel will be put into the salt by corrosion,
making necessary its cleanup by hydrogen treat-
ment and other means. A series of small-scale
fluorination tests is being made with simulated
MSRE salt to study the effects of temperature,
fluorine concentration, and fluorine flow rate on
the rate of uranium volatilization and the rate of
corrosion of Hastelloy N.
perimental results are being obtained on the
volatilization of chromium and molybdenum fluorides
(which are corrosion products) and on the behavior
of volatile fission products (Ru, Nb, I, Te) in the
fluorination of short-decay fuel. Knowledge of
fission product behavior is essential to the de-
velopment of a special fluorination procedure to
be used by the Analytical Chemistry Division in
making precise uranium analyses of fuel salt.

In addition, some ex-

28.1 FLUORINATION-CORROSION STUDY

The fluorination-corrosion tests are being made
with MSRE-type salt in a 1.87-in.-ID Hastelloy N
reactor under conditions roughly approximating
those being considered for use in the 49-in.-ID
fluorination tank at the MSRE site. The uranium

264

is being volatilized out of the salt with an inter-
mediate inert-gas sparging period to simulate
changing of the NaF sorbent beds used for re-
covery of the UF,. A gas flow rate of 146 ml/min
(STP) is being used in the 1.87-in.-ID reactor to
simulate the flow rate of 100 liters/min in the
49-in.-ID reactor.

In the larger reactor, more uranium is involved,
and the surface-to-volume ratio is lower. Hence
the change in oxidation state of the salt during
the course of UF volatilization may be different
in the two systems.

Two tests have been made, one at 450°C and one
at S00°C, each with 340 g of LiF-BeF ,-ZrF ,-UF,
(63-32-5-0.8 mole %). Salt samples were taken
during the runs at intervals (after He sparging)
determined by the necessity of changing a 12-g
NaF trap used for recovery of the UF . In each
test, 50% F,~50% He (146 ml/min) was used
initially, followed by pure fluorine (146 ml/min)
at the end of the run to ensure complete uranium
recovery. In another test at 500°C, pure fluorine
was used throughout. A trap containing 2 M
KOH was employed to sorb unused fluorine and
volatile chromium and molybdenum compounds
that passed through the NaF trap.

As shown in Table 28.1, the UF  volatilization
rate and the degree of fluorine utilization were
higher at 500°C than at 450°C. The amount of
chromium leached from the Hastelloy N reactor
increased with temperature. The effect of tem-
perature on the solubilization of iron and nickel
from the reactor is not as clear (the dip tube,
thermocouple well, and reactor bottom were made
of Ni-200). Figure 28.1 shows the relation be-
Table 28.1. Uranium Volatilization ond Corrosion Results in Fluorination Tests with

265

MSRE Salt in Hastelloy N at Two Temperatures

Fluorine Total Corrosion Volatile Material Fluerine
Temperature Concentration Time (mg in salt?®) Utilization
o) %) (hr) U Cr Efficiency
Cr Fe (%) (mg) @)
450 50 0.5 13 4 5.4 4.0
50 2.0 42 35 14.2 2.1
50 3.5 93 60 44.7 7.5
50 5.0 118 73 68.0 5.6
100 6.0 103 82 92.2 2.2
100 8.0 143 152 100 2.4 0.7
100 12.0 128 93 5.5
100 16.0 117 96 14.7
500 50 1.0 75 49 18.5 7.6
50 1.75 153 05 38.5 11.0
50 2.25 194 122 55.1 13.8
100 2.75 224 126 83.4 11.7
100 6.75 150 139 100 15 1.8
100 10.75 140 180 27
500 100 0.75 74 17 28.7 7.1
100 1.25 112 48 59.4 11.4
100 1.58 139 48 90,7 17.6
100 2.58 136 71 98.9 1.5
100 6.58 125 71 100 9 0.1
100 10.58 102 82 12

dCorrected for initial values at zero time.

tween chromium and iron leaching and uranium
volatilization for the third run (with pure fluorine);
corrosion appears to occur predominantly in the
uranium volatilization period.

Various estimates of the corrosion rate can be
calculated from the chromium data using different
time periods (Table 28.2). There is good agree-
ment between the rate calculated from the period
in which most of the chromium was leached and
that over the period where most of the uranium was
volatilized. The estimates for the complete runs
are somewhat arbitrary because fluorination was
continued after the uranium had been removed to
observe chromium behavior,

Volatility of chromium and molybdenum was
also observed in these runs. A chromium com-
pound (probably CrF) appeared to volatilize only

after most of the uranium had been removed from
the salt (Table 28.1). This chromium compound
was effectively trapped by NaF at 25°C, forming
an orange complex compound. Molybdenum (prob-
ably as MoF ) volatilized to a large extent through-
out each run and, as expected from the dissocia-
tion pressure of MoF .2NaF, was not completely
recovered in the 25°C NaF trap. Most of it passed

through to the KOH scrub solution.
The fluorine utilization efficiency figures (cal-

culated from UF ; output and fluorine input) sug-
gest that the utilization was highest during the
first 2 hr of fluorination; this is not strictly true,
since the method of calculation does not take into
account the fluorine consumed in changing the
average oxidation number of the uranium. Also,
the data do not imply that a definite consumption
Table 28.2. Estimates of Rates of Corrosion of Hastelioy N in Fluorination Tests

Fluorine Corros ion Rate (mils/hr) in Period of:
Temperature .
P Concentration Maximum Rate of 80—95% U
%) Cr Change Volatilization Complete Run
450 50 and 100 0.19 0.12 0.05
500 50 and 100 0.58 0.45 0.12
500 100 0.50 0.49 Q.07

ORNL-DWG 68-6099

100

URANIUM VOLATILIZED (%)

METAL IN SALT (ppm)

0,CORRECTED FOR Cr
VOLATILIZED FROM MELT .l

0 1 2 3 6 10 14
FLUORINATION TIME (hr)

Fig. 28.1. Relationship Between Corrosion of Hastel-
loy N and UF6 Volatilization in Fluorination of MSRE-
Type Salt at 500°C.

of fluorine has to occur before any UF is evolved.
In recent fluorination tests, some volatilization

of UF . occurred when the uranium oxidation num-
ber in the salt was as low as 4.3, This was cal-
culated on the basis of complete consumption of
fluorine introduced in the first 1.5 min of fluorine
flow.

28.2 FISSION PRODUCT BEHAVIOR -
ANALYTICAL ASSISTANCE PROGRAM

In the currently proposed analytical procedure
for the determination of uranium in MSRE salt,
which involves fluorination as the first step, a
precision of £0.1% is needed. Since it is anti-
cipated that the use of remotely controlled hot-
cell coulometric instrumentation will introduce an
inherent error of +1%, decontamination of the UF 6
NaF complex product from the fluorination step to
a level appropriate for analysis in a low-level
radiation area will be essential. In addition,
uranium losses in the waste salt and fluorination
system (traps, etc.) must be less than 0.1 wt %.
Results of preliminaty tests in which uranium has
been fluorinated from MSRE-type salt containing
volatile fission product activity have been en-
couraging.

Four fluorination tests have been completed
using specially prepared MSRE-type salt samples
supplied by the Analytical Chemistry Division.
The samples consisted of copper capsules each
containing about 46 g of LiF-BeF ,-ZrF ,-UF,
(65.4-29.4-5.0-0.2 mole %). Each salt sample was
melted out under an inert atmosphere into a suit-
able nickel fluorinating vessel and then was spiked
with about 0.1 mc of '23Ru, °5Nb, 131, or 132Te.
Solutions of these tracers were first evaporated
to dryness with the corresponding carrier (about
40 to 50 mg) onto 0.5-g quantities of LiF. Fluori-
nation was carried out by heating the salt to
550°C in an inert atmosphere and then passing a
mixture of 25% ¥ ,—75% N, through the melt for
30 min followed by pure fluorine for an additional
30 min at 100 ml/min (STP). The product gas
stream (containing F,, UF ., and volatile fission
product activity) was passed through a 20-g NaF
267

trap at 400°C to remove the fission products and
then into a 10-g NaF trap at 25°C, where the UF
product was sorbed. An additional 5-g NaF trap
was inserted downstream for determination of any
uranium breakthrough.

Based on a makeup value of 1.185 wt % uranium
in each of the initial salt samples, the total
uranium found in three tests corresponded to re-
coveries of 99.27, 100.22, and 97.65%. However,
corresponding uranium losses in the tests, based
on analyses of the traps and waste salt, were,
respectively, 0.034, 0.059, and 0.076 wt %, indicat-
ing actual recoveries greater than 99.9% (see Table
28.3).

Decontamination factors obtained for *°3Ru,
95Nb, and '3?Te during the fluorination step have
been greater than 10°, and thus these nuclides
apparently do not make a significant contribution
to the overall activity level of the UF , product.
The relative amounts volatilized from the salt

during fluorination were, respectively, about 1072,
10, and greater than 99%, with corresponding
material balances of about 95, 10, and 0.1%. Es-
sentially all the ruthenium and niobium that was
volatilized was trapped on the 400°C NaF trap;
however, no appreciable sorption of tellurium was
detected (Table 28.4). Attempts to achieve an
adequate decontamination factor (greater than 10°)
for 131 have been unsuccessful so far. In the
first two tests, where the UF , sorption trap con-
sisted of 10 g of NaF at 25°C, decontamination
factors of 14 and 8 were obtained. In the third
test, decreasing the amount of NaF to 5 g and in-
creasing the sorption temperature to 100°C re-
sulted in a higher, but still inadequate, decontami-
nation factor of 35 (Table 28.5). In this test the
measurement of the temperature of the sorption
bed may have been highly in error; therefore, a
duplicate test will be made.

Table 28.3, Uranium Recoveries and Losses During Fluorination from MSRE-Type Salt and Sorption on NaF

Uranium Uranium Uranium Losses (% of total found)
a
Run No. Makeup Value Recovery Trap Trap Waste Sait Total
(Wt %) (Wt %0) No. 1b No. 2°¢
1 1.185 99.27 0.012 0.009 0.013 0.034
2 1.185 100.22 0.051 0.006 0.002 0.059
1.185 97.65 0.026 0.022 0.022 0.076

@Based on analysis of all traps and scrubber solutions.

b4 00°C NaF trap.
€25°C NaF backup trap.

Table 28.4, Behavior of ]03Ru, 95Nb, and ]32Te During Fluorination of Uronium from MSRE-Type Salt at 550°C

Total Activity (dis/min)

Amount of UF , Product

0 Nuclides Decontamination
Nuclide MSRE Salt 400" C NaF Waste UF . Product iqs
6 Volatilized Factor
(Initial) NaF Trap Backup Trap Salt Trap )
103pu ~2 %108 <103 <10° 1.9 x 108 <10° 103 >10°
SNb ~9 % 108 3 x 107 <103 9% 107 <103 10 >10°
1320 ~2x 108 <103 <103 4.3 % 10° <103 >99 >10°

268

Table 28.5. Behavior of '3'1 During Fluorination of Uranium from MSRE-Type Salt at 550°C

Total Activity (dis/min) 131
o Decontamination
Run No. MSRE Salt 400-C Backup KOH Wa ste UF6 Volatilized Factor
(Initial) NaF Trap Trap Scrub Trap Salt Trap (%)
1 ~9 % 108 ~103 1.3 x 108 <10° 1.4 x 107 >99 14
2 ~2x 108 1.8x10* s53%x10% 2x107 2 x 10* 2.5 % 107 >99 8
3 ~1x10° 5.6x10° 7.2x107 2.7 % 10° 2.9 x 107 >99 35

29. MSRE Fuel Salt Processing

R. B. Lindauer

The modifications to the MSRE fuel processing
facility to permit processing with a 30-day decay
time have been completed. In November, it was
decided to remove the structural-metal fluorides
formed during fluorination from the salt before re-
turning it to the reactor system. Design of a salt
filter for removing the metals after reduction has
been completed, and fabrication is in progress.
The filter will have 9 ft? of Inconel porous metal
as a filter medium. The filter element is designed
for remote replacement in case of plugging or
rupture. The filter has a capacity of about 40 1b
of corrosion products, the amount formed during 50

hr of fluorination with a corrosion rate of 0.2 mil/hr.

Another change in processing plans is the elimi-
nation of the H -HF treatment of the flush and fuel

269

salts prior to fluorination. A heated salt sample
carrier was constructed as a simpler and more
economical method for verifying the oxide analyses
of fuel salt samples.

Handling of the sodium fluoride absorbers loaded
with UF, will be simplified considerably, since
it has been learned that most of the noble-metal
fission products leave the salt during reactor
operation. Molybdenum-99 would have been the
principal radiation source with short-decayed fuel.
The problem of overheating of the high-temperature
sodium fluoride trap from ?5Nb is also eliminated.
A third problem which has been eliminated is the
discharge of tellurium during processing,

Training of part of the reactor operations group
to operate the fuel processing plant is in progress.
30. Preparation of 7I.iF-233UF4 Fuel Concentrate for the MSRE

J. W. Anderson
S. Mann
S. E. Bolt

The MSRE will be refueled with 233U in 1968.
Approximately 39.5 kg of 91.4%-enriched 233U as
"LiF-?33UF, (73-27 mole %) eutectic salt will
be required. This material will be prepared in a
shielded facility because of the high 232U content
(222 ppm) of the 233Q,

The process and equipment have been described
in the previous report.’ The following sections
describe the changes in process and equipment
since the original report and the status of the pro-
gram.

30.1 EQUIPMENT CHANGES

The MSRE personnel completed a detailed evalu-
ation of their fuel requirements and loading opera-
tions for the 33U program. The predicted opera-
tional fuel load for the reactor is 35.9 kg of
uranium (~91.4% 233U), and the contingency is
3.6 kg, giving a total of 39.5 kg of 233U to be
delivered to the MSRE as the eutectic salt. This
is a decrease of about 3 kg from earlier planning. !
At the same time, the number of enriching cap-
sules was reduced from 60 to 45; so the surplus
capsules were removed from the existing array.

The unshielded glove box used at the MSRE
for drilling the uranium enriching capsule vent
and drain holes and for weighing the capsules be-
fore loading into the sampler-enricher is inade-
quate for the highly active ?33U. These opera-
tions will now be done in the TURF before the
capsules are shipped to the MSRE. A remotely

1MSR Program Semiann. Progr. Rept. Aug. 31, 1967,
ORNL-4191, pp. 252-53.

270

E. L. Nicholson
J. M. Chandler
W. F. Schaffer, Jr.

operated drilling, weighing, and packaging box
has been built and tested for this operation. The
lifting cables and keys will be attached to the
capsules before they are filled with salt, thus
avoiding a remote welding operation. After drill-
ing, weighing, and testing, the capsules will be
placed in the special holders that are used as
part of the sampler-enricher system at the MSRE,
packaged in groups of six, and shipped to the
MSRE in a shielded carrier.

The method used for bulk additions of 235U fuel
salt from the 4-in. storage containers to the re-
actor via a heated transfer line was discarded in
favor of a simpler method for the highly active
233y salt. The bulk additions will be made from
2 1/2--ir1.-diam salt cans that are vented and have
an opening in the bottom. They are essentially
‘“giant’’ enrichment capsules that are lowered
into the heated dump tank, so that the salt melts
and drains into the tank. Four 7.0-kg uranium
cans, one 3.0-kg uranium can, one 2,0-kg uranium
can, one 1.0-kg uranium can, and two 0.5-kg
uranium cans will be delivered to the MSRE. The
cans are assembled in two clusters for filling
with salt. One cluster consists of the four 7.0-kg
cans, and the other is the remaining five small
cans plus one extra 2,0-kg can which is arranged
to hold recoverable salt heels from the salt
processing system. The capsule filling furnace
in cell G was replaced with a larger furnace that
will accommodate both the capsules and 21/2-in.-
diam salt can clusters. Other equipment changes
included a new furnace liner, a new fumace stand,
can cluster dismantling tools, and a larger work
table to accommodate both the capsule drilling
and packaging box and the dismantling of the can
clusters for shipment. The cans will be weighed
in the cell before shipment by suspending them
from a wire attached to a scale located on top of
the cell.

We have determined that the density of the
stored ?3%UO, is higher than the original esti-
mate. This, plus the decrease in the total amount
of uranium required, as noted above, will allow us
to make three batches of approximately 13.5 kg U
each instead of the four batches planned earlier.
A significant saving in operating time will re-
sult from the elimination of one of the production
batches. The 45 enriching capsules will be
filled from the first batch of salt. The remainder
of the first batch, the second batch, and part of
the third batch will be used to fill the cluster of
four 7.0-kg uranium salt cans. The small cans
will be filled last, from the remainder of the third
batch.

The 233UO3 oxide can opener and dumping box
were designed using can drawings and sample
cans from Savannah River. The cans in storage
in the pilot plant were not checked closely be-
cause of the high radiation levels of the oxide.
After the box was completed, it was found that
the cans were 1.75 in. shorter than shown by the
drawings and can samples. Since modification of
the charging box to accommodate the shorter cans
was impractical, we will cement a 1,5-in. ex-
tension on each can. The fixtures for tapering
the can end and removing the weld bead, for
pressing the extension on the cans, and for curing
the cemented joint have been built and tested.
The fabrication costs and operating problems
caused by this development are minor.

30.2 EQUIPMENT AND PROCESS STATUS

The TURF building was turned over to ORNL
on September 15, 1967, and the CPFF contractor
started preparing cell G for installation of the
process equipment and completed this by mid-
October. The CPFF work consisted in installing
the instrument panel, the gas supply and electri-
cal systems to terminals inside the cell, the cell
platform, the shielding plugs for unused pipe
slots and windows, the PAR manipulator, and the
can charging chute. The process equipment was
completed in the ORNL shops by mid-October,
and ORNL forces finished the installation of the
process equipment in cell G in January. Part of

271

the processing equipment installed on and in
cell G is shown in Figs. 30.1 to 30.3. The
equipment for adding the extensions to the oxide
cans, for dismantling the capsule and can assem-
blies, and for drilling and weighing the capsules
will be located in the northwest and northeast
comers of the cell. About 130 formal drawings
plus 150 field sketches have been issued to date
for the job. We estimate that the design work is
more than 98% complete as of February 29, 1968.
Pressure vessel, criticality, building, and radio-
chemical plant safety reviews were completed,
and the process and buildings were approved for
MSRE 233U fuel production.

Equipment checkout and minor alterations and
changes were completed, and the cold test run
with 23800 (11.6 kg ?°°U) was started January
15, 1968, Some problems were encountered with
the can opener and dumping box. After a few cans
were opened, a groove rolling device accidentally
jammed the can chuck, stripping the motor drive
gears, The gears were replaced, and the re-
maining cans were opened and dumped, though
with difficulty due to binding of the cans in the
box. The box was removed from the cell, decon-
taminated, and sent to the shop for repairs and
for alterations to improve operability. The chemi-
cal part of the run started on January 24, 1968,
with the 900°C sintering treatment followed by
hydrogen reduction of the UO, to UO,. The off-
gas filter plugged during the reduction step but
was opened by blowback and has not been a
problem since the reduction step. Treatment with
H,-HF has been under way since January 27. It
was planned to follow the progress of the con-
version of UO, to UF, by titrating the off-gas to
determine HF utilization and to periodically de-
termine the freezing point of the charge as it
approached the eutectic point. The off-gas titra-
tion was hampered by HF-water condensation in
the plastic lines to the caustic scrubber bottle
and by problems with remote manipulator opera-
tion of the conventional burets in the cell. This
rig has since been replaced by a reliable titration
system, but in the early stages of the conversion
it was impossible to get reliable results with the
existing titration system. Attempts to run freez-
ing point curves on the salt in the reaction vessel
were unsuccessful because of the high thermal
capacity of the reaction vessel and furnace as
compared with the salt charge. Samples of the
272

Fig. 30.1. Control Panel and Gas Supply for

melt taken after about 290 hr of HF-H, treatment
showed that only 5 to 10% conversion of UO, to
UF , had occurred. The H,-HF sparger tube was
removed from the reaction vessel on February 14
and was found to be corroded off at the liquid-gas
interface. It was replaced, and HF utilization rose
from zero to 70%, indicating good conversion of
U0, to UF,. After about 200 hr of additional
operation, about 71% of the UO, had been reacted.
The HF utilization, as determined by titration
with the new apparatus, has decreased in the ex-
pected manner since then. Progress of the reaction
is being followed mainly by sampling the salt at
frequent intervals.

We had problems with the remote salt sampling
box because of excessive friction between the

233,

PHOTQ 91225

Fuel Salt Preparation — Cell G, Building 7930.

salt sampler probe and the sampler box. The box
was returned to the shop for repairs and for alter-
ations to improve operability. At the end of the
month, the H2-HF treatment was still under way,
and 90 to 95% of the UO , had been converted to
UF,. The progress of the reaction has been slow
owing to poorer contacting of gas and salt in the
8-in.-diam pot as compared with the 4-in.-diam
pots used during flowsheet development, and in-
terruptions of the process due to equipment prob-
lems and for salt sampling operations. We now
estimate that the chemical part of each 233U run
will take about 28 days instead of the 17 estimated
earlier. Tentative plans are to complete the cold
run in March and prepare the three batches of
2330 salt in April, May, and June.
273

PHOTO 90483A

CAN OPENER AND
DUMPING BOX

»

SALT TRANSFER
LINE

OFFGAS
FILTER

P REACTION VESSEL
= FURNACE

SALT SAMPLING
BOX ‘

Fig. 30.2. Main Reaction Vessel Furnace and Auxiliary Equipment for 233y Fuel Salt Preparation — Cell G,
Building 7930.
274

PHOTO 90464A

OFF GAS SCRUBBER " AN W L OFF GAS TITRATION
BOTTLE ! | ' EQUIPMENT

a,

ALT TRANSFER LINES

-mm

e /7 REACTION VESSEL
CAPSULE AND SALT! FURNACE
R B SALT STORAGE
; 'VESSEL FURNACE

Fig. 30.3. Auxiliary Furnaces and Equipment for 233y Fyel Salt Preparation — Cell G, Building 7930.
31.

Decay Heat Generation Rate in a Single-Region

Molten-Salt Reactor

W. L. Carter

Decay heat from fission products and 233Pa has
been calculated for a 2000 Mw (electrical) single-
region MSR operating on LiF-BeF ,- ThF4-2 3SUF,
fuel. Heat-generation rates are shown graphically
in Figs. 31.1 and 31.2 as a function of time after
reactor shutdown. The data extend from reactor
equilibrium to a decay time of about 11 years.

It was assumed that the MSR had been operating
sufficiently long for fission products and ?33Pa to
be present in equilibrium concentrations. For
233pga the characteristic equilibrium was for a
3-day processing cycle. For fission products,
equilibrium was established for three different
situations: (1) all fission products removed on a
38-day cycle through the processing plant; (2) noble
gases removed on a 50-sec cycle by sparging with
an inert gas in the circulating loop; and (3) in
addition to noble-gas sparging, noble metals re-
moved on a 50-hr cycle by reaction with piping,
heat exchanger, and reactor surfaces. In cases
2 and 3, remaining fission products were removed
by the 38-day cycle through the processing plant.
The vertical separation of the three appropriately
labeled curves of Fig. 31.1 is a measure of the
difference in decay heat from these three assumed

275

J. S. Watson

conditions., The largest difference for times soon
after shutdown occurs at about 1 hr, when the
noble-gas- and noble-metal-free fuel produces

about 33% less heat than fuel containing gross
fission products. At equilibrium for the three above
conditions, fission products produce, respectively,
144.7, 128.5, and 127.9 kw per cubic foot of fuel.

This reactor has a breeding ratio of 1.076, and
there is 205 kg of 233Pa in the fuel and processing
plant., The ?33Pa concentration in the fuel is
7.25 g/ft3 (14.5 kg total), which at equilibrium
generates 0.367 kw of heat per cubic foot; there
is 195.5 kg in the processing plant, producing
9.68 Mw of heat.

An interesting result of this study is the im-
portance of a few fission products in the total
heat generation. At equilibrium, Rb, Cs, and Sb,
respectively, account for 22.8, 21, and 17.3% of
fission product decay heat. At 1 hr decay the
amounts are I (18.6%), Kr (13%), La (10.5%), and
Y (9.6%); at 10 hr decay, the values are I (23.8%),
La (16.5%), and Y (11.4%). For much longer decay
times (e.g., 125 days), about 80% of the energy
is from Nb, Zr, Pr, and Y.
276

ORNL-DWG 68—2406A

10 — T T T T T 71TT T 717 1 T 7117 — T T T T 7777 T T T T TTT7 T T T 71717
o
6 [ -
- -
4 -~
2r -
"
< A8
10°E S10°g GROSS FISSION PRODUCTS -
8 = 8 —
[ 5 -
6r 26 —
- : 4 -
-
4 Ll
— I — NOBLE GASES SPARGED -
z ON 30-SEC CYCLE
2 GROSS FISSION PRODUCTS o 2}
ul
o . . .
Ssec 10sec 30sec Imin 5min 15 min
'04[: IO‘ ~ 1 | ||||.2 I [ 1_| | Y A W I | .
8 10 2 4 6 810 2 4 6 810 2 4 6 8-
6 TIME AFTER DISCHARGE FROM REACTOR ( hours) —
- NOBLE GASES SPARGED -
ON 30- SEC CYCLE
4 - —
o NOBLE GASES SPARGED AND —
r NOBLE METALS PLATED ON
hy REACTOR SURFACES
m 2 - —
b
z
o
2
< 3 -
=10 ]
wer N
Tg} _
z [ R
O 4k PROTACTINIUM—233
w
o |
2r -
IOz: -
5L .
6 .
| -
4
2 | ]
REACTOR POWER = 4444 MW ( Thermol)
FUEL VOLUME, = 2000 t13
ok EQUILIBRIUM 233pg = 0.256 g/liter -
- FUEL PROCESSING CYCLE = 38 days p
8 233p, PROCESSING CYCLE = 3days .
[~ -
2 -
Id 5d 204d 2mo 6mo lyr
bbb Jl |||1111l 1 [ 1 i4il l i LLllllLLJL 1

i
2 4 6 810 2

Fig. 31.1. Fissio

4 6 8102 2 4 6 8108 2 4 6 810* 2 4 6 8I0°
TIME AFTER DISCHARGE FROM REACTOR (hours)

n Product and Protactinium Decay Heat in One-Region MSR Fuel.
HEAT GENERATION RATE (watts)

277

ORNL— DWG 68-—-2107A

e e TG

R

0% — E‘\
8 — 8 — =~ —
— — -~ —
6 — GROSS FISSION PRODUCTS + GROSS 2*2pg 6 [— GROSS FISSION PRODUCTS ~
‘\\/ L S
';'4 - GROSS FISSION PRODUCTS ]
‘,‘_"2 | Imin Smin ISmin  30min __
| I
x
233
o & Pa INDECAY TANK _
— — lJIlIIIll | | | [
6 — 233pg IN DECAY TANK 0.0l 2 6 80l 2 4 6 8-
— TIME (hours)
4 S
2 P
10%— 233pg INFUEL STREAM
8 = /
6 —
4 —
2 —
lo’:
81—
6 —
4 I
2 L
IO‘:
8 I—
6 I
2 —
FUEL 233y
REACTOR POWER 4444 MW (thermal)
s FUEL VOLUME 2000 f13
10— FUEL PROCESSING CYCLE 38 DAYS p—
8 [— Pa PROCESSING CYCLE 3 DAYS —]
*E -
4 |— ]
2 — —
id 10d 304 3mo  6mo 2yr
S R T Tl aul 1 Ll o 1I|11
2 4 6 810 2 6 810? 2 € 810° 2 4 6 810* 2 6 810°

TIME AFTER DISCHARGE FROM REACTOR {( hours)

Fig. 31.2. Total Heat-Generation Rate in One-Region MSR from Gross Fission Products and 233p,,
OAK RIDGE NATIONAL LABORATORY
MOLTEN-SALT REACTOR PROGRAM

FEBRUARY 29, 1968

M. W. Rosenthal, Director
R. B. Briggs, Associate Director
P. R. Kasten, Associate Director

O n

278

RC

REACTOR CHEMISTRY DIVISION

UCC UNION CARBIDE CORP

Ut
*

-

UNIVERSITY OF TENNESSEE

PART TIME ON MSRP
DUAL CAPACITY

ANALYTICAL CHEMISTRY DEVELOPMENT

4. C, White* AC
RESEARCH & DEVELOPMENT
A. S, Meyer AC
R. F. Apple AC
C. M. Boyd AC
J. M. Dale AC
T. H. Handley* AC
D. L. Monning* AC
T. R. Musller* AC
E. Ricci* AC
F. L. Whiting uT
J. P. Young* AC
ANALYSES
L. T. Corbin* AC
G. Goldberg* AC
F. K. Heacker AC
C. E. Lomb* AC
P. F. Thomason* AC
SPECIAL ASSISTANCE

M. A. Bredig” ¢
H. R. Bronstein* C
A. S. Dworkin* C
J. H. Gibbons* P
T. J.Gofson* GE
R. L. Macklin* P
J. R. Nicholson™ GE
G. P. Smith* MEC

v
MSBR DESIGN STUDIES COMPONENTS & SYSTEMS DEVELOPMENT PHYSICS MATERIALS FUEL PROCESSING DEVELOPMENT CHEMISTRY MSRE OPERATIONS
R J. R. Weir M&C M. E. Whatley CcT W. R. Grimes* RC P. N. Haubenreich R
E.S. Betis R Dunlap Scott R A. M. Perry R HCE. MeCoy LG E L Nichalron T avbenrei
C. E. Benis* GE MSRE PHYSICS HASTELLOY N STUDIES MSBR PROCESSING
W. L. Carter* cT D. A. Canonico* M&C .
. ; o J. O. Blomeke cT
F. H. Clark 18C PUMP DEVELOPMENT B. E, Prince R A. G. Cepolina® M&C WL c;:,,, cT REACTOR CHEMISTRY COORDINATION NUCLEAR AND MECHANICAL ANALYSIS
C. W. Collins | SE R. E. Geblbach™ M&C L. M. Ferris cr H. F. McDuffie* RC C. H. Gabbard™* R J. R, Engel R
¥. K. Crowley A. G. Grindell R MSBR ANALYSIS R. W. Gunkel” MaC B. A. Hannaford cT RE Thomase RC < Fe foabhar - R Engel
S. J. Ditto 1&C P. G. Smith R J. W. Koger M&C 7. R, Hightower ¢t FA E B|“: » X C. H. Gabbard R
W. P. Eatherly ucc L.V Wilson R 0. L. Smith R A. P. Litman* M&C M S Lin cT - Blankensnip R. C. Stefy, Ir. R
A. G Grindell R H. C. Young R ¥. R, Cobb R T.S. Lundy™ M&C 1.C. Moilen cr A Wotts R
G. H. Lieweltyn GE ob .R.Co C E. Sessions MAC UE N T MSRE ON-SITE CHEMISTRY
H. A. McLoin R - D. Owens R H. T. Kerr R cNeese
M A Melain R Doyle Scott R G. M. Slaughter M&C J. P. Nichols cT R. E. Thoma** RC OPERATION
P C. E. Schilling CcT S. S. Kirslis RC
L. * GRAPHITE STUDIES - S. Ki
R L Moore? we MSBR EXPERIMENTAL PHYSICS N st F. J. Smith cr L. L. Fairchild RC R. H. Guymon R
K. A Nelms? ot R. L. Beatty M&C 1. S. Wotson cT J. L. Crowley R INSTRUMENTATION AND CONTROLS
" Robertson* R G. L. Rogen R 0. B. Cavin mac E. L. Youngblood cr PHYSICAL AND INORGANIC CHEMISTRY P. H. Horley R
f- ;- ;obe:'iofl R W. H. Cook M&C J. Beoms cT T. L. Hudson R L. g- gfl"'; '5g
. R. Rose’ . K. Adams* 18
IR R A COMPONENTS AND SYSTEMS MSBR SAFETY ANALYSIS C. R Kennedy M&C V. L. Fowler cT C. F. Boes, Jr.* RC A. 1. Krakaviak R
Dunlap S D. V. Kiplinger’ M&C J.F, Lond cT C * M. Richardsan R
p Scott N . F. Lan . J. Barten RC .
W, Tery R R b o H. F. Bauman R R. W, McClung* MAC R. 0. Peyne s T e RC T Amwine R |—  C.D. Mortin 1&C
H. L. Watts R - B. wliah'er R W. C. Robinson* M&C C. 7. Thompson ps G D. Bromton* RC Pt R J. L. Redford 1&C
L. V. Witson* R R. E. Helms R H. L. Whaley™ MEC - ©. Brun . P. Culp R. W. Tucker 18C
W. C. George R R. J. Kedl R CONSULTANT S. Cantor RC W. H. Duckworth Rob s, E. Ellis \aC
H. M. Paly R T. H. Mauney RC 1. W. Kerlin uT GRAPHITE-METAL JOINING MSRE PROCESSING E- 2 g‘:::“:iw :(c: é :- g:@ : C. C. Johnson 1&C
. : é’r“g;mes E K. V. Cook* " M&C R. B. Lindauer** CcT _g: H: Shaffer* RC Tj 0: H\I‘I’m R C. E. Kirkwoad 1&C
C A Gifford R . FJ’< Hemmand m:g R. A. Strehlow” RC C. C, Hortt R
. J. Werner PREPARATION OF 233UF . "LiF FUEL FOR MSRE R. E. Thoma** RC H. 3. Miller R
REMOTE MAINTENANCE 4 L. M. Toth RC LEp R
TECHNICAL SUPPORT . - - £+ Penton
R. Blumberg R . E. L. Nicholson CT C. F. Weaver” RC W. E. Ramsey R
P. P. Holz* R M. D. Allen . MaC J. M. Chandler cT C. E. L. Bamberger” RC R. Smith, Jr R CHEMICAL PROCESSING
J. R. Shugart R R. W. Cunningham M&C 5. Mann cT F. A. Doss RC J. L. Stepp R 1
CONSULTANT J. . Feliner MaC A. M. Rom cT H. A, Friedman® RC 5 R Wenr R R. B. Lindouer** ct
J. W. Geer™ M&C W. F. Schaffer cT L. 0. Gilpatrick* RC J. E. Walfe R
F. N, Peebles uT F. D. Harvey* M&C D. N. Hess RC 5. J. Davi R
N . J. Davis
C. W. Houck M&C B. F. Hitch RC
P. F. Huber M&C H. W. Kohn c
V. G. Lone Mec J. D. Redman* RC
E. J. Lawrence M&C G. D. Robbins RC
E.H. Lee” M&C K. A. Romberger* RC
L. C. Monley” M&C R G Ross RC DESIGN LIAISON
W. ). Mason® MeC H. H. Stone* RC C. K. McGlothlan R
B. McNabb MaC W. K. R. Finnell RC R. H. Sigler R
R. A. Padgett M&C W. Jennings, Jr. RC
N. 0. Pleasant M&C C. E. Roberts RC
R. 5. Pulliam™ MBC W. P. Teichert RC
W, H, Smith, Jr.* M&C s
L. D. Stack™ M&C
G. D. Stohler* M&C MAINTENANCE
E. M. Thomas MEC
H. R.Tinch* M&C B. H. Webster R
J. J. Woodhouse M&C IRRADIATION EXPERIMENTS t. :.. g.l:- : —
o .P. Py
:- S xj]:!‘""‘ n:g E. G. Bohlmann® RC 1R oot R
AC ANALYTICAL CHEMISTRY DIVISION - W. Willioms L e
C CHEMISTRY DIVISION ELc RC
CT CHEMICAL TECHNOLOGY DIVISION E-L: Compere R
D DIRECTORS DIVISION -G Sovage ke
GE GENERAL ENGINEERING DIVISION o R"H'" , RC
1&C INSTRUMENTATION AND CONTROLS DIVISION W Mhers RC
M&C METALS AND CERAMICS DIVISION Lo
P PHYSICS DIVISION
R REACTOR DIVISION

R

0

PN AELND -

>POOEOVUPENEIAMEILOOMPA-TMAIMOTERNLIOOLE-OACDOD

279

INTERNAL DISTRIBUTION

. Adams

. Adamson

. Affel

. Alexander

. Apple

. Baes
Baker
Ball

. Bamberger

. Barton

. Bauman

Beall

. Beatty

. Bell

—rMmem&-Tn1T1Oo0T X

. Bender

. E. Bettis

. S. Bettis

. S. Billington

. E. Blanco

. F. Blankenship

O. Blomeke

. Blumberg

L. Boch

. G. Bohimann
. J. Borkowski
. E. Boyd

Braunstein

. Bredig
Breeding

. Briggs

. Bronstein
. Browning

. Bruce

. Brunton

. Burger

. Canonico

Cantor

P ITOAOMAO®E -~ >

. W. Cardwell
. L. Carter

. I. Cathers

. B. Cavin

. Cepolino

57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.
96.
97.
98.

FACRAEO-IPrOEME-ETOPZr AN - RED-UNE-VORErE-MO S

OCOTMWOAITPTIMAITZEANITPmMmIO-T=z=0Torr

WATPr=xgx

ORNL-4254
UC-80 - Reactor Technology

. Chandler
. Cobb
. Collins

. Compere

Conlin

. Cook

. Corbin
. Cottrell
. Cox

. A,

Cristy

J. Cromer (K-25)

. Crowley
. Culler

. Cuneo

Dale

. Davis
. Davis
. DeBakker
. DeVan

Ditto

. Donnelly

Dudley

. Dunwoody

Dworkin

. Dyslin
. Eatherly
. Edlund (K-25)

Engel
Epler
Ergen

. Ferguson
. Ferris
. Fraas
. Friedman

Frye, Jr.

. Gabbard

. Gall

. Gallaher
. Gelbach

. Gibbons

. Gilliland
. Gilpatrick
280

99. H. E. Goeller 164. R. B. Lindauer
100. W. R. Grimes 165. A. P. Litman
101. A. G. Grindell 166. J. L. Liverman
102. R. W. Gunkel 167. R. S. Livingston
103. R. H. Guymon 168. G. H. Llewellyn
104. J. P. Hammond 169. E. L. Long
105. R. P. Hammond 170. M. I. Lundin
106. B. A. Hannaford 171. R. N. Lyon
107. P. H. Harley 172. R. L. Macklin
108. D. G. Harman 173. H. G. MacPherson
109. W. O. Harms 174. R. E. MacPherson
110. C. S. Harrill 175. F. C. Maienschein
111. P. N. Haubenreich 176. J. C. Mailen
112. F. A. Heddleson 177. D. L. Manning
113. R. E. Helms 178. C. D. Martin
114. P. G. Herndon 179. W. R. Martin
115. D. N. Hess 180. C. E. Mathews
116. R. F. Hibbs (Y-12) 181. T. H. Mauney
117. J. R. Hightower 182. H. McClain
118. M. R. Hill 183. R. W. McClung
119. E. C. Hise 184. H. E. McCoy
120. H. W. Hoffman 185. H. F. McDuffie
121. D. K. Holmes 186. C. K. McGlothian
122. V. D. Holt 187. C. J. McHargue
123. P. P. Holz 188. L. E. McNeese
124. R. W. Horton 189. J. R. McWherter
125. A. S. Householder 190. H. J. Metz
126. T. L. Hudson 191. A. S. Meyer
127. H. lnouye 192. E. C. Miller
128. W. H. Jordan 193. C. A. Mills
129-143. P. R. Kasten 194. R. L. Minue
144. R. J. Kedl 195. W. R. Mixon
145. M. T. Kelley 196. R. L. Moore
146. M. J. Kelly 197. K. Z. Morgan
147. C. R. Kennedy 198. D. M. Moulten
148. T. W. Kerlin 199. J. C. Moyers
149. H. T. Kerr 200. T. R. Mueller
150. R. F. Kimball 201. H. A. Nelms
151. S. S. Kirslis 202. J. P. Nichols
152. D. J. Knowles 203. E. L. Nicholson
153. J. W. Koger 204. E. D. Nogueira
154. A. |. Krakoviak 205. L. C. Odkes
155. J. S. Kress 206. W. R. Osborn
156. J. W. Krewson 207-208. R. B. Parker
157. C. E. Lamb 209. L. F. Parsly
158. J. A. Lane 210. P. Patriarca
159. C. E. Larson 211. H. R. Payne
160. E. J. Lawrence 212. A. M. Perry
161. J. J. Lawrence 213. T. W. Pickel
162. M. S. Lin 214. H. B. Piper
163. T. A. Lincoln 215. B. E. Prince
617.
618.
619.
620.
621.
622.
623.
624.
625.
626.

281

D. A. Sundberg
J. R. Tallackson
E. H. Taylor

TMETOrO-=TAE-P>OOI-OUEAOCTVDE

Orvoe—<0m=EcATTr

. Thomason
. Toth

. Trauger

. Tucker

. Ulrich

. Watkin

. Watson

. Watts

. Weaver

. Webster
. Weinberg

Weir

. Werner
. West
. Whatley

White

. Williams

. Wilson

. Young

. Young

. Young

. Youngblood
. Zapp

Biology Library

ORNL - Y-12 Technical Library
Document Reference Section
Central Research Library
Laboratory Records Department
L aboratory Records, ORNL R.C.

Asquith, Atomics International, P.O. Box 309, Canoga Park, California 91304
Bowman, Union Carbide Technical Center, 125C0 Snow Road, Parma, Ohio 44130

Brannan, Carbon Products Division, 270 Park Avenue, New York, New York 10017

aul Cohen, Westinghouse Electric Corp., P.O. Box 158, Madison, Pennsylvania 15663

216. G. L. Ragan 426.

217. J. L. Redford 427.

218. M. Richardson 428.

219. G. D. Robbins 429.

220. R. C. Robertson 430.

221. W. C. Robinson 431.

222. H. C. Roller 432.

223. K. A. Romberger 433.
224-398. M. W. Rosenthal 434.

399. R. G. Ross 435.

400. H. C. Savage 436.

401. A. W. Savolainen 437.

402. W. F. Schaffer 438.

403. C. E. Schilling 439.

404. Dunlap Scott 440.

405. J. L. Scott 441.

406. H. E. Seagren 442,

407. C. E. Sessions 443.

408. J. H. Shaffer 444,

409. E. D. Shipley 445,

410. W. H. Sides 446.

411. M. J. Skinner 447.

412. G. M. Slaughter 448.

413. A. N. Smith 449.

414. F. J. Smith 450.

415. G. P. Smith 451.

416. 0. L. Smith 452,

417. P. G. Smith 453.

418. A. H. Snell 454,

419. W. F. Spencer 455.

420. 1. Spiewak 456-457.

421. R. C. Steffy

422. C. E. Stevenson 458-460.

423. W. C. Stoddart 461-615.

424. H. H. Stone 616.

425. R. A. Strehlow

EXTERNAL DISTRIBUTION

W. O. Allen, Atomics International, P.0O. Box 309, Canoga Park, California 91304
J. G.
J. C.
G. D. Brady, Materials Systems Division, UCC, Kokomo, Indiana 46901
J. H.
W. S. Butler, Dow Chemical Company, Freeport, Texas 77541
P
D. F. Cope, Atomic Energy Commission, RDT Site Office (ORNL)

J W. Crawford, Atomic Energy Commission, Washington 20545
M. W. Croft, Babcock and Wilcox Company, P.O. Box 1260, Lynchburg, Virginia 24505
627. C. B. Deering, Atomic Energy Commission, RDT Site Office (ORNL)
628.
629.
630.

631.
632.
633.
634.
635.
636.
637.
638.
639.
640.
641.
642.
643-644.
645.
646.
647.
648.
649.
650.
651.
652.
653.
654.
655.
656.
657.
658.
659.
660.
661.
662.
663.
664.
665.
666.
667.
668.
669.
670.
671-672.
673-931.

282

D. A. Douglas, Materials Systems Division, UCC, Kokomo, Indiana 46901
H. L. Falkenberry, Tennessee Valley Authority, 303 Power Building, Chattanooga, Tenn. 37401
C. W. Fay, Wisconsin Michigan Power Company, 231 W. Michigan Street, Milwaukee, Wisconsin
53201
Gregory Flynn, General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089
A. Giambusso, Atomic Energy Commission, Washington 20545
Gerald Golden, Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, Illinois 60439
. Grigorieff, Assistant to the Executive Director, Oak Ridge Associated Universities
. Kehoe, Burns and Roe, Inc., 700 Kinderkamach, Oradell, New Jersey 07649
. Lang, Douglas United Nuclear, 703 Bldg., Richland, Washington 99352
. Larkin, Atomic Energy Commission, ORO
. Lieberman, Atomic Energy Commission, Washington 20545
. Lorenzini, Foster Wheeler, 110 S. Orange, Livingston, N. J. 07039
. Manly, Material Systems Division, UCC, 270 Park Avenue, New York, New York 10017
. Mays, Great Lakes Carbon Co., 299 Park Avenue, New York, New York 10017
. McDonald, Battelle-Pacific Northwest Laboratory, Hanford, Washington 99352
. Mclntosh, Atomic Energy Commission, Washington 20542
. Mordarski, Nuclear Development, Combustion Engineering, Windsor, Connecticut
Sldney Parry, Great Lakes Carbon, P.O. Box 667, Niagara Falls, New York 14302
Worth Percival, General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089
G. J. Petretic, Atomic Energy Commission, Washington 20545
M. A. Rosen, Atomic Energy Commission, Washington 20545
H. M. Roth, Atomic Energy Commission, ORO
R. W. Schmitt, General Electric Co., Schenectady, New York 12301
M. Shaw, Atomic Energy Commission, Washington 20545
Remo Silvestrini, United Nuclear Comoration, Grassiands Road, Elmsford, New York 10523
E. E. Sinclair, Atomic Energy Commission, Washington 20545
W. L. Smalley, Atomic Energy Commission, ORO
T. M. Snyder, General Electric Co., 175 Curtner Ave., San Jose, California 95103
L. D. Stoughton, UCC, P.O. Box 500, Lawrenceburg, Tennessee 38464
Philip T. Stroup, Alcoa, P.O. Box 772, New Kensington, Pennsylvania
J. A. Swartout, UCC, 270 Park Avenue, New York, New York 10017
R. F. Sweek, Atomic Energy Commission, Washington 20545
Richard Tait, Poco Graphite, P.O. Box 1524, Garland, Texas 75040
D. R. Thomas, Commonwealth Associates, Inc., 209 E. Washington Ave., Jackson, Michigan 49201
M. Tsou, General Motors, 12 Mile and Mound Roads, Warren, Michigan 48089
J. W. Ullmann, UCC, P.0O. Box 278, Tarrytown, New York 10591
C. H. Waugaman, Tennessee Valley Authority, 303 Power Building, Chattanooga, Tenn. 37401
D. B. Weaver, Tennessee Valley Authority, New Sprankle Building, Knoxville, Tennessee
G. O. Wessenaver, Tennessee Valley Authority, Chattanooga, Tennessee 37401
M. J. Whitman, Atomic Energy Commission, Washington 20545
H. A. Wilber, Power Reactor Development Company, 1911 First Street, Detroit, Michigan
James H. Wright, Westinghouse Electric, P.O. Box 355, Pittsburgh, Pennsylvania 15230
Laboratory and University Division (ORO)
Given distribution as shown in TID-4500 under Reactor Technology category (25 copies-CFSTI)

P EAET-ZTV-EC-E
“=swmJo» Pus—d=s