ORNL-4728.txt

MARTINMARETTA ENERGY SYSTEMS LBAARES

(THRTTATIT ORNL.4728

w/ 1

3 445k 0285149 3

MOLTEN-SALT REAGTOR
PROGRAM

Semiannual Progress Repont
Petiod Ending August 31, 1971

>AK RIDGE NATIONAL LABORATORY
CENTRAL RESEARCH LIBRARY
DOCUMENT COLLECTION
LIBRARY LOAN COPY
DO NOT TRANSFER TO ANOTHER PERSON
s thid

OAK RIDGE NATINAL LABORATORY

OPERATED BY UNION CARBIDE CORPORATION  FOR THE U.S. ATOMIC ENERGY COMMISSION
Printed in the United States of America. Available from
National Technical Information Service
U.S. Department of Commerce
5285 Port Royal Road, Springfield, Virginia 22151
Price: Printed Copy $3.00; Microfiche $0.95

This report was prepared as an account of work sponsored by the United
States Government. Neither the United States nor the United States Atomic
Energy Commission, nor any of their employees, nor any of their contractors,
subcontractors, or their employees, makes any warranty, express or implied, or
assumes any legal liability or responsibility for the accuracy, completeness or
usefulness of any information, apparatus, product or process disclosed, or
represents that its use would not infringe privately owned rights.

Y

[\

3 Y

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT
For Period Ending August 31, 1971

M. W. Rosenthal, Program Director
R. B. Briggs, Associate Director
P. N. Haubenreich, Associate Director

FEBRUARY 1972

OAK RIDGE NATIONAL LABORATORY
Qak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
- for the
U.S. ATOMIC ENERGY COMMISSION

ORNL-4728

Il

3 445k 0285149 3

TIN MARIETTA ENERGY SYSTEMS LIBRARIES

JURACTANTARIE

)

ORNL-2474
ORNL-2626
ORNL-2684
ORNL-2723
ORNL-2799
ORNL-2890
ORNL-2973
ORNL-3014
ORNL-3122
ORNL-3215
ORNL-3282
ORNL-3369
ORNL-3419
ORNL-3529
ORNL-3626
ORNL-3708
ORNL-3812
ORNL-3872
ORNL-3936
ORNL4037
ORNL4119
" ORNL4191
ORNL4254
ORNL4344
ORNL-4396
ORNL4449
ORNL4548
ORNL4622
ORNL4676

This report is one of a series of periodic reports in which we describe the progress of the program. Other reports
issued in this series are listed below. :

" Period Ending January 31,1958

Period Ending October 31, 1958
Period Ending January 31, 1959
Period Ending April 30, 1959
Period Ending July 31, 1959
Period Ending October 31, 1959

Periods Ending January 31 and April 30, 1960

Period Ending July 31, 1960
Period Ending February 28, 1961
Period Ending August 31, 1961
Period Ending February 28, 1962
Period Ending August 31,1962
Period Ending January 31,1963
Period Ending July 31,1963
Period Ending January 31, 1964
Period Ending July 31, 1964
Period Ending February 28, 1965
Period Ending August 31, 1965

. Period Ending February 28, 1966

Period Ending August 31, 1966
Period Ending February 28, 1967
Period Ending August 31, 1967
Period Ending February 29, 1968
Period Ending August 31, 1968
Period Ending February 28, 1969
Period Ending August 31, 1969
Period Ending February 28, 1970
Period Ending August 31, 1970
Period Ending February 28, 1971

L

L)

i"
}
Contents
PART 1. MSBR DESIGN AND DEVELOPMENT
1. DESIGN .. e e e e e 2
1.1 Molten-Salt Demonstration Reactor Design Study . ....... ... ... . ..o i 2
1.1.1 General . ............. PP 2
1.1.2 ReactorCore ........ . i e e e e 3
1.1.3 MSDR Off-Gas and Salt Pump-Back System . . . ............ ... . .. i 4
1.1.4 Drain Tank Cooling System . ....... ... .o s PRV 7
1.1.5 Drain Valve Cell .. .. e e e 9
1.1.6 Drain Cell Catch Pan . . ... ... e e 9 -
1.2 Some Consequences of Tubing Failure in the MSBR Heat Exchanger ........................ 9
1.3 Side-Stream Processing of the MSBR Primary Flow for Xenon and/or Iodine Removal ........... 10
1.4 MSBE DESIZN .. ..ottt e AT 11
“ 1.4.1 Reactor Core’Power and Neutron Flux Distribution .. ......... ... ... .o 11
1.4.2 Reactor Core Heat Removal ....................... P 11
. 1.4.3 Maintenance Studies . .. ................ T R AR 14
1.5 MSBR Industrial Design Study .. ... .. B PP 15
2. REACTOR PHY SICS ..o e e e e e e 16
' ]
2.1 MSR EXPERIMENTAL PHYSICS . . ..o e 16
2.1.1 HTLTR Lattice Experiments . ... ... ...t 16
2.2 PHYSICS ANALYSISOFMSBR ................. e R 18
2.2.1 Neutron Irradiation Effects Qutside the MSBR Core .. ... e 18
2.2.2 MSBE Nuclear Characteristics ............. e e 20
- 2.2.3 Fixed-Moderator Molten-Salt Reactor ... ..... ... .ttt 21
3. SYSTEMS AND COMPONENTS DEVELOPMENT .. e 26
3.1 Gaseous Fission Product Removal .......... . [ e .. 26
3.1.1 Gas Separator and Bubble Generator . ...... ... ... ... i 26
3.1.2 Bubble Formation and Coalescence Test+. ............. ... ..... S 27
3.2 Gas System Test Facility .. ... ... ... oo 27
3.3 Off-GasSystems ... ... ov i P EEE TR 28
p 3.3.1 Off-Gas System for Molten-Salt Reactors ... ....................... [ 28
b 3.3.2 Computer Design of Charcoal Beds for MSR Off-Gas Systems ....................... .29
' 3.4 Molten-Salt Steam Generator .. ... ... ..ottt e e 29
- 3.4.1 Steam Generator Industrial Program . ............. ... . i 29
3.4.2 Molten-Salt Steam Generator Technology Facility . .................0 ... . oonotn 29
3.4.3 Molten-Salt Steam Generator Test Proposals . .............. e e 30

iii
iv

3.5 Sodium Fluoroborate Test Loop ......... .. .o, e 31
3.5.1 Inspection of PKP Pump Rotary Element .. ... ... .. . ... ... ... ... .. ... . ..., 31
3.6 Coolant-Salt Technology Facility . . ... ..ottt e 34
37 MSBR PuUmps . ... e f e e 35
3.7.1 Salt Pumps for MSRP Technology Facilities . . . .. P . 35
372 ALPHA Pump ... oo e PP 35
3.7.3 Drain Tank Jet Pump System for MSDR ........... e e 36
4. INSTRUMENTATION AND CONTROLS .......... e e e e 38
4.1 Transient and Control Studies of the MSBR System Using a Hybrid Computer ......... e 38
4.2 MSRE Design and Operations Report, Part IIB, Nuclear and Process Instrumentation ............ 38
4.3 Further Discussion of Instrumentation and Controls Development Needed for the
Molten-Salt Breeder Reactor . ... ... . ... i i e e e 38
5. HEAT AND MASS TRANSFER AND THERMOPHYSICAL PROPERTIES ........................ 39
5.1 Heat Transfer . ... .. . i e e e 39
5.2 Thermophysical Properties .. .......... ... .. e SRR 41
5.3 Mass Transfer to Circulating Bubbles ... ......... e 4]
PART 2. CHEMISTRY
6. POSTOPERATIONAL EXAMINATIONOFMSRE ... ... .. .. ... . . 46
6.1 Examination of Moderator Graphite fromMSRE . ............. ... ... ... ... ol 46
6.1.1 Results of Visual Examination . . ... e v ettt e .. 46
6.1.2 Segmenting of Graphite Stringer . ................iiiniiniiniina... e .. 46
6.1.3 Examination of Surface Samples by X-ray Diffraction .............................. 47
6.1.4 Examination with a Gamma Spectrometer . ... ... e AT
6.1.5 Milling of Surface Graphite Samples . . ... ... ... ... ... . ... .. ... 49
6.1.6 Radiochemical and Chemical Analyses of MSRE Graphjte ........................... 49
6.2 Cesium Isotope Migration in MSRE Graphite ... ..ottt e et e 51
6.3 Fission Product Concentrations on MSRE Surfaces ......... ... ... ... . . i 54
6.4 Metal Transfer in MSRE Salt Circuits . . ...... O U 55
7. HYDROGEN AND TRITIUM BEHAVIOR INMOLTEN SALT ... ...ttt 56
7.1 The Solubility of Hydrogen in Molten 2LiF-BeFy ... ... .ot 56
7.2 Permeation of Metals by Hydrogen ........... ... ... ... ... ... ... ... ... e 57
7.3 Tritium Controlinan MSBR . ....................... U 59
~ 7.3.1 Mass Spectrometric Examination of Stability of NaBF,OH . .............. e 59
7.3.2 The Thermal Stability of Nitrate-Nitrite Mixtures ................ . . .. 59
7.4 . Dissociating-Gas Heat Transfer Scheme and Tritium Control in Molten-Salt
POWer SyStems . ... 60
8. PROCESSING CHEMIST RY . . .. e e e e e e e e et 62
. 8.1 The Oxide Chemistry of Pa** in Molten LiF-BeF,-ThF, ... ... ... 0. 62
8.2 The Oxide Chemistry of Pa®* in Uranium-Containing Molten LiF-BeF,-ThF, ... .............. 64
8.3 Reductive Extraction Distributions of Barium and Thorium Between Bismuth-Lead
Eutectic and Breeder Fuel Solvent ... ................ e e 67

8.4 Removal of Cerium from Lithium Chloride by Zeolite . ................ e 67

ry
|\

9. DEVELOPMENT AND EVALUATION OF ANALYTICAL METHODS

FOR MOLTEN SALT REACTORS . ... e e 69
9.1 In-Line Chemical Analysis of Molten Fluoride Salt Streams ... ............................. 69
9.2 Slotted Probe for In-Line Spectral MEasurements . . . .............ooiutoinoneaa ... 70
9.3 In-Line Determination of Hydrolysis Products in NaBF, Cover Gas ......................... 71
9.4 Determination of Hydrogen in Fluoroborate Salts . ......... ... ... ... ... ... ... ... .. ..... 73
9.5 Voltammetric and Electrolysis Studies of Hydroxide Ion in Molten NaBF, .................... 74
9.6 Electroanalytical Studies in the NaBF, Coolant Salt .............. e 75
9.7 Electroanalytical Studies of'Ni(Il) in Molten Fluoride Fuel Solvent .. ........................ 75
10. OTHER FLUORIDE RESEARCHES .. .. .. .. . it e 77
10.1  Absorption Spectroscopy of Molten Fluorides ... ..................... J 77
10.1.1 The Disproportionation Equilibrium of UF5 Solutions . ............................ 77

10.1.2 Estimation of the Available F~ Concentrations in Melts by - :
Measurement of UF, Coordination Equilibria . . ......... e e e e e 77
102 Solubility of BFs in Fluoride Melts ... ........... S S o 78
10.3  Fluorides and Oxyfluorides of Molybdenum and Niobium .. ... ... ... ... ... ... ... ...... 80
10.3.1 Mass Spectroscopy of Molybdenum Fluorides . ... .. e e 80
10.3.2 Mass Spectroscopy of Niobium Fluorides and Oxyfluorides . ......................... 80
10.4  Electrical Conductivity of Moltén and Supercooled Mixtures of NaF-BeF, .................. .. 81
10.5 Glass Transition Temperatures in the NaF-BeF, System .............. B 82
10.6 Enthalpy of Lithium Fluoroborate from 298—700°K: Enthalpy and Entropy of Fusion . ......... 85
10.7  Nonideality of Mixing in Li; BeF4-Lil ..., P 86
11. EXAMINATION OF MSRECOMPONENTS . .. ... ... . i e 89
11.1 Examination of Hastelloy N Components Exposed to Fuel Saltinthe MSRE ... ............... 89
11.2  Examination of Components from In-Pile Loop2 ...................... I 94
11.3 Observations of Grain Boundary Crackinginthe MSRE .. ... ................... . e 96
11.4 Examination of a Graphite Moderator Element ............. S 106
11.5 Auger Analysis of the Surface Layer on Graphite Removed from the Core of the MSRE . . . ... .. .. 107

PART 3. MATERIALS DEVELOPMENT

11. EXAMINATION OF MSRE COMPONENTS . . ............ B .. 89
11.1  Examination of Hastelloy N Components Exposed to Fuel Salt in the MSRE ... ... ..... . 89
11.2 Examination of Components from in-Pile Loop 2 . e L. 94
11.3  Observations of Grain Boundary Crackinginthe MSRE . .. ... ... . ... ... . ... ... ... ........ 96
11.4 Examination of a Graphite Moderator Element ... ... ... ... .. ... . ... ... . ... . ... .. ..., 106
11.5 Auger Analysis of the Surface Layer on‘Graphjte Removed from the Core of the MSRE . ... ... ... 107
12. GRAPHITE STUDIES ... ... ..\t S e 11
12.1 The Irradiation Behavior of Graphite at 715°C .. ... ... .. ... . .. . . .. 112
12.2 Procurement of Various Grades of Carbon and Graphite ... ....................... e 114
12.3  X-Ray Studies .. ... . e PRV 116
12.4 Thermal Property Testing . ...... PP e : : . 117

vi

13. HASTELLOYN ............... R SO e PR 125
13.1 Electron Microscopy Studies ................. e e e AT 125
| 13.1.1 Microstructures of Hf-Modified Hastelloy N Laboratory Melts . .. ................... w. 125
13.1.2 Ni; Ti Precipitation in Ti-Modified Hastelloy N .. ... ... ... ... ... ..... R 129
13.1.3 Strain-Induced Precipitation in Modified Hastelloy N . ........ ... ... ... ........... 130
13.2  Mechanical Properties of Unirradiated Modified Hastelloy N .. .................... e 132
13.3 Weldability of Several Modified Commercial Alloys .. ... ... ... ... .. ... .. ... .. ......... 132
13.4  Creep-Rupture Properties of Hastelloy N Modified with Niobium, Hafnium,
and Titanium . ................ J e [N 137
13.5 CormrosionStudies . ........ ... ...t e e e 138
13.5.1 FuelSalts ........... e e e e 139
13.5.2 Fertile-Fissile Salt . . ........ .. ... ... ... ... ...... e e 145
13.5.3 Blanket Salt ........... P 145
13.5.4 Coolant Salt ......... e e e e e 145
13.6  Analysis of High:Level Probe from Sump Tank of PKP-1 . . ..... e e 150
13.7 Forced-Convection Loop Corrosion Studies .. ... ... ... i, .. 150
13.7.1 Operation of Forced-Convection Loop MSR-FCL-1A .. ..................... e 150
13.7.2 Metallurgical Analysis of Forced-Convection Loop MSR-FCL-1A ......... PPN 151
13.7.3 Operation of Forced-Convection Loop MSR-FCL-2 . ............ e 151
13.7.4 Metallurgical Analysis of Forced-Convection Loop MSR-FCL-2 ... ....... el 153
13.8  Corrosion of Hastelloy Nin Steam . .......... .. .. .. oo, e P 153
13.9 Evaluation of Duplex Tubing for Use in Steam Generators ..........................0..... 156
14. SUPPORT FOR CHEMICALPROCESSING .. ....... ... ... it iiiiiaae ... 163
14.1 Construction of a Molybdenum Reductive-Extraction Test Stand e 163
14.2 Fabrication Development of Molybdenum Components ............ e ee e ae i 166
14.3 Welding Molybdenum .. ................ e P P 169
14.4  Development of Brazing Techniques for Fabricating the Molybdenum Test Loop . .............. 172
14.4.1 Resistance Furnace Brazing .. ... ...... ... i it it 172
14.4.2 Induction Vacuum Brazing . .. . .. ... ... . ... e 172
14.4.3 Induction Field Brazing .. ... ... . ... i e 173
14.5 Compatibility of Materials with Bismuth . ....... ... ... .. ... ... ... .. ... .. ... ... ..., 173
14.6 Chemical Vapor Deposited Coatings . .. .......o ittt i et et e v aeae e 176
14.7 Molybdenum Deposition fromMoF¢ ......... .. ... ... .. e 177
PART 4. MOLTEN-SALT PROCESSING AND PREPARATION
15. FLOWSHEET ANALYSIS . .ottt e e e e e, 179
- 15.1  Cost Estimate for an MSBR Processing Plant -. .. ........: e e 179
15.2 Comparison of Costs for Alternate Off-Gas Treatment Methods for an
MSBR Processing Plant . .. ... ... . . . . J 183
15.3 Effect of Noble-Metal and Halogen Removal Times on the Heat Generation
Rate in an MSBR Processing Plant . . . ... ... e e e e e e e e e 186
15.4 Long-Term Disposal of MSBR Wastes . .. .......... ... .. ... ... e 186
15.5 Availability of Natural Resources Required for Molten Salt Breeder Reactors .............. L... 189

ab'.d'
L 2}

17.

vii

16. PROCESSING CHEMISTRY ... ..ottt e e e 191
16.1  Distribution of Lithium and Bismuth Between Liquid Lithium-Thorium-Bismuth Alloys '
and Molten LiCl ...................... e e 191
16.2 Mutual Solubilities of Thorium and Rare Earths in Liquid Bismuth .. ......... .. ... ... ..... 193
16.3  Oxide Precipitation STUAIES .. ... .....ooivrune e 196
16.4 Chemistry of Fuel Reconstitution ............ .. .. . . . . . . . . 199
ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS ... ... .. ................ e 202
17.1 Lithium Transfer During Metal Transfer Experiment MTE-2 .. ... .. .. ........ e 202
17.2  Operation of Metal Transfer Experiment MTE-2B ... .. .. ............................... 204
17.3 Development of Mechanically Agitated Salt-Metal Contactors ... ... I 207
17.4  Design of the Third Metai Transfer E‘xperiment P 209
17.5 Reductive Extraction Engineering Studies . ................ T e 212
17.6  Development of a Frozen-Wall Fluorinator: Induction Heating Experiments ............. e 214
17.7 Predicted Performance of Continuous Fluorinators ........... ... ... ... ... ... ....... ... 217
17.8 Engineering Studies of Uranium Oxide Precipitation .................. ... . . ... 220
17.9  Design of a Processing Materials Test Stand and the Molybdenum Reductive
Extraction Equipment . ... .. .. .. . 222
17.10 Development of a Bismuth-Salt Interface Detector ......................... S 222

18.

CONTINUOUS SALT PURIFICATION ... ... e o... 226
-
[

L)

Ef ]
| LN

L1 §

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 20 years ago in the Aircraft
Nuclear Propulsion program to .make a molten-salt
reactor power plant for aircraft. 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 reactor would be
233(F, dissolved in a salt that is a mixture of LiF and
BeF,, but 233U or plutonjum could be used for
startup. The fertile material would be ThF, dissolved
in the same salt or in a separate blanket salt of similar
composition. The technology being developed for the
breeder is also applicable to high-performance converter
reactors. ’

A major program activity through 1969 was the
operation of the Molten-Salt Reactor Experiment. This
reactor was built to test the types of fuels and materials
that would be used in thermal breeder and converter
reactors and to provide experience with operation and
maintenance. The MSRE operated at 1200°F and
produced 7.3 MW of heat. The initial fuel contained 0.9
mole % UF,, 5% Z1F,, 29% BeF,, and 65% "LiF; the
uranium was about 33% 22°U. The fuel circulated
through a reactor vessel and an external pump and heat
exchange system. Heat produced in the reactor was
transferred to a coolant salt, and the coolant salt was
pumped through a radiator to dissipate the heat to the

‘atmosphere. All this equipment was constructed of

Hastelloy N, a nickel-molybdenum-iron-chromium
alloy. The reactor core contained an assembly of
graphite moderator bars that were in direct contact
with the fuel.

Design of the MSRE started in 1960, fabrication of
equipment began in 1962, and the reactor was taken
critical on June 1, 1965. Operation at low power began
in January 1966, and sustained power operation was
begun in December. One run continued for six months,
until terminated on schedule in March 1968.

Completion of this six-month run brought 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 concluded that
this objective had been achieved and that the MSRE
had shown that molten-fluoride reactors can be oper-
ated at 1200°F without corrosive attack on either the

-metal or graphite parts of the system, the fuel is stable,

reactor equipment can operate satisfactorily at these
conditions, xenon can be removed rapidly from molten
salts, and, when necessary, the radioactive equipment
can be repaired or replaced.

The second phase of MSRE operation began in
August 1968, when a small facility in the MSRE
building was used to remove the original uranjum
charge from the fuel salt by treatment with gaseous F,.
In six days of fluorination, 221 kg of uranium was
removed from the molten salt and loaded onto ab-
sorbers filled with sodium fluoride pellets. The decon-
tamination and recovery of the uranium were very
good.

‘After the fuel was processed, a charge of ??3U was
added to the original carrier salt, and in October 1968
the MSRE became the world’s first reactor to operate
on ??3U. The nuclear characteristics of the MSRE with
the 233U were close to the predictions, and the reactor
was quite stable.

In September 1969, small amounts of PuF; were

“added to the fuel to obtain some experience with

plutonium in a molten-salt reactor. The MSRE was shut
down permanently December 12, 1969, so that the

ix
funds supporting its operation could be used elsewhere
in the research and development program.

Most of the Molten-Salt Reactor Program is now
devoted to the technology needed for future molten-
salt reactors. The program includes conceptual design
studies and work on materials, the chemistry of fuel
and coolant salts, fission product behavior, processing
methods, and the development of components and
systems.

Because of limitations on the chemical processing
methods available at the time, until three years ago
most of our work on breeder reactors was aimed at
two-fluid systems in which graphite tubes would be
used to separate uranium-bearing fuel salts from

thorium-bearing fertile salts. In late 1967, however, a
one-fluid breeder became feasible because of the devel-
opment of processes that use liquid bismuth to isolate
protactinium and remove rare earths from a salt that
also contains thorium. QOur studies showed that a
one-fluid breeder based on these processes can have fuel
utilization characteristics approaching those of our
two-fluid designs. Since the graphite serves only as
moderator, the one-fluid reactor is more nearly a
scaleup of the MSRE. These advantages caused us to
change the emphasis of our program from the two-fluid
to the one-fluid breeder; most of our design and
development effort is now directed to the one-fluid
system.

.-’.n

oy~
e

'a

Summary

PART 1. MSBR DESIGN AND DEVELOPMENT

1. Design

Design studies were continued on the 300-MW(e)
MSDR. The core graphite was changed to slab shapes,
about 17 in. X 9% in. X 21 ft long. The slabs are held
in place by graphite posts which maintain the core
configuration yet permit movement to accommodate
temperature and radiation effects. Salt which reaches
the drain tank by entrainment in the gas effluent from
the gas separator is now returned to the primary
circulation system by two jet pumps located in a cluster
in the drain tank. The jets are actuated by salt flows
from the primary pumps and are arranged to prevent
pumpback of gas into the primary loops. The manifolds
that connect the NaK cooling circuit thimbles in the
drain tank have been rearranged to facilitate mainte-
nance, and three water tanks for heat rejection are now
provided instead of one. Seismic effects on the drain
tank and its cooling system have received preliminary
study. A basin has been provided in the drain tank cell
to catch accidental salt spills from the salt lines or
valves and to drain it into the fuel-salt drain tank.

A study was made of the consequences of the mixing
of fuel and coolant salts as a result of various modes of
failure of tubing in the primary heat exchangers in the
MSBR reference design. Four specific cases were ex-
amined: (1) doubled-ended failure near the fuel-salt
outlet; (2) double-ended failure near the fuel salt inlet;

+ (3) small coolant-salt leak into the primary system; (4)

small fuel-salt leak into the secondary system. The
study is, in many places, speculative both because of
the preliminary status of the design and the lack of
adequate physicochemical data on mixing of fuel and
coolant salts. However unpleasant some of the conse-
quences of tubing failure may be, no way of generating
either a nuclear excursion or a rupture in the primary or

-secondary loop piping is foreseen.

xi

The incentives for side-stream processing of the
MSBR fuel salt for iodine and/or xenon removal as an
alternative means for dealing with the '?3Xe poison
level are being examined. Iodine stripping would require
only small bypass flow rates, approximately 225 gpm
sufficing to reduce '2°Xe poison level to ~1% in a
100% efficient stripper. Analysis of laboratory sparging
experiments indicates only negligible movement of
iodine as HI into the graphite. Combined xenon and
iodine strippers appear to be a reasonable development
goal and a highly attractive means for achieving 0.5%
poison level with uncoated graphite.

Design studies on the MSBE were continued with
emphasis on the core heat removal and core mainte-
nance. Both prismatic and slab-type graphite elements

" were considered. Maintenance studies were also con-

ducted on repair of a leak in the primary heat
exchanger. )

A subcontract between ORNL and the Ebasco Ser-
vices Group, consisting of Ebasco, Babcock and Wilcox,
Byron Jackson, Cabot, Conoco, and Union Carbide
companies, was negotiated and signed. This industrial
group will conduct a design study of a 1000-MW(e)
MSBR plant. They have essentially completed the
selection of a reference conceptual design for further
study and evaluation. '

2. Reactor Physics

During the report period we completed and issued a
report describing the Reactor Optimum Design Code
(ROD), with instructions for its use. The code package
has been furnished to the Argonne Code Center, where
it is available upon request.

Since the Hastelloy N in the fuel system of an MSBR
should last for the life of the plant, we have calculated
the radiation damage caused by delayed neutrons
emitted outside the core and by leakage neutrons from
the core. Results indicate that helium production,
primarily from '®B(n,q) reactions, would be the most
important source of damage (assuming an initial '°B
concentration of 2 ppm). The lifetime helium pro-
duction in the heat exchanger due to delayed neutrons
would be less than 2 ppb with NaBF,-NaF coolant salt,
or close to 50 ppb with Li, BeF, enriched to 99.995%
in the 7Li isotope. In the reactor outlet line, the
terminal helium concentration would be less than 10
ppb from the delayed neutron source. Rather larger
amounts of helium could be produced by core-leakage
neutrons in parts of the heat exchangers or piping
directly facing the reactor vessel. For our present
reference design, the maximum amount of helium could
approach 1 ppm, primarily from ° ® Ni(n,a) reactions.

We calculated various reactivity coefficients for the
- MSBE and compared them with values for our reference
MSBR. The most noteworthy differences are a positive
fuel-salt density coefficient for the MSBE (negative for
"~ the MSBR) and the negative graphite temperature
coefficient for the MSBE (positive for the MSBR). Both
of these differences are associated with a greater
neutron leakage from the smaller MSBE, and result in a
much more negative overall temperature coefficient,
that is, —8 X 107%/°C vs —0.9 X 107%/°C for the
MSBR.

Investigation of batch fuel cycles for fixed-moderator
molten-salt reactors has been continued. We have found
that reactor lifetime-averaged reaction-rate coefficients
are adequate for calculating the time-dependent fuel
~ composition throughout. the lifetime for enriched-
uranium-fueled reactors, but only for the asymptotic
portion of the lifetime for recycled-plutoniim-fueled
reactors. Reaction coefficients calculated specifically
for the beginning of the lifetime are required for
plutonium calculations, because the amounts of pluto-
nium in the system at startup are sufficient to cause
significant hardening of the neutron spectrum. Calcula-
tions with plutonium feed are continuing.

Results are presented for a reactor designed as a
breeder (with continuous processing) but operated as a
converter, with enriched-uranium feed and batch
processing in four cycles over the lifetime of the
reactor. An average conversion ratio greater than 90%
and a fuel cost, excluding processing, of about 0.76
mill/kWhr were found for this reactor.

3. Systems and Components Development

Tests on the MSBE-scale bubble separator were
continued in the water loop to study the formation of
the very fine bubbles encountered with the water-
glycerol test fluid and to evaluate the separator per-

x1i

formance with these small bubbles. These tests showed
that the production of the small bubbles is influenced
mostly by the pump head and speed. Dilution tests run
with CaCl, solution showed a very sharp increase in

_separator efficiency as the concentration approached

2.7 wt %, probably the point at which significant
bubble coalescence began. Such a threshold at a lower
concentration has been discussed in the recent litera-
ture. Attempts to increase the vortex to the length
needed to separate gas from the higher viscosity fluids
led to an instability when the gas flowed at the rates
needed. However, removal of the core from the annular
recovery hub resulted in satisfactory vortex stability
over the full gas flow range. A test is being planned to
compare the bubble formation and coalescence proper-
ties of molten salt with those of demineralized water
and other solutions.

The conceptual system design description and the
quality assurance plan for the molten-salt loop for
testing gas systems (GSTF) were completed, and the
design is proceeding. '

A report is being written on the design bases for
molten-salt reactor off-gas systems.

The development of a computer program for the
design of charcoal beds for MSR off-gas systems was
started. The plan is to develop a new program by
revising and expanding existing programs. The existing
programs provide necessary information, but they are
not arranged to be of maximum usefulness to the
designer..

Foster Wheeler Corporation was chosen as the com-
pany to perform the conceptual design study of the
steam generator for use with molten-salt reactors. The
scope of work finally included in the request for

proposal package differed from that reported pre-

viously. The changes basically limit the industrial firm
to work on the steam generator and exclude the
systems design work originally requested.

Work continued on the preparation of a conceptual
system design description for the steam generator
technology facility (SGTF). This facility would be a
side loop of the coolant-salt technology facility (CSTF)
and would be used to study transients and steady-state
operation of some full- and part-length tube test
sections of a molten-salt steam generator. A study was
made of the configurations which could be used in the
SGTF and the larger (3 MW) steam generator tube test
stand (STTS). It was found that only the %- and
3, -in.-diam steam tubes could be tested in the 150-kW
SGTF at full length and that larger tubes could be
tested only under limited inlet and outlet conditions.

X
o,

However, multiple tubes and tubes up to about 1 in. in
diameter could be tested in the STTS.

The PKP pump rotary element was disassembled, and
the component parts were visually examined. Except
for the inner heat baffle plates, the appearance of the
Inconel metal surfaces indicated insignificant corrosive
attack. The inner heat baffle plates were severely
attacked, and this effect was attributed to the forma-

~ tion of corrosive products by reaction of moisture in

the incoming purge gas with puddles of NaBF, salt
which were present on the heat baffle plates as a result
of ingassing transients.

The mechanical design of the CSTF is nearing
completion. The design now includes the corrosion
product trap for studying the deposition of corrosion
products and the on-line monitoring station for study-
ing the electrochemical and surface properties of the
salt. The parts of the MSRE to be used in the
construction have been removed and cleaned in prepara-
tion for assembly. Fabrication of the drain tank and the
specimen surveillance station was completed. Fabrica-
tion of the corrosion product trap and the on-line
monitoring station was started.

The pump requirements for the CSTF and the GSTF
were reviewed, and a plan for adapting the pumps from
the MSRE program was developed. It was found
desirable to alter the MSRE Mark II pump by use of
available equipment to provide additional head.

After making temporary repairs to the ALPHA pump,
it was installed in the MSR-FCL-2 test facility and is
being operated at test program design conditions.

A study of a particular jet pump system (two pumps
in series) for returning fuel salt from the drain tank to
the fuel-salt system in a Molten-Salt Demonstration
Reactor indicated that the system was hydraulically
suitable for the application.

4. Instrumentation and Controls

The hybrid computer model for use in the control
studies of the MSBR is now operational.

Part 2 of the report -on MSRE nuclear and process
instrumentation and an updated report on instru-
mentation and controls development for molten-salt
breeder reactors were completed and prepared for
publication.

5. Heat and Mass Transfer and Physical Properties

Heat transfer. The influence of the magnitude of the
temperature coefficient of viscosity on the variation of
the ‘heat transfer coefficient along a ‘tube has been
determined with and without a hydrodynamic (un-

xiii

heated) entrance, using a proposed MSBR fuel salt as

* the working fluid. Experiments were performed at two

temperatures, 1450 and ~1100°F, for which the
temperature coefficients of viscosity differ by a factor
of 4, the higher coefficient corresponding to the lower
temperature. Results indicate that, in the transitional
flow regime of these experiments (Reynolds modulus
3000 to 4000), increasing the negative temperature
coefficient by decreasing the average temperature re-
sults in a reduction in the heat transfer coefficient near
the exit of the heated length (120 L/D) for operation
without a hydrodynamic entrance region. The influence
of temperature for operation with an entrance region
was found not to be significant beyond L/D of about
40. The observed influence of temperature on heat-
transfer development is attributed to the stabilizing
effect of heating for the case of a fluid whose negative
temperature coefficient of viscosity is relatively large.

Thermo physical properties. New measurements have
been made of thermal conductivity of the mixture
LiF-BeF, (66-34 mole %) over the range 300 to 870°C
using an improved variable-gap apparatus capable of
+10% or better accuracy. These new results suggest that
the temperature dependency of thermal conductivity
for this mixture is smaller than previous measurements
had indicated. The ratio of the thermal conductivity of
the liquid to that of the solid was determined to be
~0.75.

Mass transfer to circulating bubbles. Mass transfer
coefficients have been measured and correlated for
helium bubbles (mean diameter from 0.015 to 0.05 in.)
extracting dissolved oxygen from glycerol-water mix-
tures (Schmidt modulus range from 419 to 3446) over a
Reynolds modulus range from 8.1 X 10° to 1.6 X 10°
for both horizontal and vertical flow. At sufficiently..
high flow, the horizontal -and vertical results are
equivalent and are correlated in terms of the dimension-
less Sherwood, Reynolds, and Schmidt moduli and-the
ratio of bubble to pipe diameter. :

A criterion for predicting the condition for equiva-
lence is presented; and use is made of the criterion as a
scaling factor to correlate the mass transfer coefficients

" in vertical flow in the transitional region between

turbulence-dominated and qu1escer1t bubble rise situ-
ations.

PART 2. CHEMISTRY
6. Postoperational Examination of MSRE

A graphite stringer removed from the core of the
MSRE after being exposed to fissioning molten salt for
the total duration of MSRE operation appeared to be in
generally good condition, with sharp corners, clean
surfaces, and no visible evidence of attack by corrosion
or radiation. X-ray diffraction analysis of material from
the surface suggested that Mo, C, Ru metal, Cr,C;, and
NiTe, may have been present but that Mo metal, Te
metal, Cr metal, CrTe, and MoTe, were probably
absent. Gamma scanning of cross sections of the
stringer by a pinhole technique showed the deposition
of 125Sb and ! °®Ru to be limited to the surface and to
be very spotty and erratic.

Samples milled from the surface to the middle of the
graphite stringer were dissolved and analyzed radio-
chemically for many long-lived fission product species.
The samples were also analyzed chemically and by
delayed neutron counting for U, Li, Be, Zr, Fe, Ni, Mo,
and Cr by spectrographic methods. The observed
concentration profiles for these species were generally
in agreement with behavior previously observed for the
* surveillance specimens. The uranium analyses indicated
that there was a total of about 10 g of 233U on and in
all of the graphite in the MSRE. The spectrographic
analyses indicated traces of nickel in a few of the
outermost surface samples. The radiochemical analyses
showed that the ‘“noble metals” (*°®Ru, !23Sb,
110Ag, 95Nb, and '*7Te) were concentrated at the
surface, that the °°Sr (33-sec ®°Kr precursor) had a
much steeper concentration profile than ®°Sr (3.2-min
89Kr precursor), and that the profiles for '34Cs and
137Cs indicated appreciable diffusion of cesium toward
the fuel salt. The ®°Co analyses indicated that some
nickel had deposited on the graphite surfaces.

Careful analyses for. tritium in the MSRE graphite
yielded the surprising result that nearly 15% of the
tritium produced in the MSRE was retained by the
graphite.- About one-half of this tritium was trapped in
the outer 0.15 cm of the graphite stringer.

Concentration profiles for two cesium isotopes,
137Cs and '3*Cs, were obtained on the graphite bar
from the MSRE core center, removed in early 1971.
These profiles demonstrate diffusion of cesium atoms,
probably on internal graphite surfaces, after adsorption
at rates consistent with estimates based on data from
gas-cooled reactor studies.

The recovery of segments of surfaces from the MSRE
fuel circulating system in January 1971 has permitted
the determination of intensity of deposition of fission

product isotopes in the core, heat exchanger, and pump

bowl. Antimony, tellurium, technetium, ruthenium,
and niobium were found deposited strongly on both
metal and graphite surfaces, with intensities somewhat
greater than previously observed on surveillance speci-
mens.

Xiv

Cobalt-60, found on segments of MSRE heat ex-
changer tubing and core graphite bar, must have come
from neutron irradiation of Hastelloy N, with subse-
quent transport and deposition. Higher values on core
graphite indicates further activation after deposition.

General metal loss and deposition are indicated. The

reactor vessel walls are suggested as a source, the
transported metal possibly being sputtered by fission
fragments from adjacent fuel. -

7. Hydrogen and Tritium Behavior in Molten Salts

Modifications were made to the experimental appara-
tus in an effort to improve the reproducibility of the
data. Results for helium solubility in Li, BeF, indicate

that these efforts have been successful; although the

hydrogen solubility data still show an uncomfortable
degree of scatter, we believe this to be due to air
inleakage over prolonged periods. The effects of such
inleakage can be corrected by a straightforward but
tedious procedure.

Tests of the efficiency of hydrogen recovery in the
stripper chamber of the apparatus indicate at least 90%
of the gas can be recovered within a 2-hr period of
operation. Moreover, in contrast with our previous
(probably erroneous) result, only 5% of the hydrogen
recovered resulted from the stripper annulus col-

' lections.

Measurements of the transport of hydrogen through
Kovar metal at 495°C indicate that the pressure
dependence for the rate of transport can be described
by P?, where n = 0.560 £ 0.011 over the pressure range
1.4 torrs < P < 800 torrs. The product of the diffusion
coefficient and solubility constant characteristic of this

‘system.under the conditions cited was found to be DK

=(5.18 £0.24) X 107! moles/cm-min-torr”.

In studies with a time-of-flight mass spectrometer,
pure NaBF;OH was shown to decompose completely
under continual pumping at 100°C. The volatile
product was pure water, and the yield was precisely 0.5
mole H, O per mole of NaBF; OH. Continued heating to
800°C yielded only BF5, whose evolution began at
about 245°C, and NaF, whose vapor appeared at
725°C.

In preliminary experiments with the TOF mass
spectrometer, pure anhydrous sodium nitrate was
shown to evolve only NO and O, over the temperature
range 380 to 400°C.

The use of a dissociating-gas heat transfer system
using nitrogen dioxide (N, 0, = NO, =NO+0,) (7~
N; + O,) under development in the U.S.S.R. appears to
offer possibilities to the Molten-Salt Breeder of miti-
[ K}

gation of tritium losses from the power system. Under
dissociating conditions the effective heat capacity and
effective thermal conductivity of the gas are increased
severalfold, resulting in increased efficiency of heat
transport, higher cycle efficiency, and the p0351b111ty of
lower equipment costs.

8. Processing Chemistry

As expected, it has been found that the precipitation
of Pa** from a molten mixture of LiF-BeF,;-ThF,
produces a ThO,-Pa0, solid solution. From measure-

.ments of the distribution of Pa** to this phase it can be

predicted that in the event of an accidental contami-
nation with oxide of an MSBR fuel, the precipitated
phase would consist mainly of a UO,-ThO, solid
solution with little PaQ, dissolved in it.

In further studies of the precipitation of the much
less soluble Pa(V) oxide, its solubility. in the presence of
UF, and a UQ,-rich oxide phase has been found close
to the predicted value, based on measurements in the
absence of uranium. These results thus confirm that
Pa%* can be precipitated by oxide from an MSBR fuel
salt while precipitating no other oxide.-

A ‘comparative - evaluation of bismuth-lead eutectic
with pure bismuth for application in the reductive
extraction process was made. Distributions of barium
and thorium between the eutectic melt and LiF-BeF,-
ThF, (72-16-12 mole %) were determined at 650°C
after successive additions of lithium metal were made.
The equilibrium quotient for barium, Dg,/
(D )* =9.81 X 10%, from this experiment was 2.4
times less than that reported for pure bismuth. The
equilibrium quotient for thorium, Dpy,/(Dy;)* = 6.64 X
10%, was 3.7 times greater than reported for this salt
and pure bismuth. The indicated solubility of thorium
in the eutectic at 650°C was approximately 1500 ppm
by weight or about one-half that of thorium in pure
bismuth.

The removal of cerium from molten LiCl onto Linde
4A, sodium form, No. 8 mesh zeolite was demonstrated
at 650°C. For a I-kg batch of LiCl-CeCl;y (99.5-0.5
mole %) and successive exposures to 14-g aliquots of
dried zeolite, loading capacities of about 0.13 g of
cerium per gram of zeolite were found. Within the
precision of the data the distribution of cerium between
the molten chloride and solid zeolite phases could be
expressed  empirically as In  (ppm Ce in
LiCl)=12.163 + 1.35 In (g Ce/g zeolite) for cerium
concentrations down to 4600 ppm. Analyses of the salt
phase showed essentially complete exchange of lithium
for sodium in the zeolite.

Xv

9. Development and Evaluation of Analytical
Methods for Molten-Salt Reactors

The first successful in-line analyses of a flowing salt
stream are now being ‘performed. Trivalent uranium is
being determined in an MSRE-type salt in a thermal
convection loop using the voltammetric technique.
Concentrations of U(IIl) from 0.02 to 0.15% of the
total uranium have been measured. For these deter-
minations we are using a special voltammeter that has
been modified for operation with the counter electrode
at ground potential and can be operated by a PDP-8I
computer.. This system is operated automatically and
performs analyses at hourly intervals. During one 72-hr
period of constant U(III) concentration the precision of
the determinations was about 2%. This.operation has
uncovered two-problems not encountered in quiescent
laboratory melts: interference from particulate matter
on the melt’s surface and an unexpectedly high effect
of surface vibrations on the electrode response. These
problems have largely been resolved by shielding the-
working electrode in a nickel tube which opens below
the surface of the melt and is purged of salt by a stream
of helium until immediately before the analyses are
made. This feature should also extend electrode life and
will be incorporated in future in-line systems. :

We are developing a simple slotted optical probe that
will prove useful for the analysis of fluid streams. Its
application to MSRP streams will depend on the
discovery of an optical material that is impervious to
molten fluorides. At present, LaF; appears to be the
best candidate for NaBF, streams.

A large part of our development effort was directed
toward the analysis of NaBF,, particularly for proton-
ated species which may be of significance in the tritium
containment problem. A “dewpoint” technique appears
feasible for the in-line measurement of hydrolysis
products in coolant cover ‘gas. Both electrical and
optical methods are being tested. For calibration-of this
system and other water determinations a special Karl
Fischer method is being developed. An improved
coulometric titrator for this analysis has been fabricated
and is almost ready for use.

We are now routinely determining OH™ in NaBF,
samples to low ppm levels by measuring the absorbance
of pressed pellets at 3641 cm™". To confirm tentative
calibration factors for this infrared method we are using
an isotopic dilution technique for the measurement of
total hydrogen in NaBF,. Initial results are in excellent
agreement with those of the infrared method; however,
the blanks resulting from the extraction of hydrogen
from the Pyrex equilibration ampule are so high that
results are subject to question. Additional measure-
ments will be done in quartz ampules. _

A novel electroanalytical method is based on the
diffusion of hydrogen into a hollow palladium electrode
when NaBF, melts are electrolyzed at a controlled
potential. The measurement of the pressure generated
in the electrode is a sensitive measure of protons at ppm
~.concentrations. . The technique offers advantages of
specificity, “applicability to in-line analysis, and the
possibility of a measurement of H?/H ratios in the
coolant.

A general voltammetric study in progress on molten
NaBF, has shown a working voltage span (vs a quasi
. reference electrode) from —1.2 V, limited by the
deposition of boron, to +1.8 V where dissolution of the
platinum electrode occurs. Waves for Fe** - Fe?* and
Fe?* to Fe® are observed at approximately —0.2 and
—0.4 V respectively. The Ti** = Ti** wave occurs at
—0.5 V while the Ti** = Ti° reduction is beyond the
cathodic limit of the melts.

To complete the fundamental study of the electro-

xvi

A systematic study of the solubility of BF; in
LiF-BeF, mixtures of- varying composition is con-
tinuing. These measurements, which are also capable of
defining free F~ activity in solvent melts, continue to

“show that over the composition range 63 to 85 mole %

LiF the BF; solubility varies nearly linearly with the
thermodynamic activity of LiF. Enthalpies of solution
and mole entropies of transfer of BF; from the gas to

. solution at the same concentrations have been evalu-

analytical chemistry of nickel in LiF-BeF;-ZrF,, dif- -

fusion coefficients (D) of Ni(Il) were measured at
platinum and pyrolytic graphite electrodes. At 500°C
values of D of 1.05 X 107® and 1.07 X 107% cm?/sec
were found by voltammetry and chronopotentiometry
respectively. There was no evidence of alloy formation
when nickel was deposited on and stripped from a
platinum electrode.

10. Other Fluoride Researches

Spectrophotometric techniques using the diamond
window cell have been used to determine the ratio of
UF; to total dissolved uranium in molten mixtures of
LiF and BeF, in contact with graphite. The data
obtained for these mixtures agree well with values
calculated using activity coefficients obtained by other
workers. However, data obtained in LiF-BeF,-ThF,
mixtures by the spectrophotometric method agree
much less well with the calculated values. Attempts in
this study to unambiguously identify the stoichiometry
of the uranium carbide phase in equilibrium with the
UF;-UF, solute have not been successful.

Similar spectrophotometric techniques have been
used to define the equilibrium .

UFs* = UF,> +F~

in mixtures.- of LiF and BeF, as a function of
compositions. It appears that a quantitative ranking of
fluoride solvents as to their activity of free F~ may be
possible by use of this method.

ated.
Study of the reactions

2MoF;(s) = Mo(s) + MoF(g)

and
4 1
§—M0F3(s) ?—‘;Mo(s)f MoF,(g),

in which accurate measurements of equilibrium pressure
of the product gases were made, has permitted assess-
ment of the standard free energy changes for these
reactions and evaluation of the free energy of formation
of MoF3 and MoF,. The free energy of formation (per
fluoride bond) at 700°C [appears to be —50.5 kcal for
MoF4(g), -55.0 = 0.3 kcal for M0F3(s) and —49 1
kcal for MoF,4(g).

An improved procedure for synthesis of pure NbF,
was developed and employed during this reporting
period. The vapor products from reaction of Nb,Ojs

‘and F, have been determined over the temperature

range 375 to 550°C, and the gaseous products of
reaction of NbFs with Nb;Og have been studied over
the temperature range 350 to 750°C.

Continuing the investigation of electrical conductance
and. its relation to glass transition temperatures, mea-
surements were made of the temperature dependence of
conductance in molten and supercooled NaF-BeF,
mixtures containing 65 and 70 mole % beryllium
fluoride. Temperatures ranged from 540 to 329°C,
105° into the supercooled region, while activation
energies varied from 10.5 to 15.9 kcal/mole. Measure-
ment of glass transition temperatures Tg in this system
indicated that the region of bulk-quenched glass forma-
tion extends from 37 (*1) to 100 mole % BeF,, Tg
values falling from 129 (+2)°C-at Xp,p, =0.38 to 11.7°
at XBeF = 0.50 and remaining relatively constant
(£2°) to XB g, = 090, beyond which the thermal
effect could not be discerned.

The enthalpy of LiBF; was measured from room
temperature to temperatures above the melting point.
No evidence for a solid transition was found. The

v
I ¥

results are compared with those for the other alkali
metal fluoroborates.
The phase diagram of Li,BeF,-Lil was determined.

The interesting deviations from ideality of both the

Li, BeF, liquidus and the Lil liquidus are discussed.

PART 3. MATERIALS DEVELOPMENT

11. Examination of MSRE Components

Examination of specimens from the mist shield of the
MSRE pump bowl, a rod from the sampler cage in the
pump bowl, and rings from the control rod thimble that
had restricted salt flow showed that all components had
intergranular cracks when strained. Two samples from

in-pile loop 2 were_ strained, and some shallow cracks

were formed in both samples. Thus, all metal surfaces

exposed to the fuel salt have intergranular cracks.
Fission products on graphite spedimens were meas-

ured by two techniques. Significant quantities of

elements from the Hastelloy N and fission products

were observed by both techniques.

12. Graphite Studies’

"The large range of graphitic materials which have now
been irradiated in HFIR has permitted certain general-
izations as to irradiation behavior. Two classes, so-called
conventional materials and hot-worked materials,
appear to have quite limited resistance to radiation
damage. The other two, apparently binderless graphites
and black-based graphites, appear to offer room for
substantial improvement. Both types of material are
being studied in our own fabrication program. Speci-
mens of the binderless type that we have made have
already been irradiated to 1 X 10*? neutrons/cm? in
the HFIR, and to that level appear to be stable.

Increased efforts are being directed at sealing the
graphite against fission product gases. We continue to
have somewhat mixed results with irradiated sealed
samples, and in no case is there an unqualified
successful sample. A major effort is now proceeding on
microstructural examination of the coated and surface-
impregnated samples to determine the modes of failure.

The *thermal conductivities of potentially useful
graphites are now being measured as a function of
fluence. An apparatus to similarly follow radiation-

induced changes in thermal expansion is almost com-

plete. 7

The calculation of strain fields about an interstitial
aggregate has been completed, and electron diffraction
calculations are well under way.

Xvil

13. Hastelloy N

The effects of Hf, Nb, Ti, Si, and C on the
microstructures of several small melts have been evalu-
ated. The fine MC-type carbide was characteristic of the
alloys low in Si and containing low to intermediate
amounts of C. Higher C resulted in relatively coarse
M,C, and Si caused the formation of coarse M¢C.
Commercial alloys have these same microstructures
except for the ones containing Hf, where the precipitate
is not generally dispersed. Many of these new alloys
have exceptional unirradiated strengths. The postirradi-
ation properties generally agree with the micro-
structural observations with good properties being
associated with the fine generally dispersed MC car-
bides. : ‘

Corrosion studies have continued to show excellent
compatibility of Hastelloy N with salts composed of
LiF, BeF,, UF,, and ThF,. The corrosion of Hastelloy
N in sodium fluoroborate seems to be dominated by
impurities. Four thermal convection and two pump
loops are being used to evaluate fluoroborate corrosion.
Heat transfer measurements have been made in the
second pumped system, and they show that sodium
fluoroborate obeys the Sieder-Tate correlation over a
wide variety of conditions. ) ,

Unstressed Hastelloy N continues to show good
compatibility with steam, but two tube burst specimens
have failed prematurely when exposed to steam under
stress. The duplex nickel 280-Incoloy 800 being
evaluated for steam service shows poor creep ductility
on the nickel 280 side. This can likely be improved by
changes in the fabrication procedure.

14. Support for Chemical Processing l

We have continued fabrication of components for the
molybdenum test stand. All of the 3%-in.-diam closed-
end half sections for the feed pots and column
disengagement containers, including sections up to 12
in. long, have been fabricated by back extrusion and
machined. A roll-bonding joining technique was de-
veloped for attaching Y-, %-, and '4-in.-OD lines to the
37%-in.-OD containers where - electron-beam welding
could not be used. Leak-tight roll-bonded joints have
been made using commercial tube expanders at 250°C
in an inert gas environment.

Joining studies were continued on tube-to-header,
header-to-header, and the tube-to-tube types of joints
required for the loop. A procedure was developed for
electron-beam welding the 7;-in.-diam tube-to-header
type of joint. Three sets of back-extruded half sections
xviii

were successfully joined by either electron-beam or
GTA girth welds. Development of tube-to-tube field
welding techniques using the orbiting arc welding head
was continued by joining longer lengths of tubing in a
vertical alignment. ‘

The iron-base alloy Fe—15% Mo—5% Ge—4% C—1% B
has been selected as the filler metal for back-brazing
joints in the molybdenum loop. Several methods of
brazing using resistance and high-frequency induction as
heating sources have been investigated. Techniques for
induction brazing in a dry box and induction field
brazing have been developed.

Compatibility studies of molybdenum, tantalum,
T-111, and several grades of graphite were continued in
quartz thermal convection loops operated for 3000 hr
at 700°C maximum temperature and 100°C AT. Of the
materials tested, molybdenum and pyrolytic graphite
have been least affected by Bi—100 ppm Li solutions.
T-111 shows good resistance to mass transfer but
becomes embrittled during test. Unalloyed tantalum has
exhibited much higher mass transfer rates than the
above-mentioned materials. Grades of graphite having
available open pores tend to allow bismuth to pene-
trate, although no other evidence of interaction has
been noted.

Chemical vapor deposition studies on molybdenum
and tungsten coatings have indicated that the coatings
are less adherent to electroless nickel-coated specimens
compared with electrodeposited nickel-coated speci-
mens. Tungsten CVD coatings were successfully applied
to several roll-bonded joints, but attempts to coat a
Hastelloy N thermal convection loop resulted in cracks
at the corners of two joints where the tubing was
welded together.

PART 4. MOLTEN-SALT PROCESSING AND
PREPARATION .

15. Flowsheet Analysis

A study was completed in which the capital and
operating costs were determined for a plant that

processes the fuel salt from a 1000-MW(e) MSBR on a

ten-day cycle. The processing plant is based on the
reference flowsheet that uses fluorination for uranium
removal, reductive extraction for protactinium removal,
and metal transfer for rare-earth removal. The latest
version of the flowsheet incorporates an improved
method for recovering UF¢ from the fluorinator off-gas
.and a method for recycling fluorine, hydrogen, and HF
in the plant in order "to minimize the quantity of
radioactive waste produced.

In preparing the cost estimate, preliminary designs
were made for all major process equipment items, and
the costs of piping, instrumentation, and certain auxil-
iary items were determined by using appropriate -per-
centages of the costs of the fabricated equipment items.
The total direct and indirect costs for the plant are
$20.6 and $15 million respectively. Thus, a total plant
investment of $35.6 million is required. The resulting
net fuel cycle cost (1.12 mills/kWhr) is composed
largely of fixed charges on the processing plant (0.7
mill/kWhr), inventory charges on salt and fissile ma-
terials in the reactor and ?3?Pa in the processing plant
(0.42 mill/kWhr), and processing plant operating
charges (0.083 mill/kWhr); credit is taken for excess
fissile material produced (0.089 mill/kWhr). A 0.28
power dependence of processing plant cost on process-
ing rate was estimated; this indicates that a considerable
saving in processing cost could be achieved by using a
processing plant with a larger power generation capacity
than 1000 MW(e).

Capital and operating costs were calculated for two

methods for obtaining and disposing of hydrogen, HF,

and fluorine in a processing plant. The first method,
purchase and disposal of the gases after one cycle
through the process, requires large amounts of F,, H,,
and HF to be purchased and a large amount of
radioactive waste to be generated. The second method,
collection of the HF and H, and production of
additional H, and F, for recycle by electrolysis of the
HF, requires the purchase of only small amounts of HF
and H, and produces only a small quantity of radio-
active waste. Contributions to the fuel cycle cost for
the once-through case and the recycle case are 0.11 and
0.024 mill/kWhr respectively. The recycle case was
adopted for use in the processing plant.

Calculations were made of the thermal power that
will be produced in an MSBR processing plant by the
decay of noble-metal fission products as a function of
the residence time of these materials in the primary

- reactor system and of the fraction of the materials

accompanying the fuel salt to the processing plant. A
heat generation rate of 200 kW will result if 1% of the
noble metals reach the processing plant in the case of a
l-min holdup time, or if 20% reach the processing plant
in the case of a ten-day holdup time. Calculations were
also made of the heat generation rate resulting from

‘decay of the halogen fission products in the processing

plant as a function of the halogen and the noble-metal
removal times. The heat generation rate will be 210 kW
for a halogen removal time of ten days if the noble-
metal holdup time in the reactor is 0.1 day.
Calculations were made of the long-term hazard of
high-level = radioactive wastes produced in the three
nuclear fission fuel cycles under development. Several
aspects of the MSBR fuel cycle related to waste disposal
problems are considered.

A survey was made of published data on the
availability of and future demand for natural resources
required for an MSBR power economy. Consideration is
given to the possible impact on the MSBR concept of
the depletion of world reserves of natural resources
required by molten-salt breeder reactors.

16. Processing Chemistry

Measurements were made of the equilibrium distri-
bution of lithium and bismuth between liquid lithium-
bismuth alloys and molten LiCl, and of the mutual
solubilities of thorium and selected rare earths in liquid
bismuth. . These studies relate to the development of the
metal transfer process for removing rare earths from
MSBR fuel salt. '

Protactinium -pentoxide was selectively precipitated

‘from LiF-BeF,-ThF, (72-16-12 mole %) that contained

up to 0.25 mole % UF, by sparging the salt with
HF-H,O-Ar gas mixtures at 600°C. The equilibrium
quotient for the reaction PaFg(d) + *%H,O(g)= "%
Pa, O;(s) + 5HF(g) was determined to be about 3 at

600°C. The results of this work are consistent with.

earlier indications that Pa, O5 can be precipitated from
MSBR fuel salt as a pure, or nearly pure, solid phase.

Studies relating to the chemistry of fuel recon-
stitution were initiated. Gaseous UF4 reacted quantita-
tively with UF, dissolved in LiF-BeF,-ThF, (72-16-12
mole %) at 600°C according to the equation
UF¢(g) + UF4(d) = 2UF;(d). The dissolved UF;s ap-
peared to be stable in both gold and graphite equip-
ment. o

17. Engineering Development of Processing
' ' Operations

Additional information on the behavior of lithium
was obtained during metal transfer experiment MTE-2.
During this experiment, the lithium concentration in
the Li-Bi solution used to extract rare earths from the
lithium chloride decreased from an initial value of 0.35
to 0.18 mole fraction after about 570 liters of LiCl had
been contacted with the Li-Bi solution. The decrease in
lithium concentration was shown to be consistent with
recently obtained data on the concentration of lithium
in LiCl that is in equilibrium with Li-Bi solutions.

XiX

A new experiment (MTE-2B) is presently under way
to study further the transfer of lithium from a Li-Bi
solution containing lithium at the concentration pro-
posed for extracting trivalent rare earths from LiCl (i.e.,
5 at. %). This experiment is designed in a manner such
that data on lithium transfer can be obtained by several
independent methods. The data obtained thus far show
a decrease in the lithium concentration in the Li-Bi
solution that represents an equilibrium lithium concen-
tration in the LiCl of less than 0.3 wt ppm. Since the
corresponding quantity of reductant did not transfer to
the Th-Bi solution in the experiment, it is believed that
part of the decrease is due to a slight loss of reductant
from both the Li-Bi and the Th-Bi solutions as the
result of the reaction of reductant with impurities in
the system. It is believed that further operation of the
experiment will show lower equilibrium lithium concen-
trations in the LiCl.

The design of the third engineering experiment for
development of the metal transfer process has been
completed, and most of the equipment has been
fabricated. The main process vessels are now being
installed. This experiment will use flow rates that are
1% of the estimated rates required for processing a
1000-MW(e) reactor. Mechanical agitators will be used
to promote efficient contacting of the salt and bismuth

- phases. Tests are under way to evaluate an agitator drive

unit and nonlubricated shaft seal assembly that is
water-cooled and buffered with inert gas. Although
some leakage of gas around the seal has been observed,
it appears that the seal will be satisfactory for use with
experiment MTE-3. No evidence of stable emulsions of
bismuth in salt that had been mechanically agitated was
found. The hydrodynamic performance of mechanically
agitated salt-metal contactors was investigated using
mercury and water to simulate bismuth and molten salt
in order to determine favorable operating conditions.
Tests of several sizes of contactors with differing
agitator arrangements established that the common
factor that limits the agitator speed is entrainment of
water in the mercury circulating between the two halves
of the contactor. Consideration of the data on limiting
agitator speed, along with data from the literature on
mass transfer in this type of contactor, suggests that
optimum mass transfer performance will be obtained by’
the use of the largest possible agitator diameter. This
will also result in the lowest possible agitator speed,
which should facilitate maintaining a gastight seal on
the agitator shaft. -

‘'We have begun mass transfer experiments in which
the rate of transfer of zirconium from molten salt to
bismuth is measured by adding ®7ZrO, tracer to the
salt phase prior to contacting .the salt with bismuth
containing reductant in a 0.82-in.-diam, 24-in.-long
packed column. Three of the experiments resulted in
the transfer of 15 to 30% of the ® 7 Zr present in the salt
and gave measured HTU values that ranged from 1 to 4
ft; the lower values were obtained from data in which
we have the most confidence. These data indicate that
‘packed column contactors will be satisfactory for use in
MSBR processing systems. We believe that much of the
. scatter observed in the mass transfer data obtained thus
far is caused by small amounts of air entering the
flowing stream samplers in which samples of salt and
bismuth are obtained. An improved sampling technique
will be used in future experiments.

We have been investigating high-frequency induction
heating for possible use as a corrosion-resistant heat
source (in the molten salt) in an experiment to
demonstrate protection against corrosion in a con-

tinuous fluorinator by use of a frozen wall. Heat

generation rates were measured with four induction coil
designs in a system that used 31 wt % HNO; as a
substitute for molten salt. The measured heat genera-
tion rates were used to compute correction factors for
use in equations for calculating heat induced in similar,
but idealized, geometries. These equations were then
used to design equipment for a molten-salt induction
heating experiment for verifying the design equations,
and for testing the operation of the proposed heating
method and the means for introducing the power leads
into the fluorinator. The test vessel, which is 5 ft tall
and has a diameter of about 6.5 in., is a short version of
the proposed frozen-wall fluorinator.

A mathematical analysis for predicting continuous
fluorinator performance was completed, and calcula-
tions were made for operating conditions of interest for
MSBR processing. The model assumed that the removal
of uranium from the salt is first-order with respect to
the uranium concentration in the salt, and took into
account the effects of axial dispersion resulting from
the rising gas bubbles. Values for the dispersion
coefficient were obtained from recently developed
correlations on axial dispersion in open columns during
the countercurrent flow of air and aqueous solutions,
and the fluorination reaction rate constant was evalu-
ated using previously obtained data on the performance
of a l-in.-diam continuous fluorinator. The predicted
fluorinator heights required for removal of 95 and 99%
of the uranium from MSBR fuel salt on a ten-day cycle
are 10.2 ft (for a 6-in. diameter) and 17.8 ft (for an
8-in. diameter) respectively. The calculated results are
encouraging since they indicate that the desired ura-
nium removal efficiencies can be obtained with single
. open columns of moderate height.

XX

Fabrication and installation of equipment for a
single-stage experiment to study the. precipitation of
UO,-ThO, solid solutions from molten fluoride salts
have been completed. The major equipment items
performed satisfactorily in the two experiments carried
out to date. In the first experiment, a gas stream
containing 15% water—85% argon was fed to the
precipitator vessel at the rate of 0.5 scfh for a period of
4 hr with the salt at 600°C. Analyses of samples
indicate that 17% of the uranium initially present in the
salt was precipitated during the run and that a water
utilization of about 25% was obtained. In the second
experiment, the temperature of the salt and the
composition of the inlet gas were the same as in the
first experiment. The salt was contacted with the gas at
the rate of 0.5 scfh for 9 hr, after which the gas flow
rate was increased to 1.5 scfh for an additional 5 hr.
The total precipitation time (14 hr) extended over a
period of about ten days. During this time the run was
interrupted on several occasions for minor equipment
modifications. Analyses of samples and measurements
of HF evolution throughout the second experiment
indicate that more than 50% of the uranium was

. precipitated. ' ;

The detailed designs of the extraction column and the
salt and bismuth head pots were completed for the
molybdenum reductive extraction facility, and fabri-
cation of these items was started. Conceptual design
sketches for the equipment supports in the containment
vessel and preliminary drawings for the molybdenum
piping arrangement inside the containment vessel were
completed. Further work on these drawings will be
deferred until the piping can be analyzed for thermal
expansion stresses and the full-scale mockup can be
completed to investigate fabrication and field assembly
problems.

An eddy-current type of detector is being developed

‘to allow detection and control of the bismuth-salt

interface in salt-metal extraction columns or mechani-
cally agitated salt-metal contactors. The probe consists
of a ceramic form on which bifilar primary and
secondary coils are wound. A high-frequency alter-
nating current is passed through the primary coil, and a
current that is dependent on the conductivities of
materials adjacent to the primary and secondary coils is
induced in the secondary coil. Initial tests showed that
the probe was quite sensitive to drift in the electronic
circuit, to minor variations in line voltage, and to
changes in temperature of the electronic components.
To circumvent this problem, an improved electronic
circuit that is relatively insensitive to these effects was
devised. Calculations for predicting the performance of
interface detectors were made based on measurements
2

of either the magnitude of the current induced in the
secondary coil or the shift in phase between the
primary and secondary voltages. Optimum operating
conditions were defined, and a prototype probe for use
in the molybdenum reductive extraction column was
fabricated. Tests carried out at room temperature
confirmed the predicted performance of the probe.
Equipment has been installed for final testing of the
probe at 600°C with molten bismuth.

18. Continuous Salt Purification

Contamination was found to be the reason for the
occasional high iron concentrations reported for salt

samples from previous experiments. This contamination
apparently occurred during removal of salt from the
nickel samplers. To eliminate this problem, the sampler
design was modified, and copper was used as the
material of construction. Five iron fluoride reduction
runs were made with iron concentrations of about 700
to 900 ppm. The gas flow rates used in these runs were
only about 20% of previous values since the packed
column had become partially restricted. After the runs
had been completed, the column was removed and cut
into sections for examination. A heavy deposit of nickel
and a near-uniform deposit of iron were noted on the
packing. A column of slightly modified design has been
fabricated and installed.
Partv 1. MSBR Design and Development

R.B. Briggs  P. N. Haubenreich

The design and development program has the purpose
of describing the characteristics and estimating the
performance of future molten-salt reactors, defining the
major problems that must be solved in order to build
them, and designing and developing solutions to prob-

- lems of the reactor plant. To this end we have published

a conceptual design for a 1000-MW(e) plant and have
contracted with an industrial group organized by
Ebasco Services Incorporated to do a conceptual design
of a 1000-MW(e) MSBR plant. This design study is to
use the ORNL design for background and to incor-
porate the experience and the viewpoint of industry.
One could not, however, propose to build a
1000-MW(e) plant in the near future, so we are doing
studies of plants that could be built as the next step in
the development of large MSBRs. One such plant is the
Molten-Salt Breeder Experiment (MSBE). The MSBE is
intended to provide a test of the major features, the
most severe operating conditions, and the fuel reproc-
essing of an advanced MSBR in a small reactor with a
power of about 150 MW(t). An alternative is the
Molten-Salt Demonstration Reactor (MSDR), which
would be a 150- to 300-MW(e) plant based largely on
the technology demonstrated in the Molten-Salt Reac-

. tor Experiment, would incorporate a minimum of fuel
reprocessing, and would have the purpose of demon-

strating the practicality of a molten-salt reactor for use
by a utility to produce electricity. In addition to these
general studies of plant designs, the design activity
includes studies of the use of various fuel cycles in

molten-salt reactors and assessment of the safety of
molten-salt reactor plants. The fuel cycle studies have
indicated that plutonium from light-water reactors has
an economic advantage over highly enriched 23U in
fueling molten-salt reactors. Some studies related to
safety are in progress preliminary to a comprehensive
review of safety based on the design of the 1000-MW(e)
MSBR. : '

The design studies serve to define the needs for new
or improved equipment, systems, and data for use in
the design of future molten-salt reactors. The purpose
of the. reactor development program’is to satisfy some
of -those needs. Presently the effort is concerned largely.
with providing solutions to the major problems of the
secondary System and of removing xenon and handling
the radioactive off-gases from the primary system. Work
is well along on the design of the one loop facility for
testing the features and models of equipment for the
gaseous fission product removal and off-gas systems and
for making special studies of the chemistry of the fuel
salt. Design is nearing completion and construction-is in
progress on a second loop facility for studies of
equipment and processes and of the chemistry. of
sodium fluoroborate for the secondary system of a
molten-salt reactor. The steam generator is a major item
of equipment for which the basic design data are few
and the potential problems are many. A program
involving industrial participation is being undertaken to
provide the technology for designing and building
reliable steam generators for molten-salt reactors.
1. Desi_gn

E. S. Bettis

1.1 MOLTEN-SALT DEMONSTRATION REACTOR
DESIGN STUDY

E. S. Bettis

L. G. Alexander  H. A. McLain
C. W. Collins J. R. McWherter
W.K. Furlong  R.C. Robertson
' H. L. Watts

1.1.1 General

Design and evaluation studies of a 300-MW(e) mol-
ten-salt demonstration reactor (MSDR) were continued.
This reactor would be a first-of-a-kind prototype to
demonstrate the feasibility and to delineate the prob-
lems of construction and operation of a large-scale
molten-salt reactor power station. Insofar as possible,
all aspects of the MSDR design study have been made
conservatively and in such a manner as to require
development of little new technology. The 26-ft-diam
reactor has a sufficiently low power density (~10
W/cm?) for the radiation damage to the core moderator
to be low enough for the graphite to last the 30-year
‘design life of the plant. The MSDR reactor outlet
temperature was taken as 1250°F to increase the
margin of safety with regard to radiation damage to
Hastelloy N, although it has not been established that
this is required. This reduction in temperature, together
with use of 900°F steam rather than the 1000°F steam
proposed for the MSBR, will allow the nitrate-nitrite
salt in the tertiary system to operate at a lower
temperature, although, here again, these are not neces-

sarily requirements for an MSBR. It was also decided to
operate the steam system at 2400 psia (with reheat to
900°F), again in line with using conservative design
conditions for the MSDR. The fuel-salt chemical treat-
ment need not consist of more than fluorination, which
is an already well-developed process. Periodic discard of
the LiF-BeF, carrier salt will be used to limit accumula-
tion of fission products. The built-in safety features of
the MSDR give it a ‘“walk-away” capability — that is,
the decay heat removal system will operate even with
complete loss of station power.

The reactor vessel, the reactor containment and plant
building layouts, the fuel-salt storage tanks, and prelimi-
nary primary and secondary heat exchanger design
concepts have all been described previously.'™ The
design properties of the fuel salt and the secondary and
tertiary heat transport salts have been reported.?
During the past report period, however, the configura-
tion of the core graphite was changed, and the method
of handling the off-gas and continuously pumping salt
back from the drain tank was revised. The drain tank
cooling system, along with the drain tank valve and salt
catch pan arrangement, was also studied and probably
received the major portion of the MSDR design effort
during the past period.

1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970,
ORNL-4548.

2. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970,
ORNL-4622. :

3. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676.

[
1.1.2 Reactor Core

As described previously,? the MSDR réactor vessel is
made of Hastelloy N and is about 26 ft in diameter by
35 ft high, including the fuel-salt inlet and outlet
plenums at top and bottom. The reactor is graphite
moderated and reflected, with the fuel salt circulating
upward through the core at a velocity of about 1.5 fps.
At this low velocity it will not be necessary to seal the
graphite against penetration of xenon in order to
maintain the '3°Xe poison fraction below 0.003. Use
of the unsealed graphite will greatly simplify the
production of graphite and will lower the fabrication
costs. - :

The configuration of the core graphite has been
changed from that previously reported. Slabs of graph-
ite 1*%4 X 9% in. X about 21 ft long are now spaced

vertically in the core as indicated in the plan view, Fig.
1.1, and in more detail in Fig. 1.2. This configuration is
similar to one proposed by the Ebasco Services group in
the current industrial study of an MSBR. The methods
of assembling the core at bottom and top are shown in
elevation in Fig. 1.3. If the slabs are not available in
21-ft lengths, they can be made up of shorter lengths
joined together. Because of the relatively low power
density in the MSDR, fuel salt trapped in small cracks
at the joints would not cause excessive temperatures.
Graphite vertical posts about 27 ¢ X 2% in. in cross
section and formed as shown in Fig. 1.2 are used to
maintain the configuration of the graphite slabs within
the core, while at the same time accommodating
thermal and radiation-induced dimensional changes.
The lower 6-in. length of each post is turned to 1% in.
in diameter to fit into a hole in the bottom graphite

ORNL-DWG 7{-13125

= ; . . .I. .'.'i
inseeeenez

;m?icDMl.l'h.h.N]lihifiifi_o

SileeeeeeeZ
§_AEUIN|UIN|NIN|N"
=i | i 7‘*1

O~ REGULATING
ROD THIMBLES
DO8 S
“‘?j‘i"‘i"
== =le

Fig. 1.1, Plan view of 300-MW(e) MSDR reactor core.
ORNL-DWG 71-13126

7

L)

FahY T,

rd

|

A\

)
o

L L

7

014

Y\
o

8,3 % SALT
VOLUME

N

7 X LW

o

7

rany v

T

2

b

24\

Fig. 1.2. Reactor core cell assembly plan for 300-MW(e) MSDR.

grid plate, and an upper l-in. length has a diameter of
about 7% in. to fit into a corresponding hole in the
upper orifice plate, as shown in Fig. 1.3. The posts and
slabs thus form a cell about 11% X 11'% in. Dowels
provide interstitial spaces for maintaining the upward
flow of salt at the design velocity and establish the
salt-to-grai)hite ratio necessary for nuclear performance.
Retainer plates at the top of the posts tie the entire
core together, as indicated in Fig. 1.3. Individual orifice
plates are placed over each cell under the retainer plates
to hold down the slabs as well as to provide the
. necessary orificing of the flow.

The slabs adopted for the MSDR core graphite are
easier to fabricate than the prisms used in the MSBR
and earlier MSDR designs. The slab shape will also allow
the graphite to operate at lower temperatures and at
lower stress levels. The graphite temperature at the
center of the core has been estimated to be 1190°F.

1.1.3 MSDR Off-Gas and Salt Pump-Back System

The primary function of the fuel-salt drain tank is to
store the salt safely when it is drained from the reactor
primary system, but the tank serves some secondary
purposes as well. The revised MSDR system no longer

depends upon the drain tank as an expansion volume
for the salt since the salt-circulation pump bowls are
large enough for this purpose. The drain tank continues
to serve as a holdup and decay volume for the gases
stripped from the circulating fuel salt in normal
operation. The drain tank is well suited for this purpose
because its heat removal system can handle the approxi-
mately 6 MW(t) of heat generated in the gas.

In the MSDR fuel-salt circulation system about 10%
of the pump outlet flow is bypassed through a gas
separation loop. The salt first passes through a gas
separator in_this loop, where about 2 cfm (STP) of gas,
with 1 to 2 ft3> of salt entrained in it, is removed.
Several schemes were studied for separating the en-
trained salt from the gas, such as spray towers, cyclone
separators, etc. The very high heat release from the
radioactive gas requires large heat removal systems,
however, and the indeterminate manner in which the
solid daughters of the gaseous fission products could
coat metal surfaces makes it necessary to provide ample
surface for accumulation of the particles. After some
study it was decided to use the drain tank itself to
separate the salt from the gas, the large volume of the
tank permitting simple separation by gravity. The large
surface of the cooling thimbles in the drain tank would
ORNL-DWG 7i-13127

. \\ N EAN PR ‘\’.
. P L SO,
AT RN 5N

UPPER REFLECTOR N

/
7

TIE PLATE

ORIFICE PLATE

HOLES, SLOTS, ETC.

)

NN

N

BOTTOM GRID

|

Fig. 1.3. Reactor core cell assembly elevation for 300-MW(e) MSDR.

provide sufficient area for the solid daughters of the
noble gases to coalesce.

~ Each of the gas separators in the three salt-circulation
loops in the MSDR discharges the gas-salt mixture into
the drain tank through a 1%-in. line. The temperature
of the pipe is limited to about 1325°F by the cooling
action of the salt entrained in the gas. The line is
provided with a syphon break to prevent drainage of
the main salt-circulation loops in the event of a pump
failure.. The syphon break is in the form of a loop of
piping that rises above the level of the liquid in the
pump bowl and is equipped with -a venturi, the throat
of which communicates via a '%-in. pipe to the gas
space above the liquid in the pump bowl. During

normal operation, gas from the bowl is drawn into the
venturi and discharged along with the gas-salt mixture
from the gas separator into the drain tank. The total
flow of salt into the drain tank via the gas separators is .
estimated to be about 40 gpm. Two small jet pumps in
the 'tank return the salt to the primary loops. The
operation of the system is described below, and the
design of the jet pumps is described in Sect. 3.7.3.

An elevation of the jet pumps mounted in the center

“of the drain tank is shown in Fig. 1.4. The assembly is

installed so that it can be withdrawn from the tank if
\repair or replacement becomes necessary. One of the jet
pumps takes its suction from the bottom of a vertical
6-in. pipe at the center line of the vessel. The suction.
FILL JET

TO PRIMARY
SYSTEM

FROM TRANSFER
PUMP

W)
S
3 &
N N P77
RN AL
AT
,,,,,,

-~ -
A
Y, «
-
Ll aema

FROM PRIMARY
SYSTEM

2 B
Ry

'}\‘

SECTION A-A
{ ENL.ARGED )

12-in.-DIAM OPENING
DRAIN TANK

LIFT JET
PUMP BACK JET

RETURN TO PRIMARY FILL JET
SYSTEM

ORNL- DWG 71-13i28

-

A

RETURN TO PRIMARY
SYSTEM (3in.}) ——|

FILL JET
DISCHARGE { 2in.)

PUMP BACK JET-—_|

. JET SUPPLY LINES ~j 191t Oin.

PUMP BACK JET
SUCTION {1V in.) e

DISCHARGE (1% in.) —7}

DISCHARGE

FROM LIFT JET LIFT JET —_|

FILL JET
LIFT JET
SUCTION (15 in.)

SUCTION (1Y% in.}

ELEVATION

Fig. 1.4. Elevation of jet pump assembly in drain tank,

head on the pump varies from about +15 to —5 ft as the
level of salt in the pipe changes from full to empty.
Since the discharge from the jet is proportional to the
suction head, varying from zero at the lowest level to
about 50 ppm when the sump is full, the jet cannot
pump gas from the tank when it is empty of salt. The’
jet pump uses salt as its working fluid, the salt being
taken from a manifold connecting the three salt-circula-
tion pumps. Each manifold connection has a ball check
valve to prevent backflow in the event of a pump
shutdown. -

As is also shown in Fig. 1.4, a second jet pump is
provided in the assembly to keep the 6-in. central pipe
supplied with salt from a small sump in the bottom of
the drain tank. This jet has a capacity of about 50 gpm,
keeps the liquid pumped out of the drain tank, and,
when the tank is empty, may recirculate gas within the

tank. As described above, however, the system is

designed to prevent pumping of gas from the tank back
to the pump bowls. It provides automatic flow control
regardless of the number of primary pumps that are

running,

-

T
A ‘third jet pump is provided in the assembly to fill
the primary system with salt from the drain tank and to
transfer salt from the drain tank into the fuel-salt
storage tank. The working fluid for this third jet pump
is provided by a separate small pump located in the
chemical processing cell.

1.1.4 Drain Tank Cooling System

The drain tank cooling system described in a previous
report? has not been changed in principle, but the
thimbles have been rearranged. The change was'made to
facilitate the manifolding of the pipes which carry the
cooling NaK into and out of the thimbles. As shown in
the plan view of the top of the drain tank, Fig. 1.5,
there are now 60 sets of thimbles, each containing 6
thimbles arranged symmetrically in a circular pattern.
Figure 1.6 is an elevation of the drain tank. One of the
manifolds is shown in cross section in Fig. 1.6 to
indicate that it would be possible to remove and replace
manifolds if maintenance were required. By-manifold-
ing only six thimbles in a circular pattern, the measures

required to accommodate thermal expansion are made

‘much less stringent. Use of 60 different circuits limits

the amount of NaK that could leak from one circuit to
about 16 ft3. _

In another change in the drain tank system from that
described previously,? the one large water tank, to
which the NaK circuits reject their heat by radiation,
has been replaced by three smaller tanks. The NaK
circuits are manifolded in such a manner as to divide
the heat removal capability between the three water
pools, and with any two water tanks in operation the
temperature in the drain tank could be limited to about
1400°F. This arrangement is considered to be more
reliable than the- previous one in which a single leak
could result in total loss of water.

Studies have been initiated on the effects of seismic
disturbances on the drain tank and its cooling system.
Preliminary calculations indicate that the individual
thimbles in the drain tank would have a natural
frequency of about 1.5 cps when each thimble is
considered to be a 20-ft-long cantilevered member.

“Since earthquakes produce excitation in the range of 3

ORNL-DWG 71-13129

FROM CHEMICAL PROCESS

COOLING MANIFOLD
ASSEMBLIES

EMERGENCY
DRAIN PLUG

INSPECTION PORTS

GAS-SALT DISCHARGE

Fig. 1.5. Plan view of top of drain tank showing new thimble arrangement,.
JET PUMP ASSEMBLY LINES A

— e { a—

ORNL-DWG 7§-13130

COOLING MANIFOLD ASSEMBLY -

L ‘COOLING LINES GAS-SALT DISCHARGE
~ DRAIN LINE |
31t Oin. NO. 2
insPECTION & ¥ (. i '
PORTS‘\ l j (
4
° ,/__.._/I/ *,-/ CRUCIBLE COOLING
0
g —= NO.!
171 A E
74
™
¥l
N
N W
%
1N %
A N \
V]
"
17} £ /1
\ y IHININININIRIN NN y
: St v iy
% N
AN ¢
N
wnon ¢ 0N CORRUGATED FLOW
0 N<ri  SEPARATOR
»)
/ H
#
0
9%
1
N
N
i 4
A
L =T o
U i
JUL 4
w/
7 CRUCIBLE _COOLING .
CIRCUIT NO.2
THIMBLE ASSEMBLY l
H\ INSPECTION PORTS :
\ JET PUMP ASSEMBLY
\ CRUCIBLE COOLING. .@
CIRCUIT NO.1=—"""|
§
/ZZ\ T
(4 £\ % g '
121t 5in,

Fig. 1.6. Primary salt drain tank sectional elevation.

3 ]

PR
[ 34

to 4 cps, means for changing the natural frequency of
the thimbles have been given preliminary study. All
thimbles in each group of six would be tied or banded
together at 3-ft intervals along the length. In addition,
all 60 groups of thimbles would be tied together at the
bottom by “U’-bolt-type clevises. These ties would
restrict radial movement to make the entire array act as
a unit if seismically disturbed, but would permit
longitudinal motion resulting from differential thermal

expansion. The method of joining the units together is

indicated in Fig. 1.6.

1.1.5 Drain Valve Cell

The cell provided for the freeze valve in the fuel-salt
drain line leading from the bottom of the reactor to the
drain tank has been eliminated in'the revised MSDR
design, and the valve is now located in the drain tank
cell. This . arrangement facilitates replacement of drain
lines and associated ' piping should maintenance be
required. The drain lines now pass through penetrations
in the side .of the reactor cell, and the valves in these
lines are accessible through a separate hatch in the cover
of the drain tank cell.

1.1.6 Drain Cell Catch Pan

A catch pan is provided in the bottom of the reactor
cell to collect any accidental spillage of salt and direct it
into the drain tank through frangible disk valves. An
additional catch basin has now been added near the top
of the drain tank in the drain tank cell to catch any
leakage from the primary salt lines and direct it into the
drain tank. This catch basin extends from the periphery
of the drain tank at the top to the walls of the drain
tank cell. As presently conceived, a 6-in.-diam hole in
the basin communicates with the drain tank, the
opening being normally sealed by an inverted cap with
its lip in a trepanned groove that is filled with NaF.
Since the melting temperature of the NaF is about
1600°F, the opening would remain sealed unless
flooded with fuel salt, which would dissolve the NaF
and allow the fuel salt to run into the interior chamber
to float a graphite plug and then flow into the drain
tank. o

With catch basins provided for the reactor cell
equipment, for the salt piping in the drain tank cell, and
for the drain tank vessel itself (in the form of a cooled
“crucible” surrounding the tank), we believe that the
likelihood of an accident in which molten salt collects
in a pool in the bottom of the containment vessel
without sufficient cooling is very remote indeed.

1.2 SOME CONSEQUENCES OF TUBING FAILURE
IN THE MSBR HEAT EXCHANGER

R. P. Wichner

A study was made of the consequences of the mixing
of fuel and coolant salts as a result of various modes of
failure of tubing in the primary heat exchangers in the
MSBR reference design. Four specific cases weré ex-
amined: (1) double-ended failure near the fuel-salt
outlet; (2) double-ended failure near the fuel-salt inlet;
(3) small coolant-salt leak into the primary system; (4)
small fuel-salt leak into the secondary system. The
study is, in many places, speculative both because of
the preliminary status of the design and the lack of
adequate physicochemical data on mixing of fuel and
coolant salts. However unpleasant some of the con-

sequences of tubing failure may be, no way of

generating either a nuclear excursion or a rupture in the
primary or secondary loop piping is foreseen.

Some additional tentative conclusions of the study
are: ' '

(1) For a case 1 failure (double-ended tubing failure
near the fuel-salt outlet of the heat exchanger), 3.5
Ib/sec of coolant salt initially leaks into the primary
system. The boron poison in this leakage stream is
sufficiently large such that a low-power-level trip, if
present, is actuated essentially immediately on arrival of
the contaminated fuel at the lower axial blanket in the
reactor vessel. The worth of the boron poison is
estimated 'to be 0.18% 6k/k per pound of coolant salt
distributed uniformly over zones.I and II of the reactor
core. :

(2) Additionally, for a case 1 tubing failure, 0.96
Ib/sec of fuel salt initially leaks into the secondary loop.
However, . the bulk of the fuel-salt leakage into the
secondary system, and the coolant-salt leakage into the
primary systefn, occurs after reactor shutdown. A total
of approximately 300 lb of fuel salt leaks into the
secondary system, all but 4 Ib occurring in the 20-min
interval in which the freeze valves are thawing and the

.loops are draining. If the cover-gas pressure in the

secondary loop were set about 3 psi higher than the 10
psig in the present design, there would be no fuel-salt
leakage in this interval, and the total loss of fuel salt to
the secondary system would be approximately 4 1b for
this case.

(3) It is judged that the fuel salt which leaks into the
shell of the heat exchanger after shutdown will initially
be in the form of immiscible droplets that subsequently
freeze and settle onto the lower tube sheet. The
maximum temperature caused by this localized heat
source is estimated to be approximately equal to that
reached in central portions of the heat exchanger due to
noble-metal heating, or approximately 1660°F when
coolant salt remains in the shell.

(4) If rapid evolution of BF; from the coolant-salt
stream leaking into the: primary system is assumed, a
maximum pressure rise of 74 psi in the primary loop is
estimated for a case 1 type of failure. It is judged that
rapid evolution of BF; will not damage the primary
pumps if shutdown procedures are initiated most
promptly — as would be the case if a low-power-level

trip is incorporated into the automatic protection

system.

(5) Since there is no provision for draining coolant
salt from the shell of the heat exchanger, ultimately,
- following a case 1 failure, the entire contents of the
shell — 444 ft> (52,000 1b) — leaks into the prlmary
drain tank through the failed tube. The main con-
sequence of this large addition of coolant salt to the
fuel salt is judged to be the evolution of large volumes
of BF, in the primary system with possible deleterious
effects to the charcoal beds. No means for seriously
elevating the fuel liquidus temperature is envisioned

even by the addition of two heat exchanger shells worth-

of coolant.

(6) A case 2 failure (double ended tublng failure near-

the fuel-salt inlet) initially allows 7.4 Ib/sec of fuel salt
to leak into the secondary system. If a delayed neutron
detector on the secondary system is an integral part of
the reactor protection system and a signal from it stops
the pumps, then a total of approximately 60 Ib of fuel
salt is lost. It is judged that this fuel salt will form
immiscible droplets that freeze and settle to the lower
tube sheet.

.(7) For case 3 it was assumed that a pinhole in the
tubing allows 3 X 107? ‘lb/sec of coolant salt into the
primary system, at which rate approximately 2.6 hr are
required to add —0.3% 8k/k -- the total worth of the
control rods. A control rod location alarm is set off at
some point during the gradual inward drift of the
control rtods. If no operator action is taken, the
low-temperature trip on the core outlet initiates shut-
down approximately 26 min after full insertion. Thus,
the duration of this incident is 2 to 3 hr; however, even
a fairly insensitive BF, detector on the off-gas flow, if
one were available, would detect the presence of the
leak within minutes.

(8) In case 4 a 3 X 107 1b/sec leakage rate of fuel
salt into the secondary system is assumed to occur
through a 2-mil-diam pinhole in the tubing near the fuel
inlet. A leak rate of this magnitude would cause a flux
of delayed neutrons of approximately 70 neutrons

10

cm 2 sec™! at a detector positioned near the outlet of

the heat exchanger. This is seven times the lower
detection limit of ion chambers, and hence an auto-
matic shutdown would be initiated a few seconds after
the inception of the leak, if a delayed neutron detector
were so deployed.

1.3 SIDE-STREAM PROCESSING OF THE MSBR
. PRIMARY FLOW FOR XENON
AND/OR IODINE REMOVAL

R. P. Wichner

A state-of-the-art report, presently in preparation, on
means for dealing with the ' **Xe poison in molten-salt
reactors will contain an evaluation of the side-stream
processing approaches to the problem. These processes
differ - basically from the present reference design
method in that a single bypass flow of fuel salt is
processed in a gas-stripping device for removal of xenon
and/or iodine. Thus, the four gas-separator loops are
eliminated as are the circulating bubbles; however, use
of the latter is not precluded in side-stream.processing
concepts. '

There are a number of strong incentives for side-

..stream removal methods, especially for iodine or

combined xenon and iodine removal schemes. Present
evidence indicates that the “effective solubility” of
jodine in fuel salt is quite high,* so that some difficulty
may be expected in efficiently removing it in a stripping
device. This inconvenience is more than compensated
for by the corollary that diffusion rates of HI into the
graphite will be low. Hence, the sole significant com-
petitive removal rate is the slow decay to xenon, and
consequently only very small bypass flow rates are
required. If an efficient iodine stripper is available,
bypass flows as low as 225 gpm will capture almost all
the iodine before it decays to xenon, and if no other
control method is applied, the xenon poison fraction

will be the 0.001 resulting from the direct fission yield

of 135 Xe.

A particularly attractive situation is attained by a
combined iodine plus xenon stripper. If one assumes a
bypass flow through such a device of 10% of the
primary flow — equal to the bypass flow rate in the

4. The “effective solubility” is defined by Pyj = [I” }/Kkess
and can be shown to be a function of the proton concentration.
The chemistry of iodine stripping will be further discussed in
the next semiannual report and in the forthcoming state-of-the-
art report on 135 Xe control procedures. Experiments by Baes
and Bamberger are the basis of the iodine-stripping remarks.

-

r:
‘e

reference design through the four gas separators — and
that such a device, designed -primarily for xenon
removal, does so with high efficiency, then if addition-
ally only 10% of the iodine throughput were simultane-
ously removed, a !°°Xe poison fraction of ~0.003
would be attained without employing impervious coat-
ings on the graphite. Present understanding of the
mechanism of iodine stripping indicates that some HI
(and TI) will be removed in a xenon stripper-employing
inert helium stripping gas due to the presence of
protons and tritons in the fuel. The degree of HI
removal may be enhanced by adding some HF to the

stripping gas.

1.4 MSBE DESIGN
M.I.Lundin J. R. McWherter

1.4.1 Reactor Core Power
. and Neutron Flux Distribution

J. R. McWherter J. P. Sanders

The design studies reported previously® were contin-
ued with the reactor core concept remaining essentially
unchanged. The core is a cylinder 45 in. in diameter and
57 in. high formed of graphite bars and contains 15 vol %
fuel salt. The core is in a 90-in.-ID spherical reactor vessel.
The space between the vessel wall and the core is filled
with fuel salt. The reactor thermal poweris 150 MW.

A computer subroutine, entitled HOLES, was used to
produce contour plots of the power distribution and
neutron fluxes in the MSBE core; the calculated power

and neutron flux distributions were produced by a

finite-element, multigroup reactor physics code called
CITATION. Data from CITATION case 41 (the 45-in.-
diam, 57-in.-high cylindrical core with a 0.15 salt
fraction located in a 90-in.-diam spherical vessel) was
scaled to a reactor power of 150 MW as input for
HOLES.

Two of the plots produced are shown in Figs. 1.7 and
1.8. Several characteristics of the plots are dependent
upon the features of the CITATION code. The CITA-
TION code indicates the value of the power density and
neutron flux for a limited number of points in the axial
and radial directions, and the code approximates a
spherical geometry by using a series of stacked cylindri-
cal disks of decreasing radius. Values of the power

- density and neutron flux for coordinate points at the

center line and ‘midplane are not given since the

5. MSR Program Semiannu. Progr. Rep Feb. 28 1971,
ORNL-4676, pp. 36—40.

11

analytical and numerical solutions are indeterminate at
these points. The subroutine HOLES plots the contour
lines by making a linear interpolation along the grid-
lines between the given values.

Because of the relatively gross spacing of lattice
points, the power density and the neutron flux are not
represented accurately in the region near the boundary
between the core and the blanket. The neutron flux is

~assumed to vary linearly between the lattice points in

the core and in the blanket nearest the boundary, and

~ the power density varies across the boundary with the

change in flux and fuel-salt fraction. The approximation
of spherical geometry by stacked cylindrical disks,
which is used in CITATION, becomes evident in the
perturbations in the contour lines for power densities
near the vessel wall. Because of the symmetry of the
core, all the contour lines should have slopes which are
perpendicular to the axes. Again, the large difference in
the values of the coordinates of the lattice points
obscured this characteristic at the higher power levels
where the change in values for adjacent lattice points
was the greatest.

The values shown for power density in Fig. 1.7 are
per unit volume of the core and moderator. Discon-
tinuities occur at a radial position of 57 ¢m (22% in.)
and at an axial position of 72 cm (28", in.) above the
midplane; at these points, the salt fraction changes from
0.15 to 1.00 in the radial direction and to 0.70 in the
axial direction. :

This same subroutine was used to draw contour plots
of several combinations of the nine energy groups of
neutrons in the MSBE. To get the graphite damage flux,
data from the first (highest) group were combined with
85% of the data for the second energy group. This gives
the estimated flux above 50 keV and was used as input
for the plot given in Fig. 1.8.

1.4.2 Reactor Core Heat Removal

H. A.McLain  D. W. Wilson®

Calculations indicate that the graphite moderator
temperature at the center of the reference three-pass
MSBE core” is 100 to 150°F higher than it is at the
center of the reference MSBR core.® The maximum

-temperature at the center of the MSBE core is about

6. TVA employee on assignment to ORNL.

7. J. R. McWherter, Molten Salt Breeder Experzment Design
Bases, ORNL-TM-3177 (November 1970).

8. Roy C. Robertson (ed.), Conceptual Design Study of a
Single-Fluid Molten-Salt Breeder Reactor, ORNL-4541 (June
1971). .
12

ORNL-DWG 71-13132
L 1 4 1 1 1

'120" ] i N ; . | -

”Q,_‘--\G

1004 —___

‘»90:_\ 24

80-E

W

= _— =

AXIAL DISTANCE {cm)

RADIAL DISTANCE (cm)

PRESSURE VESSEL
BOUNDARY : : i

kil A T T
70 0 Mo 120

Fig. 1.7. Power distribution in the MSBE core.

1450°F, and the average temperature is 1390°F. For a
single-pass core, these temperatures are about 10°F
lower. A disadvantage of the single-pass core is that the
velocity required to achieve the 250°F AT across the
core is only one-third of that in the 15-ft-high MSBR
reference concept.® |

The high temperatures at the center of the MSBE core
are due to the high power density and graphite heating

rate,” 114 W per cubic centimeter of core and 11 W per
cubic centimeter of graphite, respectively. These tem-
peratures also imply that the graphite at the center of
the reactor core would have to be replaced after about

" 9. 1. R. Engel, MSR Program Monthly Report for May 1971,
MSR-71-37, pp. 4—6 (internal report).
13

ORNL—DWG 74 -13131

'120 - L

110
100.
90
80
7O
60

50

AXIAL DISTANCE (cm)

40

30

20

1 - 1 . L : .l
[~ CORE BOUNDARY

1 L L l L

PRESSURE VESSEL
BOUNDARY.

\ .
0.7 \

(neutrons/cm?- sec) x40™'?

\

\

50

RADIAL DISTANGE (cm)

0 70 80

Fig, 1.8. MSBE graphite damage flux (£ > 50 keV).

1.7 years of full-power operation using the relation
recommended by Eatherly.'® Calculations were also
made for prismatic moderator elements in a single-pass

10. D. Scott and W. P. Eatherly, “Graphite and Xenon
Behavior and Their Influence on Molten-Salt Reactor Design,”
Nucl. Appl. Technol. 8,179-89 (February 1970).

core in which the 3.64-in. distance between the flat
surfaces was reduced to 3.0, 2.75, and 2.5 in., but the
central hole diameter was held at 0.6 in. These changes
resulted in average graphite temperatures at the center
of the core of 1310, 1290, and 1260°F, respectively,
with corresponding graphite lifetimes of 1.9, 2.0, and
2.1 years. Average graphite temperatures and lifetimes
along the reactor center line for the 3-in. prismatic
element are shown in Fig. 1.9. The minimum graphite
lifetime is at the center of the core. |

Slab-type moderator elements also were considered.
At the center of the core, slabs of thicknesses of 1.0,
1.25, and 1.5 in. have average graphite temperatures of
1270, 1290, and 1320°F, respectively, and lifetimes of
2.0, 2.0, and 1.9 years, respectively.

ORNL-DOWG 71-13133

AVERAGE GRAPHITE TEMPERATURE (°F)
1100 1150 1200 1250 1300 1350
T T

125 = GRAPHITE LIFE

75 — CORE

MIDPLANE

TEMPERATURE

DISTANCE ABOVE
BOTTOM OF CORE (cm)

0 : 2 .4 6
GRAPHITE LIFE ({years)

Fig. 1.9. Average graphite temperature and life along MSBE
core center line for 3-in. prismatic element.

1.4.3 Maintenance Studies

J. R. McWherter  W. K. Furlong
H. A. McLain J. T. Meador
W. Terry

Studies were made of procedures for replacing the
core moderator elements i_ndividua]ly instead_of ‘replac-
ing the entire core assembly at one time as proposed in

14

the reference design.” The main difficulty in removing
elements individually is holding the elements remaining
in the core in their normal upright positions when some
have been removed. A permanent fixture or a re-

movable jig will have to be designed to accomplish this. -

Layout studies of the MSBE primary heat exchanger
were made with the major consideration being given to
maintenance. In one approach proposed, a tube can be
plugged in place or the entire heat exchanger can be
removed from the reactor cell for repair or replacement.
The heat exchanger is assumed to be a U-shaped unit
mounted horizontally.- Horizontal mounting of a U-
shaped unit is desirable to permit draining of the fuel
and coolant salts and to minimize the possibility of
accumulating crud in the heat exchanger tubes.

In the proposal, the unit is mounted on trunnions to
facilitate removal from the cell for repair or replace-
ment, in the uhlikely event this is required. A remote
pipe-cutting device is lowered into the cell through an
opening in the shielding above the equipment to cut the
pipes connecting the exchanger to the salt systems. The
exchanger is then rotated on the trunnions to a vertical
position and lifted out of the reactor cell through an
opening at the top of the cell. This procedure is
reversed when installing a heat exchanger with the
welding of this unit to the connecting pipes being done
remotely. Pipe springback and alignment are major
considerations in the last step. Present remote welding
technology'! indicates that the two ends of a pipe to
be welded must be aligned within Y, ¢ in.

11. MSR Program Semignnu. Progr. Rep. Aug 31, 1970,
ORNL-4622, pp. 45-50.

wri
1.5 MSBR INDUSTRIAL DESIGN STUDY
M.I. Lundin J. R. McWherter

The subcontract’? negotiated between ORNL and
the Ebasco Services group, consisting of Ebasco,
Babcock and Wilcox, Byron Jackson, Cabot, Conoco,
and Union Carbide Companies, was signed. This sub-

15

contract covers the conduct of an industrial design

study of a 1000-MW(e) MSBR plant.

12. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 36. o

ORNL-DWG 71-13134

Fig. 1.10. MSBR industrial study proposed graphite array.

Task I, the selection of a reference conceptual design,
is essentially complete. Two progress reports were
submitted.

The overall core size is the same as that in the ORNL
reference concept,® but a slab-type graphite element is
proposed. Several of these slabs are held at the top and
bottom in a hexagonal array as shown in Figs. 1.10 and
1.11. This assembly is sufficiently small to be handled
and replaced as a unit. One hundred fifty-seven such
arrays are required for the core. A 2-ft-thick radial
graphite reflector surrounds the core. A boron-contain-
ing thermal shield separates the reflector from the
pressure vessel to reduce the radiation damage resulting
from transmutation of boron in the Hastelloy N.

- ORNL-DWG 71-13135

Fig. 1.11. MSBR industrial study proposed graphite array —
zone I — section at midplane.
2. Reactor Physics

- A.M.Perry

2.1 'MSR EXPERIMENTAL PHYSICS

2.1.1 HTLTR Lattice Experiments
G.L.Ragan  O. L. Smith

The experimental program, described in considerable
detail in the preceding semiannual report,! has now
been .completed at the Battelle Northwest Labora-
tories.”> Measurements were made at four operating
temperatures: 20, 300, 627, and 1000°C. At each
temperature BNWL measured the basic lattice multipli-
cation factor k. and the reactivity worths of some 24
perturbations at the center of a lattice that was
neutronically similar to a typical MSBR design pro-
posal. These perturbations W;re chosen to reveal certain
- reactor physics characteristics of such a lattice and to
provide experimental results against which we can test
our calculational techniques.

The measured multiplication factor k.. varied from
1.029 at 20°C to 1.004 at 1000°C. Qur only precalcu-
lated value (1.020 at 627°C) was considerably higher
than the measured value (1.007 at 627°C), but the
significance of the discrepancy cannot be assessed until

1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, pp. 45—48. .

2. E. P. Lippincott, Battelle Northwest Laboratories, personal
communication to G. L. Ragan, Qak Ridge National Labora-
tory. Experimental results are being reported in Technical
Activities Quarterly Report, AEC Reactor Development and
Technology Programs, April, May, and June
BNWL-1522-3.

1971,

we are able to include the self-shielding effects of the
particles in the fuel mixture — which had to be
neglected in the precalculations. The perturbation
worths at 627°C that we calculated' showed similar
divergences from those measured, but particle  self-

. shielding effects were neglected in these precalculations

.16

also. .

In our precalculations, the XSDRN code® was used to
obtain group cross sections for each region and also to
perform multigroup one-dimensional neutronics calcu-
lations with the S, transport method. The standard
version of XSDRN uséd in the precalculations is limited
to one degree of heterogeneity per region, for example,
homogeneous fuel rods disposed in a lattice arrange-
ment. We have now modified the GAM-II part of the
XSDRN code so that it obtains group cross sections in
the resonance energy range for doubly heterogeneous
regions; for example, the self-shielding effects of ThO,
particles within the above-mentioned fuel rods may be
included.

In its original form,* GAM-II yields the lump (region
0) collision density per unit energy Fo(E). From this,
one readily.obtains the lump flux per unit lethargy,

3. N. M. Greene and C. W. Craven, Jr., XSDRN: A Dis-
crete-Ordinates Spectral Averaging Code, ORNL-TM-2500 (July
1969).

4. G. D. Joanou and J. S. Dudek, GAM-II: A B3 Code for the
Calculation of Fast-Neutron Spectra and Associated Multigroup
Constants, GA-4265 (September 1963).
Fo(u) _ EFy(E)
zt,o(u) Et,o(E) ’

polu) =

where Z, ((£) is the total cross section in the lump.
These results are made possible by eliminating explicit
reference to the external (region 1) moderator by (1)
assuming that region 1 sources are well approximated
by using the asymptotic flux and the narrow resonance
(NR) approximation for them, (2) using the flat-source
overall escape probabilities Py(u) and Pj(u), and (3)
invoking the reciprocity theorem. With the same as-
sumptions, it can be shown that the region 1 flux is
given by

Pi(u) VoI, o)
1 —Py(u)

¢, () = 1— [1—¢o(l ,

VIEI,I v

where V,, and V, are the fractional volumes of the two
regions. It should be noted that we do not assume ¢, ()
to be simply the asymptotic flux, nor is the equation

17

given limited to small fractional lump volumes. As -

illustrated by examples given below, the equation has

been tested rather extensively, with good results.

The revised GAM-II treatment in the XSDRN code

treats a doubly heterogeneous problem by assuming -

that the two degrees of heterogeneity are separable —
an assumption that will be examined below. Although

the code is more general, it will be explained on the

basis of the ThO, grains in the fuel of an HTLTR
loading. The fuel is first considered as an infinite
medium, and the GAM-II solutions for ¢y(x) in the
grains and ¢,(x) in the matrix around them are
obtained over the usual lethargy band encompassing a
given thorium resonance. The associated volume-
averaged fluxes and the regional disadvantage factors
do(u) and d(u) are then obtained, and the appropriate
disadvantage factor is applied to the cross sections of
each nuclide at each lethargy point. These disadvan-
taged cross sections are then taken to characterize a
smeared (homogenized) fuel within the fuel rods (say,
region 2) in a lattice having graphite (say, region 3)
between the rods. The GAM-II treatment is again
applied, over the same lethargy band encompassing the
given thorium resonance, to obtain ¢,(u) and ¢3(u).
Now one performs for the given thorium resonance the
appropriate® integration to obtain cell-averaged cross

5. N. M. Greene, ‘‘Spatial Shielding in the Nordheim Integral
Treatment,” Trans. Am. Nucl. Soc. 14, 364 (1971).-

sections that include grain self-shielding effects. The
same procedure is repeated for each thorium resonance,
as is customary with GAM-IL.

Instead of using the modified GAM-II treatment to
obtain grain shielding factors as was done above, several
alternative procedures may be used. Perhaps the sim-
plest of these is the NR approximation, as applied to
the grain. One assumes that the sources in the grain are
well approximated by using the asymptotic flux and the
grain potential scattering cross sections to calculate
them — just as one assumes for the moderator in the
usual GAM-II treatment. This leads to

0

$o(u) = Po(u) + [1 Po(u)] ( X

Z; olu) —

p.1

¢y(u)=1— o(u) v

for the NR approximation. Here Ep,o and Z, , are the
potential and total cross sections for the grain, all grain
nuclides being included.

Alternatively, one may assume that the NR approxi-
mation may be used for all sources due to moderators,
but that no downscattering is contributed by collisions
with resonance-absorber atoms, the so-called narrow
resonance infinite-mass (NRIM) approxnnatlon The
resulting fluxes are

P+ [1— Py(a)] (25 o/Z,,000)]

d)o(u) = : ’
1 - [1 _P (u)] [2 o(u)/zt O(u)]
Vo Z} o) — ¢o(u) Z¢ o(w)
0,0 = 1= Pif) > =
. _p,lA .

where superscripts 7 and m denote resonance absorber
and admixed moderator, respectively.

An accurate solution of the two-region problem may
be obtained with the computer code GAROL,® which
solves explicitly for the neutron-collision densities in
the two regions, avoiding most of the approximations
present in the earlier GAM-II code. While not integrated
into our cross-section averaging procedures, this code
permits checks to be made, with test problems, on the

6. C. A. Stevens and C. V. Smith, GA-ROL: A Computer .
Program. for Evaluating Resonance Absorption Including Reso-
nance Overlap, GA-6637 (August 1965).
approximations to be used in the analysis of the
HTLTR experiments.

Numerous other approximate methods have been
proposed. In a recent paper,” P. Wilti discusses several
of these and proposes one of his own that is somewhat
related to the NRIM approach. Walti gives an example
involving a carbon matrix containing thorium carbide
grains of 400-u diameter that occupy 10% of the
volume. We have used his problem to compare several
methods of obtaining grain shielding factors, with
results given in Table 2.1. Our results differ slightly
from Wilti’s because we used slightly different reso-
nance parameters. Fluxes and flux ratios are given at
‘the peaks of the two lowest thorium resonances.
Programming of the Walti method and adaptation of
GAROL to the IBM-360 were done by W. E. Thomas.

" Table 2.1. Fluxes at two thorium resonance peaksa
calculated for a test problem by several methods

"y 21.7-eV peak 23.4-eV peak

Method .

pofu)  o1(W)  doley o) o) dold
GAM-II 0.1155 0.1562 0.7398 0.0754 0.1209 0.6236
NR 0.1137 0.1544 - 0.7363 0.0720 0.1176 0.6121
NRIM 0.1116 0.1524 0.7323 0.0732 0.1188 0.6164
Wilti 0.7323 0.6158
GAROL 0.1118 0.1513 0.7392 0.0725 0.1165 0.6219

9Each resonance treated separately and based on unit asymp-
totic flux. ‘

One may have confidence, on the basis of Table 2.1

and other more extensive studies of the same sort that.

we have carried out, that the use of GAM-II to evaluate
the grain shielding factor in a heterogeneous fuel — as
we have done in our revised XSDRN code — is valid.

The results are very close to those given by GAROL,

which is an almost exact solution. Not quite so good,
but probably adequate for most practical cases, are the
other three methods tabulated.

On the other hand; the assumed separability of the
two degrees of heterogeneity certainly introduces some
error, perhaps enough to be significant. In principle,
one could consider the three distinct regions (called O,
1, and 3 above) simultaneously and either (1) generalize

7. P. Wilti, “Evaluation of -‘Grain Shielding Factors for
Coated Particle Fuels,” Nucl. Sci. Eng. 45,321-30 (1971).

18

the GAM-II treatment or (2) use a multiregion code
such as EGGNIT.® Procedure (1) was tried, but it was
soon realized that a major obstacle in these approaches
is that expressions for the collision probabilities Py, o,
Py and P, 3 are not available. (It was encouraging,
however, that approximate expressions gave results that
were close to those obtained on the basis of the
assumed separability.) Efforts therefore turned to
checking the validity of the results obtained by as-
suming separability by comparison with results ob-
tained using a transport method of solution for a
selected problem involving a double heterogeneity. This
work is still in progress.

2.2 PHYSICS ANALYSIS OF MSBR

2.2.1 Neutron Irradiation Effects Qutside
the MSBR Core

J. R. Engel

One of the basic characteristics of the current MSR
concepts is the emission of a substantial fraction of the
delayed fission neutrons in regions other than the core -

-of the reactor. Besides the effect on the dynamics of

the chain reaction in the core, the emission of neutrons
from the circulating fuel allows a number of potentially
significant neutronic processes to occur in noncore
components. Computations were performed to estimate
the magnitude and effects in the reference design MSBR
of two out-of-core neutron processes: irradiation dam-
age to the Hastelloy N structural material and neutron
activation of the secondary salt. In general, two areas
were examined: the reactor -outlet lines where the
neutron sources are most intense and the primary heat .
exchangers where there is interaction with the coolant
salt. '

Neutron sources. Delayed neutrons are emitted by at
least 25 nuclides that have fission product precursors
with half-lives ranging from a small fraction of a second
to 55.6 sec. These precursors are normally combined
into six groups, and the net yield of delayed neutrons in
each group is expressed as a fraction of the total
neutron yield from fission. Data on the.half-lives and
yields of delayed neutrons are readily available for the
various fissile materials. However, considerably less
material has been published regarding the energy
spectra of the delayed neutrons, particularly for neu-
tron energies above 1 to 2 MeV. Although relatively few

8. C. R. Richey, EGGNIT: A Multigroup Cross Section Code,
BNWL-1203 (November 1969). .
Ty

Table 2.2. Neutron activation processes in MSBR components due to delayed neutrons

Helium production in 24 . Triti
. Na inventory Htium
Component Hastelloy N in 30 EFPY in coolant salt production
(ppb) . rate
_ (€ e
"OB(n, @) ®Ni(n, a)? (Ci/day)
Heat exchanger
NaBF4-NaF 1.3 1-9 5000 0.03
Li,BeF4 (0.99995 "Li) 46 3-11 , 0.7
Li, BeF4 (natural Li) 3.3 1-9 ' 124
Reactor outlet line : 7 8-40

“Based on initial '°B concentration of 2 atom ppm.
bRange indicates effect of different assumptions for delayed-neutron source spectrum.

°In the coolant salt.

high-energy delayed neutrons are to be expected, some
neutron capture processes are quite sensitive to the
high-energy neutron flux. Consequently, calculations
were performed for two assumed shapes of the de-

Jlayed-neutron energy spectrum above 2 MeV. In one

series the spectrum was assumed to be flat out to 10
MeV, and in the second series the data below 2 MeV
were extrapolated to zero at about 5 MeV. The first
assumption almost certainly overestimates the high-
energy processes, and the second may also be con-
servatively high.

The intensities of the delayed-neutron sources were
computed from a hydraulic model’ in which the
reactor was divided into 16 regions with fission-energy
production in 10 of them. Each region was treated as a
homogeneous vessel to compute the concentration of

'delayed-neutron precursors at its outlet, and the various

streams were mixed according to the flow distribution
through the vessel. This led to a precursor concen-
tration for each delayed-neutron group at the reactor
outlet. From that point slug flow was assumed in
calculating the concentrations in other parts of the
external loop.

Processes. An important process for potential neu-
tron-irradiation damage in Hastelloy N outside the core
of an MSBR is the production of helium from (n, «)
reactions. Very low helium concentrations (as low as 35

9. R. C. Robertson, ed., Conceptual Design Study of a
Single-Fluid Molten-Salt Breeder Reactor, ORNL-4541, p. 47
(June 1971). '

10. H. E. McCoy and J. R, Weir, Jr., “Stress-Rupture
Properties of Irradiated and Unirradiated Hastelloy-N Tubes,”
Nucl. Appl. 4(2), 96—104 (February 1968).

ppb) have been found to significantly reduce the
ductility of the standard (unmodified) alloy.!®
Although the modified alloys are much less sensitive to
helium, the production rate remains of interest. For
low-energy neutrons the principal source of helium is
the '°B(n, «)’Li reaction in boron that occurs as an
impurity in the alloy. At higher neutron energies, the
reaction °®Ni(n, a)® ® Fe may also be important because
of the high nickel concentration in Hastelloy N. There
is evidence from high-flux irradiations of other helium-
producing reactions {probably involving sequential neu-
tron absorptions), but these are not likely to be
significant at the low fluxes that will exist outside the
core of an MSR. Aside from helium effects, Hastelloy N
is also subject to damage due to atom displacements
caused by neutron scattering events.

The activation processes of interest in the secondary
salt depend on the choice of coolant. With sodium
fluoroborate, the principal activity is 15-hr 2#*Na from
the reaction

23Na(n, v)**Na.

With lithium-beryllium fluoride, the several reactions
that lead to tritium production are of greatest interest.
(Other activities are very short-lived.) Tritium pro-
duction was evaluated both for a coolant salt containing
natural lithium and for one in which the lithium is
0.99995 "Li.

Computations and results. All the processes described
above were evaluated from one-dimensional neutron
transport calculations made with the computer program
XSDRN,? using 123 neutron energy groups. The results
of the various neutron absorption processes are sum-
marized in Table 2.2.
The values listed for helium buildup in Hastetloy N
are based on an initial '®B concentration of 2 atom
ppm (about 2 wt ppm of natural boron). Since the
boron reaction is essentially a thermal-neutron reaction,
the helium from this process was unaffected by the
assumed shape of the delayed-neutron energy spectrum.
On the other hand, the ®®Ni(n, ) reaction has a high
energy threshold,.so two values are listed. The first or
lower value corresponds to the neutron emission spec-
trum that went to zero at about 5 MeV, while the
higher value was obtained with the spectrum that
extended to 10 MeV. Since all of the helium values
from *®Ni are quite low, the only potentially signifi-

20

cant difference .produced by the shape of the delayed-

neutron spectrum  is in the reactor outlet line. The
results in Table 2.2 consider only the internal neutron
sources — delayed fission neutrons plus those produced
by fissions within the components. One calculation was
performed to estimate the effect of core-leakage neu-
trons on the rate of helium production. The maximum
amount of helium produced in.the outer shell of the
heat exchanger in 30 equivalent full-power years
(EFPY) was about 500 ppb, 80% from °®Ni(n, a)
reactions. Thus helium - production from core-leakage
neutrons may completely overshadow, at least locally,
the production from delayed neutrons, unless appro-
. priate shielding is provided to reduce the fast-neutron
flux from this source.

The activation processes in the coolant salt were
found to be essentially independent of both the
high-energy end of the delayed-neutron source spec-
trum and the incidence of core-leakage neutrons. The
29Na calculation is applicable only to the NaBF,
coolant, and the value listed is the total steady-state
inventory for the 1000-MW(e) MSBR reference design.
The production rates of tritium in the various coolant
salts are all much smaller than the rate of production in
the fuel salt (2400 Ci/day). In the case of coolant
containing natural lithium, however, the production in
the coolant is not negligible.

The damage rate in Hastelloy N by atom displacement
was calculated for the highest, flux regions in both the
reactor outlet line and the heat exchanger with NaBF,
. coolant. Since the neutron flux spectrum.outside the
core is quite different from any fission-reactor spec-
trum, the rate of accumulation of such damage was
expected to be different also. Con'sequently, atom
displacement rates were computed using effective

“damage cross sections” evaluated by Jenkins.'! These '

cross sections were obtained by explicit evaluation of
the energy transferred to primary recoil atoms and the
Kinchin-Pease model for secondary atom displacements.

In the calculations, the effect of the assumed delayed-
neutron spectrum was examined for the heat exchanger;
however, none of the calculations included the effects
of core-leakage neutrons. The effect of the delayed-
neutron source spéctrum was appreciable in terms of
the >®Ni(n, a) reaction rate, but not significant in terms
of overall helium production. However, all of the flux
spectra in the out-of-core components were substan-
tially “softer” than a typical test-reactor flux spectrum
(e.g., that of HFIR). Consequently, the rates of atom
displacement per unit flux (£> 100 keV) were lower
by factors of 1.2 to 1.3 than those that would be
encountered in a reactor spectrum with the same total
flux above 100 keV. Thus, the total number of
metal-atom displacements produced in these compo-
nents by operation of an MSBR for 30 EFPY could be
produced by irradiations to 2 to 4 X 10'® nvt (£> 100
keV) in a test reactor like HFIR. Although the total
number of atom displacements is not an entirely
accurate measure of metal damage (since it completely
ignores annealing effects), the low equivalent HFIR
fluences suggest that damage by this process is probably
not of major concern for MSR components.

2.2.2 MSBE Nuclear Chamcteristics
J. R. Engel

Perturbation calculations were added to a two-
dimensional, nine-group diffusion calculation for the
core of the reference concept MSBE'? to provide
estimates -of some of the reactivity coefficients of this
reactor. Although these calculations are subject to
refinement when a detailed core design is evolved, they
serve to illustrate the kinds of values that may be
expected. The results are summarized in Table 2.3,
along with previousiy published results for the single-
fluid MSBR.'®> A prominent difference between the
MSBE values and the comparable quantities for the
larger MSBR is in the reactivity effect of fuel-salt
density.. In the MSBE the density coefficient is positive,
so that voids decrease reactivity (—0.1% 6k/k for 1%
voids in the salt). This effect appears to be a con-
sequence. of the higher neutron leakage from the
physically smaller MSBE core. Also significant-is the

11. J. D. Jenkins, RICE: A Program to Calculate Primary
Recoil Atom Spectra from ENDF/B Data, ORNL-TM-2706
(Feb. 14, 1970).

12. 1. R. McWherter Molren Salt Breeder Experiment Design
Bases, ORNL-TM-3177 (November 1970).

13. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970
ORNL-4548, pp. 63—64.
v

¥

Table 2.3. MSR reactivity coefficients

Variable Coefficient
MSBE MSBR
Uranium concentration +0.33 +0.47
(8k/k)/(6 U/U)
Thorium concentration —0.38
(6k/k)/(6 Th/Th)
Fuel-salt density . +0.10 —0.04
(8k/k)/(80/p)
Temperature (6k/k)/6T,°C~"
Fuel salt -7.3x107° -32x107°
Graphite 07x107°  +23x10°°
Total 8.0Xx10° —09x107°

negative temperature coefficient for the graphite in the
MSBE. This quantity combines with the salt density
effect to produce a larger negative overall temperature
coefficient in the MSBE as compared with the reference
MSBR.

The neutron fluxes and power densities from the
above calculations were used to estimate the neutron
and gamma heating in the core graphite by comparison
with comparable quantities for the MSBR.!* Figures

ORNL-OWG 71-13365

10 \ TOTAL

N

"
2 ,
QO
e
Z \
- 8 [ 6AMMA \
q _
@
© \ \
< g
i—.
2 N \
I
= \\
E \
£ 9 h w
<]
: A .
NEUTRON \
2 ——
CORE —
0 |
0 10 20 30 40 50.. 60

RADIAL POSITION {cm)

Fig. 2.1. Radiation heating in MSBE graphite near midplane,
150 MW(t),

14. MSR Program Semiannu. Progr. Rep. Aug 31, 1969,
ORNL-4449, pp. 63-67.

21

2.1 and 2.2 show the resultant heat generation rates in
the graphite radially near the core midplane and axially
near the core center line.

.SE‘ 12 ORNL-OWG 71-13366
S CORE- i— BLANKE T ‘
S — ' | PLENUM
= . | |
R e -
o) I
= 6 \‘\\ | 1|
b—.
I 4 aN |
= NEUTRON \L\ Ll
T M— ——
T~

o ) L]

0 25 50 75 100 125

DISTANCE ABOVE CORE MIOPLANE {cm)

Fig. 2.2. Radiation heating in MSBE graphite near core axis,
150 MW(t). ‘

2.2.3 Fixed-Moderator Molten-Salt Reactor
H. F. Bauman

We have extended the capability of the ROD code for
the calculation of the lifetime average performance of
molten-salt converter reactors with batch processing. At
the end of a batch cycle, usually more uranium can be
recovered than is required to make the reactor critical
with clean salt at the start of the next cycle. Therefore,
we have added an option in the HISTRY part of ROD
to permit the required portion of the uranium (as the
mixture of nuclides present in the core at the end of the
cycle) to be used for startup and the remainder to be
fed in as required before extraneous feed (e.g., pluto-
nium) is resumed. We have also added a-provision for
changing from one extraneous fuel feed to another
(e.g., from plutonium to enriched uranium) at a
specified time in each cycle. Further, we have repro-
grammed the cost section of HISTRY to permit the
calculation of the present worth of all purchases of
fissile, fertile, and carrier salt materials over any number
of cycles at a specified discount rate. This method is
more accurate for calculating costs which change with
time (e.g., the inventory cost, when the fissile inventory
increases during the cycle) than the calculation of costs
from data averaged over the reactor lifetime. (A
description of the ROD code, including the foregoing
improvements, was published during this report
period.}3) The studies reported below used these new
calculational methods. '

15. H. F. Bauman et al., ROD: A Nuclear and Fuel-Cycle
Analysis Code for Circulating-Fuel Reactors, ORNL-TM-3359
(September 1971).
In previous reports we described the calculated
performance of fixed-moderator’®  continuously
processed molten-salt breeder reactors,'” and how a
reactor with core proportions optimized for breeding
could operate as a converter using only batch process-
ing.'® Based on these results we chose the reactor
described in Table 2.4 for further studies on the effects
of various fissile feed ‘schemes. Throughout these
studies we held the processing interval fixed at six
full-power years (four cycles covering a 30-year life at
0.8 plant factor).

The first calculations were intended to compare the
use of plutonium and enriched uranium as the fissile
material for the initial startup and’as feed to supple-
ment the bred material. The plutonium feed compo-
sition was taken to be equivalent to that produced in
water reactors (at. % 23°Pu/2%°Pu/2*!Pu/?%%Pu:
60/24/12/4); the uranium feed was assumed to be 93%
2350, In either case, we assumed that at the end of
each cycle the uranium is recovered and the salt
containing the fission products and plutonium is dis-
carded. (Uranium recovery by the fluoride volatility

16. The term “fixed-moderator™ is used to indicate that the
core is designed so that the limiting fluence for fast-neutron
damage to graphite will not be exceeded in a 30-year nominal
reactor life (24 equivalent full-power years). '

17. MSR ,Program Semiannu. Progr. Rep. Aug. 31, 1970,
ORNL-4622, pp. 26—30. '

18. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, pp. 43-45. .

Table 2.4. Characteristics of the 1000-MW(e) fixed-moderator
- molten-salt reactor® considered in fueling studies

Core diameter overall, ft 28.5

Thickness of core zones, ft

Zone 1 : : 8.46
Zone 2 ’ 3.64
Zone 3 1.97
Annulus ' 0.14
Salt volume fractions '
Zone 1 ) 0.137
Zone 2 0.111
Zone 3 0.127
Total salt volume in fuel system, fit3 2500
Fuel salt, mole % LiF-BeF,-ThF, 69-17-14
Core inlet-outlet temperature, °F 1050--1300
Thermal power, MW(t) : 2250

9A reactor with three-zone core optimized for maximum
conservation coefficient consistent with 30-year life for the
graphite, assuming continuous processing (case SCC-201).

22

process is inexpensive and efficient, but there is at
present no economically feasible process for recovering
plutonium from the fuel salt.) None of the recovered
uranium is sold except that from the last cycle. Instead
it is used for startup in the next cycle, with any excess
fed in as required as long as it lasts. We found that in
either case the calculated conversion ratio is high (over
0.90) so that 233U soon builds up to become the
dominant fissile material. Consequently, over most of
the lifetime of the reactor the concentrations of the
extraneous fissile nuclides are low compared with their
concentrations for a short time after the initial startup.
Recognition of this situation caused us to reexamine
one of the approximations in the calculations, namely,
that neutron energy spectra and effective cross sections
do not change drastically over the. time interval being
considered. '
The ROD-HISTRY computation takes an input set of
nine-group cross sections and generates a set of reac-
tion-rate coefficients (equivalent one-group cross sec-
tions) averaged over the reactor volume and over a
neutron energy spectrum corresponding to a particular
fuel composition. The program then uses these coef-
ficients to compute nuclide concentrations as a func-
tion of time and a new fuel composition that is the
average over the time interval specified for the HISTRY
calculation. This routine is iterated until the time-
averaged composition agrees satisfactorily with the
composition used in evaluating the reaction-rate coef-
ficients. (The same set of nine-group cross sections is
used throughout the iterations.) Of course, the changing
fuel composition” would actually affect the neutron

energy spectrum so that the efféctive cross sections are

not always exactly the same as they are when the fuel
composition is at the average for the time interval; the
error involved in the calculation depends on hew much
the spectrum changes during the interval.

We suspected that for the reactors started on pluto-
nium a considerable spectrum change might be induced
by the effects of the low-lying plutonium resonances. A
typical set of nine-group absorption cross sections
prepared for our study is given in Table 2.5. The energy
boundaries are shown in the table; groups 1 to 5 are
slowing-down (fast) groups while 6 to 9 are thermal
groups. The cross sections of 233U and 23* U are fairly
similar; starting up on 23°U and shifting to 233U,
therefore, would not be expected to greatly change the
reactor neutron spectrum or the fuel-nuclide reaction
rates from those calculated for the lifetime average. The
239Pu cross sections, on the other hand, are very much
higher in thermal groups 7, 8, and 9. This high cross
section makes plutonium a very reactive fuel, with low
A

23

Table 2.5. Typical nine-group absorption cross sections appropriate for the lifetime-averaged fuel composition |
of the fixed-moderator molten-salt reactor

Volume fraction fuel salt =0.12
_Carbon—to-thorium ratio = 167

Group ‘ 1 2 3
Lower boundary, eV 821,000 31,800 1230

Absorption cross
section, barns

321 . 0.2 0.3 1.0
133y 2.1 2.6 7.1
233y 14 19 6.1
239py 2.1 1.9 4.1
240p, 1.9 0.7 2.2
241py 2.0 3.0 6.1

4 5 6 7 8 9
47.9 1.86 0.768 0.180 0.060 0.0047
3.9 2.3 0.6 1.8 3.4 6.3
25.0 1340  318.0 177.0 266.0 491.0
31.0 - 66.0 49.0 175.0 270.0 560.0
35.0 74.0 41.0 2110.0 767.0 883.0"
27.0 21.0 7114.0 189.0 159.0 242.0

31.0 162.0 20.0 1149.0 850.0 1155.0

critical loadings in reactors in which the flux is well
thermalized. However, when recycled plutonium con-
taining much 2%°Pu is the principal fuel, the high
240py cross section (group 6) tends to depress the
thermal flux, hardening the spectrum and increasing the
critical loading of fissile material. Over the lifetime of a
molten-salt converter reactor that is started on recycled
plutonium but soon has much #?3U and little pluto-
nium in it, the effect of plutonium on the neutron
energy spectrum might be expected to range from quite
important to rather slight. Reaction-rate constants
appropriate for the low lifetime average concentration
of plutonium are therefore likely to be seriously in
error at the start of life when the plutonium concen-
tration may be over ten times as high.

QOur calculations had used a set of nine-group cross
sections appropriate for a fuel composition close to the
lifetime average and had computed reaction-rate coef-
ficients for the lifetime-averaged composition. Qur first
test was to recalculate the first full-power year of
operation for both the uranium-fed and the pluto-
nium-fed converters, using the same nine-group cross
sections as before but having HISTRY evaluate reac-
tion-rate coefficients for the average composition for
the first year rather than for the average over the
reactor lifetime. Some results of the two computations
are listed in the first two columns of Table 2.6. The
most salient point is that the calculated fissile loadings
for the plutonium-fed reactor differ by a factor of 4
between two computations in which the only difference
is the interval used in evaluating the reaction-rate
constants. This is because the plutonium concentrations
during the first year, being much higher than the
lifetime average, cause more flux hardening and de-
pression of the reaction-rate coefficients. Obviously the

use of a lifetime-averaged fuel composition in deter-
mining the reaction-rate coefficient entails unacceptable
error when applied to the plutonium startup cases we
had been considering.

However, the reaction-rate coefficients for the en-
riched-uranium startup case changed hardly at all, and
lifetime-averaged coefficients appear to be entirely
adequate for these calculations.

The changes in self-shielding of the plutomum of a
plutonium-fed molten-salt converter reactor are great
enough to affect the calculation of the nine-group cross
sections also. To evaluate the significance of this, we
calculated a set of nine-group cross sections for a
reactor containing about 1000 kg of fissile plutonium.
(The asymptotic fuel composition used previously in
calculating nine-group cross sections included less than
100 kg of fissile plutonium.) This set was then used in
HISTRY to calculate the first year of operation,
evaluating reaction-rate constants for the average com-
position during the year. Results are shown in column 3
of Table 2.6. The effective thermal cross sections for
the plutonium nuclides in the resonance groups were
considerably lower in this set. The spectrum-hardening
effect was therefore not as great, and the reaction-rate
coefficients were not as greatly depressed as they were
in case A32. Compared to case A24-2, however, the
coefficients are low, and the critical loadings are much
higher.

From the above considerations we conclude that
previously reported calculations for converters started
on plutonium,'® while well representing the asymptotic
portion of the feed cycle, do not adequately represent
the startup condition in which plutonium is the .
principal fissile material. The error involved in these
calculations will be minimized in future calculations in .
24

Table 2.6.. Calculated reaction-rate coefficients and fissile inventories
in batch-processed molten-salt reactors

Fuel composition assumed
in determining nine-group

cross sections

Interval for averaging

composition for reaction

coefficient calculation

Plutonium-feed cases

Identification
Reaction coefficients®
232
2334
239p,
240p,

241Pu
242Pu

Inventories at end of
first year,d kg
233
239p,
241p,

Total

Uranium-feed cases

Identification
Reaction coefficients®
232
233
235
239p,

Inventories at e.nd of

first year,d kg
233

235
239Pu

Total

Asymptotical

Life€

A24-2

1.0
59
24.3
14.8
18.8
1.9

451
559
366

1376

A21-1

1.0

6.06
5.51
25.5

649
1683
3

2335

Asymptotic?

+ First yeard

A32 A33
1.0 1.0

3.5 3.9

9.3 10.2
9.8 12.5
8.2 8.6

2.6 2.9

235 303

4190 2621
1263 ‘ 996

5688 3920
A34

1.0

6.15

5.58

26.0

644

1664

3

2311

4Composition corresponding approximately to end of life in case A24-2 (for

plutonium-feed cases) or in case A21-1 (for uranium-feed cases).

bComposition corresponding to highest fissile plutonium concentrations in

case A24-2: 762 kg 23%Pu, 228 kg 2*1Pu,
€24 equivalent full-power years, with processing at 6, 12, and 18 EFPY.
p N

IRirst equivalent full-power year of first cycle.

€Specific neutron absorption, normalized to 2327,

High Pu®

First yeard
which ROD-HISTRY averages fuel compositions over
much shorter intervals. It is evident, however, that
measures to maintain a well-thermalized flux spectrum
will probably be required to use recycled plutonium
effectively for the startup of high-performance thorium-
2331 converter reactors. |

We are continuing. to study plutonium startup, with
the object of reducing the initial fissile loading. One
method would be to increase the moderator ratio by
simply lowering the thorium concentration in the
reactor that we have been studying. In addition, we
plan to study reactors designed specifically as pluto-
nium-fed converters (in contrast to the reactor studied
thus far, whose core characteristics were optimized for
breeding with continuous processing). We would expect

Table 2.7. Performance of a 1000-MW(e) fixed-moderator
molten-salt reactor operated as a converter?
with enriched uranium as feed material

Case identification - A21-1
Feed material purchased 235pb
Initial loading, kg fissile » 2430
Lifetime purchases, kg fissile 3599
Recovery at end of life, kg
3y 1950
335y 381
Net lifetime requirements,© kg fissile 1268
Lifetime averaged conversion ratiod 0.946
Fuel costs,® mills/kWhr
Inventory
Fissile - 0.566
Salt 0.051
Salt replacement : 0.092
Fissile burnup 0.054
Total 0.763.

dReactor described in Table 2.4. Salt and plutonium dis-
carded and uranium recovered at end of four six-EFPY: cycles.
Recovered uranium used for startup and feed in. next cycle.

b939%, enrichment.

cLifetime purchases less fissile uranium récovered at end of
last cycle.

dNet, taking into account discard of plutonium and neglect-
ing any loss of uranium in each processing cycle.

¢Excluding processing costs. Obtained from present-worth
calculation of fissile, fertile, and carrier salt purchases and
fissile sales over life of reactor, with discount rate = 0.07 vear !,
compounded quarterly, and inventory charge rate =-(0.132
year ™' Values of 11.9 $/g
sile plutomum were assumed..

SuU, 13.8 $/g 232U, and 9.9 $/g fis-

25

the optimum converter reactor designs to have a
well-thermalized core flux spectrum.

The results of the enriched-uranium feed case shown
in Table 2.7 indicate that very good performance can be
obtained from the breeder reactor operated as a
converter with batch processing. The changing fuel
composition over the reactor lifetime is shown in Fig.
2.3. The conversion ratio of over 90% is good for a
reactor processed only every six years. The fuel cost of
about 0.76 mill/kWhr excluding processing is attractive.
The cost of processing and disposal of the spent fuel has
been roughly estimated to add about 0.1 mill/kWhr to
this cost. The high fissile inventory cost relative to the
burnup cost suggests that a lower thorium concen-
tration would give a lower overall fuel cost for a reactor
designed specifically as a converter. This would result in
a lower fissile inventory at some sacrifice in conversion
ratio.

ORNL-DWG 71-13367

2000

\235U — /\\ //\ 233U
1800 \ // / /
woo | .

1400 ,
1200 \ /
1000 \ :
éOO / \
600 \
/ \ / 236U
— 4——-——://" :,/ 235y
/ /\// ~L
200 V ; .
236U

0 4 8 12 16 20 24 28

TIME (equivalent full- power years)

INVENTORY (kg)

400

Fig. 2.3. Fuel nuclide inventories for enriched uranium feed
case A21-1. .
3. Systems and Components Development

Dunlap Scott

3.1 GASEOQUS FISSION PRODUCT REMOVAL
C. H. Gabbard

3.1 .1 Gas Separator and Bubble Generator

The testing of the bubble separator continued in the
water test loop.! Tests were conducted to learn more
about the formation of the very fine bubbles that were
encountered with the water-glycerol test fluid and to
evaluate the separator performance with these small
bubbles. Although much of the available evidence
indicates that the bubbles will be larger in molten salts,
the large changes observed in the circulating gas content
of the MSRE fuel salt suggest that the bubble size might
vary over a considerable range. The development of the
bubble separator is therefore proceeding on the basis of
removing the small bubbles with the best possible
efficiency. '

A series of tests with glycerol was completed which
shows that the liquid circulation pump was the source
of the very small bubbles and that the bubble size was
related to the pump head or speed. It is probable that
small bubbles formed in the pump during the operation
with demineralized water coalesced to a larger size
before reaching the separator, while the bubbles formed
in the glycerol solution were inhibited from coalescing

1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 49.

and were representative of the size generated. The test
fluid was changed from the water-glycerol mixture to a

- 31% CaCl, solution in hopes of avoiding the very small

26

bubbles while maintaining hydraulic similitude, but
little or no difference was observed.

A gamma radiation densitometer was installed on the
water test loop to measure the separation efficiency

" under sevéral conditions of operation. We measured the

performance of the separator with the large bubbles in
demineralized water and with the small bubbles in the
31% CaCl, solution, and the results are shown in Fig.
3.1. The effect of bubble size on the performance of
the bubble separator was also demonstrated by a
dilution experiment starting with the 31% CaCl,
solution. Figure 3.2 shows the results of this experi-
ment. As the concentration was reduced, a sharp
increase in separation. efficiency indicated the point
where the bubble size increased as a result of coales-
cence. The coalescence transition occurred at a concen-
tration of about 2.7 wt %, which is considerably greater
than the 0.61 wt % reported in recent literature,? but
this difference could be caused by other additives in our
test solution or by the grossly different hydrodynamic
conditions. These tests suggest that the size of bubbles
circulating in a reactor system will be determined by
the hydraulic design of the pump and circulating system
and by the coalescence properties of the fuel salt.

2. R. R. Lessard and S. A. Sieminski, “Bubble Coalescence
and Gas Transfer in Aqueous Electrolytic Solutions,” Ind. Eng.
Chem., Fundam., 10(2), 260—-69 (May.1971).

.
ORNL-DOWG 71—102214A

100 I |
90{ DEMINERALIZED WATER

50 —
31 % CaCl, SOLUTION

0 1 2 3 4 5 6
GAS INPUT RATE ({scfm) ’

Fig. 3.1. Performance of 4-in..ID standard length bubble
separator at 500 gpm. '

ORNL~DWG 71-13136

100 &
L
L\
L ]
2
~ 80
(6]
=
iy
. .
w .
L 70 \ | _— GAS REMOVAL
Lo d EFFICIENCY
‘--_-.-. L\
.60 : ——
\'
50

1.0 14 1.2 1.3
SPECIFIC GRAVITY

Fig. 3.2. Results of bubble coalescence test with CaCl,
solution at liquid flow rate = 500 gpm and gas flow rate = 5.15
scfm.

The distance between the swirl vane hub and recovery
hub of the test separator was increased from 27 to 44
in. to achieve the greater vortex length required to
separate bubbles from higher viscosity fluids. The
separation efficiency was improved substantially at low
gas flow rates by lengthening the separator, but a vortex
instability limited the gas removal capacity to below 1
scfm. Similar results were obtained with a long separa-
tor having a tapered outer casing. However, the removal

27

of the central core from the annular recovery3 hub
resulted in satisfactory vortex stability over the full gas
flow range of O to 6 scfm. When operating in this
manner the gas void diameter is between % and Y% in.
depending on the gas flow rate. Efficiency tests are
currently in progress for this mode of operation.

3.1.2 Bubble Formation and Coalescence Test

A test has been planned to compare the bubble
formation and coalescence properties of molten salts
with those of demineralized water and other fluids. The
test will consist of mechanically agitating a sealed
quartz capsule of the test fluid in a special furnace and
photographing the capsule to show the size and
behavior of bubbles of gas in the liquid during and after
the agitation. The design of the test rig is nearing
completion.

3.2 GAS SYSTEM TEST FACILITY
R. H. Guymon

The gas system test facility (GSTF) is a loop for
circulating a fuel salt typical of the MSBR for use in
deveéloping the technology of components to be used in
the primary system of a molten-salt reactor. The
performance of bubble generators and. bubble separa-
tors, being developed initially in a water loop (see Sect.
3.1.1), will be tested, along with components for
purification and recycle of off-gas. Other tests include
determining the effect of the UF;/UF, ratio on the
surface tension of the salt, studying noble gas and
possibly tritium distribution, and measuring heat and
mass transfer coefficients. The conceptual system de-
sign description (CSDD)? and the quality assurance®
for the facility were published during the period.

The facility, shown schematically in Fig. 3.3, will be
located in the enclosure previously occupied by the
molten-salt pump test stand in Building 9201-3 of the
Y-12 plant. The loop is of 5-in. sched 40 Hastelloy N
pipe, much of which was removed from the MSRE
coolant system. Molten salt at 1050 to 1300°F is
pumped at 1000 gpm and 100-ft head by the MSRE

3. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-
4676, p. 49. '

4. W. K. Furlong, Conceptual Design Description of the
Molten-Salt Loop for Testing Gas Systems, ORNL CF-71-1-40
(Jan. 25, 1971 ) for internal use only).

5. R. H. Guymon, Quality Assurance Program Plan for the
Gas Systems Technology Facility, ORNL CF-71-7-6 (July 6,
1971)(for internal use only).
INSTRUMENT
PENETRATIONS

VIEWING

SALT PUMP

ORNL—DWG 74-—13137
STACK

|

7 (O | BLOWER

" N ENCLOSURE

SURVEIL-
LANCE
STATION

B

ALT
MONITORING

( BULK SALT SEPARATOR

STATION

e

11T

FUTURE
EXPERIMENTS

VENTURI

SAMPLER

BUBBLE
BUBBLE /////r SEPARATOR

GENERATOR

a——— HELIUM

il
L s

———=— STACK-

Fig. 3.3. Gas system technology facility. |

Mark 2 pump modified as described in Sect. 3.7.1. A
line downstream of the pump allows part (approxi-
mately one-half) of the flow to bypass the main loop.
The main loop flow goes through a flow-measuring
venturi (from the MSRE), a bubble separator, and a
bubble generator before rejoining the bypass flow and
returning to the salt pump.

Helium is injected at the bubble generator and flows
around the loop and through the pump prior to being
stripped at the bubble separator. The separated gas (O
to 37 slm) and entrained salt (12.5 to 6 gpm) go to the
bulk salt separator which is an 8-ft-long section of 5-in.
Hastelloy N pipe inclined slightly from horizontal. The
salt from the bulk salt separator is returned to the loop
via the salt pump bowl, and the gas is recycled to the
bubble generator. .

Pipe nozzles are shown in the high- and low-pressure
parts of the loop to facilitate installation of future
experiments. A salt-monitoring station is included for
testing various on-line instruments for determining
chemical and physical properties of the salt.

The main loop configuration was established, the

preliminary instrument application drawings were com-
pleted, and the drain tank design was nearly complete

at the end of the period. Initial operation is scheduled
to start in the fall of 1972, using coolant salt from the
MSRE to establish a base line for later operations. After
preliminary experiments and shakedown tests are com-
plete, this salt will be replaced with a salt similar to the
fuel salt projected for the MSBE.

3.3 OFF-GAS SYSTEMS
A. N. Smith

3.3.1 Off-Gas System for Molten-Salt Reactors

‘A design basis report is being written for off-gas
systems for use with molten-salt reactors. This report
reviews the present technology of off-gas systems,
discusses the advantages and disadvantages of several
approaches for the design of the off-gas systems,
describes the design of the system chosen for a
reference, and discusses the additional information
which must be developed to establish the design. The
primary purpose of this report is to document the basis
for the reference design and to define the development
effort needed for the design of reliable off-gas systems.

'
3.3.2 Computer Design of Charcoal Beds
for MSR Off-Gas Systems

The design of charcoal beds for the delay of fission
product gases in MSR off-gas systems involves an
iterative calculation best handled by a digital computer.
Existing computer programs, which were used for the
Homogeneous Reactor Test and the Molten-Salt Reac-
tor Experiment, are arranged to calculate charcoal
temperatures and isotope exit concentrations and, as

such, are insufficient to be of maximum usefulness to

the charcoal bed designer. A computer program is
needed which includes cost and other selected param-
eters and which furnishes data necessary for mechanical
design. A study is being made to determine how the
existing programs might be revised and enlarged to
achieve such a program.

3.4 MOLTEN-SALT STEAM GENERATOR

J.L.Crowley R.E.Helms

3.4.1 Steam Generator Industrial Program

Prbposals were received from four of the eight

industrial manufacturers of steam generators that re-

ceived the request-for-proposal package. An evaluation
team studied the proposals and selected the Foster
Wheeler Corporation. A subcontract is being negotiated
with them.

" The scope of work requested of the industrial firm
was revised from that reported previously.® The
changes basically involved limiting the industrial firm to
work on the steam generator and excluding the systems

- design work originally requested. Although Task II was

most affected by this change, a brief description of all
four tasks follows. _ .
Task I — conceptual design of a steam generator for
the ORNL 1000-MW(e) MSBR reference steam cycle.
The firm will produce a conceptual design of a steam
generator to operate with the salt and steam conditions
of the MSBR reference design. The industrial firm will
consider all aspects of a conceptual design, including
unit size, control of steam temperature, and unique
support systems such as a startup system and a salt
circuit pressure relief system. o
Task II — feasibility study-and conceptual design of
steam generators using lower feedwater temperature.
This task is divided into two parts, with the execution

6. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970,
ORNL-4622, p. 40.

29

of the second dependent on the results of the first.
First, the selected firm will assess the feasibility of
molten-salt steam generators using feedwater tempera-
tures between 500 and 600°F. He will investigate both
subcritical and supercritical pressures. If the feasibility
assessment is affirmative for one or both of the above
pressures and if ORNL agrees, then the industrial firm
will prepare conceptual designs for steam generators
using these conditions and the salt system conditions of
Task I.

Task III — conceptual design of a steam generator for
a molten-salt reactor of about 150 MW(t). ORNL will
select a steam generator conceptual design from the
results of Tasks I and II. Then the industrial firm will
adapt this concept to a molten-salt reactor system of
about 150 MW(t).

Task IV — description of a research and development
program for Task III steam generator. For Task IV the
industrial firm will describe the research and develop-
ment program necessary to assure that the design of the
Task III steam generator is adequate. As part of this
task the industrial firm will review ORNL’s proposed
steam generator tube test facility and the associated
program and will also review the program of testing:
Hastelloy N and other materials in steam.

We hope to have Task I conceptual design completed
during FY 1972. Tasks II, III, and IV would be
completed during FY 1973 or early in FY 1974.

3.4.2 Molten-Salt Steam Generator Technology Facility

Work continued on the preparation of a conceptual
system design description (CSDD) for the steam gene-
rator technology facility (SGTF).” This facility would
be a side loop of the coolant-salt technology facility
(CSTF)8 and would be used to study transient and
steady-state operation of some full-length tube test
sections and portions of a tube test section of a
molten-salt steam generator. The tests performed in this
facility would better define some of the technological
uncertainties to be investigated later with a larger

‘multitube test facility (STTS). The SGTF would be

used to:

1. investigate some of the problems of generating steam
using sodium fluoroborate and feedwater at various
temperatures up to 700°F;

7. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 52.
8. Ibid., p. 57.
30

2. perform heat and mass transfer tests in various steam
generator tube configurations;

3. investigate the use of thermocouples, strain gages,
and flow elements in a molten-salt environment
typical of larger test facilities such as the STTS;

4, study flow stability in a single tube and the ability
of orifices to eliminate any instabilities for load
changes in the range of 20 to 100%;

5. explore ‘some of the problems associated with the
detection of a steam-to-salt leak and the venting of
resulting pressure;

6. obtain operating data on a scaled-down version of a
startup system.

3.4.3 Molten-Salt Steam Generator Test Proposals

Molten-salt steam - generator test configurations are
being considered for test in the SGTF and the STTS.

To date studies have been made on several tube-in-
tube configurations in which supercritical water flows
vertically in the inner tube countercurrent to an
unbaffled salt stream in the annulus between the inner
and outer tube. To select an inner tube wall thickness, a
computer stress analysis program was written to calcu-
late the internal working pressure that would produce a
total prirhary plus secondary stress intensity of 13,000
psi at 1100°F. The allowable stress intensity of 13,000
psi at 1100°F meéts the requirements of Section III of
the ASME Code and Code Case 1315-3 for Hastelloy N
material. The results of these calculations are shown in
Table 3.1. | - .

Seven tube sizes ranging from % to 1 in. in diameter
in steps of '3 in. and having an allowable working

pressure equal to or greater than 3600 psig at 1100°F
were selected from Table 3.1. These tubes were
subjected to further study. _
A heat transfer program was written to calculate the
salt and supercritical water/steam temperature profile
for the tube-in-tube molten-salt steam generator ‘test
section. The program calculates the steady-state heat
transfer, steam pressure drop, water/steam temperature,
salt temperature, and temperature difference across the
inner tube for a finite element length of test section.

. The salt temperature (TFI), the inside wall temperature

of the inner tube (TWI), the water/steam temperature
(TSO), and the temperature drop across the inner tube
wall (DTW) as a function of test section tube length (ft)
are then displayed by the Calcomp plotter as shown in
Fig. 3.4. The inner tube inside diameter (DI), the inner
tube outside diameter (DO), the outer tube inside
diameter (D1I), the outer tube diameter (D10), the
total heat transferred (kW), the salt flow rate (WF), the
steam flow rate (WS), the inlet pressure (psia), pressure
drop (psi), the finite element length (ft), and the date
are printed by the Calcomp plotter above the curves.

Using a feedwater flow rate that would produce 20
fps and a salt flow rate .of six times the feedwater flow
rate, the outer tube was sized to give salt velocities of
about 20 fps. Preliminary results of calculations for
seven tube-in-tube steam generator configurations are
shown in Table 3.2. For full-length tubes that receive
700°F feedwater and produce 1000°F steam, only the
'4- and the *g-in.-diam tubes could be tested in the
150-kW SGTF, while larger tubes would be tested in the
3.0-MW STTS. Sections of larger tubes could be tested
in the SGTF but not for full inlet and outlet steam
generator conditions. .

Table 3.1. Maximum working pressure for Hastelloy N tubing in steam generators at 1100°F

Tube QD - Maximum pressure (psig) for tube wall thickness (in.) of —

(in.) 0.0350¢  0.0490 0.0650 0.0830  0.0950

0.1090  0.1200 0.1340  0.1480 0.165 0.180

0.2500  3180.4 40972 50024 57662  6125.6
0.3125 25859  3437.6 4283.1 50715  5501.2
03750 22002 2953.4 37255 44810 4918.0
04375  1915.6  2588.4  3292.1  4000.3  4423.4
0.5000  1692.6 2298.3  2945.6  3599.5  4001.4
0.5125  1654.3 22482 2379.3  3588.8  3926.1
0.6250 13745 . 1878.6 24228 29943  3351.3
0.6875  1256.3 1721.0 22258  2765.0  3096.3
0.7500 ~ 1156.7 15877 2058.0 25589  2876.2
0.8750 998.4  1374.5 1788.0 22323  2516.4

1.0000 878.1 1211.6 1580.1 1978.9 2235.3

6393.5

5905.6 6240.0 6368.2 " 6481.9

5360.7 5771.8 5978.8  6211.5 6406.4  6498.8
4867.6 5309.8 5527.9 58233 6116.3  6347.2
44324 48750 5100.6 5418.0 5748.6 60_68.6
4358.7 4794.8 5020.6 5349.0 5675.8 59975
3743.6 4160.0  4379.2  4698.9 5051.9 5418.0
3468.6 3807.8 40799 43922 47424  5113.7
3229.5 3611.1 3815.4 4118.2 4461.6  4831.4
2835.4 3183.7 3371.9  3653.9 3978.3 43349
2525.1 2843.7 3017.1 3278.5 3582.1  3920.1

"
ORNL-DWG 71-13138

Dl (in.) 0.245 KW 149.58 INLET PRESSURE 3800.00

DO {in.) 0.375 WF 4812.00 PRESSURE DROP 101.06
D11 (in.) 0.495 WS 802,00 ELEMENT LENGTH (ft} 1.00
DIO (in.) 0.625 NB 2 DATE 8-30-71
1200
TFI
L
1100 ' . //
/ TWI
/ //TSO
1000 %
'l.;l_' / . ///
= 800 //, 200
£ 7 |
oq: —'_____-_____.._-
w700 -
< —— —1150 &
600 <
// N =
—1 t00 §
. DTW
500 S @
<1
[
—1 50
400 <].
300 o]
0 5 10 15 20 25 30 35

TUBE LENGTH (ft)

Fig. 3.4. Tube-in-tube molten-salt steam generator test sec-
tion. '

3.5 SODIUM FLUOROBORATE TEST LOOP
A. N. Smith

The as-removed appearance of the PKP pump rotary
element and the appearance of certain other test loop
components were described in the previous report.
During this period, the pump rotary element was
dismantled, and the component parts were examined
visually after washing with hot water. Except for the
inner heat baffle plates, the appearance of the Inconel
metal surfaces was such as to imply insignificant attack
during the fluoroborate service. The inner heat baffle
plates were badly corroded, and their condition was
attributed to corrosive material produced by the reac-
tion of moisture in the incoming purge gas with puddles
of salt which were left on the baffle plates during the
ingassing transients in May and June 1968. More
detailed information on observations and conclusions
resulting from examination of the rotary element parts
is presented below. No further work is planned on this
test program. A report? is being prepared summarizing
the work on the sodium fluoroborate test loop.

31

3.5.1 Inspection of PKP Pump Rotary Element

The impeller and upper impeller housing (see Fig. 3.5)
were removed from the PKP pump rotary element, and
the upper labyrinth seal ring and the heat baffles were
removed from the impeller housing. All parts were then
examined visually after washing with hot water to
remove residual deposits of NaBF, salt. The following
is a summary of our findings. -

1. The vanes on the back side of the impeller and the
grooves inside the upper labyrinth seal ring showed
little sign of wear, based on the appearance of sharp
edges and corners. '

2. The diameter of the slinger ring appeared to match
the curvature of the high-nickel, gray-black lumps
previously described.19 '

3. The appearance of the inner heat baffle plates was
markedly different from that of the outer heat baffle
plates. Except for a slight blackening of the surfaces,
the outer plates suffered very little attack. On the other
hand, the inner baffle plates (see Fig. 3.6) had been
severely attacked as evidenced by holes completely
through in some places and by severe pitting in other
places. The attack was not homogeneous on any one
plate or from one plate to another. The top surface of
the top plate was severely pitted, but the other surfaces
which combined to form the chamber above the top
plate (shaft, inner surface of impeller housing support
cylinder, and lower surface of the cooling oil chamber)
were relatively free of attack. All baffle plates except
the one below the top plate had holes corroded
completely through. The inner heat baffles fitted very
tightly against the inner surface of the upper impeller
housing support cylinder, so that probably most of the
gas flow was past the inner annulus between the baffle
plates and the pump shaft.

4. A groove was found in the surface of the pump
shaft at a point adjacent to the inner heat baffles. The
groove was about Y, in. wide and about % in. in
maximum depth. It was smooth in contour and was not
pitted. '

5. Metal surfaces had a coarse, grainy appearance but
were not severely pitted except on the inner heat baffle
plates. | .

6. Contrary to pump assembly drawing F-2882-K,
there were no holes in the impeller housing support
cylinder just above the inner heat baffles. The absence

9. A. N. Smith, Experience with Fluoroborate Circulation in
an MSRE-Scale Loop, ORNL-TM-3344 (to be published).

10. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 55.
32

Table 3.2. Results of calculations for test sections operating with salt and feedwater velocities at 20 fps

2.20.1

Test section number 1.20.1 3.20.1 . 4.20.1 5.20.1 6.20.1 7.20.1
OD of inner tube, in. 0.250 -0.375 0.500 0.625 0.750 0.875 1.000
1D of inner tube, in. 0.152 10.245 0.334 0.407 0.510 0.579 0.640
Wall thickness of inner tube, in. 0.049 0.065 0.083 0.109 0.120 0.148 0.180
OD of outer tube, in. 0.375 0.625 0.750 1.000 1.125 1.250 1.500
ID of outer tube, in. - 0319 0495 0666 0810 0.995 1.152 1.310
Wall thickness of outer tube, in. 0.028 0.065 0.049 0.095 0.065 0.049 0.095
Inlet feedwater temperature, °F - 700 700 700 . 700 700 700 700
Outlet steam temperature, °F 1001.6 1008.9 1001.4 1002.1 1002.1 1002.2 1002.2
Salt outlet temperature, °F - 850 850 850 850 850 850 850
Salt inlet temperature, °F 1144.1 1145.8 1143.1 1142.4 11416 1140.6 1139.4
Maximum temp drop across inner tube wall, °F 1329 146.3 158.4 174.4 176.6 185.3 191.9
Inner tube wall temp at maximum ATW 798.0 796.0 790.4 787.9 787.4 787.4 787.8
Feedwater flow rate, Ib/hr 309.0 802.0 1491.2 2214.8 3483.0 4488.6 5469.8
Salt flow rate, Ib/hr 1854.0 4812.0 8944.2 13,288.8  20,898.0  26,931.6 32,818.8
Maximum allowable working pressure for

inner tube at 1100°F 4097 3725 3600 3743 3611. 3654 3920
Water/steam pressure drop, psi 73.0 95.6 109.9 128.6 1449 164.2° 187.4
Tube or test section length, ft 16 33 53 76 106 139 176
Total heat transferred, kW 56.1 146.5 270.0 400.2 627.9 806.7 980.5
Heat transferred to point of max temp

drop across inner tube wall, kW 28.5 81.5 141.0 212.0 336.7 442.6 572.4
Distance from inlet to position of '

maximum tube wall AT, ft . 8 17

28 41 58 78 101

of these holes means that all of the pump shaft purge
gas had to flow past the five inner heat baffle plates in
order to get to the main part of the pump bowl.

We have drawn the following conclusions from these
observations: ' ‘

1. Except for the inner heat baffles, corrosive and
erosive attack was slight, which is consistent with
observations made of pump bowl surfaces.1?

2. The high-nickel, gray-black lumps were formed by
deposition on the surface of the slinger ring, probably
by precipitation of corrosion products on the cooler
shaft surface as was discussed earlier.1©

3. The severe heterogeneous attack suffered by the
inner heat baffles came about in the following manner.
NaBF, salt was forced into the baffle region during the
ingassing transients in May and June 1968.1! Scattered
puddles of salt were retained on the baffle surfaces
because the latter were not perfectly flat. The incoming

11. MSR Program Semiannu. Progr. Rep. Aug. 31, 1968,
ORNL-4344, pp. 76-77. . .

purge gas was in intimate contact with the puddles of
salt, and the moisture in the purge gas, estimated to
have been somewhere between 1 and 20 ppm by
volume, combined with the puddles of salt to form
reaction products which corroded the adjacent metal
surfaces.

4. The degree of attack on the upper baffle plates
was at least as severe as that on the lower baffle plates.
If we assume a significant temperature gradient between
the bottom and top plates, the implication would be
that the water-salt reaction products are highly corro-
sive even at relatively low temperatures. Direct measure-
ments of the baffle plate temperatures were not made,
but it is estimated that the temperature of the top
baffle plate was S500°F or less. Additional tests are

needed to determine the variation of corrosiveness with

temperature, since there are important general implica-
tions for reactor systems using NaBF,. For example,
efficient cleaning procedures will be needed to ensure
that NaBF, salt residues are not left on the outside
surfaces of pipes and equipment where exposure to
moist atmospheres is probable, and where the metal
33

temperature is high enough so that corrosion would be caused by material, presumably a mixture of salt, metal,
significant. and corrosion products, which became wedged between
. ' 5. The groove in the pump shaft was apparently the shaft and one of the inner heat baffles.

ORNL—DWG 71-43439

NOTE :
1 TEMPERATURE OF SALT IN PUMP
TANK WAS NORMALLY 1025 °F

100 ¢m>/min.
INERT GAS
SHAFT PURGE

.

3 .
_ 900 cm /m'n-—* 2 THICKNESS OF HEAT BAFFLES WAS
\ 0.032 in. EXCEPT TOP ONE WAS
MAIN FLANGE \ N 0.067 in. A 0.065-in. SPACER
\ & : WIRE WAS USED BETWEEN BAFFLES
COOLING OlL PASSAGES \§ '
\ P 4
\ [
INNER HEAT BAFFLES _____§ %/ 7777
| & 600 / .
OUTER HEAT BAFFLES\\\\\\ cm3/ min. |~ GROOVE FOUND (N SURFACE OF SHAFT
: \ \ APPEARED TO BE CURVED IN CROSS-
\ § N\ SECTION AND ABOUT % in. WIDE BY
PUMP TANK ABOUT % in. MAXIMUM DEPTH
Q APPARENTLY CAUSED BY MATERIAL
4 WEDGED BETWEEN HEAT BAFFLES
. - _ AND SHAFT
o
)

74
7

- SUPPORT RINGS —1

.}

™~ SLINGER RING

\

sIX Ya —in.~diam WEEP HOLES —

. FOUR ¥ —in.-diam WEEP HOLES — | ]

- SPLASH PLATE — 1]

IMPELLER HOUSING | —]

UPPER LABYRINTH SEAL RING

L/l X R

N )
N

SUPPORT CYLINDER —] :
' ' T —PUMP SHAFT
UPPER IMPELLER HOUSING —_] — .
. . % v
\ N —— WINDOWS (2%, in. WIDE BY 1Y% in.
7 HIGH) TO PERMIT FOUNTAIN-FLOW
NORMAL SALT \\\\ /\ SALT TO FLOW BACK TO MAIN PART
LEVEL IN \ ) OF PUMP TANK (TOTAL OF FOUR)
PUMP TANK ‘
J //

_ ¢ OF VOLUTE

| ™ UPPER LABYRINTH SEAL LEAKAGE
{(FOUNTAIN FLOW)

O
%

| V)

VOLUTE _ %
¢
#
¢
¢

P _ IMPELLER /l/

lwe— § OF 3-in.—diam PUMP SHAFT

Fig. 3.5. Cross section of PKP pump. NaBF,4 circulation test, PKP loop, 9201-3.
PHOTO 7E67(A

NOTE: SEMI-CIRCULAR NOTCHES IN PERIPHERIES WERE MADE DURING REMOVAL OF BAFFLES FROM PUMP

IMPELLER END

-CEECCIIPN -

Fig. 3.6. Inner heat baffles from PKP pump showing posttest appearance of upper surfaces. NaBF 4 circulation test, PKP loop,

9201-3.

3.6 COOLANT-SALT TECHNOLOGY FACILITY
A. L. Krakoviak

The mechanical design of the coolant-salt technology
facility is nearing completion. The design now includes
two side-stream loops and a specimen surveillance
station. One of the side-stream loops supplies salt to a
corrosion product trap to investigate on-line removal of
Na3CrFg from sodium fluoroborate by cold zone
deposition as outlined in the “Conceptual System
Design Description.” The trap was designed to operate
from 750 to 1150°F at a salt flow rate of 0.2 gpm for a
salt system cycle time of approximately 8 hr. The cold
trap was also designed to investigate the recombination
of BF3 vapor (12 to 25% BFj in the off-gas stream
from the pump bowl) with the relatively cool salt melt
(~850°F) in the cold trap. This system is expected to
reduce the BF; content in the off-gas stream to <0.5%
BF;. Provisions have been made in the cold trap to
condense and collect the acid liquid!?2 from the off-gas
stream.

The second side-stream loop was designed for on-line
monitoring of the salt by electrochemical means (devel-
oped by the Analytical Chemistry Division). The loop
consists of a 3-gal-capacity vessel with associated piping
which will operate at a flow rate of ~2 gpm while
maintaining a constant salt volume of 1.2 gal in the
vessel. The salt-monitoring vessel was designed to

12. MSR Program Semiannu. Prog. Rep. Aug. 31, 1970,
ORNL-4622, pp. 41-44.

provide: (1) a quiescent liquid surface of sodium
fluoroborate representative of the main loop, (2)
operation at constant temperature and salt level in the
vessel, (3) salt residence time of approximately 30 sec,
(4) on-line removal, insertion, and adjustment of depth
of immersion of the electrode, and (5) visual access to
the gas-liquid-electrode interface.

Fabrication of the drain tank and the material
specimen surveillance station for the CSTF is complete,
and fabrication of components for the two side-stream
loops has started.

The MSRE coolant system piping, pump bowl, and
level indicator, which were removed for reuse in the
CSTF, were cleaned with oxalic acid to remove the
residual LiF-BeF, salt. However, removal of the resid-
ual tritium contamination from the inside surfaces of
the pump bowl was not complete. The greatest contam-
ination was associated with oily and carbon-like coat-
ings on surfaces,!? and a standard smear test of a
coated area in the pump bowl cylinder opposite the
shield plug showed ~800,000 dis/min of transferable
tritium. After cleaning this surface with abrasive cloth
and dilute nitric acid, transferable tritium was reduced
10 12,000 dis/min. Using this procedure, the piping and
pump bowl nozzles within approximately 6 in. of
proposed welds were cleaned to tritium levels below
2000 dis/min (MPC for uncontrolled area) and are
ready for assembly.

13. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 58.

e 0y

3.7 MSBR PUMPS

A. G. Grindell
W.R. Huntley L. V. Wilson
H. C. Savage H. C. Young

3.7.1 Salt Pumps for MSRP Technology Facilities

The coolant pump tank and volute that were removed
recently from the MSRE and the spare rotary element
from storage will be used in the coolant-salt techniology
facility. When used with a 10.33-in.-diam impeller, this
combination will provide the required 90-ft head and
800-gpm flow at 1765 rpm by operating slightly toward

" the shutoff side of its approximate MSRE operating

point of 85-ft head and 900-gpm flow rate. This
operation is not expected to impose excessive radial
hydraulic forces on the pump shaft. The available 75-hp
motor will be loaded to approx1mately 45 hp for this
service.

The rotary element will be. reconditioned with new
bearings, and the shaft seals will be relapped. It will
then be operated in a cold shakedown stand to check
shaft seal leakage rates prior to installation in the
coolant-salt technology facility.

The head and flow requirements for the gas system
test facility will be provided by installing an available
11%-in.-diam impeller of MSRE coolant-salt pump
design in the MSRE Mark 2 fuel-salt pump. This
impeller will provide 1000 gpm and approximately
100-ft head with 100-bhp input at near 1800-rpm pump
speed. Since the inlet and discharge openings in the
coolant-salt pump impeller are smaller than those in the
fuel-salt pump impeller, a stationary filler piece will be
installed in the volute inlet. A spare MSRE coolant
pump drive motor, nominally rated for 75 hp at 1200
rpm, but wound for 1800-rpm operation, will be used.
This oversize motor can sustain.continuous operation to

113 hp at 1800 rpm. The oversize frame was selected

early during MSRE design to standardize the motor-to-
bearing housing mounting dimensions for both the
1200-rpm fuel pump and 1800-rpm coolant pump.

The rotary element from the Mark 2 pump will be
reconditioned. New bearings will be installed, the shaft
seals will be.relapped, and the rotary assembly will be
balanced dynamically. The element then will be oper-
ated in the cold shakedown stand to check shaft seal
leakage rates prior to installation in the gas system
technology facility. During the assembly, measurements
will be made of radial load vs shaft deflection and shaft
critical frequency of the Mark 2 pump shaft with the
11%,-in.-diam impeller. '

Spare parts for the two pump assemblies have been
inventoried, and spare hollow metal O-rings made from

-Hastelloy N tubing have been ordered. The Y%-in.-OD

by 0.035-in.-wall tubing for the O-rings will be formed
by drawing %-in.-OD by 0.072-in.-wall Hastelloy N
tubing that is on hand.

3.7.2 ALPHA Pump

Initial operation of the ALPHA pump!? with high-
temperature salt in the MSR-FCL-2 test facility demon-
strated the need for several modifications to improve
reliability; the design of these modifications is in
progress. In the meantime, the pump is being operated
in the MSR-FCL-2 at the specified design conditions.

Prior to installation in the salt facility, the pump was
subjected to a cold shakedown test to ensure that
bearings and rotating and static seals function properly.
The results of several tests indicate steady-state oil
leakage rates past the rotating shaft seals ranging from
0.3 to 3.5 cc/day for a pump speed of approximately .
4000 rpm and a seal differential pressure of 7 psi.

Considerable difficulty was experienced with the
hollow metal O-ring gaskets which seal the removable
pump rotary element to the pump bowl. In the present
design the space between two concentric gaskets is
buffered with helium. The gaskets are of hollow

“stainless steel tube construction, have a toroidal diam-

eter of % in., an overall diameter of 6 in., and are
plated with silver -or gold. In an initial test with
silver-plated gaskets, after carefully hand polishing the
sealing surfaces in the matching flanges, a total leak rate
of approximately 6 cc/hr past both rings was attained
with the buffer zone pressurized to 6.5 psig. The plating
flaked from gold-plated gaskets during installation, and
so gold plating is not receiving current attention. The
total leak rate was reduced in a later installation by
carefully hand polishing the silver-plated gaskets: the
total leak rate of approximately 1 cc/hr was attained
with the buffer zone pressurized to 8 psig. This leak
rate was considered satisfactory because it is unlikely
that either air or moisture can enter the pump in a
significant quantity past the buffer gas barrier. How-
ever, because of the difficulty in attaining and maintain-
ing the polished surfaces needed to achieve a satisfac-
tory seal with the hollow metal O-rings, we plan to
replace them with a buffered ring joint seal utilizing a
solid metal gasket.

14. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 59.

’
The ALPHA pump developed a gas leak across the
lower shaft seal after 56 hr of intermittent operation
and a total of 510 hr at elevated temperature in the
MSR-FCL-2. Disassembly and inspection of the pump
revealed than an O-ring made of Buna-N had failed. The
O-ring is part of the shaft seal which was purchased
from the Sealol Corporation. The lower shaft seal
operates with oil lubricant contacting its top side and
gas contacting its lower side. The gas is principally
helium containing approximately 0.1% BF; whose
source is the sodium fluoroborate in the pump tank.
The dilute BF; attacked the Buna-N material and
produced the gas leak. The split purge flow of helium in
the shaft annulus was not operated in this application
because it had not been found necessary to the
successful operation of a similar seal on the LFB pump
in the MSR-FCL-1. Impurity control and facility
operation are also simplified when the split purge is not
used.

The mechanical design of the lower shaft seal is being
‘changed to (1) replace the present elastomeric static
seal with a metallic bellows and (2) to eliminate the
press fit of the seal cartridge in the inner bearing
housing. The press fit has caused galling of the static
sealing surface in the inner bearing housing. In addition,
the cross section of the pump shaft will be strengthened
in the vicinity of the lower shaft seal and bearing. These
improvements in the pump will be accomplished during
a future scheduled shutdown.

In order to proceed with the test program for the
MSR-FCL-2, temporary repairs were made to the pump.
An O-ring made of Viton, which is more resistant to
BF; than Buna-N, was installed in the present lower
shaft seal. The inner bearing housing was machined and
plated with copper to provide a good sealing surface for
the seal .cartridge. After a cold shakedown test the
pump was reinstalled in the MSR-FCL-2 facility and is
presently supplying 850°F salt at the design flow rate
of 4 gpm at 4800-rpm shaft speed. '

3.7.3 Drain Tank Jet Pump System for MSDR

A study was made of a jet pump system to return fuel
salt from the drain tank to the fuel-salt system of the
Molten-Salt Demonstration Reactor (MSDR). The salt
going to the drain tank is part of the effluent from the
gas separator used to remove the gaseous .fission
products from the fuel salt. It is estimated that the flow
~ of fuel salt into the drain tank will be between 27 and
36 gpm when all three fuel-salt pumps and their gas
injector-separator systems are operating normally.

36

REACTOR

The jet pump system is shown schematically in Fig.
3.7. During normal operation, jet pump A, having a
pumping capability of a-little greater than 36 gpm, will
take salt and gas from the bottom of the drain tank and
discharge it along with its driving salt, 0,4, into the
inner vessel. It will probably be necessary to install a gas
separator in the discharge line of jet pump 4.

Jet pump B is the primary salt return pump and is
designed to have a suction flow that is strongly
dependent upon the salt level in the inner vessel. The
characteristics of jet pump B are shown in Fig. 3.8,
where Q;p is the flow return from the drain tank and
includes the driving flow, 0, 4, for jet pump A. During
normal operation the salt level, Hy, in the inner vessel
will be between 6 and 10 ft above the center line of jet
pump B. If the salt level in the inner vessel should drop
to about 5 ft below the jet pump center line, the
suction flow will go to zero and the pump will be
operating very near cavitation. The purpose of having
this pump characteristic is to prevent the unwanted and
uncontrolled reintroduction of gas from the drain tank
into the fuel-salt system should the salt flow from the
separator be drastically reduced. The jet pumps have
some overcapacity should the flow from. the gas

ORNL-DWG 71—13140

GAS SEPARATOR
AHy =150 ft /
:4| ' TO HEAT

EXCHANGER

FUEL
- N\ _PUMP
1

—r

FROM

| 27-36 gom
@n=84¢gpm

65 ft

70 ft

Fig. 3.7. Jet pump system for returning salt from drain tank
to fuel-salt system in a molten-salt demonstration reactor
(MSDR). Schematic of the jet pump drain tank fuel-salt system
showing installation of jet pumps 4 and B and giving elevations
and driving flows.
oA

separators go above the design flow. The jet pump
<characteristics are based on dimensionless experimental
data from pumps tested by NASA.15 |
The study was continued to obtain an indication of
the effects of plating or erosion and corrosion on the

ORNL-DWG 71-1314%
80

70 CCNDITION "a

) / CONDITION "b"
60 : / DESIGN
// CONDITION "gd"

50 / A CONDITION ¢

/ OPERATING
40 / RANGE

ST
V4

@gg (gpm)

-5 0 5 10 15
My, HEAD ABOVE JET PUMP "B" (ft}

Fig. 3.8. Jet pump for returning salt from drain tank to
fuel-salt system in a molten-salt demonstration reactor {(MSDR).
Graph of delivery capacity of jet pump B at its initial design
area ratio of 0.197 and at four different off-design conditions.

15. N. L. Sanger, Cavitating Performance of Two Low-Area
Ratio Water Jet Pumps Having Throat Lengths of 7.25
Digmeters, NASA-TN-D-4592 (May 1968).

operational characteristics of jet pump B. The (unused)
jet pump would have a nominal design area ratio
(driving nozzle area/mixing throat area) of 0.197. Four
conditions were investigated:

a. an area ratio of 0.207 caused by an increase in the
nozzle diameter by erosion and corrosion,

b. an area ratio of 0.207 caused by a decrease in the
mixing throat diameter by plating,

c. an area ratio of 0.187 caused by a decrease in the
nozzle diameter by plating,

d. an area ratio of 0.187 caused by an increase in the
mixing throat diameter by erosion-corrosion.

The driving nozzle diameter for the 0.197 area ratio
was about 0.500 in., and the corresponding mixing
throat diameter was 1.126 in. The four conditions listed
above represent corresponding diametral changes as
follows: :

a. nozzle diameter increase of 0.012 in.,
b. mixing throat diameter decrease of 0.027 in.,
¢. nozzle diameter decrease of 0.013 in.,

d. mixing throat diameter increase of 0.030 in.

The results are shown in Fig. 3.8 from which it can be
seen that (1) an increase in the nozzle diameter
increases jet pump delivery just as does a decrease in the
mixing throat diameter, and (2} a decrease in the nozzle
diameter decreases jet pump delivery as does an increase
in the mixing throat diameter. It can also be seen in Fig.

. 3.8 that conditions ¢ and b would result in an

uncontrolled delivery of radioactive gas to the fuel-salt
system if the inflow of salt to the drain tank from the
gas separator were reduced to about 6 gpm, which
corresponds to a (gp of approximately 15 gpm.
Conditions ¢ and d limit the pumping-capability of the
two jet pumps to about 40 to 45 gpm.

There are alternative configurations and character-
istics for jet pumps that may furnish adequate opera-
tional margins when the long-term erosion, corrosion,
and plating characteristics of container material in fuel
salt are understood well enough to warrant additional
study.
4. Instrumentatidn and Controls

S. J. Ditto

4.1 TRANSIENT AND CONTROL STUDIES OF THE
MSBR SYSTEM USING A HYBRID COMPUTER

O.W. Burke

The hybrid computer model of the MSBR system as
described in the February, 1971, semiannual progress
report! is operational. In order to accommodate the
calculational stability and time requirements of the
digital portion of the hybrid computer, the model
operates in a time domain that is 20 times real system
time.

The model has been checked out, and a number of
transients have been run in order to test its validity. At
this point the model includes no controllers. The next
order of business is to incorporate system controllers
into the model and to check the performance of these
controllers. .

The model will be used in the development of an
overall control scheme for the MSBR plant. Of partic-
ular interest is the effect of variable flow rate in the
secondary salt loop and the use of bypass valves in the
secondary loop. The model will also be used to aid in

the determination of the range of thermal transients .

which could be expected in the salt systems under
anticipated transient conditions. This information is
needed as_input for the design studies of the steam
generator.

4.2 MSRE DESIGN AND OPERATIONS REPORT,
PART IIB, NUCLEAR AND PROCESS
INSTRUMENTATION

R. L. Moore P. G. Herndon

This repoft is the second of two parts describing
MSRE nuclear and process instrumentation. In the first

I. MSR Program Semiannu. Progr. Rep. Feb. 28, 1_971,
0RNL—4676, p.61. .

38

~part,> Chaps. 1 and 2 provide broad and .quite general

descriptions which are intended to convey the criteria
and philosophy used as the basis for the design of the
MSRE instrumentation and controls systems. Chapter 2
also contains detailed descriptions of ‘the nuclear
instrumentation, the health physics, process, and stack
radiation monitoring systems, the data logger computer,
and the beryllium monitoring system. Chapters 3

* through 7, which are contained in this volume, provide

detailed descriptions of the MSRE process instrumen-
tation and of the electrical control and alarm circuitry.?

The draft of the report has been completed, reviewed,
and submitted for minor editing and publication.

4.3 FURTHER DISCUSSION OF
INSTRUMENTATION AND CONTROLS
DEVELOPMENT NEEDED FOR THE
MOLTEN-SALT BREEDER REACTOR

R. L. Moore

Previously published information* concerning the
development and evaluation of process instrumentation
applicable to molten-salt breeder reactors was updated.
Areas where instrumentation techniques and com-
ponents tested during operation of the Molten-Salt
Reactor Experiment may be applicable to the MSBR

“are described, and recommendations for further devel-

opment are stated. In this study to date, no problems
are. foreseen that are beyond the present state of the
art.’ .

This report is now in publication.

2. J. R. Tallackson, MSRE Design and Operations Report,
Part IIA, Nuclear and Process Instrumentation, ORNL-TM-729.
3. Abstract of ORNL-TM-729, Part IIB.

4. J. R. Tallackson, R. L. Moore, and S. J. Ditto, Instrumen-
tation and Controls Development for Molten-Salt Breeder
Reactors, ORNL-TM-1856 (May 1967).
. 5. Abstract of ORNL-TM-3303.

« (3
a

5. Heat and Mass Transfer and Thermophysical Properties

H. W. Hoffman

5.1 HEAT TRANSFER
J. W. Cooke

In studies of heat transfer to a proposed MSBR fuel
salt (LiF-BeF,-ThF,-UF4; 67.5-20-12-0.5 mole %) we
observed that, for the Reynolds modulus range from
2000 to 4000, the heat transfer coefficient varies along
the test section in a manner which appears to be related
to a delay in transition to turbulent flow. It has been

-suggested that this delay in transition is abetted by the
stabilizing influence of heating for the case of a fluid

having a large negative temperature coefficient.! Addi-
tional experimental results using the proposed MSBR
salt have been obtained which tend to confirm this
theory.

In Fig. 5.1, the viscosity u and its first derivative
du/dt are piotted as a function of temperature for the
subject molten-salt mixture. Note that the temperature
coefficient |du/dt| decreases by a factor of 10 from 900
to 1450°F. If, indeed, the flow is stabilized by a large
negative value of du/dt, the stabilization should become
less pronounced as the fluid temperature increases; the
variation in the local heat transfer -coefficient with
length should be less. To verify this hypothesis,
additional heat transfer measurements were taken at an
average fluid temperature of 1450°F and compared
with existing data at 1100°F. These data are tabulated
in Table 5.1.

1. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, pp. 64-67.

39

J. J. Keyes, Ir.

It was not practical to operate the gas-pressurized
heat transfer system in such a manner that exactly the
same mass flow rate and average fluid temperature were
maintained for consecutive runs with and without an
unheated entrance length. This was particularly true at
the higher temperature where both the mass flow rate
and frictional pressure losses were so small that moni-
toring the flow rate to hold the Reynolds modulus
constant was not possible. In addition, the fluid
temperature (and thus the fluid viscosity) varied from
inlet to outlet of the test section, particularly during
the higher heat flux runs.

ORNL-DWG 71-13195
80 0.30

I
o
0,
Q

N
2 \\%-‘ |

s
e/ /df>\'--.._\"""---._._

\ |
\
IELTING POINT | B

du/dt {Ib/ft-hr. °F)

a, VISCOSITY (Ib/ft.hr)
o
o

0
800 900 1000 1100 1200 {300 1400 1500 1600
TEMPERATURE (°F)

Fig. 5.1. The viscosity and its temperature dervative as a

‘function of temperature for the molten-salt mixture LiF-BeF,-

ThF4-UF4 (67.5-20-12-0.5 mole %).
40

Table 5.1. Data for heat transfer experiments using the salt mixture LiF-BeF,-ThF4-UF, (67.5-20-12-0.5 mole %)

/A z
Run T Tout ATy q R, Heat — — - h ‘ = b
No.4 (°F) (°F) C°F) (Bt; };IO‘-st;t balance Re Ver  Mnu [Btu hr ! ft™2 (°F)7!] Ns.r
15 1077.2 1099.9 105.8 0.99 1.01 3370 15.0 204 937 7.96
16 1082.7 1106.6 85.8 0.99 1.08 3264 14.7 25.1 1153 9.94
22 1089.3 1135.5 170.7 2.0 1.13 3591 138 25.1 1157 9.87
23 1114.2 1163.6 189.8 2.0 1.09 3749 1279 234 1076 9.40
104 1399.5 1445.7 78.9 0.82 1.02 3260 6.00 20.22 930 10.93
105 1410.0 1459.0 99.3 0.82 1.04 3152 5.84 16.21 750 - 8.80
106 1390.6 15103 2469 1.96 1.12 3409 556 17.28 795 9.17
107 13854 1506.3 272.8 2.02 1.10 3457  5.62 16.31 750 8.57

2Even runs with unheated entrance length; odd runs without.
bo - o 1/3 0.14
NS-T—NNU/NPI' (p'/lu'S) -

To compare the local heat transfer coefficients at the
two temperatures with or without an entrance region, a
method was developed to normalize the data using
Hausen’s equation to include the dependence of local
heat transfer coefficient on Reynolds modulus and fluid
properties: '

u 0.14
Nyy =0.116 (Vg 213 — 125)Npr1/3<r> . (5.

s

This equation correlates molten salt as well as oil and
water heat transfer data in the transitional flow regime.
The normalizing procedure followed three steps: (1)
Each local heat transfer coefficient for the individual
runs was adjusted to the same Reynolds and Prandtl
moduli based on the average fluid temperature in the
test section; (2) each local coefficient for a pair of
consecutive runs (with and without an entrance region)
was then adjusted to an average of the two Reynolds
and Prandtl moduli for the pair of runs; and (3) finally,
the coefficients for two pairs of runs at two tempera-
ture levels were adjusted to the average Reynolds
modulus for the two pairs of runs. In most cases, these
adjustments amounted to only a few percent in the A,
value.

Variations in the ratio of the normalized local heat
transfer coefficient h,, to the heat transfer coefficient at.
the exit, h.,; (entrance length case), with distance
from the inlet to the heated test section are shown in
Fig. 5.2. Included are results for two values of heat flux
and Reynolds modulus.

A small but significant improvement can be seen in
the heat transfer development with length at the higher
fluid temperature, as evidenced by the more rapid

approach to constant h, for the case without an
entrance length. The influence of heat flux on the heat
transfer development cannot be conclusively deduced
from Fig. 5.1, however, because of the simultaneous
variation of Reynolds modulus in this transitional range
for which relatively small changes in the Reynolds
modulus are significant.

ORNL~-DWG 7{-1{3196

1.6 : ,
Nge= 3500
2
14 [\ g/a= 2x10% Btu/hr- ft
\ — — 1430°F
. 12 N\ {450°F
= \\Q .
£
\’f 1.0 \ N”H
h N \ =T ]
—T
~ —
S —
WITHOUT
0.6 !
1.6 | I
: ' Nge= 3250 _‘
2
‘.4 g/2 = 1x10% Btu/hr- ft
\\ — — 1095°F
-— o
Qg {.2 1450°F
~ N WITH 1
< 1.0 .\\
N
0.8 A\ WITHOUT ~—— st =
~. —_— e — .T —
0.6
0" 20 40 60 80 100 120

X/0

Fig. 5.2. Variation in the ratio of local to exit heat transfer
coefficients with distance from the inlet to the heater section
with and without an entrance length. The exit coefficient with
an entrance region was used to obtain this ratio.

e
a

LY

a

Experiments aimed at a study of the effect of wetting
on molten-salt heat transfer using the proposed MSBR
fuel salt have been delayed due to inability, at the
present, to understand the wetting behavior of this salt
in static capsule experiments. The salt was observed to
be wetting on Hastelloy N initially, whereas our
previous experience with the salt mixture LiF-BeF,-
Z1F,-ThF,-UF, (70-23-5-1-1 mole %) suggested that

41

nonwetting should be observed initially, and that a

reducing agent (e.g., Zr or Be) would be required to
produce wetting.? Since the MSBR salt exhibited

wetting under all conditions, we believe that it is

necessary to explain the behavior and to investigate
techniques for inducing nonwetting which can be
applied in the heat transfer system.

%

5.2 THERMOPHYSICAL PROPERTIES
J. W. Cooke

Measurements of the thermal conductivity of the salt
mixture LiF-BeF, (66-34 mole %) have been made over
the temperature range from 300 to 870°C (458°C mp)
using the improved version of the variable-gap tech-
nique.® The preliminary results (omitting two data sets
in the solidus region which are being analyzed) are
shown in Fig. 5.3. These new results fall midway

between previous uncorrected and corrected results,*>5 .

but the temperature dependency in the liquidus region
is less than our previous measurements had indicated.
The ratio of the liquid to solid conductivity is around
0.75 compared with the average value of 0.86 + 0.13
published for a group of salts.® The estimated maxi-

. mum uncertainty in the present results is expected to

be less than +10%.

5.3 MASS TRANSFER TO CIRCULATING BUBBLES
T.S. Kress

The first phase of an experimental program designed
to measure mass transfer rates between small bubbles
circulating with a turbulent liquid has been completed.

2. MSR Program Semiannu. Progr. Rep. Feb, 28, 1971,

ORNL-4676, pp. 67—68.

3. MSR Program Semiannu. Progr. Rep. Feb. 28, 1970,
ORNL-4548, p. 89. :

4. MSR Program Semignnu. Progr. Rep. Feb. 28, 1969,
ORNL-4396, p. 125.

5. MSR Program Semiannu. Progr. Rep. Aug. 31, 19689,
ORNL-4449, p. 92. : :

6. A. G. Turnbill, Australian J. Appl. Sci. 12, 30-41
{(January 1961). ‘ :

ORNL-DWG 71-13197

(x10 °)
|
[}
5 |
o I MELTING POINT
: |
E 2 } {
B | .
> l _—-6—-—-—01"'""__— oo *—
L I
> I
c 8 .
o |
o .
=
o)
(]
a
= q
o
wl
I
’_
0
400 500 600 700 800 900

TEMPERATURE (°C)

Fig. 5.3. The thermal conductivity of the molten-salt mixture
LiF-BeF; (66-34 mole %) using an improved variablegap -
conductivity apparatus.

Mass transfer coefficients have been measured for
helium bubbles of mean diameters from 0.015 to 0.05 -
in. extracting dissolved oxygen from five different
mixtures of glycerol and water (Schmidt modulus range
from 419 to 3446) over a pipe Reynolds modulus range
from 3.1 X 10® to 1.6 X 10° for both horizontal and
vertical flow in a 2-in.-diam conduit. Extensions of this

~program are projected for different pipe diameters and

different interfacial conditions. Additional experiments
are anticipated that will establish mass transfer rates
applicable to such flow. “discontinuities” as elbows,
tees, valves, venturis, etc. .
One of the first conclusions of these experiments is
that the only effect on mass transfer rates of bubble

volume fractions up to 1% is in the highly predictable

change in surface area. No significant influence of
bubble volume fraction on the mass transfer coeffi-
cients (which are based on unit area) was detected.

An examination of all the experimental data has
revealed that the mass transfer coefficients depend on
the flow orientation (horizontal or vertical). At suffi-
ciently high flows, however, the horizontal and vertical
results are observed to be indistinguishable within limits

- of the experimental error. To establish a criterion for

the Reynolds modulus that marks this equivalence of
horizontal and vertical flow, use was made of an
expression developed previously’ for the relative im-

7. MSR Program Semiannu. Progr. Rep. Feb. 28, 1969,
ORNL-4396, pp. 124--28.
42

portance of turbulent inertial forces compared with
gravitational forces,

FIR = 0316# Re“/4
i/ g 2p3D4 VS )

A conservative value, Fl-/}“‘&r = 1.5, has been estab-
lished above which it is valid to assume that horizontal
and vertical flow mass transfer coefficients are the
same. Equation (5.2) is a correlation of the horizontal
flow data and the vertical flow data for which F /F >
1.5:

Sh/Sc!'/? = 0.34Re?-** (a,,/D)'-? , (5.2)
‘which has a standard deviation in In (Sh/Sc'/?) 0f 0.19
and an index of determination of 0.86. Figure 5.4
shows a comparison of the horizontal flow data with
Eq.(5.2).

As flow was decreased to a value for which the ratio
of liquid axial velocity to bubble terminal velocity (in
the gravity field) was about 3, severe stratification of
the bubbles made operation in horizontal flow imprac-
tical. The vertical flow coefficients underwent a transi- -
tion from the behavior represented by Eq. (5.2) to
approach asymptotes characteristic of bubbles rising

A

through a quiescent liquid. These asymptotic values, &,
are’ well described by the correlation recommended by
Resnick and Gal-Or,?

k, =0.154 (Dpg/u)/? \/dvs/2-. (5.3)

In addition, the variable ratio F;/F, proved to be a
good linear scaling factor for calculating the vertical
flow coefficients in the transition rtegion. A recom-
mended correlation for the vertical flow coefficients in
the transition region is

1

(10F;/F,)/1.5
ka|Tv (10F;/F,)/1.5

k. =
1+ (10F/F)/15 |’

Vv

where k;, and k, are to be determined from Egs. (5.2)
and (5.3), respectively. The curves calculated from this
equation are shown in Fig. 5.5, together with horizontal
and vertical flow data for a 25% glycerin mixture. The
horizontal flow regression equation plotted in Fig. 5.5
includes data for glycerm compos1t10ns of 12.5, 25, and
37.5%.

8. W. Resnick and B. Gal-Or; “Gas-Liquid Dispersion,”
295393 in Advances in Chemical Engineering, Vol
Academic, New York, 1968.

pp.
7,

ORNL-DWG 71— 79944

10
[ HORIZONTAL FLOW
—  REGRESSION LINE FORCED TO FIT Sc’2
—  d, BUBBLE
5 MEAN SCHMIDT NO., Sc
~ DIAMETER (in) 370 750 2013 419 V4
L 0.015 ] . . F A
— 0.02 s v a4 » .7
3 0.025 B s 0 e ¥ ¥4
' 0.03 .0 o o 2] ;P
) 0.035 a v < S s/ L}S
& y# |- BOGLCERNE 125 25 375 0 _ |4 &4 i
.o N o
—_ Lo
= ol e X
73} 5 R 0’04 -
7P
Y- -
§ / -, .
o 7
> /<Sh=0.335 ReO-9 sc¥2 (%
10° :
103 2 5 10? 2 5 10° 2 5 108

PIPE REYNOLDS NO., Re= VO/v’

Fig. 5.4. Typical experimental mass transfer coefficients for horizontal and vertical flow in a 2-in.-diam conduit-.
e

43

ORNL—DWG 7{—7994A

10
CALCULATED ASYMPTOTES Iys
5 La= 0.04
— I T —— |
g ; _ —~~ T P 1 )0.03
- - e ——] 7 , y
- ¢>———__-__¢.._:\\\\ X // , ,0.02
=, L TN \ L ‘;A | 0.015
Q : v [ ]
: Nz,
L VERTICAL FLOW 7 J A
S (USING FI//'; AS SCALE FACTOR) o™ A%%/LOCUS OF
< ! ' NS A=
l.(}_) . w= OZ",(
= O
2 7
- 05— K
74 | 25% GLYCERINE FROM HORIZONTAL FLOW
< SCHMIDT NO. = 750 : REGRESSION EQUATION g7
N QP SR N Sh=0.00538 Re'-22 57! (Z22) |
< BUBBLE 1 | | 70.00 0
MEAN
0.2 |- DIAMETER HORIZONTAL  VERTICAL
‘ (in.) FLOW FLOW
0.015 . o
0.02 - . o
AR
.04 v v
O.i " 1 b " 1 Lt 1
10° 2 5 10% 2 5 10° 2

PIPE REYNOLDS NO., Re =VD/v

Fig. 5.5. Comparison of experimental horizontal flow mass transfer data with the recommended regression correlation.
The preceding correlations apparently apply only for
mobile interfacial conditions as evidenced by the high

exponent for the Schmidt modulus. For rigid interfaces,
this exponent is expected to be '43 and the coefficient
multiplying the equation should be different. In the
absence of experimental results, a tentative correlation

for rigid interfacial conditions mlght be 1nferred from .

Eq (5.2) to he:

Sh=0.25Re®24sc!/3(q,,/D)'-0 (5.4)
in which the coefficient‘ 0.25 was obtained from the
coefficient of Eq. (5.2), 0.34, by multiplying it by the
ratio of rigid to mobile coefficients. of equations®

applicable to bubbles moving steadily through a liquid. -

Equation (5.4) should be used with caution because it
has not been justified experimentally or theoretically. A
similar transformation of Eq. (5.3) would be required
to obtain the rigid-interface values of the vertical flow
asymptotes.

9. A, C. Lochiel and P.>H. Calderbank, ‘*‘Mass Transfer in the
Continuous Phase around Axisymmetric Bodies of Revolution,”
Chem. Eng. Sci. 19,471-87 (1964).

44

Nomenclature

0 = molecular diffusion coeffic1ent
D = conduit dlameter
d,; = bubble Sauter-mean diameter
F = .gravitational force on a bubble (buoyancy)

F = mean inertial force on a bubble due to turbulent
. fluctuations

g = gravitational acceleration
k, = low flpw asymptotic value of &,

k;, = horizontal flow axially averaged mass transfer
coefficients

= vertical flow axially averaged mass transfer coeffi-
_ cients -

Re = pipe Reynolds modulus (¥VDp/u, in which V is the
liquid axial mean velocity)

Sc = Schmidt modulus (u/p’l@z)
Sh = Sherwood modulus (kD/ 19)
= llquld viscosity
p = liquid density
v = liquid kinematic viscosity (u/p)
te

‘_P»art 2. C_fiefiiis»tfirym.

W. R. Grimes

The chemical development activities described below
are, in the main, conducted to provide chemical
information required in the development of molten-salt
reactor systems. :

An appreciable, though diminishing, fraction of the .

. effort is still devoted to postoperational examination of

specimens and materials from MSRE. Such examina-
tions, several of which are detailed in this document,
should be largely concluded during the next reporting
period.

The behavior of hydrogen and tritium in molten salt,
metals, and graphite constitutes a substantial fraction of
the rescarch and development effect. Solubility of
hydrogen and its isotopes in molten fluorides is under
study, as are rates of diffusion of hydrogen through
pertinent metals and possible chemical exchange
schemes capable of holding up the tritium to allow time
for its controlled recovery. '

Studies of chemical separation of protactinium during
this period have been confined to those based upon
selective precipitation of oxides. It now appears that
such a separation process is chemically feasible; we
anticipate that, insofar as small-scale studies are con-
cerned, these researches will be concluded during the

coming few weeks. Small efforts continue on improve-
ments to the reductive extraction process for rare earths
and, more significantly, upon methods for removal of
rare-earth species from lithium chloride.

Study of possible mechanisms of the intergranular
attack upon the metallic components in MSRE now
constitutes an appreciable fraction of the chemical
research program. These studies, all initiated in recent
months, have not yet produced information to report.

The principal emphasis of analytical chemical devel-
opment programs has been placed on methods for use
in semiautomated operational control of molten-salt
breeder reactors, for example, the development of
in-line analytical methods for the analysis of MSR fuels,
for reprocessing streams, and for gas streams. These
methods include electrochemical and spectrophoto-
metric means for determination of the concentration of
U* and other ionic species in fuels and coolants, and
adaptation of small on-line computers to electro-
analytical methods. Parallel efforts have been devoted
to the development of analytical methods related to
assay and control of the concentration of water, oxides,
and tritium in fluoroborate coolants.

45
6. Postoperationai Examination of MSRE

6.1 EXAMINATION OF MODERATOR GRAPHITE
- FROM MSRE

S.S.Kirslis  F. F. Blankenship

A complete stringer of graphite (located in an axial
" position between .the surveillance specimen assembly
and the control rod thimble) was removed intact from
the MSRE. This 64.5-in.-long specimen was transferred
to the hot cells in building 3525 for examination,
segmenting, and sampling. Most of the examinations
and analyses were similar to those given to the graphite
surveillance specimens previously removed from MSRE
during its scheduled shutdowns. In addition to these
“standard” methods, surface samples from the stringer
were examined by Auger spectroscopy and by a special

x-ray diffraction technique, cross sections of the

stringer were gamma scanned with a special pin-hole
collimating device, and samples milled from the stringer
were carefully analyzed for tritium.

6.1.1 Results of Visual Examination

Careful examination of all surfaces of the stringer
with a Kollmorgen periscope showed the graphite to be
generally in very good condition, as were the many
surveillance specimens previously examined. The cor-
" ners were clean and sharp, the faint circles left upon
milling the fuel channel surfaces were visible and
appeared unchanged, and the surfaces, with minor
exceptions described below, were clean. The stringer
bottom, with identifying letters and numbers scratched
on it, appeared identical to the photograph taken
before its installation in MSRE.

Examination revealed a few solidified droplets of
flush salt adhering to the graphite, and one small (0.5
cm?) area where a grayish-white material appeared to
have wetted the smooth graphite surface. One small
crack was observed near the middle of a fuel channel.
At the top-surface of the stringer-a heavy dark deposit

- was visible. This deposit seemed to consist in part of

salt and in part of a carbonaceous material; it was

. sampled for both chemical and radiochemical anaiysis.

a6

Near the metal knob at the top of the stringer a crack in
the graphite had permitted a chip (about ! mm thick
and 2 cm? in area) to be cleaved from the flat top
surface. This crack may have resulted from mechanical
stresses due to threading the metal knob into the
stringer (or from thermal stresses in this metal-graphite
system during operation), rather than from radiation or
chemical effects.

6.1.2 Segmenting of Graphite Stringer

‘Upon completion of the visual observation and
photography and after removal of small samples from
several locations on the surface, the stringer was
sectioned with a thin carborundum cutting wheel
(Canosaw) to provide five sections of 4 in. length, three
thin (10 to 20 mil) sections for x radiography, and
three thicker (30 to 60 mil) sections for pin-hole
scanning with the gamma spectrometer. Each set of
samples contained specimens from near the top, middle,
and bottom of the stringer. The large specimens, from
which surface samples were subsequently milled, in-
cluded (in addition). two samples from intermediate
positions.

]
le

6.1.3 Examination of Surface Samples
by X-Ray Diffraction

Previous attempts to determine the chemical form of
fission products deposited on graphite surveillance
specimens by x-ray reflection from flat surfaces failed
to detect any element except graphitic carbon. A
sampling method which concentrated surface impurities
was tried at the suggestion of Harris Dunn of the
Analytical Chemistry Division. This method involved
lightly brushing the surface of the graphite stringer with
a fine Swiss pattern file which had a curved surface. The
grooves in the file picked up a small amount of surface
material which was transferred into a glass bottle by
tapping the file on the lip of the bottle. In this way
samples were taken at the top, middle, and bottom of
the graphite stringer from the fuel channel surface,
from the surface in contact with graphite, and from the
curved surface adjacent to the control rod thimble.
Samples were also taken of the ““dirty” graphite and salt
on the top end surface of the stringer.

Although each sample contained only about a dozen
tiny particles and weighed less than 0.1 mg, the various
samples read 1 to 4 R/hr at a distance of 6 in. with a
soft-shelled “Cutie Pie” outside the glass bottle wall.

47

This radiation field made transfer of the particles into -

glass capillaries for powder x-ray diffraction examina-
tion a difficult operation. However, three (0.2 m)
capillaries were packed and mounted in holders which
fitted into the x-ray camera.

Short-exposure x-ray photographs were taken w1th
precautions to minimize blackening of the film by the
radioactivity. In spite of the precautions, samples from
the fuel channel surfaces yielded very dark films which
were difficult to read. Many weak lines were observed
in the x-ray patterns. Since other analyses had shown
Mo, Te, Ru, Tc, Ni, Fe, and Cr to be present in
significant concentrations on the graphite surface, these
elements and their carbides and tellurides were searched
for by careful comparison with the observed patterns.

In all three of the graphite surface samples so far
analyzed, most of the lines for Mo,C and Ru metal
were certainly present. For one sample, most of the
lines for Cr,C; were seen. (The expected chromium
carbide in equilibrium with excess graphite is Cr3C,,
but nearly half the diffraction lines for this compound
were missing, including the two strongest lines.) Five of
the six strongest lines for NiTe, were observed. Mo
metal, Te metal, Tc metal, Cr metal, CrTe, and MoTe,
were excluded by comparison of their known pattern
with the observed spectrum. These observations (except
for that of Cr,C;) are in gratifying accord with

expected chemical behavior and are significant in that
they represent the first experimental identification of
the chemical form of fission products known to be
deposited on the graphite surface.

Several hot cell techniques were tried for mounting
the x-ray diffraction samples in glass capillaries or on
quartz fibers, but they proved unsuccessful. To avoid
undue radiation exposure to analytical chemistry per-
sonnel, the remaining samples were not analyzed.

6.1.4 Examination with a Gamma Spectrometer

In order to provide an inexpensive guide for later
radiochemical work with milled stringer samples, a
simple technique was developed (with the help of E. 1.
Wyatt and H. A. Parker of the Analytical Chemistry
Division) for taking gamma scans of a number of
locations on the cross-section samples of the graphite
stringer. The sample, approximately 2 X 2 X 0.050 in.,
was placed over a Y, ¢-in.-diam hole through the center
of a large face of a 4 X 4 X 8 in. lead brick. The hole
was positioned over the center of the Nal crystal of a
gamma spectrometer. The sample was shielded above by
a Y-in.-thick sheet of plate glass with a scratched cross
on the bottom surface positioned above the hole in the
lead brick. By means of a 12-in. plastic ruler taped to
the specimen, any desired region of the graphite cross
section could be positioned over the hole in the lead
brick for gamma scanning. With this relatively rapid
technique, 18 to 20 positions were gamma scanned on
each of the cross-section samples from the top, middle,
and bottom of the graphite stringer.

The distribution of the four main activities observed
(*®%Ru, '238b, ' **Cs, and '3 7Cs) in the three samples
is shown graphically in Fig. 6.1. The numbers on each
map are counts per minute above background and are
proportional to the radioactivity at each position for a
given isotope. (For comparison between isotopes,
counts per minute are not proportional to disintegra-
tions per minute because of different counting effi-
ciencies.) In broad outline, the results were as expected.
Heavy surface depositions of '°®Ru and '?°Sb were
observed with little penetration to the interior. The Cs
isotopes were found at deeper locations since they
possessed gaseous precursors (5.27-day '??Xe and
4.2-min '3 7Xe).

The volume of graphite scanned by this techmque was
quite small (% ¢ in. diam X 0.050 in.); this permitted a
more fine-grained look at fission product distribution
than was possible by the milling techniques previously
used for sampling graphite surveillance specimens.
Correspondingly, the surface concentrations of all
48

MIDDLE
'OGRU

0.85
o]
0.3 0.08
0.43(7)

MIDDLE
1255,

0.20 0.30
0.08 0.85
0.38 0.97
0.80
0.4 0.87
2.75

MIDDLE
134cs

ORNL-DWG 71-13198

Q.55
0.25
[¢]
-
o}
BOTTOM
10ép,

©
o}
o}

.87

3.55
333
328

BOTTOM
1255y

0.60
0
0.20

4.70

BOTTOM
137cg

0.15

Fig. 6.1. Gamma scans of MSRE graphite stringer cross sections.

BOTTOM
1345

(o (¢

isotopes varied considerably between the two fuel
channel surfaces examined. Usually, but not always,
there was less of each isotope deposited near the
surfaces in contact with adjacent graphite stringers or

" with the control rod thimble. Individual (reproducible)

134Cs points were often considerably higher or lower
than adjacent points on a line between one fuel channel
surface and the other. The variations at the graphite
surface are believed to indicate spotty rather than

.uniform deposition of fission products on the specimen

surface. The interior variations are believed to indicate
variations in porosity of the interior graphite. These
observations and hypotheses are in accord with auto-
radiographs of graphite surveillance specimens pre-
viously examined which indicated spotty distribution of
total activity.

6.1.5 Milling of Surface Graphite Samples

As was done for the five previous sets of graphite
surveillance specimens removed from the MSRE, sur-
face samples for chemical and radiochemical analyses
were milled from the five 4-in.-long segments from the
top, middle, bottom, and two intermediate locations on
the graphite stringer. The milling techniques developed
for previous surveillance specimens could not be used
because of the peculiar geometry of the graphite
stringer. A new technique was developed using the large
milling machine in building 3525 hot cells and a
rotating milling cutter 0.619 in. in diameter. The
specimen was clamped flat on the bed of the machine,
and cuts were made from the flat fuel channel surface
and from one of the narrower flat edge surfaces on both

sides of the fuel channel. The latter surfaces contacted

the similar surfaces of an adjacent stringer in the MSRE
core. After the sample was clamped in position, the
milling machine was operated remotely to take succes-
sive cuts 1, 2, 3, 10, 20, 30, 245, 245, and 245 mils
deep to the center of the graphite stringer. The
powdered graphite was collected during and after each
cut in a tared filter bottle attached to a vacuum cleaner
hose. The filter bottle was a plastic cylindrical bottle
with a circular filter paper convering a number of
drilled holes in the bottom. This technique collected
most of the thinner samples, but only about one-half of
the larger 245-mil samples. After each sampling the
uncollected powder- was carefully removed with the
empty vacuum cleaner hose.

Before samples were cut from the narrow flats, the
corners of the stringer bars were milled off to a width
and depth of 66 mils to avoid contamination from the

adjacent stringer surfaces. Then successive cuts 1,2, 3,

49

10, 20, and 30 mils deep were taken. Finally, a single
cut 62 mils deep was taken on the opposite flat fuel
channel surface. Only the latter cut was taken on the
two stringer samples from positions halfway between
the bottom and middle and halfway between the
middle and top of the stringer.

"Four blank samples were milled by identical proce-
dures in the hot cells from an unirradiated spare stringer
of CGB graphite to indicate the amount of contamina-
tion received by the samples in the process of being
milled in the highly contaminated hot cell.

The sample bottles were reweighed and delivered for
radiochemical analysis.

6.1.6 Radiochemical and Chemical Analyses
of MSRE Graphite

The milled graphite samples were dissolved in a
boiling mixture of concentrated H,SO, and HNO;
with provision for trapping any volatilized activities.
The following analyses were run on the dissolved
sarriples:

1. Gamma spectrometer scans for '°®Ru, '%38b,
134cs, 137Cs, Y10Ag, '%%Ce, °°Zr, *°Nb, and
$9Co '

2. Separate radiochemical separations and ahalyses for
89gr, 2981, 127Te, and *H. A few samples were
analyzed for °° Tc.

3. Uranium analyses by both the fluorometric and the
delayed neutron counting methods.

4. Spectrographic analyses for Li, Be, Zr, Fe, Ni, Mo,
and Cr. (High-yield fission products were also looked
for but not found.)

Uranium and spectrographic analyses. In Table 6.1 are

‘given the uranium analyses (calculated as *?°U) by

both the fluorometric method and the delayed neutron
counting method. The type of surfaces sampled, the
number of the cut, and the depth of the cut for each
sample are also shown in the table. Samples between 19
and 29 were inadvertently tapered from one end of the
specimen block to the other so that larger-than-planned
ranges of cut depth were obtained. Samples from 3 to
18 were taken from the topmost graphite stringer
spécimen, those from 19 to 31 and sample 36 were
from the middle specimen, and those from 32 to 48
were taken from the bottom specimen. '

In view of the fact that the uranium concentrations
were at the extreme low end of the applicable range for
the fluorometric method, the agreement with the
delayed neutron counting method was quite satis-
50

Table 6.1 Chemical analyses? of milled samples

: '233U, ppm, .
S i Cut and Depth, Total U, ppm, delaved Li. pom Be: ppm 7t oom Metal,
ample typeb mils fluorometric y ? PP PP » PP ppm
neu_tron

1 1 Blank 0-6 <10 <2 - —

2 2 Blank 6-30 1 0.1 £0.05 <2 <1 — -

3 1FC 0-2 28 355+1.5 360 320 1600 —

4 2FC 1-3 21 269+1.2 340 170 - -

5 3FC 3-6 8 115+ 0.8 - - - -

7 5FC 16-36 2 26 +£0.2 - - - _
11 9 FC 556-801 3 2603 310 200 - 820 Fe
12 1E 0-2 <1(M 22.7x3.5 250 " 180 - High Fe
13 2E 2-3 <30 - B8.61:23 110 50 — ) -
14 3E 3-6 9 55+0.7
17 6 E 3666 3 5003
18 1 Deep 0-62 2 24 +£0.1 40 20 - -
19 . 1FC 0-5 21 21.5+ 0.6 220 150 970 Mo,1100 Ni
20 2FC 1-7 9 10.1 £ 0.8 190 100 - —
21 . 3FC 3-10 4 7104
23 SFC 16—40 <1 1.0+ 0.7
26 8 FC 311-556 <2 0.8x0.6 10 610
27 1E 0-3 3 3.8+0.2 150 90 1400 High Fe
28 2E 1-5 <8 §9:0.1 230 110
29 3E 3-8 3 5.7+04
31 1 Deep 0-62 4 33x0.1 80 50 70 150 Ni
32 1 Deep 0-62 14 13.0+0.2 120 90 80 180 Ni
33 1FC . 0-0.2 12 18.1+1.3 1400 290 High High Fe + Mo
34 2FC 0-3 36 292+13 410 240 500 2900 Ni
35 3 FC 3-6 18 20009
36 5E 16-36 1 2.510.1
39 6 FC 36-66 3 3.5+ 0.1
43 ‘1E 0-1 51 118+ 6 1000 400 8000 High Fe
44 - 2E 1-3 46 366 1.6 40 270 550 220 Ni
45 3E 3-6 8 7.8 0.7
48 6 E 3666 2 3.60.2 :
49 1 Blank 0-2 ' <10 <2 - . —
50 2 Blank 2-30 <1 0.1 +£0.04 <7 <03 <40 <70 Ni

Dashes in the body of the table represent analyses showing none present Blanks indicate the analyses were not done.
5The number is the number of cut toward the interior starting at the graphite surface. “FC™ stands for a fuel channel surface,
“E” for a narrow edge surface, and “Deep” for a first cut about 62 mils deep from a fuel channel surface.

€“High” indicates an unbelievably high concentration (several percent).

"

factory. The data suggest that the sizable variations
between different surfaces (e.g., the three deep cut
samples 18, 31, and 32) were real. Sizable variations
also exist in uranium concentrations in the deep interior
of different regions of the stringer. These values range
from 2.6 ppm at the top to 0.8 ppm at the middle to 3
ppm at the bottom.

The concentration profiles indicated by the data in
Table 6.1 were generally similar to those previously
observed both on the surface and in the interior.
Concentrations dropped a factor of 10 in the first 16
mils. A rough calculation of the total 233U in the
MSRE core graphite indicates about 2 g on the surface
and about 9 g in the interior of the graphite. These low

values indicate that uranium penetration into modera-
tor graphite should not be a serious problem in
large-scale molten-salt reactors.

The fact that fluorometric values for total uranium
and the delayed neutron counting values for 233U
agreed (with the 233U value usually larger than the
total U value) indicates that little uranium remained in
the graphite from the operation of the MSRE with
2351 fuel. Apparently, the 238U and 233U previously
in the graphite underwent rather complete isotopic
exchange with 233U after the fuel was changed.

The spectrographic results also given in Table 6.1
were disappointing in several respects. The sensitivities

- for all species were lower than hoped. Thus only the

TR
e (4

51.

strongest spectra (Be and Li) could be measured with -

any reliability, and results for Zr, Fe, Ni, and Mo were
highly susceptible to sample contamination. While the
U-to-Li-to-Be ratios were approximately the same as for
fuel salt (1 to 14.5 to 8.7) for about half the samples
analyzed, the Zr results were more erratic. It had been
hoped that high yield stable or long-lived noble-metal
fission products (Mo, Tc, Ru, and Te) might be
detected in these samples, but they were not. The
finding of 150 to 2900 ppm Ni in a few of the surface
samples is probably real. The main conclusions from the
spectrographic analyses are that adherent or permeated
fuel salt accounted for the U, Li, and Be in half the
samples and that a thin layer of Ni was probably
deposited on some of the graphite surface.
Radiochemical analyses. Since the graphite stringer
samples were taken more than a year after reactor
shutdown, it was possible to analyze only for the
relatively long-lived fission products. However, the
absence of interfering short-lived activities made the
analyses for long-lived nuclides more sensitive and
precise. The radiochemical analyses are given in Table
6.2, together with the type and location of surface

" sampled, the number of the milling cut, and the depth

of the cut for each sample.

The species 25Sc, 1%%Ru, '1%Ag, °°Nb, and !?"Te
showed concentration profiles similar to those observed
for noble-metal fission products (°°*Mo, 12°Te, 32 Te,
103Ru, '°6Ru, and ®>Nb) in previous graphite surveil-
lance specimens, that is, high surface concentrations
falling rapidly several orders of magnitude to low
interior concentrations. The °3Zr profiles were of
similar shape, but the interior concentrations were two
or three orders of magnitude smaller than for its
daughter ®SNb. It is thought that 3 Zr is either injected
into the graphite by a fission recoil mechanism or is
carried with fuel salt into cracks in the graphite; the
much larger concentrations of **Nb (and the other
noble metals) are thought to result from the deposition
or plating of solid metallic or carbide particles on the
graphite surface. The °°Sr (33-sec ?°Kr precursor)
profiles were much steeper than those previously
observed in ‘surveillance specimens for ®°Sr (3.2-min
89Kr precursor), as expected. An attempt to analyze
the stringer samples for ®° Sr also was unsuccessful. It is
difficult to analyze for one of these pure beta emitters
in the presence of large activities of the other.

The hot cell facility in which the graphite samples
were milled was heavily contaminated with ! **Eu and
LS4 By, Correspondingly, the blank samples (Nos. 1, 2,
49, and 50) did not contain significantly less of these
activities than the stringer-samples. Although the blank

samples were also contaminated with ®°Co, the concen-
trations in the outermost stringer samples were signifi-
cantly higher. It probably was formed by an n,p

reaction from stable ®°Ni. Its presence confirms the

spectrographic analyses indicating that nickel was pres-
ent on the stringer surfaces.

Surprisingly high concentrations of tritium were
found in the moderator graphite samples (Table 6.2).
The tritium concentration decreased rapidly from about
10'" dpm/g at the surface to about 10° dpm/g at a
depth of Y ¢ in. and then decreased slowly to about
half this value at the center of the stringer. If all of the
graphite in the MSRE contained this much tritium, then
about 15% of the tritium produced during the entire

" power operation had been trapped in the graphite.

‘About half the total trapped tritium was in the outer

Y, ¢-in. layer,

Similarly high concentrations of tritium were found
in specimens of Poco graphite (a graphite characterized
by large uniform pores) exposed to fissioning salt in the
core during the final 1786 hr of operation. Surface
concentrations as high as 4.5 X 10'® dpm/g were
found, but interior concentrations were below 10®
dpm/g, much lower than for the moderator graphite.
This suggests that the graphite surface is saturated
relatively quickly but that diffusion to the interior is
slow. . -

If it is assumed that the surface area of the graphite
(about 0.5 m?/g) is not changed by irradiation (there
was no dependence of tritium sorption on flux), there

‘was one tritium per 100 surface carbon atoms. Since

the MSRE cover gas probably contained about 100
times as much hydrogen (from pump oil decomposi-
tion) as tritium, a remarkably complete coverage by
chemisorbed hydrogen is indicated.

6.2 CESIUM ISOTOPE MIGRATION
IN MSRE GRAPHITE

E.L.Compere S.S. Kirslis

Fission product concentration profiles have been
obtained on the graphite bar from the center of the
MSRE core which was removed early in 1971. The bar
had been in the core since the beginning of operation; it
thus was possible to obtain profiles for 2.1-year '34Cs
(a neutron capture product of the stable '??Cs daugh-
ter of 5.27-day '33Xe) as well as the 30-year '*7Cs
daughter of 3.9-min '37Xe. These profiles, extending
to the center of the bar, are shown in Fig. 6.2.

Graphite surveillance specimens exposed for shorter
periods, and of. thinner dimensions, have revealed
Table 6.2. Radiochemical analyses of graphite stringer samples

Sample - Cut and Depth, : | - Disintegrations per minute per gram of graphite on 12-12-69

No. type,a mils lZSSb 106Ru 127Te 95Nb llOAg 99'1"'C l3"-lcs 137CS 9031. 144Ce - 9SZI 152Eu 154Eu 60C0 3H
1 Thin blank 0-6 <8E35 <4E6 <7E9 <9ES <3E5 <8E8 5.79E5 6.22E6 <2ES8 2.93E7 3.12E7 1.06E7 9.15E5
2 Thick blank -~ 6-30 - <8E9 8.53E5 <7E8 <9E4 <3E4 <7F4 . 1.56E6 <2E7 1.85E6 1.83E6 6.81E5 2.38E6
3 1FC.T 0-2 3.1E10 7.31E10 B8.95E10 <4E12 <1E9 ~330 <3E8 1.0E9 8.7E9 8.49E9 <1.8FE10 <4.8ER <SE8 3.63E8 5.92E10
4 2FCT 1-4 1.14E9 - 3.74E8 3.42E9 <2.7E11 2.84E8 ~13 1.06E7 2.39E8 1.1E10 4.26E8 <1.1E9 1.76E7 2.37E7 4.4E7 1.21E10
5 3FC,T 3-7 552E8 1.08E9 1.81E9 <1El11 - 2.75E7 <7E6 2.39E8 3.88FE9 1.72E8 <2E9 1.50E8 1.51E8 3.77E7 6.31E9
6 4 FC,T " 6-17 1.76E8 3.72E8 5.23E8 <3E10 <6E7 4.39E6 3.49E8 3.38E9 5.27E7 <3E8 <1E7 <1.3E7 1.51E7 1.75E9
7 5FCT 16-37 1.07E8 2.54E8 3.50E8 <I1E10 <3.6E6 8.81E6 3.96E8 1.24F9 3.82E7 <2E8 <7E6 <5E6 1.14E6 1.85E9
8 6 FC,T 3667 591E7 202E8 1.96E8 <9E9 <3.3E6 2.03E6 2.69E7 2.78E8 3.79E7 <2E8 <9E6 <4.8E6 1.70E6 1.03E9
9 7 FC,T 66-312 <7E6 290E7 1.39E7 2.73E10 <9E5 3.32E7 3.00E8 7.23E7 1.69E8 295E8 <3E6 <3E6 <1.7E6 1.36E9
10 8 FC,T 311-557 <8Eé6 <2E7 4.89E6 <54E9 <I1E6 247E8 146E8 1.12E7 - <4.1E7 <1.2E8 <3E6 <2E6 <7E5 6.32E8
11 9 FC,T 556—-802 <I1E7 <3.3E7 2.00E7 <T7E9 <3E6 3.04E8 1.64E8 2.12E6 <35E7 <2.0E8 <4E6 <1.6E6 <2E6 8.16E8
12 1E, T 0-2 3.02E9 1.82E10 1.07E10 1.59E12 3.62E8 1.40E8 7.46E8 5.86E9 4.84E9 1.09E10 <8E7 <4.2E6 2.25E8 1.78E10
18 1 Deep,T 0-62 145E9 6.08E9 6.52E9 1.23E11 <3E7 <l1E7 2.14E8 1.13E9 <2E8 -~ <6ES8 <1E7 <7.9E7 1.05E7 3.48E9
19 1 FCM 0-5 249E9 9.20E9 1.21E10 1.57E12 5.22E8 3.95E8 1.10E9 1.13E10 6.68E9 1.25E10 <6E7 <4E7 2.73E8 5.72E9
20 2FCM 1-7 _ 1.50E¢9 1.20E10 1.03E10
22 4FCM 6-20 _ 8.33E9
23 5 FCM 16—40 . 1.28E6 _ 3.19E9
24 6 FCM - 36-70 : v 7.95E8
26 8 FCM 311-556 <BE6 <3.9E9 5.79E6 <3El10 <3Eé 2.85FE8 <2E8 1.22E7 5.58E7 <2E8 - <3E6 <7.8E5 6.11E8
27 1 EM 0-3 9.10E9 7.13E9 291E9 1.22E12 2.54ES8 2.86E8 8.50E9 8.67E9 4.52E9 8.50E9 <4E7 <3E7 1.77E8 1.56E10
28 2EM 1-5 493E7 1.83E8 2.39E8 2.24E11 2.37E7 2.83E7 5.28E8 6.94E9 4.06E8 <I1E9 7.07E6 .7.34E6 >2.30E9
36 S5EM 16-36 1.07E7 7.34E7 4.79E7 9.85E10 <4E6 340E7 5.14E8 2.18E8 <6ES8 3.71E6 3.53E6 .2.99E9
" 31 1 Deep,M 0-62 3.91E9 1.86E10 9.08E9 4.90El11 <9E7 9.72E7 9.13E8 2.26E9° 1.10E8 <2E9 = <1E8 <6E7 1.24E8 5.50E9
32 1 Deep,B 0-62 145E9 4.62E9 5.84E9 2.16El11 <6.3E7 1.96E8 1.13E9 243E9 6.14E8 1.13E9 <3E7 <1.6E7 4.30E7 8.45E9
33 1FCB 0-0.2 296E11 4.19E11 B.22E11 2.13E13 <2E9 1.37E9 4.69E9 1.07E10 5.09E10 <1E11 <3E9 <3E9 6.82E9 1.59El1l
34 2 FC,B 0-3 5.76E9 1.15E9 1.75E10 3.04E12 1.57E9 1.76E8 6.43E8 B8.13E9 4.67E9 1.14E10 . <2.7E7 1.63E8. 4.31E10
35 3FC,B 3-6 1.27E9 3.29E9 3.22E9 3.78E11 2.76E8 . 1.17E7 2.66E8 451E8 <2E9 <1.8BE7 1.94E7 1.26E10
38 SFC,B 16—36 3.27E8 8.38E8 7.27E8 <8.3E10 <5.7E6 2.20E7 1.17E9 <1E8 <6E8 ‘ 6.26E6 1.96E7 3.46E9
40 7 FC,B 66-311 3.13E7 1.04E8 5.20E7 <I1.0E10 <2E6 1.03E8 1.33E8 <3E7 <2E8 <1.6E6 3.23E6 5.69E8
42 9 FC,B 556—-801 <I1E7 <39E7 6.19E6 <4E10 <2E6 3.13E8 2.93E8 1.69E7 6.91E7 <3.2E8 <1E6 - 4.21E6 7.23E8
43 1EB ~ 0-1 - 3.05E10 6.94E10 8.06E10 8.49E12 <7.8E8 8.57E8 2.83E9 1.11E10 2.42E10 4.58E10 <3.5E8 <2E8 1.27E9 9.18E10
44 2EB 1-3 345E8 8.30E8 1.06E9 1.50E12 5.07E8 3.07E7 2.71E8 9.76E8 <4E9 <1.6E7 1.34E8 1.46E10
49 Thin blank 0-2 1.33E7 5.09E7 14E11 <9ES <1.0E6 1.54E6 1.77E7 3.25E7- 2.10E9 " 1.94E7 2.3E7 8.56E6 4.65E6
50  Thick blank 2-30 2.39E6 B8.61E6 7.02E6 <3.1E9 <2ES5 <1.3E5 2.90E5 1.80E6 <I1E8 4.69E6 4.96E6 1.07E6 1.79E6
51 Top knob 2.24E10 1.94E10 <1E12 <2E8 <8E7 1.0E8 <7ES§ <1E10 <3E7 2.47E9
KD  Top chips’ 0-30 2.83E10 2.19E11 8.8E10 5.82E12 <5.7E8 3200 <24E8 1.0E9 1.17E10 <3E9 <3.2E10 1.6E8 1.06E8 1.27E10

“*The iumber is the number of cut toward the interior starting at the graphite surface. “FC” stands for a fuel channel surface, “E” for a narrow edge surface, and “Deep” for a first
cut about 62 mils deep from a fuel channel surface. T, M, and B represent samples from the top, middle, and bottom specimens of the graphite stringer, respectively.

o L4 . » L] PR 2 4 3 ) .' .

cs
a2 @

1018 ORNL-DWG 71-13199
E L o =
i _ACCUMULATED '*"xe =
o 1/ DISINTEGRATION (est) —
o 10" &= —
= =\ =
G =\ .
e S \ ]
o
5 1016 \\ 13705 =
E := ‘ I —
L. l | |\ ]
o [
- 1015 b1 \‘ -
& Ly =
- =\ I =
e . F .
= 14
5 10" 134, =
5 F -
5 BAR CENTER LINE -
213 —
102 L ]

0 100 200 300 400 3500 600 700 800
DEPTH (mils)

Fig. 6.2. Concentration of cesium isotopes in MSRE core
graphite at given distances from fuel channel surface.

similar profiles for '37Cs.! Some of these along with
profiles of other rare-gas daughters were used in an
analysis by Kedl® of the behavior of short-lived noble
gases in graphite. Xenon diffusion and the possible
formation of cesium carbide in molten-salt reactors
have been considered by Baes and Evans.?

An appreciable literature on the behavior of fission
product cesium in nuclear graphite has been developed
in studies for gas-cooled reactors by British investigators
(Dragon Project), Gulf General Atomic workers, and by
workers at ORNL. .

The profiles shown in Figure 6.2 indicate significant
diffusion of cesium atoms in the graphite after their
formation. The 5.27-day half-life of ' 33 Xe must have

1(@). S. S. Kirslis et al., “Concentration Profiles of Fission
Products in Graphite,” in the following MSR Program Semi-
annu. Progr. Reps: Aug. 31, 1970, ORNL-4622, pp. 69-70;
Feb. 28, 1970, ORNL-4548, pp. 104—105; Feb. 29, 1967,
ORNL-4119, pp. 125-28; Aug. 31, 1966, ORNL-4037, pp.

53

resulted in a fairly even concentration of this isotope
throughout the graphite and must have produced a flat
deposition profile for '33Cs. This isotope and its
neutron product ' **Cs could diffuse to the bar surface
and could be taken up by the salt. The '**Cs profile
shows that this occurred. The ! **Cs concentration that
would accumulate in graphite if no diffusion occurred
has been estimated from the power history of the
MSRE to be about 2 X 10'* atoms per gram of
graphite (higher if pump bowl xenon stripping is
inefficient), assuming the local neutron flux was a
minimum of four times the average core flux. The
observed '3?Cs concentration in the bar center where
diffusion effects would be least was about 5 X 10'*
atoms of '3%Cs per gram of graphite. The agreement is
not unreasonable.

Data for 30-year '®7Cs are also shown in Figure 6.2.
For comparison, the accumulated decay profile of the
parent 3.9-min '®7Xe is shown as a dotted line in the
upper left of the figure. This was estimated assuming
that perfect stripping occurred in the pump bowl with a
mass transfer coefficient from salt to central core
graphite of 0.3 ft/hr.* and a diffusion coefficient of Xe
in graphite (10% porosity) of 1 X 1075 ¢m?/sec.’

Near the surface the observed '*7Cs profile is lower
than the estimated deposition profile; toward the center
the observed profile tapers downward but is about the
estimated deposition profile. This pattern should de-
velop if diffusion of cesium occurred. The central
concentration is about one-third that near the surface.
Steady diffusion into a cylinder® from a constant
surface source to yield a similar ratio requires that
Dt/r* = ~0.14. For a cylinder of 2-cm radius and a salt
circulation time of 21,788 hr, a cesium diffusion
coefficient of about 7 X 10™° ¢cm?/sec is indicated.

Data developed for cesium-in-graphite relationships in
gas-cooled reactor systems’ at temperatures of 800 to

1100°C may be extrapolated to 650°C for comparison.

180—84. (b} D. R.Cuneo et al., “Fission Product Profiles in -

Three MSRE Graphite Surveillance Specimens,”” MSR Program
Semiannu. Progr. Reps: Aug. 31, 1968, ORNL-4344, pp. 142,
144-47; Feb. 29, 1968, ORNL-4254, pp. 116, 118-22.

2. R. J. Kedl, A Model for Computing the Migration of Very
Short Lived Noble Gases into MSRE Graphite, ORNL-TM-1810
(July 1967).

3. C. F. Baes, Jr., and R. B. Evans Ill, MSR Program
Semiannu. Progr. Rep. Aug. 31, 1966, ORNL-4037, pp.
158-65. : ‘

The diffusion coefficient thereby obtained is slightly
below 10~ ?; the diffusion coefficient for a gas (Xe) is
about 1 X 107° cm?/sec. Some form of surface
diffusion of cesium seems indicated. This is further

4. R. B. Briggs, Estimate of the Afterheat by Decay of Noble
Metals in MSBR and Comparison with Data from the MSRE,
MSR-68-138 (Nov. 4, 1968).

5. R. B. Evans III, J. L. Rutherford, and A. P. Malinauskas,
Gas Transport in MSRE Moderator Graphite, 1I. Effects of
Impregnation, IIl. Variation of Flow Properties, ORNL-4389
(May 1969). : »

6. J. Crank, p. 67 in The Mathematics of Diffusion, Oxford
University Press, London, 1956.

7. H.J. deNordwall, private communication.
substantiated by the sorption behavior reported by
Milstead® for the cesium-nuclear graphite system.

In particular Milstead has shown that at temperatures
of 800 to 1100°C and concentrations of 0.04 to 1.6 mg
of Cs per gram of graphite, cesium sorption on graphite
follows a Freundlich isotherm (1.6 mg of Cs on 1 m?* of
graphite surface corresponds to the saturation surface
compound CsCg). Below this, a Langmuir isotherm is

54

6.3 FISSION PRODUCT CONCENTRATIONS
ON MSRE SURFACES

E.L.Compere E.G. Bohlmann

The recovery of segments of surfaces from the fuel
circulating system of the Molten-Salt Reactor Experi-

‘ment in January 1971 has permitted the direct com-

indicated. In the MSRE graphite under consideration, -

the !'37Cs content was ~10'® atoms per gram of
graphite, and the 133 and 135 chains would provide
similar amounts, equivalent to a total content of 0.007
mg of Cs per gram of graphite. At 650°, in the absence
of interference from other adsorbed species, Langmuir
adsorption to this concentration should occur at a Cs
partial pressure of about 2 X 107'® atm. At this
pressure, cesium transport via the gas phase should be
negligible, and surface phenomena should control.

To some extent, Rb, Sr, and Ba atoms also are
indicated to be similarly adsorbed and likely to diffuse
in graphite.

It thus appears that for time periods of the order of a
year or more the alkali and alkaline earth daughters of
noble gases can be expected to exhibit appreciable
migration in the moderator graphite of molten-salt
reactors.

8. C. E. Milstead, “Sorption Characteristics of the Cesium-
Graphite System at Elevated Temperatures and Low Cesium
Pressure,” Carbon 7, 199-200 (1969).

parison of the intensity of deposition of several fission
product isotopes at a number of points around the
circuit. Activity determinations were available on seg-
ments of a central graphite bar and a control rod
thimble from the core center, on a segment of heat
exchanger tube and shell, and on specimens cut from
the pump bowl mist shield. Most of these determina-
tions have been described by members of the Reactor
Chemistry and the Metals and Ceramics Divisions in this
and the precedmg reports.

In Table 6.3 we present these data expressed as atoms
per square centimeter for comparison with each other,
and with a maximum mean deposition intensity given
by the MSRE inventory divided by total area of
graphite and metal in contact with flowing fuel salt.

Comparisons between graphite and metal for various
isotopes in various core regions do not indicate any
marked difference in affinity or sticking factor between
the two substances, except that possibly ?°Tc is more
strongly deposited on metal. Deposits on graphite
appear more intense at the bar center. For '?°Sb,
strong depositions occurred in almost all locations,
particularly on metal — it appears necessary for the
outer graphite regions to be lower than elsewhere to

Table 6.3, Fission product concentration on graphite and metal surfaces in MSRE (10'4 atoms/cm?)

Inventory/total flow area 16 9 52 127 3050
Graphite  Metal Graphite Metal Graphite Metal Graphite Metal Graphite Metal

Core center

Top 6,9 24 6,13 17,26 150

Middle 22 15 6 3 _ 40 - 15 100 90 " 1500

Bottom : 9 36 4 10 11 18 50 120 2500
Heat Exchanger _ 7

Tube . 15 7 23° 1200

Shell ' 29 29 2200
Pump bowl

Subsurface 31 5 14 22 1500

385,41

Interface ' 86,54

e @
‘» q

stay within inventory. Tellurium, an antimony daugh-
ter, is somewhat simjlarly though perhaps less intensely
deposited.

Ruthenium-106 was found more strongly deposited in
the core than in the heat exchanger and most strongly
at the gas-liquid interface in the pump bowl (as was
125gp),

Because over a dozen half-lives had elapsed before
counting, the ®5Nb data are doubtless less precise but
appear similar to the ruthenium pattern.

Generally these data show. somewhat higher inten-
sities of deposition than reported® for surveillance
specimens; turbulence and mass transfer rates may have
been less for surveillance specimens.

The mass transfer coefficients cited by Briggs'®
indicate that deposition should vary with flow condi-
tions from region to region, and also should depend on
sticking factors presumably characteristic of the surface
material and substance transported. Mass transfer coef-
ficients for xenon atoms in various regions of the MSRE
were given as: core center 15%, 0.3 ft/hr; core average,
0.06 ft/hr; heat exchanger, 0.7 ft/hr; piping, 1.2 ft/hr;
reactor vessel, 0.6 ft/hr; and 0.02-in. bubbles in salt, 2
ft/hr. The same proportions between regions should
hold for colloidal particles (<1 u) of similar size, shape,
and density. -

There is nothing in the data presented here to show
directly whether deposition occurred by an atomic or
particulate mechanism. The relative scatter of the data
does not permit us to relate them to the mass transfer
coefficients of the regions.

6.4 METAL TRANSFER IN MSRE SALT CIRCUITS

E.L.Compere E.G.Bohlmann

Cobalt-60 is formed in Hastelloy N by neutron
activation of the minor amount of *®Co (0.09%) put in
the alloy with nickel; the detection of 6°Co activity in

9. F. F. Blankenship et al., “Examination of the Fourth Set
of Surveillance Specimens from the MSRE,” MSR Program
Semiannu. Progr. Rept. Feb. 28, 1970, ORNL-4548, pp.
104—-111,

10. R. B. Briggs, Estimate of the Afterheat by Decay of
Noble Metals in MSBR and Comparison with Data from the
MSRE, MSR-68-138 (Nov. 4, 1968).

55

bulk metal serves as a measure of its irradiation history,
and the detection of °Co activity on surfaces should
serve as a measure of metal transport from irradiated
regions. Cobalt-60 deposits were found on segments of
coolant system radiator tube, heat exchanger tubing,
and on core graphite removed from the MSRE in
January 1971.

The activity - found on the radiator tubing (which
received a completely negligible neutron dosage) was
~160 dpm/cm?. This must have been transported by
coolant. salt flowing through heat exchanger tubing
activated by delayed neutrons in the fuel salt. The heat
exchanger tubing exhibited subsurface activity of about
3.7 X 10® dpm/cc metal, corresponding to a delayed
neutron flux in the heat exchanger of about 1 X 10t°,

- If metal were evenly removed from the heat exchanger

and evenly deposited on the radiator tubing throughout
the history of the MSRE, a metal transfer rate at full
power of about 0.0005 mil/year is indicated.

Cobalt-60 activity in excess of that induced in the
heat exchanger tubing was found on the fuel side of the
tubing (3.1 X 10° dpm/cm?) and on the samples of
core graphite taken from a fuel channel surface (5 X
10% to 3.5 X 107 dpm/cm?). The higher values on the
core graphite and their consistency with fluence imply
that additional activity was induced by core neutrons
acting on 8°Co after deposition on the graphite.

The reactor vessel (and annulus) walls are the major
metal regions subject to substantial neutron flux. If
these served as the major source of transported metal,
and if this metal deposited evenly on all surfaces, a
metal loss rate at full power of about 0.3 mil/year is
indicated. Because deposition occurred on both the
hotter graphite and cooler heat exchanger surfaces,
simple thermal transport is not indicated. Thermo-
dynamic arguments preclude oxidation by fuel.

One mechanism for the indicated metal transport
might have 10% of the 1.5 w/cc fission fragment energy
in the annular fuel within a 30-u range deposited in the
metal, and a small fraction of the metal sputtered into
the fuel. About 0.4% of the fission fragment energy
entering the metal resulting in such transfer would
correspond to the indicated reactor vessel loss rate of
0.3 mil/yr. If this is the correct mechanism, reactors
operating with higher fuel power densities adjacent to
metal should exhibit proportionately higher loss rates.
7. Hydrogen and Tritium Behavior in Molten Salt

7.1 THE SOLUBILITY OF HYDROGEN
IN MOLTEN 2LiF-BeF,

D. M. Richardson = W. Jennings, Jr.
A.P. Malinauskas

The efficiency with which a dissolved gas can be
stripped from a liquid depends in part on the solubility
of the gaseous species in the fluid. Therefore, the rate
of removal of tritium in the MSRE circuits, both
intentional (in the helium sparge) and unintentional
(through the heat exchangers), is governed to some
extent by the solubility of the gas in the pertinent
molten salts. For this reason, hydrogen solubility
studies in molten salts have been undertaken.

The experimental method and apparatus have been
described earlier.! In brief, the apparatus is of the
two-chamber type which was first employed by Grimes,
Smith, and Watson,?2 but modified to accommodate
effects which arise because of the ability of hydrogen to
migrate through metals. The molten salt under investi-
gation is saturated with hydrogen in one of the two
chambers, and, after saturation has been attained, some
of the fluid-is transported to the second chamber. In
this second chamber the hydrogen is stripped from the
salt, and in auxiliary apparatus the amount dissolved in
the known volume of liquid is determined.

1. J. E. Savolainen and A. P. Malinauskas, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 115—
17.

2. W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys.
Chem. 62, 862 (1958).

Several modifications were made to the apparatus
during this reporting period in an effort to improve the
reproducibility of the data. The extent to which these
alterations have aided the experimental method is indi-
cated by the results for helium which are given in Table
7.1. These data are still of a preliminary nature. They
were taken while a considerable inleakage of argon,
which is used as a cover gas for the circulating pump in
the strip section, was known to occur.

The hydrogen data, which are also presented in Table
7.1, again display an uncomfortable degree of scatter.
We now believe this scatter to be a manifestation of air
inleakage over prolonged periods of time, with resultant

. interactions between the oxygen and the hydrogen-

56

metal-salt system. The lattermost hydrogen result which
is tabulated was obtained after sparging the salt with
hydrogen over an extended period of time. This run was
also the only one of the three involving hydrogen in
which complete removal of hydrogen in the stripper
chamber was indicated. However, the difference in
purity between this run and that immediately preceding
is due almost entirely to helium of a still undetermined
origin. '

Shortly after the last hydrogen run listed in Table 7.1
had been completed, two hydrogen recovery tests were
performed. These tests involved admitting a known
amount of hydrogen to the xenon strip gas in the
stripper chamber (which contained 622 cm® Li, BeF,
at 600°C), circulating the mixture through the salt,
then attempting to recover the hydrogen admitted. The
fraction of original hydrogen recovered as a function of
time is shown graphically in Fig. 7.1. The purity of the
gas collected at the termination of recovery run 1 was

- g
e X

57

Table 7.1. Solubility of helium and hydrogen in Li; BeF,

Gas Tem(;‘:ecr';lture Salt ';rcai:lgt;erred ia:l go(}lgltasc;;(; Purity 103k 2
He. - 605 585 5.239 0.252 6.17
He 605 622 2.619 0.536 5.99
He 600 287 2.020 . 0.337 6.22
He 601 287 1.382 0.553 7.07
H, 604 579 3.454 0.732 11.51
H, 604 634 8.026 0.819 27.17
H, 600 622 2.506 0.525 5.46

2The Ostwald coefficient K . is defined as the ratio of the concentration of the
gas in the dissolved state to its concentration in the gas phase (assuming ideal gas

phase behavior).

ORNL-DWG 71-13200

1.00
//
a 7~
x ”
Lt 7
3 7
g / /s
o / ,/
o~ 0.95 r
=2
] -
= -~
g 4
i 7 ’
/
/
0.90 L

0 " 100 200 300
ELAPSED TIME (min) :

Fig. 7.1. Fraction of original hydrogen recovered from the
solubility apparatus strip section as a function of time since
admission of the hydrogen into this region. Upper curve,
recovery run 1; lower curve, recovery run 2. Smoothed results
obtained from both the annulus and the stripper chamber
collections were combined to yield the results presented here.

97.3% H,, whereas that of recovery run 2 was 95.1%
H,. In both cases it was assumed that this purity
obtained throughout the experiment, and was charac-
teristic of the gas collected through the stripper annulus
as well as the stripper section itself. On this basis the
amount of hydrogen collected through the annulus
region only amounted to about 5% of the total col-
lected. This is in marked contrast to the result reported
earlier! when no salt was present in the stripper
chamber; we now feel that the earlier value was grossly
in error. The important point is that we can recover at
least 90% of the hydrogen from the stripper chamber
within a 2-hr period. Further experimental work is
currently in progress.

7.2 PERMEATION OF METALS BY HYDROGEI\i

J. H. Shaffer
W. Jennings, Jr.

A.P. Malinauskas
W. R. Grimes

Accurate evaluation of tritium migration in'an MSBR
will depend, in part, on estimates of its permeation
rates through the metal walls of the fuel and coolant
systems. These estimated transport rates are currently
based on diffusion coefficients and solubility constants
derived from relatively extensive data that are available

‘in the open literature. However, the extrapolation of

these transport rates to low hydrogen pressures shows
significant departures from their classical dependence
on the square root of the hydrogen pressure. This
dependence on square root of H, pressure at higher
hydrogen pressures is due to the dissociation and
association of hydrogen as the rate-controlling step.
Departures. from this pressure dependence at lower.
hydrogen pressures are presumed to be due to surface
adsorption phenomena as the rate-controlling event.
Thus, estimates of gas transport rates at very low
tritium pressures in an MSBR (assumed to be dependent
on v/P) may be erroneous. This experimental program
has been directed toward an evaluation of hydrogen
permeation rates through metals at hydrogen pressures
less than 10 torrs where departures from the P1/2
dependence have been reported.

The amount of hydrogen g which will pass through an

- area A of metal per unit time is given by the relation

a=-DAg, 1

where D is the diffusion constant (area/unit time) and ¢
is the concentration of hydrogen (cm® @ STP per cm?
of metal) at any point x within the metal. For a round
tube having length / and a radius r = x, the hydrogen
flux can be expressed as

_ _ de '
J—q./21rrl-—DE. (2)

Upon separation of variables and integration between
limits of radii (a, ) and concentrations of hydrogen on
the metal surfaces (C,, Cp),

q = [2nD/In (b/a)](Cq — Cb) - (3)

The dissolved gas concentrations can be interchanged
with gas pressure by the Freundlich relation to yield the
equation

q = [2nIDK/In (b/a)] (P} - P}) , | “4)

where K is the gas solubility constant.

By material balance the amount of hydrogen ¢ pass-
ing through the metal tube per unit time may also be
expressed as

U T )

For ideal gas behavior

0Py Va; 0Py Vb;

ofr T RTa; ar § RTb;’ ©)

where i and j denote the incremental isothermal sec-
tions of each gas reservoir and T is the absolute
temperature. By combining Eqs. (4) and (6), changes in
the hydrogen pressure associated with the diffusion
process are related to the hydrogen pressure by the
equation

opP, Va;
4=~ Z gy, = 2rlDK/In 6/)) ;- P§) , (7)

for the decrease in pressure P, by diffusion into a
volume at pressure Pp,. Conversely,

_ oPy ’Vbj

9= ,-ZR—n;,- = [2nIDK/In (b/2)] (P - P}), (8)

S8

for the increase of pressure Pj as hydrogen diffuses
through the metal cylinder from pressure P,,.

The diffusion cell consisted of a Pyrex glass envelope
into which a short section (1.73 in. X 0.018 in. wall) of
"4-in.-OD Kovar metal (54% Fe, 28% Ni, and 18% Co)
had been fused. Glass tubes connected to the ends of
the envelope in a manner which permitted access by gas
pressure to either side of the metal surfaces. The cell
was mounted horizontally within a mullite tube in a
l-in.-diam X 12-in. tube furnace. Four thermocouples
in wells within the glass envelope were used to establish
the operating temperature of 495°C and to define the
thermal gradient of the furnace. Gas pressures were
determined from an electronic pressure meter (MKS
Baratron type 77) using two differential-pressure-sens-
ing heads having O- to 10- and O- to 1000-torr ranges.
Hydrogen used in the system was passed through a
palladium diffusion unit for purification. The vacuum
system was of conventional design with mercury diffu-
sion and oil pumps. '

In the first set of experiments, P, was maintained at
zero and the quantity ¢ evaluated from incremental
changes in P,. An average value for ¢ at an average value
for P, for each experiment was determined by the
method of least squares. In the second set of experi-
ments, P, was held constant and ¢ was evaluated from
incremental changes in Pp, where P, = 0 at ¢ = 0.
Extrapolations of the data from each experiment to ¢ =
0 yielded instantaneous gas transport rates at Py = 0 at
discrete values of P,. Thus, under both experimental
conditions the equations defining the pressure depend-
ence of the diffusion process could be reduced to

Ing=nlnP;+ 1n2nl/in b/a)DK . €)]

The results obtained by least-squares regression analyses
from some 30 experiments according to this equation
yielded a value of » =0.560 £ 0.011 and a value of DK =
5.18 X 107'" moles/torr"-cm-min (limits of standard
error =4.95to 5.42 X 107'1) at pressures for P, of 1.4

‘to 800 torrs. The direction of hydrogen flow through

the test cell had no significant effect on the experi-
mental results.

Although the results of these experiments demon-
strate some deviation from the P1/2 pressure depend-
ence of g, the experimental apparatus was not suffi-
ciently sensitive for evaluation at much lower hydrogen
pressures contemplated for an MSBR. Consequently,
further work on this program has been discontinued
pending results from a similar study using mass spectro-
metric techniques and much lower pressure ranges.

e
“r s

7.3 TRITIUM CONTROL IN AN MSBR

The tritium generated in an MSBR must be prevented
from escaping into the environment. Some removal
(and containment) of tritium as T,, HT, or TF from the
primary fuel system is expected. However, the ease of
diffusion of hydrogen isotopes through metal (and
especially through the large area of thin metal in the

" heat exchangers) ensures that a large fraction of the

tritium will proceed through the heat exchangers and
eventually to the outside world unless it can be held up
by chemical means or by efficient diffusion barriers.

7.3.1 Mass Spectrometric Examination
of Stability of NaBF;OH

C.F.Weaver J.D.Redman

If hydrogenous material (i.e.,, NaBF;OH) can be
incorporated in reasonable quantity in the NaF-NaBF,
secondary coolant, the tritium diffusing from the fuel
circuit of an MSBR could be held up by chemical
exchange

T + NaBF;OH = NaBF;0T + H

for a period sufficient to permit its controlled removal
from the coolant circuit. This possibility has received,
and is receiving, considerable attention.

We have investigated with a TOF mass spectrometer
the thermal behavior of several of the pertinent mate-
rials. A previous report3 has described the behavior of
NaBF,, NaOH, and the products of reaction of H,0
with NaBF,. We have studied the thermal stability of
NaBF;O0H and the composition of gas produced during
its decomposition over the temperature interval 25 to
800°C.

The starting material was shown by x-ray diffraction
to be nearly pure NaBF;OH. The material began to
decompose at 90°C, and H,O was the only gaseous
product. The reaction, as suggested previously,? is
probably: '

2NaBF;OH(s) = Na, B, F4O(s) + H,O(g) .

The H, O was shown to be free from impurities by mass
spectrometric analysis.

59

This reaction was followed to completion at 100°C
by constantly monitoring the effusing water vapor.
Nearly 100% agreement between the theoretical mass of
H, O and that accounted for by the mass spectrometer
indicated that, indeed, 1 mole of water was produced
by 2 moles of decomposing NaBF;OH. At 373°C, 1.3
X 1072 g of H, O effused in 30 min through an orifice
3.16 X 107* cm?® in area. By use of the Knudsen
equation, the pressure was calculated to be 2.44 + 0.04
X 107* atm. This pressure value was used to calculate
the standard free energy change for the above reaction:

0
AF373°K =—RT Inp

=7.9 kcal/mole .

Further heating of the residue to 800°C yielded only
BF; and NaF vapors. The BF; effused over the range
245 £ 5 to 800°; NaF vapor appeared at 725°C.

7.3.2 The Thermal Stability of Nitrate-Nitrite Mixtures
C.F.Weaver J.D.Redman

A study of the nitrate-nitrite mixtures was initiated as
part of an evaluation of their potential use as coolants
in molten-salt reactors. In such salt mixtures as HITEC,
NaNQO,-NaNO;-KNO; (40-7-53 wt %), the reaction

H, + 2KNO; = 2KNO, + H,0

(which has a large negative standard free energy
change) may afford the means to convert elemental

' tritium to tritiated water, and thus provide a means to
-control its distribution.

3. C. F. Weaver, J. S. Gill, and J. D. Redman, MSR Program -

Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 85—87,
93. -

4. L. Kolditz and Cheng-Shou Lung, “Condensation of Fluor-
oborates,” Z. Chem. 12,496 (1967).

Since nitrates and nitrites are known to decompose
thermally, it is useful to identify the vapor products of
the above reaction to determine what additional species
are formed. In order to establish the needed back-
ground for interpretation of such data, the separate
components were first investigated.

Sodium nitrate was dehydrated in a Pyrex flask by
passing dried helium gas through the molten salt for
several hours at temperatures just above the melting
point, 306.8°C. The product was colorless and did not
adhere to the container. Samples of the salt were
transferred to a nickel Knudsen effusion cell. The gases
from the heated cell were analyzed with a TOF mass
spectrometer. The only vapor products at 380 to 400°C
were NO and O,. These permanent gases do not allow
accurate pressure measurements because of background
interference, but approximate values were NO = 1.6 X
107% torr, O, = 2.1 X 107? torr. The presence of
oxygen leaves little doubt that tritium will be oxidized.
The behavior of KNO3; and NaNO, is being investigated
by experiments now in progress. In subsequent experi-
ments we expect to observe the vapors over HITEC salt
with and without H, .

7.4 DISSOCIATING-GAS HEAT TRANSFER SCHEME
AND TRITIUM CONTROL IN MOLTEN-SALT
POWER SYSTEMS

E. L. Compere

The use of a dissociating-gas heat transfer system using
nitrogen dioxide (N, 04 = NO, =NO+0,)(?—>N, +
0,), under development in the U.S.S.R., appears to
offer some interesting possibilities to the Molten-Salt
Breeder in terms of mitigation of tritium losses from
the power system. Under dissociating conditions -the
effective heat capacity and effective thermal conduc-
- tivity of the gas are increased several fold, resulting in
increased efficiency of heat transport and the possi-
bility of lower equipment costs.

In the Soviet fast reactor proposal,> pressurized nitro-
gen dioxide is the primary fluid, removing heat from
fuel elements and driving a gas turbine. Thermal and
radiation decompositions are by implication not signifi-
cant, and '*C production was not mentioned. Similar
usage in molten-salt systems, though replacing both a
secondary salt and a steam system with the dissociating-
gas system, would involve the possibility of leakage of
nitrogen oxides and oxygen into the primary salt circuit
and subsequent reaction with graphite. Although such
reactions would doubtless not be extremely violent —
probably the gases would behave about like pressurized
air — the use of nitrogen dioxide to transfer heat
directly from*molten-salt fuel will not be further con-
sidered here.

For a Molten-Salt Breeder Reactor, the dissociating
gas system could drive a gas turbine, replacing the steam
turbine system; the primary coolant ‘presumably would
be lithium beryllium fluoride or some other such salt.
Heat would be transferred to nitrogen dioxide at 100
atm or more, possibly reaching temperatures of 600 to
700°C. At temperatures above about 500°C, appreci-
able (and rapidly reversible) dissociation into NO and
0, occurs (absorbing about 14 kcal/mole). As the gases
pass through the turbine and cool, exothermic recombi-

§. A. K. Krasin et al.,, Atomnaya Energiya 30, 180-85
(February 1971); translated by F. Kertesz, The BRG-30, Exper-
imental Power Plant with a Gas-Cooled, Fast Reactor and a
Dissociating Heat Transfer Agent, ORNL-tr-2500.

60

nation occurs at a rapid rate until most is recombined.
The reaction of NO and O, is kinetically third order;
the rate increases slightly as temperature is decreased.
The nitrogen dioxide associates (rapidly and reversibly)
to dinitrogen tetroxide at temperatures of 100 to
200°C, with release of an additional 6 to 7 kcal/mole
NO,. The dinitrogen tetroxide can be condensed with a
fairly low heat of vaporization. (The heat of vaporiza-
tion of N,O4 is 9.1 kcal/mole at the atmospheric
boiling point of 21°C, and of course diminishes to zero
at the critical temperature of 158°C and 99 atm.) The
ease of vaporization at low temperatures and the rela-
tively high heat of dissociation at high temperature may
facilitate the recovery of thermal energy from the
molten-salt system. .

Though both NO, ‘and NO are thermodynamically
susceptible to decomposition into the elements, rates
appear acceptably slow. Any decomposition of NO,
appears first to produce NO, which may decompose
further, so NO is the critical substance. The thermal
decomposition of nitric oxide as reported by Yuan,
Slaughter, Koerner, and F. Daniels® and Fraser and
Daniels? proceeds by both a second-order homogeneous
mechanism for which the rate at 600°C is indicated to
be ~4 X 1077 X P, moles liter™! hr™' and a
heterogeneous, zero-order reaction for which the rate of
600°C 'is indicated to be less than about 3 X 107¢
moles/m™ hr™'. Though not trivial, such rates may
well be tolerable, depending on processing and makeup
requirements in particular systems. In any event the
heat of decomposition of NO, ~20 kcal/mole, is of the
same order as heat transferred, and there is no increase
in molecules on decomposition. Consequently any de-

composition into N, and O, would require release of .

these gases and makeup, but little hazard should ensue.
The materials of construction should be stainless

steels, or chromium containing alloys such as Inconel.

(or Hastelloy N) which can be heated in pressurized air
to the temperatures in question. Oxide films will build
up on the surface in the usual way and inhibit corrosion
processes.

Tritium entry into the power system would be inhibi-
ted by the films. It will become diluted by the entire
gas phase, reducing its activity. Furthermore, any trit-
ium entering the system should be oxidized and re-

6. E. L. Yuan et al., “Kinetics of the Decomposition of Nitric
Oxide in the Range 700—1800°,” J. Phys. Chem. 63, 952-56
(1959).

7. J. M. Fraser and F. Daniels, *“The Heterogeneous Decom-
position of Nitric Oxide with Oxide Catalysts,” J. Phys. Chem.
62, 215-20 (1958).

o T
™A

61

tained in the form of TNO; and T,O ‘vapors. These
should react relatively slightly with oxidized walls, so
little escape is indicated by permeation of corrosion
hydrogen through the metal. The gaseous T, O or TNO,
in the system would occupy the full volume (~10°
liters) of the heat transfer system, which thereby would
have appreciable capacitance before processing was
required. (The ~2400 Ci/day of total tritium produc-
tion is 0.04 moles/day, most of which should not even
penetrate into this system.)

Processing to remove such tritium as enters the
nitrogen dioxide system could probably be accom-
plished by various schemes, batchwise or on side
streams: adsorption, freezing, or distillation suggest
themselves. Addition and removal of H, O could readily

provide an isotope dilution effect if desired. Thus the

major modes of escape of tritium from the power
system which remain are by leakage, either through
turbine seals or into the condenser water. It would
appear necessary to prevent leakage of NO, and such
burden as it has of tritium into condenser water, and to
contain or recover any leakage past turbine seals. This
appears required in corresponding steam systems also,

but tritium is thought to enter the steam systems

considerably more readily.

In either system the diversion of tritium from the
power system implies a requirement that it be recovered
from the primary circuits. Recovery from the primary
system is not discussed here.

In summary, possible advantages of a nitrogen dioxide
power system thus include:

1. The ability to block a major tritium path to the
environment.

2. The recovery of such tritium as may enter the
system in a concentrated form,

3. The lower equipment costs because of improved
_thermal transport properties.

4. Improved thermodynamic efficiency.

5. Less restriction on properties of intermediate cool-
ant salt, since vapor generation can be accomplished
with possibly greater ease. '

Hazards and uncertainties include:

1. Routine handling of toxic or noxious gases under
pressure.

2. The question of the irreversible decomposition of

nitrogen oxides into the elements.

3. The relation between chemical reaction rates, heat

transfer, and flow rates when all rates are high.

4. The effectiveness of the oxide films inApreventing
tritium passage has not been evaluated.

5. Recovery methods for recovery of such tritium as
enters the system are undefined.

6. Cost comparisons with other systems (steam) have
not been made.

On balance the use of nitrogen dioxide as a dissotiat-
ing-gas- thermal fluid for a Molten-Salt Breeder Reactor
appears to merit further consideration.
8. Processing Chemistry

8.1 THE OXIDE CHEMISTRY OF Pa%*
IN MOLTEN LiF-BeF,-ThF,

C.E.Bamberger C.F. Baes, Jr.
R. G. Ross D. D. Sood'

A study was initiated previously? to determine the
precipitation behavior of Pa(IV) in LiF-BeF,-ThF,
(72-16-12 mole %) when oxide is introduced. This
information is needed not only in connection with.the
development of possible MSBR fuel reprocessing meth-
- ods based on oxide precipitation, but also because it is
obviously relevant to .the determination of the oxide
tolerance of MSBR fuels. :

The results reported previously should correspond to
the equilibrium

PaF4(d) + ThO,(ss)
<= Pa0,(ss) + ThF4(d), (1)

where d denotes a dissolved siaecies and ss a solid
solution phase. Such an exchange reaction is to be
expected on the basis of the similarity in size of the
ions Pa** and Th*" and the fact that regular solutions of
UQO,-ThO, are formed from fluoride melts upon oxide
additions.> In order to confirm the above presumed
reaction (1) and to estimate its equilibrium quotient

1. Guest scientist from Bhabha Atomic Research Centre,
Bombay, India.

2. C. E. Bamberger, R. G. Ross, and C. F. Baes, Jr., MSR

Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676,
pp. 120-22.

3. C. E. Bamberger and C. F. Baes, Jr., J. Nucl. Mater. 35,
177 (1970).

62

Pa — XP302XThF4

: (2)

XThO 2XP3F4

where X denotes mole fractions, it was necessary to
determine the composition of the solid phase. Accord-
ingly, we decided to isolate and examine directly the
few milligrams of oxide phase present.

The flanged top with the stirrer assembly was
removed from the cooled equilibration vessel and
replaced by another top provided with a 1-in.-OD nickel
filter to be used for removing the molten fluorides at
temperature. The system was reheated to 570°C under
flowing H, and kept at temperature for 40 hr before we
attempted to remove the molten fluorides. This permit-
ted redissolution of any solid phase that might have
precipitated on cooling. Only 10% of the fluoride phase
was subsequently removed, probably due to plugging of
the filter. The system was again cooled to room
temperature and the procedure repeated with another
filter. This time we successfully removed all but a small
fraction of the fluoride phase. After cooling, the solid
residue was loaded into a '%4-in. nickel tube with a
fritted bottom, and the remnant of the fluoride phase
was forced out of. the compartment at temperature
using H, pressure on the inside and vacuum on the
outside of the compartment. :

A portion of the oxide residue was repeatedly washed
with hot “Verbocit”® solution, followed by treatment

4. “Verbocit” solution: (developed by the Analytical Chem-
istry Division, ORNL) 50 g sodium versenate + 20 g sodium
citrate in 200 ml saturated boric acid solution. Final pH = 9. .

-
with 7N nitric acid solution at 70°C. It was then
washed with acetone and dried. Gamma counting
indicated the solids to contain 27.9 mole % as PaO, . No
significant amount of elements other than Pa, Th, and
O were found by spectrographic analysis and analysis
for fluoride. The results of these analyses were (in
ppm): Li = 10, Be < 1, Ni= 30, Cr = 30, Fe'=200, and
F = 1700. X-ray diffraction showed, surprisingly, that
two fcc phases were present, giving:

I. Major phase (60 to 90 mole %),

63

o =5.5926 £ 0.0003 A .

II. Minor phase (10 to 30 mole %),

an = 5.5429 £0.0024 A .

No Pa was dissolved during the treatment with HNO;.
Since the amount of fluorine found was quite low
relative to the Pa present, it was concluded that no
significant amount of a Pa oxyfluoride phase was
present. ' ‘

The purified oxide sample was examined with the
Scanning Electron Microscope (SEM). The presence of
two distinct phases was confirmed, the most abundant
phase consisting of large well-formed octahedral crys-
tals, 5 to 25 u on a side, and the other of small, <1 u,
irregularly shaped particles or aggregates. The composi-
tion of individual crystals of Phase 1, the composition
of the smaller particles or aggregates containing Phase
II, and the average composition of a large field were
measured by x-ray fluorescence with the SEM. The
results obtained, normalized to the average composi-
tion measured by 7y counting, were:

Phase I (large crystals), 17.5+ 1.8% Pa .
Phase II (small crystals),43.9 + 4% Pa .

Use of the above results to calculate the relative
abundance of each phase gives 61 mole % of Phase I and
39 mole % of Phase Il, which is in fair agreement with
earlier estimates based on the intensities of lines
obtained by x-ray diffraction.

Due to the marked difference in morphology of the
phases it seems reasonable to assume that Phase I, the
well-formed large crystals, was the phase present at
equilibrium and very probably is a PaO,-ThO, solid
solution. Furthermore, the composition found for
Phase I is consistent with that calculated for the
equilibrium Pa0,-ThO, solid solution by material
balance. (The latter estimate fell in the range Xy, =

0.13 to 0.33 depending on the value assumed for the
solubility product of ThO, and the amount of oxide
present initially.) Phase II probably precipitated as an
artifact of oxidation by impurities during the filtration
procedure and could consist of mixed aggregates of
Pa0,, , and ThO,, where x <0.5.

The composition of Phase I (Xp,5, = 0.175), which
is assumed to have been at equilibrium with a melt
containing Xp,, =2.7 X 107 at 567°, gives 073 =
94 4.

Assuming, as was the case for the UO,-ThO, solid
solutions, that the entropy change for the exchange
reaction (1) is zero, we obtain

' Xp.o. X
log @72 =log( 740, T"“) =1.66(10%T). (3)
ThOgXPaF4

This estimate, along with the estimated solubility
product of ThO,, leads to the predicted Pa*'-0%"
precipitation behavior shown by the smooth curves in
Fig. 8.1. The calculation was performed as follows:
First the mole fraction of PaO, in the equilibrium

oxide solid solution was calculated for each point from

Q;’f;] [(eq.} 3] and the observed mole fraction of PaF,:

Q'l;?l (XPaF4/XTh’F4)
[1+ 058 (Xeur s rae) |

This, in turn, was introduced into the following
material balance expression for the total oxide present:

XP302 =

T _
no2 2(”Th02 * nPaOg) t (Xp2-)ng s

which, in terms opra02 and QThO2 , becomes

where

ngi = total moles of oxide,

nr0, Mpa0 nF_=moles of ThO,, Pa0,, and of
LiF + BeF, + ThF,,
64

Y _

PaF4 = initial mole fraction of PaF,; before oxide
addition,
Xg2-=mole fraction of dissolved oxide.

If it is assumed that 2.5 millimoles of dissolved oxide
was initially present, the previously reported results
(the points so corrected in Fig. 8.1) agree well with the
calculated curves.

Work presently continues to obtain additional Q,';,‘;
values with other solid solutions more dilute in Pa and
to test further the assumption that the PaO,-ThO,
phase is a regular solution. The data obtained thus far
indicate that in the event of an accidental contamina-
tion with oxide of an MSBR or an MSCR, the precipi-
tated phase would consist mainly of a UO,-ThQ, solid
solution with little PaO, dissolved in it.

o ORNL-DWG 71-8629
- N
\ \ AN
\\ \ \ \l
s I\ NN \\
E' \\ \z
> N ™~ _730°C__|
s \\ N .
E \ \ 9 \“-—-,_9_?3"(:
= {
£ T~
- \ 70
. \ ~~357°C
| ]
+ ‘
er& \\
L 2 \
\\ Pa’*t 563°¢C
¥
\\
[0}
[0} 25 50 75 100

2

0°” ADBED {millimoles /kg)

Fig. 8.1. Concentration of PaF4 in LiF-BeF,-ThF4 (72-16-12
mole %) in equilibrium with Pa0,-ThQ, solid solutions. The
abscissa shows the total oxide added to the system (as ThO3).
The lines were calculated with log Q%‘,l = 1.66/T(°K) and
assuming that 2.5 mM Q2 were initially present.

8.2 THE OXIDE CHEMISTRY OF Pa** IN
URANIUM-CONTAINING MOLTEN
LiF-Bng -ThF4

R.G.Ross.” C.E.Bamberger C.F.Baes,Jr.

Previous measurements® on the precipitation of Pa(V)
by oxide in molten LiF-BeF,-ThF, (72-16-12'mole %)
led us to predict that protactinium could be efficiently
precipitated from an MSBR fuel without precipitating
any uranium (as UO,) by the controlled addition of

oxide under midly oxidizing conditions. This prediction
was based on measurements of the assumed equilibrium

'hPay05(c) + % ThF4(d) =PaFs(d) + % ThO,(c),
log O =log (X, /X345 ) (1)
- =594 -998 (103/T),

combined with the previously measured equilibrium®

UF4(d) + ThO,(ss) = UO, (ss) + ThF,(d) |
log @ = log (‘XUOQ ] XTh_F4/XT hOgXUF4) ()

= (2.376 £ 48) - 10°/T,

where c, ss, d, and X denote, respectively, crystalline

" and solid-solution phases, dissolved species, and mole

fractions which gave

"4 Pa, Os5(c) + % UF,4(d) = PaF5(d) + %, UO, (ss) »

- 5/4
log Q3 =log (XPanauoz/XU/m) (3)

=594 —7.01 (103/T),

where (5 relates the concentration of PaF; in the salt
phase (in mole fraction units) to the activity of
U0, (auog) when Pa, Qs is present. The value of Q3
estimated in this way was sufficiently small to suggest
that the concentration of PaFs could be lowered by
oxide precipitation to a value as low as 10 ppm in the
presence of 0.3 mole % UF, without precipitating UO,.

In order to confirm that Q5 is indeed small enough to
achieve such a selective precipitation of protactinium,
reaction (3) was measured directly by equilibrating
molten LiF-BeF,-ThF, (72-16-12 mole %) containing
initially 0.3 mole % UF,; and 0.06 mole % PaF; (2300
ppm) with precipitated Pa(V) oxide and UQ,-ThO,
solid solution formed in situ by the addition of ThO,.
Agitation was provided by sparging argon through the

~melt, which was contained in a copper liner, with a

conical bottom, in a nickel vessel. A small amount of
CuO was added to the system to establish a relatively
oxidizing condition and thus assure that all the protac-
tinium was in the pentavalent state. Filtered samples of
the melt were obtained using copper filter sticks of -

5. R. G. Ross, C. E. Bamberger and C. F. Baes, Ir., MSR

Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, p.

92.
‘6. C. E. Bamberger and C. F. Baes, Jt., J. Nucl. Mater. 35,
177 (1970). '
-y

65

various porosities, concordant results from which indi-
cated that no contamination of the filtered samples
with oxides had occurred. The samples were analyzed
for uranium (sufficient 233U was present for determi-
nation by delayed-neutron counting) and for protac-
tinium (by gamma spectroscopy). The results obtained
were used to calculate values of Q5 at various tempera-
tures which are shown in Fig. 8.2. In these calculations
of Q, the activity of UO, in the oxide phase was
estimated from the concentration of UF,; and the
previously determined values of Q, and yyp,. Also
shown in Fig. 8.2 are values of (; derived by
combining the previously measured equilibria (1) and
(2). A least-squares‘fit of all the weighted data gives

3
log Qs = (4.49 £ 0.87) — (5.69 * 0.86) % @

and is shown by the line in Fig. 8.2.

ORNL-DWG 71-13201
TEMPERATURE (°C)
750 700 650 600 550
2
[ I I T I

10~! \\ é
N

° :i‘\
6 \ T.
\ p
\ij
a‘i‘s" 9 u\
< N
u:ig ® Vq
3 L \
x @ N
]
o \

10~2 <\'
8 ' N
N\
° \
N
4 \\
095 1.05 1.15 .25
100O/ 7(°K)

Fig. 8.2. Values of @5 = Xpana%/32/X%/1?4 measured

directly (e) and derived from measurements in the absence of
uranium (¢) vs 103/T(°K). The line represents the least-squares
fit of all the data.

Combination of the present direct measurements for
Qs with the expression for @, gave the additional
values for Q; shown in Fig. 8.3. Also shown are the

- previous direct measurements of @;. The revised

expression for @y, indicated by the line in Fig. 8.3, is

3
log 0; = (449 + 0.87) — (8.66 £ 0.86) % )

The measured values of Q5 thus are consistent with
those predicted over a year ago® from measurements of
Q,; and Q,. It should be noted in Figs. 8.2 and 8.3 that
most of the values of Q3 and Q; obtained from
measurements in the absence of uranium show a
positive deviation from the least-squares fit. Presently,
we believe that this could be caused by the presence of

ORNL-DWG 71-13202
TEMPERATURE (°C)

750 700 650 600 550
| | [ | |
-4
10 \.L
8 [ \

LY
SE "\\
S AN
N

: | \
0.95 1.05 .45 (.25
1000/ 7 (o

Fig. 8.3. Valuesof 0, = XPan/X"Sr/t14F4 measured directly ()
and derived from measurements in the presence of uranium (e).
The line represents the combination of the expression for Q3
(Fig. 8.2) and for Q,.
appreciable amounts of Pa*" in the previous determi-
nations of (J;. [t was assumed in these experiments,
wherein nickel oxide was added, that all the protac-
tinium was oxidized to the pentavalent state. In view of
the uncertainty in the potential of the Pa**/Pa® couple,
this may not have been the case. When improved values
of the redox potential become available it will allow us
to correct these data for this effect, although it is
expected that the correction will not be a large one.

The values of Q; and Q3 have also been measured by
Mailen’ but using much lower initial concentrations of
protactinium (100 ppm). His results show more scatter,
probably reflecting the greater difficulty of achieving
equilibrium with the much smaller amount of Pa(V)
oxide phase present. The resulting plots of @, and Q;
- vs 1/T cross ours at ~670°C, but they are much steeper,
indicating much more positive AS values and much
more negative AH values for reactions (1) and (3). We
feel that the results reported here are more accurate not
only because they show less scatter, but also because
the resulting AS values are much more consistent with
what one would predict from estimates of the absolute
entropies for the reactants and products.

The value of Q3 obtained by us at 600°C (0.009)
agrees within a factor of 3 with a value of 0.027 derived
from the results of Tallent and Smith (Part IV, this
report), who measured the equilibrium

1/2Pa2 05 (C) + SHF(g) = Pan(d) + 25H2 O(g) . (6)

This agreement is notably good since, in order to derive
a value of @3 from these measurements, we have
combined their measured equilibrium quotient for
reaction (6) with the equilibrium quotient for the
reaction

66

% U0, (c) + SHF(g) = % UF,4(d) + 2.5H,0(g)  (7) |

estimated® from measurements in a different solvent
(0.67 LiF-0.33 BeF,).

The consistency of the present results provides

confirmation that Pa(V) can be precipitated by oxide
from uranium-containing melts while precipitating no
uranium as an oxide phase. This is illustrated in Fig.
8.4, where the concentrations of PaF; in an LiF-BeF,-
ThF4-UF, (72-16-12-0.3 mole %) solution, at various

7. 1. C. Mailen, MSR Program Semiannu. Progr. Rep. Feb. 28,
1971, ORNL-4676, p. 245.

8. C. F. Baes, Jr., in “Symposium on Reprocessing of Nuclear
Fuels,” Nucl. Metallurgy, Vol. 15, CONF 690801, 1969.

ORNL-DWG 71-13203

1000 AN A

500 \

Neoe N
200

i
|
I
I
|
i
|
|
|
|

N N
_ 550°C \
E
= {00 N \\
— AN AN
o N\ N y
2, N, N N
N\ N\ '
5 \\ N
N LN !
A |
N \ I
N
20 , ' \\\ =
|
\\ :
10 N
0.05 0. 4. 0.2 0.5 1.0

a
uo,

Fig. 8 4 Concentratlons of PaFs calculated by means of log
Xpar,atio,/ XifE,) = 449 — 5.69 (103/T) for the solvent

LiF- Bng-ThF4-UF4 (72-16-12-0.3 mole %) at equilibrium with
P302.5(C).

temperatures, calculated by means of Eq. (4), are
plotted as a function of the activity of oxide ion in the
melt. In an MSBR -fuel (Xyp, == 0.003) we estimate
that ayo, must exceed ~0.95 before a UQ, phase
(actually a UO,-ThQ, solid solution rich in UO,) will
precipitate.® Thus, for example, at 550°C withay g, =
0.75 the concentration of PaFs; in equilibrium with
Pa, O; should be ~15 ppm.

The consistency of the values of Q, and Q53 with each -

other also confirms the assumption inherent in their
definition, namely, that the Pa(V) oxide phase involved
is Pa;Os, a fact which has not yet been established
directly because of the difficulty of isolating a suffi-

ciently large and pure sample of the protactinium oxide

phase.

Pyl
a3

8.3 REDUCTIVE EXTRACTION DISTRIBUTIONS
OF BARIUM AND THORIUM BETWEEN
BISMUTH-LEAD EUTECTIC AND
BREEDER FUEL SOLVENT

D.M. Richardson J. H. Shaffer

The eutectic mixture of bismuth and lead (56.3-43.7
mole %) offers the advantages of a low melting point
(125°C) and zero volume change on freezing; it should
also be somewhat less aggressive than pure bismuth
toward container materials. These properties were
sufficiently attractive to merit further examination.
Distributions of barium and thorium between the
eutectic melt and LiF-BeF,-ThF, (72-16-12 mole %)
were determined at 650°C in a 4-in. IPS, SS 304L vessel
with a mild steel liner. Barium fluoride was added to
the fuel solvent salt to make 9.1 X 1072 mole % (with
133Ba). Reductions were performed by successive
additions of 0.25 or 0.5 g of clean lithium metal;
samples of salt and metal phases were taken after a
minimum of 5 hr of argon sparging at 1 liter/min.

The distribution data obtained are shown in Fig. 8.5
together with the derived lines (with theoretical slopes).
These lines arbitrarily pass through the log-average of
the first 11 data points. This weighting of the sample
results is based on the assumption that phase segrega-
tion in the frozen metal pellets caused the measured
barium and thorium concentrations to be low in the
later stages of reduction.” The equilibrium quotient for
barium was: Dg,/(Dy;)* =9.810 X 10?, 2.4 times less
than reported for this salt and pure bismuth.'® The
equilibrium quotient for thorium with eutectic metal
was: Dpp/(Dy;)* =6.635 X 10®, 3.7 times greater than
reported for this salt and pure bismuth.' "

The barium activity of the metal phase was essentially
constant following the last three additions of 0.5 g of
lithjum and would indicate that the solubility limit of
thorium in the eutectic had been reached. Unfortu-
nately the sample data scattered badly, and additional
special samples were not taken. Examination of the
data points suggests the thorium solubility in the
eutectic may be approximately 1500 ppm at 650°C.

9. D. M. Richardson and J. H. Shaffer, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL4676, p. 131.

10. J. C. Mailen, F. J. Smith, and C. T. Thompson, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p.
290. -

11. L. M. Ferris, .J. J. Lawrance, and J. F. Land, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1969, ORNL-4396, p.
284.

67

ORNL-DWG 71-13204
cpm/g of metal, 1338qg

50 100 200 500 1000 2000
100 I -
BO ID T ¥ T LI
_fi=9.810x102 i
&0 Li Y P A
. .:01/ A &
40 [ J L
3 e 3 —
. /" A A )/
5 V‘. r ‘/‘ -
PR 4K a 4
20 =]
D
™ =6.635x108
(o))
o RN
50 100 200 500 1000 2000
ppm Th

Fig. 8.5. Reductive extraction from LiF-BeF,-ThF,4
(72-16-12 mole %) into bismuth-lead eutectic (56.343.7 mole
%) at 650°C.

This is about one-half the solubility of thorium in pure
bismuth at 650°C.'?

It is apparent from these data that the Pb-Bi eutectic
is less valuable for this particular separation (though it
is quite acceptable) than is pure Bi. It is clear that if the
low freezing point, low volume change upon freezing,
and probable less corrosivity of this eutectic are of real
value additional separation studies should be under-
taken.

8.4 REMOVAL OF CERIUM FROM LITHIUM
CHLORIDE BY ZEOLITE

- D.M. Richardson  J. H. Shaffer

The observation that cerium (as Ce®) is partly
removed from molten LiCl solutions by finely pow-
dered Linde Zeolite 4A was reported earlier.'® The
same results have now been demonstrated repeatedly in
an experiment where the dried zeolite was exposed,
using filter tubes, in the form of commercially supplied
No. 8 mesh spheres. Fission product processing feasi-
bility is indicated, at least for cerium, by observed
loadings of as much as 0.13 g of cerium per gram of dry
Zeolite 4A. : :

The initial salt was approximately 1 kg of LiCl-CeCl;
(99.5-0.5 mole %) with several millicuries of '**Ce.

12. C. E. Schilling and L. M. Ferris, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1968, ORNL-4344, p. 297.

13. D. M. Moulton and J. H. Shaffer, MSR Program Semi-
annu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 109.
Purifications and subsequent exposures to zeolite were
conducted in the same 4-in.-IPS nickel vessel.

Numerous 20-mil aliquots of Linde 4A, sodium form,
No. 8 mesh zeolite were equilibrated to constant weight
(five weeks) over CaCl,*6H,0 solid and saturated
aqueous solution at 24.5°C. Each aliquot was dried in a
filter tube for 18 hr at 375°C with argon purge. The
filter tube was then transferred under continuous argon
purge to the extraction vessel and slowly lowered into
the molten chlorides at 650°C. Salt was then pushed
alternately into and out from the tube by pressure
fluctuations. Inflow was damped by autopressurization
of argon in the top of the tube and was terminated
upon contact of salt with an electric probe. Contact
times were approximately 20 sec per stroke; contacted
volumes were 10% of salt inventory per stroke. Al-
though 50 strokes were usually made per aliquot, no
significant increase of loading was noted from 20 to
100 strokes. |

The exposed zeolite was recovered as loose spheres,
unaltered in appearance except for a film of salt on the
surface. It was very radioactive-and could be monitored
with a survey meter. Since the specific activity involved
an indeterminate amount of salt, however, the amount
of cerium removed per aliquot was computed from
physical inventory, and the decreasing radioactivity of
filtered salt samples removed after each batch exposure.
The data obtained are shown in Table 8.1. Within the
precision of the data the distribution of cerium between
the molten chloride and the solid zeolite phases can be
expressed empirically by the equation

In (ppm Ce in salt) = 12.163 + 1.35 1n (g Ce/g zeolite)

for concentrations down to 4600 ppm in the salt phase.
Although an effective experimental procedure was

68

established, cumulative errors introduced in material
balance calculations prevented further evaluation of the
rare-earth removal process in this experiment.

Chemical analyses for sodium in the salt showed a
steady increase and indicated, as shown in Table 8.1,
that exchange with lithium was essentially complete.
This finding is contrary to the reported equilibrium
quotient Dy,/Dy; = 78.4 for this zeolite in LiCl at
650°C,'* and suggests that a processing plant for a
molten-salt thermal reactor would require that Zeolite
4A be in the lithium form, with 7Li. Analyses for
aluminum in the salt were several hundred ppm but
showed no rising trend. :

Zeolite 4A is initially floated by molten lithium
chloride, but filling of the void space between crystal-
lites with LiCl causes the spheres to sink to the bottom.
Complete filling of the void space between crystallites
would result in the consumption of approximately 0.16
g of lithium per gram of dry, lithium form, Zeolite 4A.
Molecular inclusion of lithium chloride within crystal-
lite cavities would increase this consumption by possi-

- bly 50%.' * This consideration involving conservation of

materials would need to be included in the final
evaluation of a zeolite process. _

Further experiments are intended to measure the
removal of other fission products, both divalent and
trivalent, from molten LiCl by zeolite. Wide variations
may be found since the possible mechanisms include
ion exchange, reaction with the zeolite structure, and
formation of molecular inclusion complexes.

14. W. A. Platek and J. A. Marinsky, J. Phys. Chem. 65, 2118
(1961). .
15. R. M. Barrer and W. M. Meier, J. Chem. Soc. 299 (1958).

Table 8.1. Removal of cerium from molten LiCl by Zeolite 4A

Number of exposures to

14 ppm Ce Grams of cerium removed Fraction of total
g of No. 8 mesh dry . K .
" Linde Zeolite 4A in salt per gram of zeolite sodium exchanged
1 14,200 0.133 0.98
2 11,500 0.125 0.65
3 9,000 0.115 0.73
4 7,800 0.054 0.92
5 6,200 0.071 0.96
6 4,600 0.067 0.72

9. Development and Evaluation of Analytical Methods

for Molten-Salt Reactors

A. S. Meyer

9.1 IN-LINE CHEMICAL ANALYSIS .
OF MOLTEN FLUORIDE SALT STREAMS

- J.M.Dale A.S.Meyer

The first successful in-line analysis of a constituent in
a flowing molten fluoride salt stream is now in
operation. Automated analyses for U(III) in LiF-BeF,-
Z1F,-UF, (65.4-29.1-5.0-0.5 mole %) in the NCL-21
thermal convection salt loop are being routinely per-
formed by our PDP-8I computer voltammeter system.
The loop was set up and is being used by the Metals and
Ceramics Division' to study the effect of the melt on
metal specimens in Hastelloy N. ‘ '

The NCL-21 loop was filled on July 20 and held
under isothermal conditions at about 650° until July
26, at which time the cold leg temperature was
established and Hastelloy N specimens were inserted
into both the hot and cold leg of the loop for corrosion
studies. Analyses of the U(III) concentration of the
melt were started on July 22 and were continued on a
routine basis until the specimens were removed from
the loop on August 23. The original measurements in
the melt indicated that about 0.02% of the total
uranium was present as U(III). The U(III) concentration
showed a gradual increase until a value of 0.05% was
reached at 475 hr of loop operation. At this point the
rate of formation of U(III) increased, and the concen-
tration reached a value of about 0.15% when the metal
specimens were removed from the loop.

1. J. W. Koger, Part IV, this report.

No major problems have been encountered with the
computer or the voltammeter in this system. Prior to
installation on the loop the voltammeter was modified
by T. R. Mueller of our instruments group to enable the
counter electrode of the system to be used at ground
potential. This was done to help suppress any electrical
noise that might be contributed to the voltammetric
wave by surrounding equipment. However, on occasion
the response from the electrodes has been noisy and
erratic, making it impossible to locate the E'% of the
U(IV) - U(III) reduction wave with any degree of
confidence. It appears that this is caused by some
material floating -on the surface of the melt as it comes
in contact with the electrodes. In order to eliminate this
problem a new electrode assembly was designed which

“allows the salt to be flushed from around the electrode

69

with helium. This assembly was installed in the loop
after the metal specimens were removed. The electrode
is encased in a '4-in. nickel tube, open at the bottom,
which protrudes below the surface of the melt. During
computer-controlled runs the salt enters the electrode
compartment from below the melt surface and remains
there for the time of measurement. The salt is then
flushed from the compartment, and the electrode
remains in a helium gas envelope when not in use. This
appears to keep floating material away from the
electrode and also should increase the lifetime of the
electrodes or electrode ared-defining insulators which
may be added at a later time. It was hoped that this
new assembly would also decrease the surface vibrations
around the electrode. It was observed during operation
of the analysis system with the original electrode that
70

the derivative voltammogram which is used to locate .

the E' is highly susceptible to small variations in the
height of the melt. If the salt loop is given a small push
during a run the derivative wave oscillates with the wave
created on the melt surface. The new assembly does not
completely eliminate this effect, but it does dampen the
surface wave in the vicinity of the electrode, and
equilibrium is restored within a few seconds.

After the system is started the analyses are performed
completely unattended through computer control. At
initiation of the analyses a signal from a logic circuit
that we added to the computer starts a timer which
causes the.pressure in the electrode assembly to be
released and salt enters the electrode compartment. The
 computer then operates the voltammeter to perform
five analyses for U(III), and the results plus some
diagnostic information are printed out on a Teletype
after each analysis. At the end of the run the average
U(III) concentration and standard deviation are calcu-
lated and printed out. When the run is completed the
timer shuts off causing the salt to be flushed from the
electrode assembly. At the present time a set of five
analyses is made every hour, although the number of
analyses and the intervening period may be adjusted in
the computer program for the particular needs of the
experiment. The computer program which we have
modified to fit our particular situation was originally
developed by M. T. Kelley and R. W. Stelzner.? As an
example of the precision of the analysis system, a
standard deviation of 2.1% was obtained for the results
of the hourly analyses made over a one-weekend period
when the U(III)} concentration appeared to be relatively
stable.

One useful addition to the analysis system is now
being considered which involves the measurement of
the salt temperature. During the first month of opera-
tion the temperature of the melt where the analysis is
made varied from about 610 to 680°C. Because the
temperature is needed in the computer program to
calculate the U(III) concentration, it is planned to
monitor an amplified thermocouple output with the
computer so that the true temperature at the time of
analysis is used in the calculations. It is also planned to
investigate the extension of the analysis system to the
monitoring of the corrosion products in the melt.

9.2 SLOTTED PROBE FOR IN-LINE
SPECTRAL MEASUREMENTS

J.P. Young

A device has been recently developed by Wilks
Scientific Corporation called a “Miran” infrared

~ probe.* This probe, somewhat similar to a pH electrode

in shape, makes use of multiple internal reflections of a
light beam of a selected infrared wavelength range
within the probe to direct a beam of light, which enters
the top, through the probe around the bottom and
again through the probe finally exiting in a different
direction at the top. A detector monitors the exiting
beam. Since internally reflected light is subject to
absorption by the medium which surrounds the point
of these reflections, the multiple-reflected beam is
attenuated if the probe is placed in an absorbing
solution, and absorption measurements are possible. In
considering the usefulness of such a device it seemed
apparent that the utility could be greatly enhanced if
one cut a slot (see Fig. 9.1) of appropriate width in
some portion of the internal beam path and in a

-

direction normal to the beam path. With a slot of -

perhaps 0.2 to 1.0 cm in width a probe could be made
with a controlled and relatively much longer path
length than that which could be realized with only
internal reflection. With a slot the optical design of the
probe could be simplified to require a minimum of two
or three internal reflections to direct a beam of light
into and out of a cylindrical probe; the base of the
probe would be prismatic, conical, or perhaps hemi-
spherical, with the slot in the apex of the prism or on
the side of any of the shapes. By incorporating a
suitable light source and detector with a suitable slotted
optical probe, absorbance measurements could be made
in any region of the electromagnetic spectrum. The

device could also be used for other optical measure-

ments such as turbidity. The design of this probe would
lend itself to a sealable insertion into a pool of liquid or

a flowing stream. It should be remembered that internal

reflection can occur only when the index of refraction
(n) of the medium through which the light is traveling
(probe) is larger than n of the surrounding medium (the
liquid sample). Over the common range of values of n
encountered, a difference of about 0.5 in the two values
of n seems adequate; however, through the use of a
specially designed probe bottom a smaller difference
may be satisfactory. :

By the use of a probe made from Plexiglas and using
solutions of a blue dye, copper phthalocyanine sulfo-
nate, placed in the slot, absorbance data were collected
and compared with similar data obtained from the

2. M. T. Kelley, R. W. Stelzner, and D. L. Manning, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL-4449, p.
157.

3. Wilks Scientific Corporation, South Norwalk, Connecticut;
Bulletin M1-1-10M, July 1971.
ORNL DWG 71-11827A

~=— SLOT

Yoin. TYP.

Fig. 9.1. Slotted optical probe for spectral analysis.

solution with a Cary spectrophotometer, model 14M.
The light source used with the probe was a 100 uwatt
He-Ne laser line at 633 nm, and the detector was a
“Densichron” phototube detector. The comparison is
shown in Table 9.1.

In-line applications for spectral analysis by means of a
slotted optical probe appear numerous. For molten-salt
reactor applications, it is possible that an. LaF; slotted
probe would be useful in monitoring protons by means
of the BF;OH™ absorption in molten NaBF,; the
existence of such an absorption was reported previ-

71

Table 9.1. Absorbance of Cu phthalocyanine solution

Absorbance/1 ecm

Sample
Cary spectrophotometer Slotted probe
H,0 0.0 0.0
1 ' 0.188 0.187
0.384

2 0.382

ously.® With a diamond slotted probe, several spectral
measurements could be made directly in MSBR circulat-
ing fuel. Suitable laser light sources are available for
either of these applications; the refractive index of
either material is quite adequate for proper internal
reflections within the probe.

9.3 IN-LINE DETERMINATION OF HYDROLYSIS
PRODUCTS IN NaBF, COVER GAS

R.F. Apple  A.S.Meyer

Earlier analytical studies®*® on the off-gas of the
NaBF, circulating test loop have shown that water can
be present in significant quantities as hydrolysis prod-
ucts of BF; in the cover gas of the proposed MSBR
coolant salt. An in-line method for the measurement of
this “water”” would be useful for the detection of leaks
in steam generators, the investigation of the role of the
cover gas in the containment of tritium, the develop-
ment of methods for the removal of the corrosive
hydrolysis products before they can reach sensitive
components of the flow control system, and studies of
coolant reprocessing systems.

Conventional methods for the determination of water
in inert gases are subject to interference from BF; and
possibly HF in the cover gas. At the suggestion of
Dunlap Scott” we are investigating “dewpoint tech-
niques” for the determination of these hydrolysis
products. We anticipate that this empirical technique
may require extensive calibrations because the conden-
sation temperature of a given concentration of “water”
may well be influenced by the partial pressures of both
BF, and HF.

4. J. B. Bates, H. W. Kohn, J. P. Young, M. M. Murray, and
G. E. Boyd, MSR Program Semiannu. Progr. Rep. Feb. 28,
1971, ORNL-4676, pp. 94-96.

5. R. F. Apple and A. S. Meyer, MSR Program Semiannu.

" Progr. Rep. Feb. 28, 1969, ORNL-4396, p. 207.

6. R.F. Apple, J. M. Dale, L. J. Brady, and A. S. Meyer, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622,
pp. 116—18. ' \

7. Reactor Division.
In feasibility studies we are using a simulated cover
gas prepared by bubbling mixtures of BF3 and helium
through a saturator that originally contains water. When
a steady state is reached the saturator contains liquid
hydrates of BF3, and the exit gas contains hydrolysis
products in relatively low concentrations. By direct
Karl Fischer titration (subject to significant interference
from BF;) we have estimated *“water” concentrations
of about 200 ppm at a saturator temperature of 45°C
and 3000 ppm at 75°C. In our experiments the
dewpoint detector is either contained in the same oven
that maintains the saturator temperature or is con-
nected to the oven through a short nickel tube that is
heated above the saturator temperature. Under these
. conditions the observed dewpoint should coincide with
the temperature of the saturator.

Our initial approach was to attempt to detect the
formation of a conductive film of BF; hydrates by
measuring the resistance between closely spaced plat-
inum electrodes inserted in the gas stream. The elec-

72

trodes are mounted through an insulator (temperature -

controlled by an air stream) which provides a surface
for the condensatlon of the hydrolysis products. At
present we have used Teflon, Teflon with its surface
modified by treatment with a sodium solution, and
high-fired alumina as insulating materials. Measurements
on alumina appear more consistent; however, this may
be the result of improved electrical instrumentation
used in the later tests. This technique gives definite
indication of the condensation of a film of hydrates.
When the probe (electrode assembly) is about 10°C
above the saturated cover gas temperature its resistance
is about S00 k€2 for 20-mil electrodes spaced about %, 4
in. apart. The resistance falls to about 300 £ when the
temperature is reduced to 10°C below that of the cover
gas. In its present state the electrical probe does not
appear to be practical because of its extremely slow
response. Several hours are required to approach these
equilibrium resistance values..

We are now testing an optical method for dewpoint
detection. The design of our optical cell is based on a
suggestion of S. S. Kirslis.® The cell utilizes a cylindrical
copper bleck with one face polished and gold-plated.
This polished face is sealed into an observation chamber
through which the cover gas sample flows. The external
end of the copper block is penetrated to near the
polished surface by holes for cooling air and a thermo-
couple. The inlet stream is directed against the polished

8. Reactor Chemistry Division.

surface, and films of condensate are observed through a
sapphire window. With this cell condensation is ob-
served within about 15 min after the block is cooled to
the dewpoint. Over the range of 45 to 75°C dewpoints
are observed at about 1°C below the temperature of the
cover gas. It is likely that this difference reflects the
difficulty in detecting the initial formation of a film of
condensate visually. With commercial instruments (not
necessarily applicable to our corrosive gases) dewpoints
can be measured to a few tenths of a degree. A
complete evaluation of the precision and accuracy of -
the method must await calibration using gas streams
whose absolute water content is determined by some
alternate method. The Karl Fischer titration applied
after the water is separated from BF; and HF by
azeotropic distillation with pyridine appears to be the
most promising method. For the quantities of water
involved it will be necessary to perform the titrations
with coulometrically generated reagents.® D. J. Fisher
and T. R. Mueller!® have designed an improved
coulometric titrator for this application. The instru-
ment has been fabricated and is now being tested.
Initial measurements on standards have shown about
1% reproducibility for the titration of milligram quanti-
ties of water. With minor modifications in the elec-
tronics and in the design of the titration cell we expect
to achieve similar precision in the titration of 100-ug
quantities. This apparatus together with its metal
azeotropic still will be sufficiently portable to permit
on-site analysis of water in engineering streams.

A preliminary estimate of the potential sensitivity of
the optical dewpoint method has been made by a
dilution technique. When the concentration of hydroly-
sis products in a stream saturated at 59°C (estimated
“water’”” concentration, 1000 ppm) was diluted fivefold
with a gas stream that contained the same concentra-
tion of BF;, the observed dewpoint decreased by 24°C.
This response is roughly equivalent to that of un-
complexed water at room temperature. From these
results we estimate the precision of the method to be
about 10% for visual detection with perhaps a fivefold
improvement with automatic photometric instruments.
Commercial photometric instruments would require the
replacement of their conventional glass optics.

We believe that the optical cell will prove useful for
various development studies and may provide informa-

9. A. S. Meyer and C. M. Boyd, “Determination of Water
with Coulometrically Generated Karl Fischer Reagent,” Amzl
Chem., 31,215 (1959).

10, Analytlcal lnstrumentatlon Group of the Analytlcal
Chemistry Division,
tion on the gas phase equilibria between H, O and BF;.
In application to pumped systems, however, it will

_probably be subject to interference from oil which

enters the system through the shaft seals. We will

73

therefore continue development on the resistance

method of dewpoint detection, which should be un-
responsive to organic films, and we will test other
insulating materials and electrode configurations.

9.4 DETERMINATION OF HYDROGEN‘
IN FLUOROBORATE SALTS

J.P. Young J. B. Bates
M. M. Murray  A. S. Meyer

From the results which were reported earlier* for the
interaction of QH~ with NaBF,, it was obvious that
protons (as BF3OH7) could be detected in fluoroborate
salts. The detection limits and the quantitative nature
of such a determination, however, were unknown. Work
this period has been directed toward these latter ends.

It was found that pellets pressed from mixtures of
NaBF;OH and NaBF, which were ground in a micro
ball mill yielded infrared spectra in which "the ab-
sorbance of the NaBF;OH peak at 3641 cm™! varied
linearly with the added concentration of this species. In
a 45-mg pressed pellet the resultant factor is 63 ppm H
per absorbance unit. The method is very sensitive; much
less than 1 ppm H could be detected. It is assumed that
the method will be equally sensitive, on a molar basis,
to deuterium. ‘

Although a good calibration curve was obtained, the
method of preparing the samples did not guarantee that
these “standards™ were of known hydrogen concentra-
tion. If some of the added BF;OH™ were lost in the
grinding and pressing, or, rather, if it were converted to
some species which did not exhibit the sharp peak at
3641 cm ™!, the experimentally observed factor would
be higher than the actual factor. Indeed, one such effect
was observed in high concentrations of NaBF;OH or if
the mixtures were ground for too short a time. Under
such conditions, rather than one sharp peak, two sharp
peaks are observed in the spectrum of the pellets at
3641 and at 3621 cm™'. The former peak is as seen for
BF;0H in solid solution in NaBF,4. The latter peak is
at the same position as the BF;OH™ peak in pure

‘NaBF;OH. On mild heating of the pellets, 70°C for 18

hr, the peak at 3621 cm ™' disappears with little if any
enhancement of the 3641 cm™! peak. Apparently when
these two peaks are observed, two kinds of BF;OH ™ are
present in the NaBF, matrix: a bound or dissolved
species and an undissolved and probably segregated

species. On heating the latter species it decomposes-to
form water and some form of oxyfluoroborate.

From the above it is seen that one particular kind of
nondissolved OH species can be identified. It is reason-
able to assume that all OH species could be observed by
their absorbance in this region of the infrared spectrum
so that their presence would be detected. Rather than
rely on this assumption, however, an independent
method involving isotopic dilution of H with D was
devised so that the spectrally observed BF;OH, or H,
concentrations could be compared with H concentra-
tions found by mass spectrometry. If one allows a

quantity of hydrogen in a particular species to react

with an approximately equal amount of deuterium, as a
volume of D, gas, in a sealed system at equilibrium, one
should find a large and proportionate amount of H in
the D, gas. Such an equilibration was studied by
allowing several molten samples of H-containing NaBF,4
or NaF-NaBF, to react with a known amount of D, gas
in sealed Pyrex ampules for various periods of time at
440°C. The proton concentration of the original salts
was determined by the infrared pellet technique using
the factor derived from the ground NaBF;OH-NaBF,
mixtures. The results of the comparison are shown in
Table 9.2.

Table 9.2. Determination of hydrogen in NaBF, samples
by infrared and mass spectral methods

H found, ppm
Sample g
Mass spectral Infrared
Eutectic 9.94 20
21.6% 24
21.76 18
NaBF4 7.7% 6.4
7.8

216-hr D, equilibration.
b48-hr D, equilibration.

The correlation -of the results is quite gratifying, but
there is an attendant high blank correction.in the mass
spectrometric results that is worrisome. We have found
that the blank correction, equivalent to 10 to 12 ug of
H, is due to Pyrex; the mechanism of the exchange is as
yet unknown. The blank is reduced by at least an order

- of magnitude if SiO, ampules are used. Further isotopic

exchange studies of fluoroborate melts contained in
Si0, ampules will be made.
Various parameters of the infrared method of H
determination in fluoroborates have been evaluated.
Too short a time of mixing with the ball mill, less than
approximately 4 min, or hard grinding yields absorption
peaks which give low results, Variation in the relative
humidity to which samples were exposed did not affect
the intensity, and therefore concentration, of the
BF;OH ™ absorbance. Even though samples were stored
at ' 100% relative humidity for 48 hr and showed
obvious signs of caking from moisture pickup, this
adsorbed water was removed in the pelletizing process
at least down to the equivalent of 4 ppm H. Since a
sample of less than 4 ppm H has not yet been observed,
it cannot be said as yet whether the 4 ppm is a real
concentration in NaBF,; or whether it represents the
extent of some hydrolysis reaction occurring during
pelletization. It would appear, however, that the former
view is more likely.

Most of the fluoroborate samples that have been
previously melted yield proton concentrations between
20 to 30 ppm. The sample shown in Table 9.2 which
has 7 ppm has never been melted. When this sample is
melted under what is thought to be inert conditions the
hydrogen content increases two- to threefold. This
apparent affinity for H, or more correctly OH since the
species being observed is BF3OH™, coupled with the
partial immiscibility of fluoroborates with BeF,-
containing melts suggests the possibility of extracting
OH™ from molten fluoroberyllates into fluoroborates
for the subsequent spectral determinations. Preliminary
studies using molten LiF-BeF, equilibrated with D, 0
have shown a very small but definite transfer of
deuterium into the fluoroborate-rich phase, as
BF;0D", which must have occurred by extraction of a
species from the molten fluoroberyllate phase. Further
studies of this will be made.

9.5 VOLTAMMETRIC AND ELECTROLYSIS
STUDIES OF HYDROXIDE ION IN MOLTEN NaBF,

D. L. Manning  A.S. Meyer

Experiments were conducted to observe the voltam-
metric behavior of hydroxide ion in molten NaBF, at
approximately 400°C. Hydroxide added as NaOH or
NaBF; OH (200 to 800 mg per 80 g of melt) produced
an increase in current over background on the voltam-
mograms over the potential range ~—0.5to —1.0 Vvsa
Pt quasi-reference electrode. Considerable noise was
encountered which is characteristic of bubble formation
at the electrode surface. This is not surprising since
hydrogen is evolved when OH™ is reduced at the
indicator electrode (OH™ + le = 0% + 4H;). After a

74

few hours, however, the current essentially returned to

~ background level. This is evidence that appreciable

concentrations of hydroxide ion are not stable in
NaBF, melts at least under our experimental conditions
in which the melt is contained in a graphite crucible
that is sealed in a quartz and Pyrex envelope. Voltam-
'metry at platinum or pyrolytic graphite electrodes does
not appear applicable to low levels of hydroxide ion,
since background voltammograms do not reveal any
wave that can be related specifically to OH reduction.
By purposely increasing the OH concentration, how-
ever, one can observe the potential range at which this
substance is reduced.

We are now testing a palladium tube electrode so that
the hydrogen formed at the electrode surface could
diffuse through the palladium into an evacuated volume
and be observed as a pressure change of the system.
This would appear to have two distinct advantages; it
would be specific for hydrogen and would allow for
increased sensitivity through controlled potential elec-
trolysis. Also the separated hydrogen could be trans-
ferred to an internal proportional counter for the
measurement of its tritium content. The technique
could thus be adapted to an in-line method for H/H
ratios in the reactor coolant.

To test the feasibility of this technique, a melt was
prepared of commercial nonrecrystallized NaBF, (esti-
mated proton level 20 to 40 ppm). A palladium tube
electrode 'y ¢ in. diam, 0.004 in. wall, 2 in. long was
closed at one end and fastened to a Y4-in. nickel tube
riser. The electrode was immersed to a depth of ~1 cm.
The evacuated volume associated with the electrode
assembly is of the order of 10 cc. Upon applyihg a
negative voltage step to the electrode, a pressure
increase was observed as the hydrogen which is formed
at the surface of the electrode diffused through the
palladium into the evacuated volume. Initially the
pressure increased from background (<2 u) to ~100 u
upon applying a 1-V step for 1 min. This is an
indication that we were definitely observing the reduc-
tion of protonic species present in the melt. After about
48 hr, however, the pressure measurements had de-
creased to ~30 u for the same potential step and time

~of electrolysis. This agrees with previous voltammetric

observations which seemed to indicate that appreciable
concentrations of OH are not stable in NaBF, melts
under our experimental conditions. An addition of
~500 mg of NaBF;0H to the molten NaBF, resulted
in an increase in the pressure measurements to about
160 u.

The results are encouraging. It is believed that the
electrolysis technique at palladium will result in a
-

specific method for detecting protonic species in
fluoroborate melts. We plan to continue these experi-
ments. However, to eliminate any effect of silica and
also to achieve higher temperatures, further experi-
ments will be conducted in a nickel cell enclosure.
Initial measurements indicate that useful voltammetric
waves for the reduction of OH can be obtained on the
palladium electrode.

9.6 ELECTROANALYTICAL STUDIES
IN THE NaBF, COOLANT SALT

D.L.Manning F.R. Clayton'!
G. Mamantov!?

Voltammetric studies are in progress in molten
NaBF, at ~420°C. Platinum and graphite cells are used
to contain the melt which is enclosed in aPyrex jacket
to maintain an inert atmosphere. A cover gas of helium
was maintained under static conditions at a pressure of
approximately 4 psig. Voltammograms were recorded at

“platinum and pyrolytic indicator electrodes (~0.1 cm?

area) vs a platinum quasi-reference electrode. The
platinum quasi-reference is supposedly poised at the
equilibrium potential of the melt. Background voltam-
mograms recorded on Harshaw NaBF, as received
showed an iron impurity wave at ~—0.4 V vs the
platinum quasi-reference. The concentration -of iron
calculated -from the voltammogram agreed well with the
chemical analysis of ~200 ppm. Reduction waves were
not observed that could be attributed to the reduction
of nickel or chromium. Chemical analyses gave 16 and 6
ppm of Ni and Cr, respectively, which are below
practical limits for well-defined voltammograms. We
have not yet investigated stripping techniques for these
jons. A small amount of iron(IIl) was added to the melt
to gain some idea of the potential of the Fe*" — Fe?*

reduction. This wave was observed at a half-wave

potential of ~—0.2 V. It appears that the iron impurity
in the NaBF, is predominantly Fe(II). On platinum, the
melt exhibits a cathodic current limit at ~—1.2 V which

appears to be due to the reduction of boron, while the

anodic limit appears at ~+1.8 V due to the dissolution
of the platinum. : '

The magnitude of the iron impurity was significantly
decreased by recrystallizing the NaBF, from 0.1 M HF
as suggested by L. O. Gilpatrick (Reactor Chemistry
Division). The iron was decreased from ~200 ppm to

11. Student participant, University of Tennessee, Knoxville.
12. Consultant, Department of Chemistry, University of
Tennessee, Knoxville.

75

20 ppm with one recrystallization, and a reasonably
well-defined reduction wave at this level was observed.

A voltammetric study is in progress on titanium(IV)
added as K, TiF to molten NaBF,. Melts of Ti(IV) in
NaBF, appear to be stable in graphite. The Ti(IV) >
Ti(III) reduction wave is observed at ~—0.5 V. It may
prove difficult to separate the reduction waves of
Ti(IV) and Fe(Il). A Ti*" - Ti° reduction wave is not
observed in NaBF,; because the cathodic limit of the
melt occurs first. | N

9.7 ELECTROANALYTICAL STUDIES OF Ni(II)

IN MOLTEN FLUORIDE FUEL SOLVENT

D. L. Manning

Further voltammetric and chronopotentiometric
studies were made at fast scan rates and short transition
times on Ni(II) in molten LiF-BeF,-ZrF, at S00°C. The
diffusion coefficient for Ni(II) was evaluated by the
two methods. For linear sweep voltammetry, the
equation for the reversible deposition of an insoluble
substance is '

i, = 2.28 X 10° n3/24D1/2Cv1/2

where l, = peak current (#A), n = electron change, C =
concentration (mM), A = electrode area (cm?), D =
diffusion coefficient (cm®/sec), and v = scan. rate
(V/sec). The voltammograms were recorded at a pyro-
lytic graphite electrode (PGE) with an area of 0.1 cm?
and at a concentration of nickel of 6.7 mM. Plots of {

vs v1/2 were linear to 50 V/sec. From the slope of the
line, the value of D is 1.07 X 107 cm?/sec. Although
an unsheathed pyrolytic graphite electrode was used (rg
= 0.05 cm), the equation for linear diffusion can be
utilized as pointed out by Delahay,'® since the quan-
tity (1/ro XD/nv)1/2 is less than about 0.2 at the scan
rates employed.

Chronopotentiograms were recorded at the PGE at
current densities ranging from 5 to 30 mA/cm?.
Corresponding transition times varied from 0.76 to
0.028 sec. Within experimental error, the iy71/? prod-
uct was constant at 469 * 0.25 X 107™* A cm™2
sec"1/2, From the Sand equation

1/2

13. Paul Delahay, ‘“New Instrumental Methods in Electro-
chemistry,” pp. 11546, Interscience, New York, 1954.
where iy = current density, 7 = transition time, F =
Faraday, and at a Ni(Il) concentration of 2.6 X 1073
moles/cm?, an average D value of 1.05 X 107® c¢m?/sec

76

was obtained. This is in excellent agreement with the

value obtained from voltammetry. Within the precision
of the measurements, the same results were obtained
using a platinum indicator electrode.

Nickel(II) when reduced at platinum did not appear
to form an Ni-Pt surface alloy under the experimental
conditions employed. Log (i, - i) vs £ plots were linear
over the range ~0.5 to 0.9 ip with the theoretical slope

for a two-electron change.'4 Also' from the chrono-
potentiograms, the ratio of forward to reverse transition
times was approximately unity, in agreement with the
predicted value for the reversible deposition of an
insoluble substance.!?

14. Gleb Mamantov, D. L. Manning, and J. M. Dale, J.
Electroanal. Chem. 9, 253 (1965). :
15. W. H. Reinmuth, Anal. Chem. 32, 1514 (1960).

. ?

P, |
10. Other Fluofide Researches

10.1 ABSORPTION SPECTROSCOPY OF
MOLTEN FLUORIDES

L.M.Toth L.O.Gilpatrick
10.1.1 The Disproportionation Equilibrium
of UF; Solutions

Investigation of the disproportionation equilibrium
reported earlier! for dilute U(III)-U(IV) mixtures in
molten fluoride solution

4UF; + 2C(graphite) = 3UF, + UC, (1)

has continued. Using LiF-BeF, mixtures as reference

“solutions the following UF,/U, ,,, concentration ratios

were measured where U, .., = 0.07 mole % (Table
10.1). These data have been used to calculate equilib-
rium quotients for the above equilibrium in both 66-34
and 48-52 mole % LiF-BeF, mixtures. It has been
found that at 650°C the equilibrium quotient at the
66-34 mole % composition (designated Q¢ 54) is 6.25
times that of the 48-52 mole % composition, or Q¢4 34
= 625 (Q,5.5,- Activity coefficient data reported
earlier® "yields Q¢ 34 = 7 Q45.5, in good agreement
with the data reported here.

However, when ternary solvent systems are compared
with the reference system of LiF-BeF,, less satisfactory
agreement has been found. For example, Q¢ 34 =~ 0.5

1. L. M. Toth, MSR Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, p. 118.

2. C. F. Baes, Ir., MSR Semiannu. Progr. Rep. Feb. 28, 1970,
ORNL-4548, p. 149.

77

Oymspr: Where Qpopp is the equilibrium quotient
measured for thé disproportionation equilibrium in
LiF-BeF,-ThF, (72-16-12 mole %). Currently accepted
activity coefficients yield Q. ¢ 34 =2 OmsBr-

The reverse of reaction (1) has been studied in an
effort to unambiguously identify the carbide phase in
the equilibrium. As indicated in the previous report’
exact reproducibility of the UF; equilibrium concen-
tration was not achieved because pure carbide reactants
were not available. Efforts have been directed at
obtaining UC, U,Cj, and UC, of greater purity with
which to further test the back reaction. Some of these
results indicate that the carbide phase may be U,Cj,
but the question remains unsettled.

The work on the disproportionation equilibrium will
continue in order that these questions may be settled.

10.1.2 Estimation of the Available F~ Concentrations
in Melts by Measurement of UF,
Coordination Equilibria

Due to insufficient experimental data, activity coeffi-
cients in ternary fluoride systems are less accurately
known than in binary systems such as LiF-BeF,. It has
been noted earlier that the equilibrium involving the

Table 10.1. Equilibrium UF4/U total ratios for LiF-BeF,
mixtures in graphite ~

Solution (mole %) Temperature (°C)

(LiF-BeF3) 650 550
66-34 0.025 0.004

48-52 0.13 0.03
two identified species of U(IV) in molten fluorides at
550°C,

UFg* = UF,> +F", (2)
provides a practical means of measuring fluoride ion
concentrations by measurement of the relative amounts
of UFg*™ and UF,® with absorption spectroscopy. In
this procedure the concentration of seven and eight
coordinated U(IV) are measured directly by using
€1050my = 14 and 20.8 liters mole™ cm™ for seven
and eight coordinated U(IV) at 550°C respectively. For
LiF-BeF, (66-34 mole %) the (UFg*")/(UF,3") ratio is
0.667 and for LiF-BeF, (48-52 mole %) it is 0.177.2
With these two reference points, other solvents, for
example, LiF-BeF,-ThF, (72-16-12 mole %) can be
scaled. It is interesting to note that UFg* /UF,% >
0.667 for LiF-BeF,-ThF, (72-16-12 mole %) indicating
that this proposed MSBR solvent is more F~ rich (or
basic) than the reference composition of LiF-BeF,
(66-34 mole %). These results are in good agreement
with the equilibrium quotient Q measured for reaction

(1) in the MSBR solvent, where Q is expected to

decrease as the F~ concentration increases.

10.2 SOLUBILITY OF BF; IN FLUORIDE MELTS

| S.Cantor R.M. Waller

The general objective of this investigation is to relate .

the thermodynamic behavior of a solute gas, BF54, to
the thermodynamic properties of molten fluoride sol-
vents. The guiding hypothesis for this investigation is
that BF3, with its unsaturated valence (i.e., its Lewis
acidity), will readily interact with the more loosely
bound fluoride ions (often referred to as “free”
fluoride) in the melt. In this part of the investigation,
we are measuring BF; solubility in the system
- LiF-BeF, ; depending on melt composition, this system
would be expected to exhibit a large variation in ““free”
fluoride concentration because LiF is a fluoride donor
and BeF,; is a strong fluoride acceptor.

This work is also relevant to molten-salt reactors.
First, ‘boron, with its high absorption for thermal
neutrons, can be introduced into the fuel salt as BF,
for purposes of reactor control.* (The BF; could be
readily removed from the fuel salt by inert-gas sparg-

3. L. M. Toth, MSR Semiannu. Progr. Rep. Aug. 31, 1970,

ORNL-4622, p. 105.

4. J. H. Shaffer, W. R. Grimes, and G. M.- Watson, Nucl. Sci.

Eng. 12,337 (1962).

78

ing.) Second, if there were a leak in the primary heat
exchanger, NaBF, in the coolant would decompose to
NaF and BF;. The degree of decomposition depends on
concentration (of NaBF, in the fuel salt), on tempera-
ture, and on the vapor space available for the escape of
BF;. BF; solubility measurements in a few melts
composed of NaF-LiF-BeF,-ThF,;-UF, could provide
the information required to determine how these
factors affect the decomposition of the coolant.

The apparatus and procedures of measurement have
been previously described.® Since the last report,®
measurements in four compositions have been carried
out: measurements have been completed in mixtures
with 63 and with 80 mole % LiF, nearly completed in
mixtures with 85 mole % LiF, and begun in mixtures
with 48 mole % LiF. A summary of the solubility data
in six melt compositions is given in Table 10.2.

In the last report® we noted that, in the composition
range 63 to 80 mole % LiF, the Henry’s law constant
for BF3 solubility varied linearly with the thermo-
dynamic activity of LiF. As Fig. 10.1 shows, the results
at 85 mole % LiF are in reasonable accord with this
linear relationship at 700°C; at 600° the deviation of
this point from the line may be an artifact of the long
extrapolation (180°C) of the data. The lines in Fig.
10.1 do not go through the origin, probably because the
solubility of BF; is also dependent on the activity of
BeF,; this dependence is much weaker than for LiF,
but is certain to be of greater importance in composi-
tions where the activity ratio of BeF, to LiF is
considerably greater than one,

For the six melts given in Table 10.2, the temperature
coefficient of solubility is negative and is not very
dependent on composition. The solubility per unit
pressure decreases by about one-half for every 60°C
(~100°F) rise in temperature. The enthalpy of solution
is calculated from the temperature coefficient of
Henry’s law:

RdanH
S dayn)

where R is the gas constant, 1.987 cal mole™ °K™!,
and T is temperature in degrees Kelvin. Values of AH
are listed in Table 10.2; with the possible exception of
one solvent (85 mole % LiF), the enthalpies of solution
are nearly identical. '

5. S. Cantor and W. T. Ward, MSR Program Semiannu. Progr.
Rep. Aug. 31, 1970, ORNL-4622, pp. 78-79.

6."S. Cantor and W. T. Ward, MSR Program Semiannu. Progr.
Rep. Feb. 28, 1971, ORNL-4676, pp. 98—-100.
¥ h

79

Table 10.2. Solubility? of BF3 in molten LiF-BeF, solvents; enthalpy and entropy'of solution

Solvent Temperature

Henry’s law constant — temperature equation;b

(o]

composition range measured K Hc in mole fraction BF 3/atm; K fiH &) A‘S; ¢ (alt 1?9;(;(]_(1)
(mole % LiF) °Cy Temperature in °K caljmose (cal mole )

63 469-654 InK;=-15458 + 8120/T -16.1 —-13.9

66 520-725 =—15.076 + 7903/T —-15.7 —-13.1

70 543-733 = —-14.940 + 8145/T -16.2 -12.8

75 641859 = —-14.518 + 8071/T -16.0 -11.9

80 713-866 =—14.018 + 7793/T —15.5 . —10.9

859 780-940 = —-13.283 + 7179/T . -14.3 -9.3

9Pressure range, 1.3 to 3.0 atm.

bl east-squares fit of the data.

“The experimental error in K 17 1S approximately +5%.
dMeasurements in this solvent composition not completed.

ORNL-DWG 71—-13205

0.007 : |
_ (mole % LiF)
®35
0.006 - 8Os
r | Y
3 /
u® e
@ 0.005 '
S
8
2 600°C
@ 0.004 7
o
E 704/
/ |
5 0.003 {mole % LiF)
b= 85 b
3 * 80
z 66 )
3 63./ ' -
- 0.002 7
w
x / 700°C
5 /
w ,/ 70
0.001 7%
/S Te3
/ 7
y .~
L
0

0 0.2 04 0.6 0.8 1.0
' ACTIVITY OF LiF

Fig. 10.1. Henry’s law solubility of BF5 vs_ activity of LiF in
Molten LiF-BeF, at 600 and 700°C {B. F. Hitch and C. F. Baes,
I1.,, Inorg. Chem. 8, 201(1969)].

A second and more interesting thermodynamic quan-
tity derived from the solubility data is the molar
entropy associated with the transfer of gas from the

vapor phase to the melt at equal concentrations; this
quantity, usually symbolized AS, is a measure of the
change in entropy of the gas caused solely by the
interaction of the melt with the gas. For noble gases
dissolved in fluorides,”*® AS, is zero or slightly
negative. Such values may be interpreted to mean that
the vibrational entropy associated with the dissolved
state of the noble gases is about the same or slightly less
than the translational entropy of the noble gases in the
gaseous state. In the case of BF;, AS, would not be
expected to be near zero; BF5, upon dissolution, should
lose some of its internal degrees of freedom (rotation
and vibration); hence, AS . should be negative. The data
thus far (see last column of Table 10.2) give substance
to the expectation; AS ., averages about —12 cal mole ™
°K™'. There appears to be a trend in AS, with mole
fraction; the lower the LiF concentration, the more
negative is AS.. A tempting but highly speculative
explanation for this trend is: the higher the local
coulombic field (i.e., the higher the concentration of
Be** ions) in the melt, the greater the hindrance to
rotation of the dissolved boron species (which is almost
certainly BF,"). We shall have to obtain data at
different mole ratios of LiF to BeF, to determine
whether or not the trend of AS, with composition
holds.

7. W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys.
Chem. 62,862 (1959). '

8. G. M. Watson, R. B. Evans III, W. R. Grimes, and N. V.
Smith, J. Chem. Eng. Data 7, 285 (1962).
80

10.3 FLUORIDES AND OXYFLUORIDES
OF MOLYBDENUM AND NIOBIUM

C.F.Weaver J.D.Redman

Study® of the fluorides and oxyfluorides of molyb-
denum and niobium has continued at an appreciably
reduced level during the past six months; major
emphasis has been placed on mass spectrometric investi-
gation of these materials. Though several unanswered
questions remain in each case, no studies of molyb-
denum fluorides or of niobium fluorides in molten salt
(2LiF-BeF,) solutions were performed during this
period.

10.3.1 Mass Spectroscopy of Molybdenum Fluorides

The thermal decomposition of solid MoF; and the
reaction of Mo with gaseous MoF have been described
previously. One of the reactions observed

2MoF;(s) = Mo(s) + MoF(g)

is of interest because the pressure of gaseous MoF, will
yield the standard free energy change for this reaction,
and, when combined with the standard free energy of
formation of gaseous MoF,'° will give the standard
free energy of formation of solid MoF; . The pressure of
MoF, at 700°C (973°K) was 0.79 X 107 atm, yielding
27 kcal for the standard free energy of the reaction
written above. From this information, the standard
formation free energy per bond for solid MoF; is —55 %
0.3 kcal compared with —50.5 kcal per bond'® for
MoF(g) at the same temperature. In addition the
reaction

Y, MoF; (s) = “5Mo(s) + MoF4(g)

at 700°C yielded a pressure of 4.6 X 107® atm and
hence a AF® of 24 kcal. This number combined with
the AF/® of MoF3(s) yields AF{’ of MoF,(g) of —49 +
1 kcal per bond. The errors were estimated from
pressure calibration measurements.

10.3.2 Mass Spectroscopy of Niobium Fluorides
and Oxyfluorides

Mass spectrometric studies of NbF, previously re-
ported have clarified the procedure for producing this
material in high purity. The literature suggests'' and
we have confirmed'? that the reaction :

4NbFg + Nb = 5NbF,

at 200°C with excess NbFs provides a useful synthesis
of NbF,. Mass spectroscopic studies indicated that the
tetrafluoride was unstable with respect to dlspropor-
tionation at 200°C by the reaction

NbF,4(s) = NbF5(s) + NbFs

unless an overpressure of NbFs was maintained. This
instability of NbF, was not noted below 100°C. The
best procedure for this synthesis, therefore, is to react
the niobium metal at 200°C with an excess of liquid
NbF; in an isothermal container until the reaction is
complete. At that time liquid NbFs is present with
NbFs vapor at nearly 1 atm. After the reaction is
complete the temperature is lowered to less than
100°C. Then the top of the capsule is cooled with
dry ice while the bottom remains near 100°C; the
excess NbF; is evaporated from the NbF, and is frozen
in the upper cold end of the capsule. The pure NbF,4
remains in the lower end of the capsule.

The following additional experiments were performed
to improve our understanding of the production and
reactions of niobium and oxyfluorides. '

A study of the vapor products from the reaction of
Nb,Os and F, was made over the temperature range
375 to 550°C. Fluorine was introduced into a nickel
reaction-effusion cell containing pure, outgassed
Nb,Os. The residue remaining after several hours of
fluorination was a mixture of Nb;O,F and Nb;;0,,F,
as identified by x-ray diffraction analysis. The vapor
species consisted of NbFs, NbOF;, and O,. The
current results, together with those from earlier studies
of the thermal decomposition of NbO,F, indicate that
the dominant reactions were:

2Nb205 (S) + 4F2 (g) = 3Nb02 F(S) + NbFs (g) + 202 (g)

9. C. F. Weaver, J. S. Gill, and J. D. Redman, MSR Program
Semzannu Progr. Rep. Feb. 28, 1971, ORNL-4676, pp 85-87,
93.

10. C. F. Baes, Jr., Reactor Chem. Div. Annu. Progr. Rep.
Dec. 31, 1966, ORNL-4076, p. 49.

11. F. P. Gortsema and R. Didchenko, Inorg. Chem. ‘4,
182-86 (1965).

12. C. F. Weaver and J. S. Gill, MSR Semiannu. Progr. Rep.
Aug. 31, 1970, ORNL-4622, p..71.

A
[ 8]

'

81

and
4NbO, F(s) = Nb3 O,F(s) + NbOF;(g) .

The pressures of NbOF; from pure NbO,F and from
Nb,Os + F, are shown below. Since the pressure

glass-forming beryllium fluoride systems to their glass
transition temperatures. The relationship of increasing
activation energies with decreasing temperature to the
concept of a vanishing free volume' ? or configurational
entropy'?:!® at temperatures above 0°K has been
discussed before'® and results reported for one super-

Pressure (torr) at temperature (°K)

Source

623° 648° 673° 723° - 773° 823°
NbO,F 1.8x 107* 62%x 1070 41x107° 1.1x107?
Nb,Os + F, 14 x 1072

calibrations have a usual reliability factor of 2 these
pressures are essentially equal, as the above equations
require. At 550°C (823°K) the partial pressures of
NbFs and F, were 1.7 X 10™ and 4.5 X 1072 torr,
respectively. The partial pressure of O;, obscured by
background interference, was definitely greater than the
F, pressure. The standard enthalpy, AH®, of the second
reaction above was calculated from the mass spectro-
metric data of both experiments to be 30 = 1.5
kcal/mole of NbOF;.

In addition, a study of the vapor products of the
reaction of Nb,Os(s) and NbFs(g) was made over the
temperature range 350 to 750°C. Below 550°C the

~ vapor was a complex mixture of NbFs (monomer and
dimer), NbF;, niobium oxyfluorides, and an oxy-

fluoride polymer. The oxyfluoride polymer fragment
Nb, OF;* has not been observed before, and we do not
know the precursor formula. Analogy with other

“fluoride polymers suggests that the precursor lost one

fluorine atom during ionization and that its formula
was Nb,OF,. Above 550°C the vapor was simpler as
shown below.

Pressure (torr X.10°)

Species
P 600° 625° 650° 700° 750°
NbFs 24 3.6 1.8 26 4.3
NbOF; 20 130 130 66

Nb,OF;" precursor 072 06 . 036 024 14

| 10.4 ELECTRICAL CONDUCTIVITY OF MOLTEN
AND SUPERCOOLED MIXTURES OF NaF-BeF,

G.D. Robbins  J. Braunstein

Measurement of electrical conductances in the molten
NaF-BeF, system has continued. with a view toward
relating the temperature dependence of conductance in

68%x 1075 18x 10" 14x107°

cooled NaF-BeF, mixture (60 mole % BeF,).! 7 These
measurements have been extended to mixtures con-
taining 65 and 70 mole % beryllium fluoride where
conductance has been determined as a function of tem-
perature. Temperatures ranged from 540 to 329°C, 105
degrees into the supercooled region.

The results are presented in Table 10.3. For the 65
mole % BeF, mixture the measured resistance varied
approximately as a linear function of the reciprocal

-square root of the measuring frequency from 10 to 80

kHz. The data were obtained with an R-C series-
component balancing-arm bridge and an all-metal con-
ductance cell, previously described.!® Due to the very
low resistance of the melts, the bridge (containing the
leads used in the measurements) was calibrated as a
function of frequency with calibrated resistances (2 to
10 ) in series with a 20-uF capacitor. The corrections
applied to the data ranged from +0.10 €2 at 10 kHz to
+0.19 Q at 80 kHz. The values of resistance, extrapo-
lated to infinite frequency R,,, are given in the table,
together with the magnitude (AR, ,_.) and relative
resistance change (AR/R_) between 10 kHz and R, .
Also listed in Table 10.3 are data for resistances of the
70 mole % BeF, mixture. Here the measured resistances
(corrected for the bridge calibration) appeared inde-
pendent of frequency from 10 to 80 kHz within the
limits shown.

Arrhenius-type plots of the logarithm of measured
resistance vs the reciprocal absolute temperature are

'13. M. H. Cohen and D. Turnbull, J. Chem. Phys. 31, 1164
(1959).

14. J. H. Gibbs and E. A. DiMarzio, J. Chem. Phys. 28, 373
(1958).

15. G. Adam and J. H. Gibbs, J. Chem. Phys. 43,139 (1965).

16. G. D. Robbins and J. Braunstein, MSR Program Semi-
annu. Progr. Rep. Aug. 31, 1970, ORNL-4622, pp. 98—100.

17. G. D. Robbins and J. Braunstein, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 109—-110.
82

Table 10.3. Resistances of molten and supercooled
NaF-BeF, mixtures

NaF (35 mole %)-BeF, (65 mole %)

T O R (%) AR | o o () AR/R . (%)
454.7 167 0.13 7.8
454.6 1.69 0.13 7.7
444.5 1.85 020 10.8
444.0 1.90 0.15 7.9
436.7 2.03 0.20 9.9
431.3 2.22 0.15 6.8
432.8 2.15 0.17 7.9
4195 2.59 0.17 6.6
398.2 3.51 - 0.18 5.1
381.0 4.60 0.22 4.8
3714 5.40 0.21 3.9
358.9 6.90 0.19 2.8
349.2 822 0.23 2.8
339.7 9.80 0.37 3.8
329.4 11.74 0.50 4.3

NaF (30 mole %)-BeF, (70 mole %)
T(°C) R (Q) T 0 R(2)
560.3 1.22 (20.03) 483.8 2.46 (£0.02)
540.5 1.43 (0.03) 479.3 2.49 (£0.02)
528.4 1.59 (£0.02) 479.2 2.55 (£0.02)
521.9 1.67 (20.03) 479.2 2.50 (£0.03)
519.6 1.67 (£0.03) 474.5 2.55 (£0.02)
512.4 1.84 (+0.03) 470.4 2.73 (x0.02)
505.0 1.94 (+0.03) 461.7 3.02 (20.02)
504.8 1.92 (x0.03) 460.5 3.09 (x0.02)
504.0 2.00 (+0.03) 444.1 3.51 (20.04)
495.5 2.13 (x0.03) 434.6 4.25 (£0.03)
4940 2.21 (£0.03)

shown in Fig. 10.2. The data were taken in the order
indicated by the numbered points, for NaF (35 mole
%)-BeF, (65 mole %) over a four-day period, and the
NaF (30 mole %)-BeF, (70 mole %) data over a span of
16 days. Attempts to supercool these mixtures to lower
temperatures were unsuccessful.

On termination of these experiments, cell constants
were determined as a function of depth of electrode
immersion (the major variable) and melt volume in
aqueous 1.000 demal KCl solution. These will be used
to calculate the composition dependence of specific
conductance from the data given here and that previ-
ously reported.'? :

Preliminary Arrhenius coefficients (activation ener-

gies),

dlogR
aQl/7) ’

Ej =4.575

taken graphically from Fig. 10.2 range from 11., to
15., kcal/mole for NaF-BeF, (65 mole %) and from
10., to 12.s kcal/mole for NaF-BeF, (70 mole %) over
the temperature intervals indicated. Based on a glass
transition temperature of 117°C (see next section),
T/Tg (°K) for NaF-BeF, (65 mole %)= 1.87 to 1.54;
T/Tg for NaF-BeF, (70 mole %) = 2.14 to 1.81. Over
the same 7/Tg range in the pyridinium chloride-zinc
chloride system,'® activation energies were 6.5 to 10.5
kcal/mole for PyCl-ZnCl, (65 mole %, interpolated)
and 5. (extrapolated in T) to 7.5 kcal/mole for
PyCl-ZnCl, (70 mole %, interpolated). This roughly
parallels the increase of activation energy with de-
creasing temperature observed here. At 450°C the
activation energies for these NaF-BeF, mixtures are
approximately 1 to 2 kcal/mole higher than for the
corresponding LiF-BeF, mixtures.

10.5 GLASS TRANSITION TEMPERATURES
IN THE NaF-BeF, SYSTEM

G.D. Robbins  J. Braunstein

The investigation of glass transition temfieratures ‘in
the NaF-BeF, system has continued, their relation to

~ electrical transport in molten alkali fluoride-beryllium

fluoride systems having been discussed before.!® Pre-
viously?® we presented results of measurements of glass
transition temperatures 7g at a number of composi-
tions; 7g’s for additional compositions, have been
obtained and some of the previous experiments re-
peated.

The lower portion of Fig. 10.3 shows a composite of
previously reported experimental glass transition tem-
peratures plus new data for compositions of 35, 70, 85,
and 90 mole % beryllium fluoride. These were deter-
mined by differential thermal analysis®' employing
techniques already described.?® The open symbols
represent the temperatures at which a transition was
observed in plots of temperature difference between an
Al, O; reference and the fluoride glass sample vs the
temperature of the fluoride sample. Duplicate values are
indicated by ticks on the symbols. The average value of
Tg at a given composition is represented as a solid

18. A. J. Easteal and C. A. Angell, J. Phys. Chem. 74, 3987
(1970). '

19. G. D. Robbins and J. Braunstein, MSR Program Semi-
annu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 99.
" 20. G. D. Robbins and J. Braunstein, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 110-112.

21. Obtained with the apparatus of L. O. Gilpatrick, whose
technical advice is gratefully acknowledged. '

1
83

ORNL-DWG 71-13206

t (°C)
. 550 500 450 400 350
— 12
170 — 10
—| 8
0.8 s
—1s5 _
x 8
— 3
0.4
— 2
0.2
. LLIQUIDUS '
)2 | | | | | | | | | 1
1.20 1.25 1.30 1.35 1.40 1.45 1,50 1.55 1.60 1.65 1.70

1000/7 (o)

Fig. 10.2. Logaiithm of resistance vs reciprocal temperature for NaF (35 mole %)-BeF; (65 mole %} and NaF (30 mole 7%)-BeF,
(70 mole %). '

symbol with an error bar of #2°C attached, and straight the lower composition limit for forming reproducible
line segments connect these solid symbols. - glasses is between 36 and 38 mole % BeF,.
Investigation of a sample of 35 mole % BeF, Glass transition temperature drops with increasing
quenched from 850 to 0°C produced the tranmsition  beryllium fluoride content until the liquid composition
temperatures shown (the numbers indicate the orderin ~ from which the glass was quenched approaches the
which they were observed), the temperatures shifting to  primary phase field of BeF, (the approximate phase
lower values on successive cycling between 50 and  diagram®?*® is shown in the upper portion of the
150°. Heating to higher temperatures produced very  figure). Little variation of 7g with composition is
little thermal effect on ‘crystallization of the super- observed between 50 and 85 mole % BeF,. The size of
cooled liquid, yet the crystalline solid transitions were the thermal effect associated with the glass transition in
sizable. This indicates the formation of very little glass  this region, however, decreases with increasing BeF,
and mostly crystalline phase on the initial quench, with  content. While the 0.3° temperature change at 85 mole
precipitation of more crystalline material on each of the % is measurable, the temperature change at 90 mole %
50 to 150° cycles. Previously, under these experimental '
conditions, we have failed to observe a glass transition
temperature .following quelllching of a 30 mo]e % Bek, 22. R. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor
mixture, while a reproducible 7g was measured at 38 Materials, ORNL-2548, pp. 34—35 (November 1959).
mole % BeF,. Hence, it would appear that, at least 23. D. M. Roy, R. Roy, and E. F. Osborn, J. Am. Ceram.
under the bulk quenching conditions employed here,  Soc. 36,185 (1953). .

84

ORNL-DWG T71-13207

600 \\/
//_-—.
500 \ /’
K 400
300
200 X
= =
je—— NO GLASS
140 |
|
S I
:m 130 ——+ 123 g\ GLASS
I 340 \ |
AELEEN |
120 Y _ . . . x }c |
\éxé—-—é‘—-"‘}f‘/g\é\ E}B |
. ‘ I %}A |
110 : '
0.2 0.3 0.4 0.5 0.6 07 0.8 0.9 1.0
""BaF2

Fig.'10.3. Phase diagram and glass transition temperatures for the system NaF-BeF .

BeF, is léss than 0.1°C. Accordingly, the 90 mole %
values are less certain than the other data, and several
different experiments at this composition have been
performed. These are labeled A, B, and C in Fig. 10.3
and represent three different equilibration times and

two different batches of homogenized starting material. .

The average Tg of 116° at XBeF, = 0.90 is probably
uncertain by *6°C. Glass transitions in pure beryllium
fluoride could not be observed, presumably due to the
similarity in heat capacity of liquid and glass. However,
since BeF, is a known glass former, it appears that the
region over which bulk-quenched glasses are formed
.extends from 37 (*1) to 100 mole % BeF,, that is,
including the primary phase fields of BeF,, NaBeF,,
and Na, BeF,, but not NaF.

In relating the temperature dependence of activation
energy for electrical conductance to glass transition
temperatures, one desires a 7g corresponding to the
glass-to-liquid transition where the liquid phase is that

in which conductance has been measured at a higher
temperature. However, the apparently small (or negli-
gible) variation of glass transition temperatures from
0.50 to 0.90 mole fraction beryllium fluoride raises the
possibility of metastable phase separation of the under-
cooled liquid into two liquid phases. Among the three
previous investigations of phase relations in the NaF-
BeF, system,2%27 there is considerable discrepancy.
Roy et al.>® prepared 10-g quenches of glasses from 60
to 100 mole % BeF,; Vogel and Gerth,2® who formed
their glasses by pressing them between SiO, pellets,

24. H. Rawson, pp. 235-48 in [norganic Glass-Forming
Systems, Academic, New York, 1967.

25. D.M. Roy, R. Roy, and E. F. Osborn, J. Am. Ceram. Soc.
33,85 (1950).

26. W. Vogel and K. Gerth, Glastech. Ber. 31,15 (1958).

27. M. Imaoka and S. Mizusawa, J. Ceram. Ass. Japan 61,13
(1953), as reported in ref. 24.
85

Table 10.4. Appearance of NaF-BeF, glasses

Mole % BeF,

Visual appearance

Microscopic appearance?

43 Clear glass

50 Clear glass

55 Clear glass

70 White, bright
translupent,
lightly
opalescent

85 White, opalescent

90 (A) White, opaque

90 (B,O) White, opalescent

100 Dull-to-dirty

Clear glass
Clear glass
Clear glass
Clear glass

Some clear-brown pieces, some with clear
matrix and spatse brown (oxide)
incipient crystallization

Clear matrix with brown (oxide)
incipient crystallization

Same as 90 (A)

Clear matrix with many gas

bubbles, no evidence of oxide or
precipitation

4G. D. Brunton, Reactor Chemistry Division, private communication.

were only able to produce clear glasses over the range

40 to 72 mole % BeF,, glasses of higher BeF, content
being opalescent or opaque; Imaoka and Mizu-
sawa???7 formed molded glasses from 45 to 75 mole
% BeF,. Furthermore, Vogel and Gerth®® reported
microphase separation, as determined by electron
microscopy, down to 49 mole % BeF, of region 0.5 to
1 . _

We have attempted to determine if the bulk-quenched
glasses produced under our experimental conditions
exhibited phase separation. Table 10.4 gives a summary
of the visual and microscopic®® appearance of glasses
examined in this investigation. At compositions 55
mole % BeF, and less the glasses appeared clear on both

a macroscopic and microscopic scale. From 70 to 90

mole % BeF, the glasses appeared opalescent-to-opaque
to the eye, with the material causing this opalescence of
region size below the resolving power of the polarized
light microscope (approximately 0.1 ). Microscopic
examination of samples at 85 and 90 mole % BeF,
revealed incipient crystals of brown material dispersed
in the clear glass matrix, the number density of the
brown particles being greater at the higher beryllium
fluoride content and its appearance resembling beryl-
lium oxide.2® The sample of pure beryllium fluoride

28. G. D. Brunton, Reactor Chemistry Division, private
communication.

appeared dull due to dissolved gas bubbles and did not
show the incipient (oxide) crystallization. Electron
micrographs of small gold-plated specimens from these
samples have been obtained.>® A preliminary interpre-
tation of the micrographs indicates little or no surface
structure on the cleavage surfaces of most of the glasses
(resolution =~0.5 wu), with some interesting (and dif-
ferent) patterns from the two 90 mole % samples. Thus,
the cause of the visually observed opalescence has not
been provable by either polarized light or electron
microscopy.

10.6 ENTHALPY OF LITHIUM FLUOROBORATE _
FROM 298 to 700°K: ENTHALPY AND
ENTROPY OF FUSION

A.S. Dworkin

We have examined the effect of the differences in
structure and cation size on the enthalpy and entropy
of fusion and transition in the alkali fluoroborates
sodium through cesium.?® Lithium fluoroborate with
its much smaller cation differs from the other fluoro-
borates in that it has at room temperature a hexagonal
structure type, probably that of Si0,.>! One recent

29. L. D. Hulett, Analytical Chemistry Division.

30. A. S. Dworkin and M. A. Bredig, J. Chem. Eng. Data, 185,
505 (1970). ' _

31. G. Brunton, private communication, 1971.
86

Table 10.5. Heats and entropies of melting and transition of alkali metal fluoroborates

T
oIr

AH,,

AS,,

ASm + AST?

(°]?) (kcal/mole) (eu/mole} ( K) (kcal/mole) (eu/mole) (eu/mole)

T AH,, aS,,
LiBF, 583 3.46 5.94
NaBF; 679 3.25 4.78
KBF, 843 430 5.10
RbBF, 855 4.68 5.5
CsBF, 828 4.58 5.5

5.9
516 1.61 3.1 7.9
556 3.30 5.9 11.0
518 2.86 5.5 11.0
443 . 194 4.4 9.9

paper®? reports that there is no transition in LiBF,4
between room temperature and the melting point, while
another®® reports a transition at about 115°C. We
therefore have measured the enthalpy of LiBF; from
room temperature up to about 40° above its melting
temperature to complete our thermochemical studies of
the alkali metal fluoroborates.

We found no evidence for a transition either in the
enthalpy measurements or in a separate thermal analysis
experiment. LiBF, is hygroscopic and must be handled
in a dry, inert atmosphere as it was in our work and in
that of ref. 32. However, this was not the case in the
work reported in ref. 33 where in fact the LiBF; was
ground to a fine powder in the open and even heated in
an open system for the differential thermal analysis
experiments. '

A chemical analysis performed in the Analytical
Chemistry Division showed that our sample contained 3
mole % LiF. Qur measured data were therefore cor-
rected for this impurity. The following equations
- represent the corrected enthalpy data for LiBF; in
cal/mole:

Hp—H,,,=-11440+3189T+ 1.0
X 1072 7% +5.50 X 10° T7" (298 to 583°K),

AHfusion = 3460 cal/mole; ASfUSiOII
=594 cal deg™" mole ™! (583°K),

Hyp—H,y, =1 1,490 + 40.1 T (583 to 700°K) .
Our thermal analysis experiment showed two breaks at
304 and 300°C. The breaks most probably indicate the

32. S. Cantor, D. P. McDermott, and L. O. Gilpatrick, J.
Chem. Phys. 52,4600 (1970).

33. R. J. Marano and E. R. Shuster, Thermochim. Acta 1,
521 (1970).

LiBF; liquidus and.the eutectic in the LiBF,-LiF
system. From the liquidus temperature, the mole %
LiF, and the heat of fusion of LiBF,;, the true

temperature of fusion is calculated to be 310°C. The

eutectic concentration can also be estimated to be ~5
mole % LiF.

Table 10.5 compares the enthalpies and entropies of
fusion and transition for the alkali metal fluoroborates.
The high-temperature forms of the fluoroborates of
sodium through cesium acquire considerable entropy at
the transition temperature, probably due to anionic
rotational or librational disorder.?® Their entropy of
melting, therefore, is lower than that for LiBF,,
although the sum of the entropy of transition and
melting (AS,, + AS,,) is higher than that for LiBF,.
NaBF, differs in both high- and low-temperature
structure from the other fluoroborates as well as from
LiBF,. This is reflected in its intermédiate value for
AS,, + AS,,.

10.7 NONIDEALITY OF MIXING IN Li,BeF;-Lil
A.S.Dworkin M. A. Bredig

The small deviations from ideality of mixing large
BeF,? ions with small F ~ ions and the larger deviations
of mixing similarly large but much more polarizable I~
ions with F~ jons have been discussed elsewhere.*® It
was of further interest to examine mixtures of I~ with
BeF,?” ions of similar size but different polarizability.
An estimate of the degree of dissociation of the BeF, 2~
ion in dilute molten iodide solutions can also be made
from a study of this system. Indeed, the phase diagram
of Li, BeF4-Lil determined by thermal analysis showed
interesting deviations from ideality. The system would
be truly binary only if the fluoroberyllate ion remained

34. M. A: Bredig, Chem. Div. Annu. Progr. Rep. May 20,
1971, ORNL-4706, p. 156..

(L]
«

undissociated. This does not appear to be the case in its
dilute solutions in Lil (Fig. 10.4). The dashed lines
represent initial slopes of an ideal Lil liquidus for one
particle per Li,BeF,, n = 1 (undissociated BeF, %), and
for n = 5 (complete dissociation to 1Be** and 4F~
ions). The experimental liquidus appears to start with a
slope approaching the » = 5 line but then strongly
curves to and apparently beyond the n = 1 line. Thus,
the Li,BeF, seems to show a high degree of dissoci-
ation only in very dilute solution. At the lowest

" concentration measured, about 3 mole % Lf;B¢F4 , only

about 10 to 15% dissociation is indicated. The rate at
which the degree of dissociation decreases with in-
creasing Li,BeF, concentration is difficult to estimate
because of the nonideality of mixtures of I~ and
BeF,? ions. This nonideality is apparent from the
phase diagram at and near the eutectic composition.

The much greater dissociation reported earlier in an
MSR monthly report was most likely due to the
presence of oxide in the Lil (~1 mole %) which may
have reacted with the Li,BeF, according to BeF,2™ +
0* - BeO + 4F  to give an effect similar to
dissociation. The experiment was repeated with single-
crystal . Lil which, when analyzed after the melting
point determination, was shown to contain only 0.1
mole % oxide. The small halts reported at 380 to 385°C
in the original experiments were much weaker but still
present in the later experiments. Whether these halts
may be attributed to a ternary eutectic with the
remaining oxide impurity or to the stabilization of
another form of Lil, perhaps tetrahedrally coordi-
nated,>®> which undergoes a phase transition to the
common octahedral one at about 385°C is still under
investigation. ’ :

The Li, BeF,4 liquidus is ideal initially, that is, forn =
1 particle (I”7) per Lil. Positive deviation, excess partial

ORNL-DWG 71-125364A

480
460 Z
. 440 S n=1
e RN
> 420 n=t—" ]
\Q
\\\ // n
400 — , o - -
380
0 20 40 ) 80 100
LI289F4 mole% Lil Lil

Fig. 10.4. The system Li; BeF4-Lil.

free energy of mixing, #figB eF, ~ 0, begins at about
15 to 20 mole % Lil. This behavior may be compared
with that previously discussed®® for fluoride-iodide
mixtures, especially those with a common bivalent
cation, Ca®*. There, positive ugan initially was much
smaller than ,uffziF in LiF-Lil mixtures. An explanation
was based on the negative contribution to ,ugal:2 from
the increase in polarization  (in configurations
Ca**I"F") of the I~ by Ca*" in the relatively strong
field of the F~ ions, which are more numerous in CaF,
than in LiF.** Applied to Li, BeF,-Lil mixtures, F~
jons even though bound to Be®* in BeF4% are more
numerous for configurations Li*I 'F~ by a factor of 4
than in LiF-Lil. This may give the strong negative
balance to positive excess free energy in Li; BeF,-rich
mixtures. ' -

35. Cf. W. Riihl, Z. Physik 143, 599-603 (1955).
Part 3. Materials Development

J. R. Weir, Jr.

The areas of materials research and development
discussed in Part 3 include the postoperation examina-
tion of components from the MSRE, the development
of radiation-resistant impermeable graphite, investiga-

tion of radiation damage and compatibility of Hastelloy

N in various environments, and materials work in
support of the chemical processing equipment develop-
ment.

One of the most important problems discovered by
examining components from the MSRE is a shallow
intergranular penetration into the Hastelloy N that
produces intergranular cracking to a depth of several
mils. This phenomenon only occurs in metal exposed to
the fuel salt, and it is of concern whether it is associated
with a corrosion mechanism or the penetration of
certain embrittling fission products. Work is also con-
tinuing on the fission product and corrosion product
deposition on graphite removed from the reactor.

The graphite program involves an assessment of the
resistance of commercial graphites to the dimensional
and structural instabilities produced by radiation
damage. Recent irradiation results indicate that graph-
ites can be fabricated at ORNL with behavior similar to
the best commercial graphites. This result is encourag-
ing in that it increases our confidence that we under-

stand enough about the radiation damage processes to.

further improve the important properties of the moder-
ator. Maintaining low permeability (during irradiation)
to gaseous fission product intrusion by sealing the

' 88

surfaces of graphite with pyrolytic carbon is proving

difficult. We are increasing our effort in this area by .

examination of the microstructural mechanisms of fail-

ure of the sealant and are investigating other sealants

such as barren salt.

The resistance of small heats of Hastelloy N modified
with Ti, Nb, Zr, Si, and C to irradiation-induced
embrittlement continues to be studied. We are utilizing
the microstructures to evaluate the type of carbide
precipitate developed in the various compositions, since
there is a reasonably good correlation between the
existence of MC-type carbides and good radiation
resistance. Our compatibility programs involve deter-
mining the corrosion resistance of Hastelloy N in fuel
salt, the coolant salt (NaBF,-NaF), and steam. The
possibility of using a duplex tube of nickel on the
coolant-salt side and Incoloy 800 on the steam side of a
steam generator is being investigated. This combination
should produce the optimum corrosion resistance for
both fluids.

The major effort in our work in suppart of chemical
processing equipment involves the development of
fabrication procedures to build a complicated reductive
extraction processing unit of molybdenum. Procedures
have been worked out for most of the components of
the system. Other possible materials that appear to be
candidates for this application are tantalum and graph-
ite. We are evaluating these materials and a few brazing
alloys for their resistance to bismuth corrosion.

ot
11. Examination of MSRE Components

H. E. McCoy

As described in the last semiannual report,! Hastelloy
N components of the MSRE appeared superficially to
have suffered very little corrosion, but when specimens
of a control rod thimble and heat exchanger tubes were
tested in tension at room temperature, shallow cracks
appeared at surfaces that had been exposed to the fuel
salt. These cracks were intergranular, extending to a
depth of roughly one to two grains, and did not occur
on the other sides of the specimens, which had been
exposed to air or coolant salt. It thus appeared that the
Hastelloy N exposed to the circulating fuel had been
attacked or embrittled along the grain boundaries by

‘some unknown mechanism.

As part of our efforts to resolve the effects of the fuel
salt on Hastelloy N, we have proceeded to test pieces
from the fuel sampler and part of the control rod
thimble that were exposed to fuel salt under different
conditions. We have also tested pieces of Hastelloy N
that were exposed to fuel salt in in-pile loop 2 in 1967
and have 'reexamined surveillance specimens removed
trom the MSRE core at various times during its
operation.

Examination of a graphite moderator element was

extended to obtain more information on the elements

at or near the surface.

11.1 EXAMINATION OF HASTELLOY N
COMPONENTS EXPOSED TO FUEL SALT
IN THE MSRE

B.McNabb  H. E.McCoy

The control rod thimble and heat exchanger tubes
that showed cracks under tensile strain at room

89

temperature’ had been exposed to fuel salt in the main
circulating stream. Other pieces available to us that had
been exposed to fuel salt under somewhat different
conditions were rods and mist shield from the fuel
sampler in the pump bowl and a section of the control

rod thimble that had been under a loose sleeve.

The mist shield was located in the pump bowl, and its
purpose was to keep salt spray from the region where
salt samples were being taken. The sampler cage is
located in the center of the spiral mist shield. The mist
shield is a spiral fabricated of '-in. sheet that has an
inside diameter of about 2 in. and an outside diameter
of about 3 in. The spiral was about 1% turns, and the
outside was exposed to agitated salt and the inside to
salt flowing at a much slower rate. The outside was
exposed to salt up to the riormal liquid level and to salt
spray about this level. The inside was exposed to salt up
to the normal salt level and primarily to gas above this
level. The general appearance of the mist shield was
described previously.?

Four specimens approximately 1 in. (vertlcal) X Y% in.
(circumferential) X ‘% in. thick were cut from the mist
shield. Their locations were (1) outside of spiral
immersed in salt, (2) outside of spiral exposed to salt
spray, (3) inside of spiral immersed in salt, and (4)
inside of spiral exposed primarily to gas. Bend tests
were performed on these spécimens at 25°C using a
three-point bend fixture.> They were bent about a line

1. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 147, 156.

2. E. L. Compere and E. G. Bohlmann, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 76—83.

3. H. E. McCoy and J. R. Weir, The Effect of Irradiation on
the Bend Transition Temperatures of Molybdenum- and Nio-
bium-Base Alloys, ORNL-TM-880, pp. 7— 10 (July 1964).
90

parallel to the %, in. dimension and so that the outer

surface was in tension. The information obtained from
such a test is a load-deflection curve. The equations
normally used to convert this to a stress-strain curve do
not take into account the plastic deformation of the
part. The stresses obtained by this method become
progressively in error (too high) as the deformation
progresses. 4

The mist shield was made from “4-in.-thick sheet of
Hastelloy N heat 5075 with 16.2% Mo, 6.7% Cr, 4% Fe,
0.44% Mn, 0.06% C, 0.58% Si, 0.27% V, 0.08% W,
0.02% Al, 0.02% Ti, 0.07% Co, 0.01% Cu, 0.003% P,

0.006% S, and 0.001% B. The yield stress at room

temperature was certified as 53,000 psi, ultimate tensile
strength 116,000 psi, and elongation 49% for the
original material. Table 11.1 shows the results of bend
tests on the specimens from the mist shield after
removal from the pump bowl. Note that the yield and
ultimate stresses are about twice as high as those
normally reported. These values - can be used only for
comparison of the relative effect of the environment on
the properties. The unirradiated control test of % in.
sheet of a similar material (heat N3-5106) did not fail
when subjected to the same bend test as the mist shield
specimens, indicating that the mist shield specimens
were.embrittled somewhat even though all of them did
strain greater than 10% in the outer fibers before
failure.

Figure 11.1a is a macrophotograph of bend specimen
S-52 from the inside top of the mist shield. The black
scale has popped off most of the bend area, exposing
shiny surface. A weld bead used to hold the spiral in a
fixed position is visible on the lower right of the
picture. The specimen failed during the bend test, but
was the most ductile of the mist shield specimens.
Figure 11.1b is a polished cross section of specimen
S-52. The fracture and the small cracks on- the tension
side are evident. Higher magnification photo-
micrographs of sample S-52 are shown in Fig. 11.2 for
the tension and compression sides. The cracks are about
1 mil deep on the tension side, and no cracks formed on
the compression side. Figure 11.3 is a macrophotograph
of the tension side of specimen S-62 from the outside
top portion of the mist shield that was exposed to salt
spray. Several cracks are evident. .

Photographs of specimen S-60 from the inside liquid
region of the mist shield are shown in Fig. 11.4.
Extensive cracking is evident in the macrophotograph
(Fig. 11.4a). The polished cross section in Fig. 11.4b
shows numerous cracks to a depth of 5 to 7 mils. A

macrophotograph of specimen S-68 from the outside

liquid region of the pump bowl is shown in Fig. 11.5.
Numerous cracks are visible near the fracture.

One of the Y-in.-diam sampler cage rods was tensile
tested at room temperature to determine the change in
mechanical properties. The rod was made from Hastel-

Table 11.1. Bend tests at 25°C on various MSRE components

. Yield Maximum?® .
Specimen ) Strain .
number stress tensile stress %) Environment
(psi¥) (psi) g
x 102 x 103
§-52 Mist shield top inside . 97 269 46.9¢ Vapor region shielded
S-62 - Mist shield top outside 155 224 10.7¢ Vapor region salt spray
5-60 Mist shield bottom inside 161 292 31.6° Liquid region shielded salt flow
S-68 : Mist shield bottom outside 60 187 17.7¢ Liquid region flow rapid salt
N3-5106 Unirmadiated control 128 238 40.5
test, % in. thick :
HTY8487 Unirradiated control 110 147 33.8
test, 0.065 in. thick .
HTY8487 Control rod thimble under 88 134 33.6 Restricted salt flow OD
spacer .
HT5060 Control rod thimble spacer 79 105 - 338 Restricted salt flow ID
sleeve

Fast salt flow OD

~ 9Based on 0.002% in. offset of crosshead travel.

bMaximum tensile stress was controlled by fracture of sample or by strain limitation of test fixture.

cSpecimen broke.

IR T
91

PHOTO 263271

(8)

Fig. 11.1. Bend specimen S-52 from the vapor region inside of the mist shield. (a) Photograph showing fracture and tension side
of specimen. Black scale has popped off exposing shiny surface in some areas. ~5.5X. (5) Photomicrograph of half of bend specimens
showing shallow cracks in tension side of bend. The lower edge of the picture does not extend to the compression side of the

specimen.

loy N heat 5059 with 16.9% Mo, 6.62% Cr, 3.92% Fe,
0.35% Mn, 0.07% C, 0.59% Si, 0.21% V, 0.07% Co,
0.01% Al, 0.01% Ti, 0.04% W, 0.001% P, and 0.003% S.
The room-temperature yield stress was certified as
51,200 psi, the ultimate tensile stress as 115,300 psi,
and the elongation as 51% for the original material. The
yield stress of the sampler cage rod from the pump
bowl was 42,450 psi, the ultimate tensile stress 93,241
psi, and the elongation 35.7%. As seen in Fig. 11.6a, the
rod ruptured below the center, probably near the
average liquid-gas interface where the deposits on the
rod appeared thickest. Numerous cracks can be seen
near the rupture in Fig. 11.6b, a greater enlargement,
and individual grains appear to have dropped out. The
cracking occurred almost to the bottom of the rod,
even extending into the area covered by the grips where
the stress would be much lower. The cracking dimin-
ished toward the upper part of the rod above the liquid

level, and cracks were not apparent at 5.5X magnifi-
cation at the top specimen grips. Figure 11.7 is a
photomicrograph of the edge near the fracture area,
which was in the liquid zone, showing the extensive
cracking. Figure 11.8 is a photomicrograph of a section
from the vapor region about 1 in. above the rupture,
and the cracks are seen to be less deep than in the
section below the liquid level.

The control rod thimble was a tube having dimensions
of 2 in. OD X 0.065 in. wall thickness. To keep the
thimble centered in the graphite moderator, spacer
sleeves were used periodically. They were joined to the
thimble by weld beads on the thimble that were
deposited through a clearance hole in the sleeve. Thus
the sleeve was restrained in the vertical direction, but
was free to move some in the radial direction. The
maximum and minimum clearances (according to shop
drawings) between the sleeve and the thimble were

(6)

Fig. 11.2. (a) Photomicrograph of tension side and rupture of specimen S-52 from inside vapor region of mist shield. Crack depth
~1 mil. 100x. (b) Photomicrograph of compression side of bend near rupture. As polished. 100X.

e

93

Fig. 11.3. Macrophotograph of the tension side of specimen S-62 that was taken from the outside top portion of the mist shield.

This area was exposed to fuel salt mist. 5.5x.

0.015 and 0.000 in., respectively. The annular region
would likely be exposed to almost stagnant salt.
Previous microprobe examination of the control
thimble under the sleeve failed to reveal any chromium
gradients such as were noted in other parts of the core.*
We decided to test a ring from under the spacer sleeve
to see if cracking occurred there as it did in the thimble
exposed to flowing salt. Figure 11.9 is a macro-
photograph of the rupture of the control rod thimble
under the spacer sleeve. Numerous cracks can be seen
over the entire gage length, so it appears that rapid flow
of fuel salt is not required to cause cracking under these
conditions. Figure 11.10 is a photomicrograph of the
same area as shown in the above figure, showing the
cracking on the side exposed to fuel salt and fission
products. A ring from the spacer sleeve was cut at the
same time as the thimble and tested in the same
manner. Figure 11.11 is a macrophotograph of the

4. B. McNabb and H. E. McCoy, MSR Program Semiannii.
Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 147-51.

rupture area of the spacer sleeve OD, which was
exposed to flowing fuel salt. Only a few cracks are
visible near the rupture, but the spacer sleeve failed at
much lower ductility than the thimble. Figure 11.12 is
a photomicrograph showing the ID that was exposed to
the same stagnant fuel salt as the OD of the control rod
thimble and the OD that was exposed to flowing fuel
salt. There does not appear to be much difference in the
cracking pattern on the OD and the ID. The spacer
sleeve was made of Hastelloy N heat 5060. Table 11.2 is
a tabulation of the tensile data on the control rod
thimble rings and the spacer sleeve. The spacer sleeve
(heat 5060) had higher strength and lower ductility
than the control rod thimble, either under the sleeve or
exposed to rapidly flowing salt.

The observations are consistent with our general
conclusions that all Hastelloy N surfaces exposed to
fuel salt crack. The samples from the mist shield show
that the severity of the cracking is less where the service
environment is vapor or mist than where 100% liquid is
the environment. The specimens from the control rod
thimble and the spacer showed that the degree of
cracking was not dependent upon the salt flow rate.

94

PHOTO 2633 - 74

(&)

Fig. 11.4. (2) Macrophotograph of the tension side of bend specimen S-60 from inside liquid region of mist shield exposed to
liquid fuel salt. ~5.5X. (b) Photomicrograph of bend and rupture area of S-60 showing extensive cracking on tension side of bend.
The lower edge of the picture does not extend to the compression side of the specimen.

11.2 EXAMINATION OF COMPONENTS FROM
IN-PILE LOOP 2

B.McNabb  H. E. McCoy

Loop 2 was an in-reactor thermal convection loop
constructed of Hastelloy N and containing fuel salt of
composition ?LiF-BeF,-ZrF,-UF, (65.4-27.84.8-2.0
mole %).% It had a peak operating temperature of 720°C
and a minimum temperature of 545°C, and it operated
1102 hr with fuel salt at these conditions. Average
power densities of 150 w/cm? of salt were attained, and
8.2 X 10'® fissions/cm® were produced before oper-
ation was terminated by a leak.

5. E. L. Compere, H. C. Savage, and J. M. Baker, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1967, ORNL-4191, p.
176.

Some of the components were made available for
testing to determine if cracking would occur under
stress as had been observed for some MSRE compo-
nents exposed to fuel salt and fission products. A
rectangular strip approximately 1 X ' X % in. thick
was cut from the top of the core vessel and bend tested
in the same way as the specimens from the mist shield.
The apparent yield stress was 104,000 psi, the ultimate
tensile stress was 225,000, and the outer fiber strain
was 41.5%. Figure 11.13 is a macrophotograph of the
bend specimen, showing that it did not fail. This
contrasts with the failure in a similar test of the mist
shield, which had been exposed for a much longer time
to flowing salt and fission products. The peak operating
temperature for the loop was approximately 720°C,
and this bend specimen came from near the hottest
region. Figure 11.14 is a photomicrograph of the top
bend-tested specimen, showing the tension side of the
95

- Fig. 11.5. Macrophotograph of the tension side bend specimen S-68 from the outside liquid region of mist shield. Exposed to
agitated fuel salt. Extensive cracking is confined to the rupture area. ~5.5X.

PHOTO 2634~ T4

T

Liquio VAPOR

(a)

(61

Fig. 11.6. (a) Tensile-tested sampler cage rod from pump bowl. Rupture occurred near the average liquid-gas interface. ~1x. (b)
Macrophotograph of rupture area showing extensive surface cracking in the liquid region. ~5.5X . Reduced 37%.

Fig. 11.7. Photomicrograph of edge near the fracture area of MSRE sampler cage rod. Note extensive cracking and grain pullout
at surface. The fracture occurred in a section of the rod that was near the average liquid level. 100X . As polished.

bend with small cracks extending to a depth of about
0.5 mil.

A ring was cut from the cold leg return line, Y-in.
pipe (corresponding to the coldest part of the loop,
545°C), and crushed in the bend rig. It appeared to
have good strength and did not crack externally, as
shown in Fig. 11.15. No analysis of the bending data
was attempted. Figure 11.16 shows microphotographs
of the cross section of the ring. The small cracks
extending to a depth of about 1 mil covered the entire
interior surface. It appears that cracking did occur on
the fuel-salt side of the Hastelloy N loop material when
it was stressed after exposure, even with its short
exposure time and low temperatures (545°C). However,
the mechanical properties were not greatly affected.

11.3 OBSERVATIONS OF GRAIN BOUNDARY
CRACKING IN THE MSRE

H. E. McCoy

The first indication that grain boundary cracking was
occurring in the Hastelloy N in the MSRE was obtained
from surveillance specimens exposed in the core. When
the third set was removed and examined in April 1968,
we first noticed intergranular cracking on specimens
stressed in tension.® Since that time we have examined
more components and find that the cracking is quite

6. H. E. McCoy, An Evaluation of the Molten-Salt Reactor
Experiment Hastelloy N Surveillance Specimens — Third Group,
ORNL-TM-2647 (January 1970).
97

Fig. 11.8. Photomicrograph of edge in the vapor region of tensile-tested sampler cage rod. Cracking was not as severe in this area

asiin the liquid region. Gly regia. 100X.

general in the fuel circuit.” Photographs of samples that
were strained at room temperature will be used to
demonstrate the cracking, since grain boundary crack-
ing is a normal process at elevated temperatures. Thus
cracks are also present in the samples tested at elevated
temperatures, but it is impossible to separate those
caused from exposure to the fuel salt and those due to
deformation. An even more careful analysis has shown
that the cracks were visible in many of the samples when
they were removed from the MSRE, and straining only
opened the cracks to make them more visible.

Several photographs of tested surveillance specimens
have been shown previously,5® ~1© but several of the
specimens were repolished and photographed more
extensively after we realized the generalness of the
problem. Figure 11.17 shows polished sections of
Hastelloy N specimens tested at 25°C after heating at
650°C for 4800 hr. (The samples were exposed to fuel
salt most of this time. The time at temperature is used

because it is likely the parameter that controls the
diffusion of fission products inward and also is impor-
tant in moving chromium to the surface to resupply any
areas depleted by corrosion.) The unirradiated specimen
was exposed to static fuel salt containing depleted
uranium in a control facility, and the irradiated
specimen is from the core of the MSRE. There are more

7. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, pp. 1-20.

8. H. E. McCoy, An Evaluation of the Molter-Salt Reactor
Experiment Hastelloy N Surveillance Specimens — First Group,
ORNL-TM-1997 (November 1967).

9. H. E. McCoy, An Evaluation of the Molten-Salt Reactor
Experiment Hastelloy N Surveillance Specimens — Second
Group, ORNL-TM-2359 (February 1969).

10. H. E. McCoy, An Evaluation of the Molten-Salt Reactor
Experiment Hastelloy N Surveillance Specimens — Fourth
Group, ORNL-TM-3063 (March 1971).

Fig. 11.9. Macrophotograph of MSRE control rod thimble Fig. 11.11. Macrophotograph of tensile-tested ring from
from under spacer sleeve exposed to stagnant fuel salt on OD.  spacer sleeve on control rod thimble. Exposed to rapidly
Tensile tested at 25°C. 5.5X. flowing fuel salt on OD shown and stagnant fuel salt on the ID

R-56054

Fig. 11.10. Photomicrograph of control rod thimble from under spacer sleeve. Exposed to stagnant fuel salt on OD (top) and to
cell air on the ID (bottom). As polished. 40X .
=

0.035 INCHES
T~ 100X

Fig. 11.12. Photomicrograph of tensile-tested ring from spacer sleeve on control rod thimble. Exposed to rapidly flowing fuel
- salt on the OD (bottom) and to stagnant fuel salt on the ID (top).As polished.

Table 11.2. Tensile data on control rod thimble rings at 25°C at a crosshead speed of 0.05 in./min.

Reduction Crosshead
Specmed Material Condition Y‘ff*;‘,}”e“ U"‘“‘(;t:“"°“ in area travel
) (in)
2 Y-8487 Irradiated 54,400 105,100 23.2 0.54
3 Y-8487 Irradiated 53,300 102,500 29.7 0.51
4 Y-8487 Irradiated 60,500 110,800 28.5 0.42
6 Y-8487 Unirradiated 52,000 114,400 4.5 1.20
7 Y-8487 Unirradiated 56,900 117,500 46.0 117
8 Y-8487 Unirradiated 58,100 151,100 43.0 L18
Under spacer Y-8487 Irradiated 60,800 116,000 329 0.45
Spacer sleeve Y-5060 Irradiated 80,900 155,200 18.0 0.20

“Based on an offset of 0.002 in.

Fig. 11.13. Macrophotograph of tension side of bend specimen from top of core vessel of in-pile loop 2, exposed to flowing fuel

salt, Specimen did not fail. 5.5x.

cracks in the irradiated specimen, but there are so few
that it is not surprising that we did not feel that they
were significant when first observed.

Similar photographs of samples removed after 22,533
hr at 650°C are shown in Fig. 11.18. By now the
frequency of cracking in the samples from the MSRE
was quite high. These samples have an enlarged section
at each end for pulling. Note that the cracks extend
along the radius even though the strain is decreasing
rapidly as the cross section increases.

A ring was cut from the control rod thimble after it
had been at temperature for 26,000 hr and had received
a thermal fluence of about 2 X 10?! neutrons/cm?. A
portion of the stressed tube is shown in Fig. 11.19. The
outside that was exposed to the fuel salt is heavily
cracked, but the crack tips are generally quite blunt,
indicating that they did not propagate as the tube was
strained. The other side of the tube was oxidized by
exposure to the cell environment of N, containing 2 to
5% 0,. The oxide cracked, but the cracks do not
penetrate the tube wall. Thus this component was
heavily irradiated, but only the side exposed to fuel salt
cracked.

Several pieces of tubing were cut from the primary
heat exchanger and examined. One piece was strained

and numerous cracks formed on the fuel-salt side (shell
side), but not on the coolant-salt side (tube side) (Fig.
11.20). A close examination of the unstressed tubing
showed that the cracks were visible. This is shown more
clearly in Fig. 11.21, a photomicrograph at higher
magnification. This component revealed that neither
irradiation nor straining were requirements for the
cracking, but that exposure to the fuel salt was
necessary.

We have collected the information that we presently
have on the depth and number of cracks formed in
various samples (Table 11.3). Several points seem
important:

1. The maximum depth of cracking did not seem to
increase systematically with exposure time over the
range 4800 to 26,000 hr. The depth seems to have
been limited to one grain diameter.

)

. The frequency of cracks increased dramatically with
increasing exposure time and reached different
values in different components because the grain size
was different.

3. The cracks were present when the material was
removed from the MSRE, and they were made more
.

salt. Maximum crack depth about 0.5 mil. As polished. 100X.

visible by straining the sample. They have blunt tips,
indicating that they do not penetrate much as the
sample is strained.

Layers were removed from several samples using
methanol-10% HCI as an electrolyte. These solutions
were analyzed by several methods and did indicate that
fission products had penetrated the metal. A typical set
of results for a heat exchanger tube is shown in Fig.
11.22. However, it is difficult to attach much signifi-
cance to these results since the cracks extended to
depths greater than our calculated sample depth (based
on the amount of Ni). We were able to find all fission
products with sufficient half-lives to still be present in
detectable amounts. Thus, these concentration profiles
may not represent solid-state diffusion, but rather the
deposition of materials in existing grain boundary
cracks.

Compere compared the concentrations of several
fission products found on the Hastelloy N surfaces with
the total inventory. The inventory values were com-
puted taking account of the power history, and the

101

Fig. 11.14, Photomicrograph of tension side of bend specimen from top of core vessel of in-pile loop 2, exposed to flowing fuel

concentration per unit area was computed by assuming
that the fission products were distributed uniformly
over the entire metal area of 7.9 X 10° cm?. The
measured concentrations were obtained by summing
the amounts found in successively removed layers.
Some samples inherently yielded better data because
the surface being dissolved was better defined. The
surveillance specimen and the heat exchanger tube were
good specimens, but the other samples had various
problems. For example, the samples from the control
rod thimble were small strips cut from a thin-walled
tube. The edges contributed appreciably to the nickel
content, but did not contribute to the fission product
concentration. The depth of penetration was based on
the nickel content of the solutions, so the measured
fission product concentrations are actually lower than
those present on the single surface that was exposed to
fuel salt. The fission products for which 10% or more of
the inventory was found on the metal surface are shown
in Table 11.4. There is much scatter in the data, and
there does not seem to be any systematic variation of a

Fig. 11.15. Macrophotograph of cold leg return line of in-pile 10p 2 exposed to flowing fuel salt on ID. The ring was crushed in
abend test at 25°C. Specimen did not fail. ~5.5X.

PHOTO 2483-71

'“I
|
t

le loop 2, exposed to flowing fuel salt at

Fig. 11.16. Photomicrographs of cross section of crushed ring from cold leg o
545°C. On ID maximum crack depth ~1 mil. As polished. 100X . Reduced 56.5%.

103

UNIRRADIATED

IRRADIATED

SURVEILLANCE - FIRST GROUP

Fig. 11.17. Photomicrographs of Hastelloy N tested at 25°

'C after 4800 hr at 650°C. The top specimen was in static fuel salt

containing depleted U, and the lower specimen was in the core of the MSRE. The true specimen diameters are 0.070 to 0.090 in.

particular fission product as one progresses around the
primary circuit. However, significant fractions of the
elements shown in Table 4.4 were found on the metal
surfaces.

We noted that interfered with the
dissolution process. To concentrate the materials that
were present along the grain boundaries, we preoxidized
a surveillance sample in air at 650°C, strained it about
half way to fracture, and dissolved successive layers. We
found that Te, Ce, Sb, Cs, Sr, and S were higher than
noted previously for samples not oxidized

Two general mechanisms seem possible based on these
observations. First, the fission products (or S from the
salt and the pump oil) could be diffusing along the grain
boundaries. All of the elements mentioned above that
were concentrated along the grain boundaries have low

oxide films

melting points and would reduce the melting point of
the Hastelloy N. Several of these elements (namely, Te,
Sb, and S) are also known to embrittle pure nickel,'!
and Te has been shown to embrittle a Ni-Cr-Co-Mo
alloy.!? Thus penetration of these elements along the
boundaries could produce a localized region with a
lower melting point and low ductility. To open the
boundary up as a crack during service would require
that this region become completely molten or that a

11. C. G. Bieber and R. F. Decker, “The Melting of Malleable
Nickel and Nickel Alloys,” Trans. AIME 221, 629 (1961).

12. D. R. Wood and R. M. Cook, “Effects of Trace Contents
of Impurity Elements on the Creep-Rupture Properties of
Nickel-Base Alloys,” Metallurgia 109 (March 1963).
UNIRRADIATED

IRRADIATED

SURVEILLANCE - FOURTH GROUP

Fig. 11.18. Photomicrographs of Hastelloy N tested at 25°C after 22,533 hr at 650°C. The top specimen was in static fuel salt
containing depleted U, and the lower specimen was in the core of the MSRE. The true specimen diameters are 0.070 to 0.090 in.

R- 55182

Fig. 1
fuel salt, and the bottom surface was exposed to the cell atmosphere of N containing 2 to
in a high radiation field. The wall thickness is about 0.065 in.

19. Photomicrograph of ring from control rod thimble that was tested to failure at 25°C. The top surface was exposed to
0. Part was at 650°C for 26,000 hr

105

R - 55178
UNSTRESSED
‘ _
STRESSED

Fig. 11.20. Photomicrographs of tubing from the primary heat exchanger. It was at 650°C for 26,000 hr. The lower specimen
was strained to failure at 25°C. The original wall thickness was 0.042 in. Fuel salt was on the upper side of each tube, and coolant

salt was on the lower side.

™

e
5

primary heat exchanger. The upper surface was exposed to coolant salt

Fig. 11.21. Photomicrograph of a tube from the MSRE
and the lower surface to fuel salt.

106

Table 11.3. Crack formation in Hastelloy strained to failure at 25°C

. Time of )
Sample description exposure Cracks Depth, (mils)
(hr) Counted Per inch Average Maximum

Surveillance, heat 5085 4,800 24 19 2.5 8.8
Control 1 1 5.7 5.7
Surveillance, heat 5085 15,300 178 134 1.9 6.3
Control 0
Surveillance, heat 5085 22,500 213 176 5.0 7.0
Surveillance, heat 5085 ’ 140 146 3.8 8.8
Surveillance, heat 5065 240 229 5.0 7.5
Control, heat 5085 4 3 1.5 2.8
Thimble : 26,000 91 192 5.0 8.0
Heat exchanger (stressed) 26,000 219 262 5.0 6.3

(unstressed) 100 228 2.5 3.8

(unstressed) 135 308

%0ne crack was 15 mils deep; the next largest was 7 mils.

stress be imposed. The first possibility seems unlikely,
and the design conditions of the items that we
examined required that they either be unstressed
(surveillance specimens) or under compression (thimble,
heat exchanger tube). Another mechanism is one of
selective corrosion. The accepted corrosion mechanism
for the fuel salt and Hastelloy N is the selective removal
of Cr by the reaction '

UF,4 + Cr (in alloy) = UF; + CrF, (in salt).

Much of our experimental corrosion work was done at
higher temperatures, and all of it was done in systems
where the concentrations of UF; and UF, were
uncontrolled and not known accurately. Thus, although
we did not observe selective grain boundary attack in
our tests, the possibility of its occurring cannot be ruled
out. In fact, the grain boundaries are generally more
reactive in metals, and selective attack would be
expected under some conditions. When the concen-
tration of UF3 was high no corrosion would occur, and
when it was low all elements would be removed. At
some intermediate concentration selective removal
along the grain boundaries could occur.

An example of selective grain boundary attack of
type 316 stainless steel in LiF-BeF, is shown in Fig.
11.23. This experiment was run to determine whether a
Mo coating could be produced on stainless steel to

make it usable in Bi systems. The specimen was
suspended in salt at 650°C, some MoF4 was injected as
a gas, the melt held at temperature for 48 hr, and the
salt” discharged. The top layer is quite high in Mo, the
intermediate layer contains salt and some metallic
elements, and the substrate is type 316 stainless steel.
The grain boundary attack is quite evident.

We cannot presently distinguish between the various
mechanisms that have been proposed. However, numer-
ous experiments are in progress to define and learn how
to control the process.

11.4 EXAMINATION OF A GRAPHITE
MODERATOR ELEMENT

B. McNabb  H. E. McCoy

The graphite moderator element removed from the

MSRE after shutdown appeared unchanged by the

irradiation or fuel-salt flow, as reported earlier.!?
Core-drilled samples were taken from the fuel channel
of this element for Auger and spectrographic analyses;
the Auger analysis of the surface layer is repofted in the
next section. Small samples (% in. diam) from about
the midplane of the core were used as electrodes for the

13. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 139—43,

ye
e

CONCENTRATION RELATIVE TO Ni

ORNL-DWG 71- 7106
HEAT EXCHANGER TUBE

23
10° e Ni
e e b

= —o—sFe 7
02 L _

s——t—e—e— Mn

rTTTII

@)
[ TTTTT

W
/'./ .
o]
Q
L

[ 1Ll

BRI
/
/
®
Lo .
({e)
S
| b L

o)
On
I
v
|

(0))

T 11111

AT
| ®

yd

: /
|

i

»

- .
»
L

————— e — — —— — . oef

10 NN Brcs ™=

:\\ e 1255 -

B "‘\ 0 ° ‘ N

-7 \ b _

‘ E \. \: l E

108 L ‘\\3 osr

= 127Te E

-\, -

00 ENG T

10—10 0\\.'\.. ?SNbé

@) 1 .2 3 4
DEPTH -(r_nils)

Fig. 11.22. Concentration profiles from the fuel side of an
MSRE heat exchanger tube measured about 1.5 yr after reactor
shutdown. .

107

" mass spectrograph analysis of fission products. Table

11.5. shows the results of successive analyses, using
increasing amounts of current. These data are quite

- useful, but several factors limit their accuracy. The

penetration of the spark into the graphite was likely a
hemisphere, so some new surface continued to be
included .as the penetration depth increased. The
calculated depths are estimates of the maximum depths
sampled. The sensitivity of the analysis increases with
increasing current. The size of the area sampled is also
of concern, particularly at the lower penetration
depths. This graphite has pores about 0.1 ¢ (1000 A) in
diameter, and 'the compositions of the first samples
(taken from areas a few angstroms in diameter) could
be quite dependent upon the location. These factors
make it impossible to obtain a quantitative concen-
tration profile, but it appears that the fission products
are concentrated at the surface and decrease rapidly

with depth.

11.5 AUGER ANALYSIS OF THE SURFACE
LAYER ON GRAPHITE REMOVED FROM
THE CORE OF THE MSRE

R. E. Clausing

Quantitative analysis of thin layers of fission products
deposited on the graphite and metal surfaces of the
MSRE would provide information valuable for pre-
dictions of the requirements for afterheat removal in
future molten-salt reactors. Auger electron spectros-
copy offers unique capabilities for such analyses. The
construction of equipment for this purpose was de-
scribed previously.'® Thé equipment has been modified .
to permit ten samples and standards to be loaded for
examination at the same time, thus facilitating the
comparison of samples with other samples or standards.
The measurements and analysis of data have been
delayed partially because of unexpected problems with
the multiple sample system and partially because of the
complexity of the spectra.

The measurements on four samples from the graphite
moderator element 1184-C-19 removed from the MSRE
after shutdown are partially complete. They give
spectra qualitatively similar to that reported previously

14. MSR Program Semiannu. Progr. Rep. Feb. 28, 1971,
ORNL-4676, pp. 143—45.
108

Table 11.4. Concentrations of several fission products on the surfaces of Hastelloy N
compared with the total inventory

. Depth of

Sample location penetration — 13C40ncentra:i:5n of nuclicllz :ompared l\zifith inventogry

(mils) Te Cs °Sb Ru Ru SN Prc
Control rod thimble (bottom) 2.4 0.43 0.84 0.85 0.40 0.13 0.37 0.32
Control rod thimble (middle) 0.1 0.14 0.24 0.35 . 0.15 0.11 0.26 1 0.19
Surveillance specimen 3.5 0.001 0.35 1.04 . 0.006 0.087 0.006 0.30
Mist shield outside, liquid 6.0 0.23 0.035 0.74 0.069 lO.iO 0.067 0.19
Heat exchanger shell 42 035 0017 068 0027 0.5 0.085 027
Heat exchanger tube 4.3 0.67 0.006 1.13 0.028 0.14 0.070 0.31

Table 11.5. Concentration in weight percént of the elements on the surface
of MSRE graphite stringer at the reactor midplane

Mass

Maximum penetration depth (A)

Element
~2.5 ~25 800 900 2000 3000 4000
Fission products detected
87 Rb 0.0018
89 Y 0.0044
90 - Sr 0.0067
Natural Mo 0.87 0.58 0.20 0.03 0.02 0.1l 0.07
Fission Mo 3.6 2.4 0.59 0.09 0.26 0.30 0.24
93 Nb 0.003 0.002 0.001
99 Tc¢ 0.57 0.39 0.09 0.02 0.05 0.05 0.04
101,102,106 Ru 1.4 095 0.19 0.04 0.11 0.09 0.09
103 Rh 0.34 0.22 0.055 0.01 0.03 0.02 0.02
110 Pd 0.11 0.053 0.01 0.02 0.015 0.01
109 Ag 0.0073 ' ’ 0.003
114 Cd 0.0046 - 0.015 0.007
125 Sb 0.01 0.0057 0.005 0.004
130 Te 0.41 0.30 0.046 0.009 0.02 0.02 0.01
133 La 0.004 0.006 0.003 0.002
138 Ba 0.005 ' :
140 Ce 0.007 0.01 0.006 0.005
141 Pr 0.003 0.003 0.003
144 ‘Nd 0.01 0.01 0.006
147 Pm 0.0008
Nonfission elements detected
Ni 0.9
Fe .03
Cr 0.05 '
Mn 0.005

v!
109

Y-101661

foo0m.

0.035 INCHES
00X

to030 ™.

P

Fig. 11.23. Type 316 stainless steel exposed to LiF-BeF containing MoF for 48 hr at 650°C.

for MSRE sample 6 (ref. 15) but differing considerably
in detail. For example, on one sample, MSRE 7, the
amounts of thodium and technetium are at least twice
that on MSRE 6. (Rhodium was found to be present on
MSRE 6 but had not been identified in the previous
report.)

Table 11.6 shows some sputter profile data from
MSRE 7. This sample came from the center of the flow
channel near the reactor midplane. The concentration
of fission products is only given in units relative to the
carbon. The determination of the actual concentrations
of various elements present must await comparison with
standards. It appears, however, that technetium, rho-

15. Ibid., pp. 145-47.

dium, and molybdenum are all present in large per-
centages (>5%) and niobium in a smaller but detectable
amount. The analysis of ruthenium is difficult because
of interference from the carbon peaks and other fission
products, but there is reason to suspect that it may also
be present in relatively large amounts in that the carbon
peak shapes are somewhat distorted. A similar situation
exists for tellurium, whose primary peaks fall at the
same energies as the secondary oxygen peaks. Although
not included in the table for this reason, it appears that
significant amounts of tellurium are also present. The
large oxygen and nitrogen peaks may indicate that a
large portion of the fission products is present as oxides
and nitrides. Small amounts of iron and nickel have
been detected on MSRE sample 13, which is from
another part of the same moderator element.

110

Table 11.6. Auger electron intensity as a function of depth below the original surface
of sample MSRE 7 from a MSRE moderator element

Accumu_lated Estimated Technetium, Technetium L o . -
argon ion depth sulfur, molyb denur; Niobium . Rhodium Nitrogen Oxygen
bombardment (atom molybdenum (200 eV) (300 eV) (383 eV) (510 eV)
(x 10'5/cm?) layers) (148 eV) (182 V)
75 15 19 40 nd® 9 12 25
75 C15 na® 35 nd 8 14 na
116 23 16 55 2 7 12 23
150 - 30 26 50 nd 11 .17 26
225 45 33 55 2 7 8 18
525 - 105 15 5 nd nd 9
1160 230 13 2 "nd 9
2325 485 5 7 2 nd nd 3
4725 945 3 8 2 nd 1 3

2nd = not detected.

bna = not analyzed.

v!
12, Graphite StudiésQ

W. P. Eatherly

The purpose of the graphite studies is to develop
improved graphites suitable for use in molten-salt
reactors. The graphite in:these reactors will be exposed
to high neutron fluences and must maintain reasonable
dimensional stability and mechanical integrity. Further,
the graphite must have a fine pore texture that will
exclude not only the molten salt but also gaseous
fission products, notably '?° Xe.

The general irradiation of commercial graphites and
experimental samples obtained from the Y-12 Develop-
ment Division has spanned a broad range of raw
materials and fabrication techniques. It has been found
that these materials could be divided on the basis of
their geometric behavior under irradiation damage into
four classes: conventional, apparently binderless, hot-
worked, and black-based graphites. At the present time
it appears that only the binderless and black-based
materials offer the opportunity of significant improve-
ment. Our own fabrication studies have therefore been
based on these two approaches. The first ORNL-
fabricated ‘binderless” graphites have now been irradi-
ated to 1 X 1022 neutrons/cm?® (£ > 50 keV) inithe
HFIR and are stable to this level. These results are most
encouraging in supporting our general concepts as to
the microscopic physical phenomena controlling dam-
age. > ‘ _

In view .of the initial successes of the graphite
fabrication studies, we have recently decided to reduce

“effort. in this area pending further irradiation results.

This permits an increased effort on various techniques

111

for sealing graphite ‘against fission product gases. A
considerable portion of the ‘present progress report
centers accordingly on improved or alternative process-
ing techniques and ‘evaluation of the microstructure on
previously irradiated specifnens. The previous materials
were fabricated in fluid bed furnaces, a technique
involving complex gas reaction kinetics and limiting in
the size specimens that could be coated or impregnated
with pyrolytic carbon. For these reasons a new furnace
was constructed utilizing a fixed substrate, and initial
coating and impregnation runs are now under way.
" Additionally, a second look is being taken at im-
pregnation with frozen salt as a means of pore blockage.
Although there is reason to question whether the
desired diffusion rates and irradiation stability can be
obtained, the technique is so simple in appllcatlon to
justify at least a cursory exploration.

‘During this reporting period the apparatus to measure
thermal conductivities of HFIR-irradiated samples was

‘completed and the first. irradiation performed. A
‘thermal expansion apparatus which will permit accurate

determination of the linear expansion coefficient as a
function of temperature is almost completely ¢on-

'structed. This information is desired not only for

engineering purposes but also “because of its close
alliance to damage.

Progress continues on the theoretical interpretation of
damage clusters as seen in the electron microscope. The
strain field calculation has been completed for intersti-
tial aggregates, and diffraction effects for crystal di-
rections have been completed. At least qualitatively the
calculations are in excellent agreement with observed
effects.

12.1 THE IRRADIATION BEHAVIOR OF .
GRAPHITE AT 715°C

C.R.Kennedy  W.P. Eatherly

We have extensively studied the neutron irradiation
behavior of graphite at 715°C, primarily to determine
the incremental or decremental changes in life ex-
pectancy resulting from a wide variety of starting
materials and fabrication methods. Over 40 experi-
mental and commercial grades of graphite were irradi-
ated in the HFIR to fluence levels up to 4 X 10%?
neutrons/cm?® (>50 keV), invariably producing signifi-
cant volume expansions. The changes in linear di-
mensions and bulk density were the prime measures in
evaluating the materials.

The irradiation behavior of the graphites demon-
strated that all of the various grades could be separated
into four basic types of materials:

1. Conventional graphites. These graphites all consist
of materials made from calcined coke or graphite flour
as filler- bonded with either thermoplastic or thermo-
setting hydrocarbons. The general irradiation behavior
is characterized by an immediate densification followed
by a parabolic expansion, as shown in Fig. 12.1. The
isotropic materials have a lower tendency for densifi-
cation than the anisotropic graphites made with acicular
cokes, and both demonstrate that the maximum densifi-
cation varies inversely with the original density. A
considerable degree of anisotropy in the densification
process appears to depend on both the method of
fabrication and the crystalline structure of the filler
particles, which usually is reflected in their morphol-
ogy. The more isotropic graphites always densified
more in the extrusion direction for extruded grades and
normal to the forming force for the molded grades.

The linear dimensional changes depended upon the
anisotropy of the graphite, the crystallite size, and the
density changes. The growth rates at fluences where the

112

ORNL-DWG 71-6910

RY -12-31
8 : /
- BS
io RY-IZ-ZBf/%
z 4 y ’YMI:}
z ROB-C /
; j YM350-10
8 Axm-nx-—.y //
g 0 —\‘ L / /
5 N /
i : % 2020, 9950,
= T BY-1 ’ ’
: /
o
> NF /!
4 SM-ITX 9972 | A
ATJ-SN\.__/7
AGOT =
\\_/425_
y | _
0 10 20 {x102")

FLUENCE (neutrons/cm?/sec)

Fig. 12.1. Volume changes of conventional graphites, 715°C.

Conventional graphites all have virtually the same
lifetime expectancy with no apparent effect of molding,
extrusion, type of filler, Thermax additions, type of
binder, or the type of impregnant.” The choice of
graphites within this category would depend upon the
necessity of isotropy and other properties, such as
thermal conductivity, mechanical properties, and coeffi-

~cient of thermal expansion.

density change was constant agreed with the "Reus'sm'

one-dimensional constant stress model.' The effect of
crystallite size on the growth rate was in excellent
agreement with that found by Bokros.?

1. A. Reuss, Z. Angew. Math. Mech. 9, 49, (1929).’

2. 1. C. Bokros, G. L. Guthrie, and A. S. Schwartz, Carbon 6,

55 (1968).

2. Apparently binderless graphites. The raw materials
and the fabrication procedures for most of these
graphites are proprietary. They are characterized by an
apparently binderless single-phase microstructure with a
fairly fine optical domain size. These graphites are all
very isotropic, have a high coefficient of thermal
expansion, and generally have a lower crystalhte size
than the conventional materials.

Their irradiation behavior shown in Fig. 12.2 is

characterized by a delay or an elimination of the |

densification process, yielding a more dimensionally
stable graphite with about twice the lifetime of the
conventional graphites. The linear dimensional changes
are isotropic, reflecting only the bulk density changes.
Thus, the individual crystallite growth rates cannot be

calculated directly from extrapolated orientation de-

pendence.
',

113

CORNL-DWG 71-6915

Y
/ F S /
8 f // /I.
&
// / AXF |
5. 1374V
~ /A/y r
Y
g . ZAH%G‘I /
s %o gf e
£ 0 23 /‘ /
o —_—/,
e a—b— a2
7 a AXF-UFG
o AXM
_ AP-03
4 oHL 18
& H-395
-8
0 10 20 30 (x102"

FLUENCE (neutrons/cm2/sec)

Fig. 12.2, The volume changes of apparently binderless
graphites,

3. Hot-worked and mesophase graphites. These
graphites by virtue of hot working are highly aniso-
tropic. Although the original densities may not be very
high, the void structure for densification has apparently
been eliminated by the hot working. None of these
materials densified under irradiation, as shown in Fig.
12.3. The volume remained constant for a short period
and then began to expand parabolically with fluence.
The life expectancy of these grades is similar to that of
the conventional grades; however, the linear di-
mensional changes are quite large due to the anisotropy
and would not be acceptable for most designs. The
linear dimensional behavior, excluding density changes,
can be represented by the Reuss model.

4. Carbon-black grades. These are grades in which all

the filler is a carbon black. They are characterized by a
structure of randomly oriented, ultrafine optical do-
mains with no anisotropy and a fairly small crystallite
size. The coefficient of thermal expansion is generally
large, the thermal conductivity is low, and the mechani-
cal properties are fairly good for the generally low
bulk density. The densification due to irradiation is
quite rapid, as shown in Fig. 12.4, possibly reflecting
high negative crystalline growth rates. The expansion
after densification, unlike that for all other materials, is
linear with fluence and seems to be independent of
grade and heat treatment. The initial increase in
density, however, depends upon final heat treatment
temperature, as shown in Fig. 12.5, and there is some

LENGTH OR VOLUME CHANGE [100 In (1'+Av/v0)} OR [100 In (1,+A.e/_eo)]'

ORNL-~DWG 71- 6909

‘ /
15
A SPECIMEN LOST
SYMMETRY
PARALLEL
H 413
_ H 414
,/ R
PARALLEL *
HOT WORKED / /|
l V °
YM350-1V
I/YMI—I';‘a s
5
./ {
[¢]
@ [o]
/ FE . './
VOLUME CHANGES o/
NORMAL
0 4 H 413
?‘Q.
e a H414
ed H 415
'E(\’B\ \_/
S N i
NORMAL 8%
HOT WORKED \\9 YMI=13
$ yM350- 11
-5 — 21
0 t0 20 (x 10%h

FLUENCE (neutrons/cm?/sec)

Fig. 12.3. Dimensional behavior of hot-worked and meso-
phase graphites, 715°C.

evidence that the densification is binder-dependent as
well. These materials are all isotropic, and the linear
dimensional changes are again a result only of bulk
density changes. The lifetime expectancy of these
materials is equal to or -greater than that of the
apparently binderless grades, such as represented by
Poco-AXF in Fig. 12.4.

It appears, therefore, that graphite with lmproved
lifetimes may be developed from either the apparently
binderless graphites, carbon-black grades, or from com-
binations of the two. The disadvantage of the low
thermal conductivity of the carbon-black grades can be
moderated by combinations of green coke and carbon-
black filler materials. Also, as discussed in an earlier
report,® the addition of carbon blacks to green filler

5

3, C. R. Kennedy and W. P. Eatherly, MSR Progfam

_ Semiannu, Progr. Rep. Feb. 28, 1971, ORNL-4676, pp.

169-70.
ORNL-DWG 71-6916

. &
2 -
> T-V a
=~ B - .
S . A /\ i
{ F b . - /a'
= 3 T-1TX
< // = ///
=] o .
'8'_ . Va// SA-45 _v. %OCO-AXF
§ o T M
2 o o '
< . 2 _____-g—" [}
5 E’_ 9 . / /A
= " = “A-680
2 A /
Y o_a A Pt
g/i/u
0 10 20 30 (102h)

FLUENCE (neutrons/cm2/sec)

Fig. 12.4. The effect of irradiation on various carbon-black
grades and on Poco-AXF at 715°C.

ORNL-DWG 71-6918

A
— ’ Y &
&£ _~"A-68I
>~ 4 A 2800°Cc —|
g / &
+ A
‘E’ 4. a
o s
(o)
E Lo | |1 | Lt
2 of— 2600%C -
E o] . g::h—»i. / .
o 2200°C -6—8—T—38 / a
E :1’4/8' -
2 2000°C -8 /“’
1w a
- -4 | ‘%/Z}’

1500°ca——?fi 2

o] 10 20 30
FLUENCE (neulrons/cmalssc)

(x1021)

Fig. 12.5. The effect of final heat treatment on the dimen-
sional stability of a carbon-black grade irradiated at 715°C.

produces a graphite with a more contiguous structure.
We are presently investigating these types of raw
material systems and are now irradiating grades with
representative structures for evaluation.

Three ORNL materials have received their first

irradiation. These were all made using green Robinson |

coke as the filler with and without Thermax additions,
and with coal-tar pitches 15V and 350 as binders. The
densities were reasonable; however, the structures con-
tained a fine network of lamellar voids normal to the
molding direction. This structural fault has been elimi-
nated in newer moldings. The results to maximum

114

2 sec™! as seen in

fluences of 1 X 10?2 neutrons cm”~
Fig. 12.6, indicated that the behavior is quite similar to : )

the apparently binderless grades exhibiting very small : -
isotropic expansion. There appears to be no immediate
influence of the binder or the addition of up to 40%

Thermax to the filler material.

wi

oL

ORNL-DWG 71-12644

° 8]
1 — :
- S ey LPOCO AXF
2o g—"] A yd
: fl A
©
2 ' e
T [ e —
Q
E-2— 8, AD/DO
% o . 122 ALL ROBINSON COKE -=15V BINDER
J=3— & A 116 60% ROBINSON COKE,
40% THERMAX —45V BINDER
—4)— a L] 157 60% ROBINSON COKE,
40% THERMAX-350 BINDER
% I N I N N N IO O B

0 2 4 8 8 10 12 14 16 8 20 22x10%

FLUENCE (neutrons/cmz/sec)

Fig. 12.6. The results of the first irradiation of ORNL
graphites, 715°C.

12.2 PROCUREMENT OF VARIOUS GRADES OF
CARBON AND GRAPHITE "

W_H. Cook W.P. Eatherly

The various grades of carbon and graphite that we
have received since January 1, 1971, are summarized in
Table 12.1. One grade, A676, has an apparent density
of 1.87 g/cm?®, but the remainder have apparent
densities from 1.79 to 1.47 g/cm?. On this basis, they
are not outstanding; however, they generally represent
some special parameter(s). :

The first four grades listed in Table 12.1 represent
studies primarily on isotropic fillers bonded with
synthesized, experimental binders of acenaphthylene
and isotruxene.

The graphite-fiber-reinforced-grade PTE is not ex-
pected to have any potential as a moderator material,
but it may be useful as a high-temperature structural
material in regions of low flux. .

Grade A680 is the low-fired lampblack-base material
that we had requested to replenish our supply of this
material to continue our studies previously reported.*

o @

4. 0. B. Cavin, W. H. Cook, and J. L. Griffith, MSR Program
Semiannu, Progr. Rep. Feb. 28, 1971, ORNL4676,p. 171.
Table 12.1. A summary of the various types of carbon and graphite samples received since January 1, 1971

Total : Max. Apparent '
Grade N.O 0 _ weight Manufacturer Fabrication Filler Binder Impregnant firing density Dlmm?nsmns Remarks
pieces method temp. 3 (in) - .
(g} o (g/cm?)
(0
675-11-4 4 ) Extruded Low-fired ACN-  Butadiene ( A} Had radial cracks
Santa Maria- ITX? gas method
676-01-4 4 : Extruded  Low-fired ACN-  Butadiene ' 0.128 ID X Had radial cracks
' : Poco, PCD-OQ ITX gas method :
~ MEDD? ’ 2800 . 0.400 OD X
118 6 o Molded® Low-fired ITX? ) 1.7 ( 0.500 Lot No. 1 of filler
_ ~ Poco, PCD-0Q R
132 -6 . Molded®  Low-fired ITX : 1.79/ . Lot No. 3 of filler
/ Poco, PC5-0Q : \ / :
PTE 1 588 CPD4 e e Pitch® e 1.62 5.9diam X 0.76 thick  Graphite fiber reinforced
A680 1 1358) - ' , 1.47 0.86 X 5.66 X 11.58
Q322 | 1,394 Stackpolef e Carbon black Pitch® e 1.62 7.4 diam X 1.4 thick
A676 1 " 1,188 1.87 1.0X4.8x% 7.9
JA-S { 584 : 158  17%¢ x 3% x 4% Preliminary
Airco Speerf e Grade JM-15 e e 1 3 5
1 996 _ 1.67 176 % 3% x 8% Cracked
R18 3 11,477 . | 1.7% 5% diam x ~5% long
Calcined 5/ 3 1 -
R20 1 3,824 5% diam x ~5% 1
- 4 GLRCH Molded  Robinson Pitch® Pitch® 2700 4 ¢ . 0B clam A ™o i One
R22 1 3,614 Cokel : 1.68 = 5% diam X ~5'%% long
R24 3. 10,769 , | 1.67 5% diam X ~5Y% long

. 9Materials Engineering Development Department of the Y 12'Plant, Qak Ridge, Tennessee.

BACN = Acenaphthylene and ITX = isotruxene.

¢Molded at 1400°C and 1600 psi.

@Parma Technical Center, Carbon Products Division of the Union Carbide Corporatlon Parma, Ohio,

€Not available or additional details not available. :

f Stackpole Carbon Company, St. Marys, Pennsylvania.

&Airco Speer Carbon Products, St. Marys, Pennsylvania.

MGreat Lakes Research Corporation, Elizabethton, Tennessee.

!Made from the Robinson Raw Coke manufactured by the Carbon Products Division of Union Carbide Corporation, Lawrenceburg, Tennessee, under AEC Contract
AT(29-2)-1955.

/Data supplied by the manufacturer.

SIT
Grade Q322 is similar to the previously described
graphitized lampblack grade A671. Grade Q322 is a
short section from a 7%-in.-diam cylinder and was
offered as physical evidence that base stock sizes as
required by the MSBE can be fabricated using a
lampblack filler. _

Grade A676 is grade A681, the graphitized lamp-
black-base body we have been studying, that has been
impregnated to change the nominal apparent densities
from 1.60 to 1.87 g/em?.

The grade JA-S is in the early stages of development
and uses an isotropic to near-isotropic filler designated
as grade IM-15. :

The series of grades R18, R20, and R24 represents
~ the start of a study with varying amounts of pitch
- binder with an isotropic filler made from calcined
Robinson (air-blown) raw coke. The latter is a product
from a sizable production (24 tons) from a pilot-scale
coker. . :

Arrangements have been essentially completed to
obtain specimens of carbon materials from Poco Graph-
ite, Inc., spanning the temperature range 1400 to
2500°C. The low-fired materials would be grade AXF
precursor fired to 1400, 1800, 2000, 2200, and
2500°C.. We plan to exténd the temperature range to
3000°C by refiring a Poco-supplied stock piece (fired
to 2500°C) to 3000°C and machining specimens from
it. Since these specimens represent material at inter-
mediate stages of a production process, certain agree-
ments are being arranged to protect their proprietary
position.

The purpose of the experiment is to take the most
stable graphite known to us and examine the effect of
heat treatment on its behavior. Both the plasticity and
crystalline size are sharp functions of heat treatment
temperatures, and both are believed to be important in
determining the lifetime of graphite under irradiation.
The resulting data, aside from their intrinsic interest,
will also afford a qualitative check of our proposed
radiation damage model. :

We have received a graphite flour, grade 1074 (~23.0
kg), from the Great Lakes Carbon Corporation. This
material is slightly anisotropic. It is being evaluated for
use in both the Molten-Salt Reactor and High-
Temperature Gas-Cooled Reactor Programs.

12.3 X-RAY STUDIES
O. B. Cavin

[sotropic polycrystalline graphites continue to be the’

most irradiation stable; therefore, the x-ray-determined
isotropy parameters on new materials continue to be of

interest. We continue to determine these parameters on
newly fabricated bodies made at both ORNL and the
Y-12 installations, as well as potential new grades that -
are received from the manufacturers. _ :

Three samples of mesophase graphite were received
for evaluation. The x-ray-determined crystalline pre-
ferred orientation parameters are shown in Table 12.2.
It was thought that the mesophase bodies should be
nearly isotropic; however, this is not the case.

The ACF4Q is a low-fired Poco grade that has not
been above 1400°C and is the starting material for the
AXF grade. Three sphere samples were cut with their
axes mutually perpendicular, and the results are shown
in Table 12.2. The sum of the three independently
determined R, values is 1.997, which is very near the
theoretical value of 2.0. This material is very isotropic.

A block of grade JA-5 graphite was received from
Airco Speer Carbon Products for property evaluation.
Three sphere-type samples for anistropy determinations
were cut from this block with their axes mutually
perpendicular. From these three samples we then

Table 12.2, X-ray anisotropy parameters

Sample number R R®> BAF
H-413 0.640 0.680 1.126
H414 0.643 0.678 1.109
H415 0.636 0.682 1.145
ACF4Q No. 22 0.665 0.667 1.014
ACF-4Q No. 42 0.667 0.666 1.014
ACF4Q No. 72 0.665 0.668 1.014
JA-5,S-1 0.663 0.668 1.016 -
JA-5,8-2 0.668 0.666 1.012
JA-5,8-3 0.670 0.665 1.031
JA-5,Y-12 0.659 0.670 1.034
2020A 0.681 0.659 1.135
2020B 0.634 0.683 1.155
2020C 0.674 0.663 1.067
HS-82 No. 26 0.668 0.666 1.012
HS-82 No. 66 0.663 0.669 1.017
HS-82 No. 76 0.678 0.661 1.106
226A 0.662 0.669 1.021
1226B 0.666 0.667 1.006
1226C 0.655 0.673 1.084
2044A 0.658 0.671 1.082
2044B 0.643 0.679 1.248
2044C 0.681 0.660 1.067
HS-17A1 0.656 0.672 1.049
HS-17A2 0.676 0.662 1.086
HS-17B 0.696 0.652 1.289
HS-17C 0.684 0.658 1.165

“R |, refers to the sphere stem axis.
br) =1 -Ry/2.
e

determined independently the anisotropy value in each
of the three directions. A sample of the same grade but

a different block was submitted by L. G. Overholser of |
Y-12. The results obtained from these samples are listed

in Table 12.2. The anisotropy values indicate that the
material was fabricated by molding and that the axis of
sample S-1 is parallel with the fabrication axis. The axis
of Overholser’s sample was also parallel with the
fabrication axis but had a greater degree of anisotropy.
This indicates either a nonuniformity in the graphite or
it came from a position in the block near the end where
die effects, if any, would be greatest. There was some
question as to the maximum temperature to which JA-5
had been; therefore, all four samples were fired at
2800°C for 1 hr. The anisotropy. values obtained from
the fired samples are within experimental error of those
in Table 12.2.

Each of the other grades of graphite reported in Table
12.2 has been around for some time, but anisotropy
values had not been determined.. These samples with
axes mutually perpendicular were used to determine the
anisotropy values independently in the three directions.
One of the samples of HS-17A had a high density, and

so a sphere was machined on each end of a l-in.-long.

sample. The sample marked A2 was the sphere having
the higher density and has a much different anisotropy
value than the Al sample. This apparently came from
near the surface where impregnation effects would be
most pronounced.

The sum of the three independently determined R,
values for each grade is very near the theoretical value
of 2.0, but there is some variation due to the
inhomogeneities of the graphite.

We are also studying the effects of firing different
graphites and filler particles to successively higher
temperatures. These will be reported in the near future.

12.4 THERMAL PROPERTY TESTING
J.P.Moore © D.L.McElroy T.G. Kollie

A guarded longitudinal heat-flow technique for
measuring thermal conductivity A, electrical resistivity
p, and thermopower S on small (1 X 7 cm) rods from
300 to 1000°K was tested with an Armco iron
standard. An error analysis indicates a most probable
determinate uncertainty in A of £1.6%. The A results on
the standard agree with the assumed values to this
uncertainty, except at 970°K where a 1 or 3%
difference is noted depending on the method of
calculation. The X of the standard was verified from 80
to 400°K in a low-temperature guarded longitudinal

117

ORNL- DWG 74-9971

144 ‘ l
AXF—50Q
e 51124 z
1.3 — P <8 |
o pti14 _ //'
H-337 %/
1.2 A p 784 - s
s p775 P
o p 761 /,J/ L
1.4 *, ¢ COMPARATIVE S Ao
|>/’ /////flf e
= z, A
: / /
ry
= 1 21
= 1.0 /l/ / b
@ ./
k=] -
] g
I ’/ A
209 i o
; A 7
- L-® .
0.8 01’ 134
ar
. 0//
0.7 7 ’d
¢ * A//
¢
0.6 ',./‘8/
0.5
300 400 500 600 700 800 900 100C

TEMPERATURE {X)

Fig. 12.7. The thermal resistivity (?\_1) of unirradiated
AXF-5Q and H-337 graphites as determined in a guarded
longitudinal heat-flow apparatus. Comparative heat-flow meas-
urements on other H-337 samples are shown.

apparatus,” and the two approaches agree to *0.2% in
the temperature range of overlap.®

Measurements of A and p were performed on unirra-
diated AXF-5Q and H-337 graphite to 1000°K in this
apparatus. The p and A values decrease with increasing
temperature. Figure 12.7 indicates a linear dependence
of A~ on temperature above S00°K for these graphites.

Three samples of each of these graphites were
irradiated in HFIR for two cycles at 550, 650, and
750°C. These samples are ready for remeasurement of A
and p from room temperature to just below the relevant
irradiation temperatures. ,

Assembly is 75% complete on a 1200°K quartz
dilatometer to measure the temperature dependence of
the temperature coefficient of thermal expansion of
these graphites and the influence of irradiation on this
property. The apparatus is being interfaced with a
computer-operated data-acquisition system to control
the experiment and to obtain the length, temperature,
and electrical resistivity of the specimen.

5. M. J. Laubitz and D, L. McEloy, Metrologia 7, 1-15
(1971). A

6. J. P. Moore and D. L. McElroy, Proceedings XI Inter-
national Conference on Thermal Conductivity, Albuquerque,
New Mexico, September 28 —October 1, 1971.
12.5 HELIUM PERMEABILITY MEASUREMENTS
ON VARIOUS GRADES OF GRAPHITE

W.H. Cook J. L. Griffith

Table 12.3 is a summary of the permeability coeffi-
cients for helium determined for a few of the grades of
graphites involved in the irradiation studies. Bulk

118

density values have been included  for comparison
purposes. These grades of graphite are being studied for
their unique starting materials and/or fabrication tech-

~niques used. The permeability determinations are part
_of the characterizing data that are measured on ma-

terials in the irradiation studies. Since none of these is
optimized for the Molten-Salt Breeder Reactor (MSBR)

Table 12.3. Helium permeability parameters of various grades of graphite?

Bulk Bod Kod Kye at 1 atm?
Grade Source? Spec. No. Orientation® density (cm?) (cm) (cm?/sec)
(g/fcm3) X 10712 x 1077 A X 1072
SA-45¢ CPD 102 AG 1.54 354 3.75 154
A-680F SCC 111 wG 149 332 2.24 123 -
A-681¢ SCC 111 WG 1.61 291 1.58 101
H-395 GLCC 51,62 . 1.78(2) 10.3(2) 3.96(2) 9.32(2)
HL-18 ACSP 82 AG 1.87 9.94 3.67 9.22
WG 1.86(2) 8.66(2) 4.43(2) 10.0(2)
ATISG CPD 41, 46 AG 1.86(20 14.8(2) 2.57(NH 8.10(7)
WG 1.81(2) 25.7(3) 4.28(3) - 17.7(3)
P-03 PCC 51,61 1.83(2) 8.34(2) 3.04(2) 7.25(2)
9948 ASCP 101,102 WG 1.91(2) 7.99(6) 3.75(6) 5.20(6)
HS-82 SCC 101 AG 1.71 1.09 0.586 2.27
' 82 WG 1.71 241 1.25 2.72
ATIS CPD 43,45 ~AG 1.85(2) 0.256(7) 0.098(7) 2.30(7)
- WG 1.83(2) 4.33(9) 0411(9) 2.36(9)
WG 1.85 -0.508(3) 0.171(3) 042(3)
AXZ-5Q Poco 101, 111 1.55(2) 9.86(2) 9.55(2) 18.5(2)
AXM-5Q Poco 101,111 1.75(2) 3.01(2) 2.78(2) 4.96(2)
AXF-5Q Poco 82,112 1.83(2) 1.21(2) 1.10(2) 2.17(2)
AXF-5QBG Poco 832 1.83 0.472 T 0.499 0.962
831 1.90 0.0511 0.0885 0.162
2 1.90 0.0438 0.0685 0.123
1 1.90 0.0378 0.0655 0.120
5 cyl¥ 1.94(5) 0.0315(5) 0.0331(5) 0.064(5)
81 1.92 0.0066 0.0096 0.018

2The numbers enclosed in pareni:heses indicate the number of values averaged.
bASCP: Airco Speer Carbon Products, a subsidiary of Air Reduction Company, Inc.
CPD: Carbon Products Division of the Union Carbide Corporation. -

GLCC: Great Lakes Carbon Corporation.
- PCC: Pure Carbon Company.

Poco: Poco Graphite Inc.
- §CC: Stackpole Carbon Company.

€AG: Across grain, perpendicular to the general # axis orientation.

WG: With grain, pargllel with the general ¢ axis orientation.
UK e = Boln) py+ % KoV .

where Ky = permeability coefficient for helium at room temperature, 28°C (cm'2 [sec),

B¢ = viscous permeability (cm?),
n = viscosity of gas, He (poise),

{p) = average pressure across specimen (dynes/cm?),

Kg = slip coefficient (cm),

V' = avérage molecular velocity of gas, He (cm? /sec).

€A lampblack-based material fired to 2800°C or higher.
YA lampblack-based material fired to ~1000°C.

& Average of five hollow cylinders (nominally 0.40 in. OD X 0.125 in.ID X 1.5 in. long) from the same block:

@

'Y

[
i

LK}

L]

119

ORNL-DWG 7T1-13690

(x107") N
;_-’/2 uim MEAN PRESSURE
N ! e —ed—e—t-—{—AXZ-5Q
N Ep—— Ir--' -
I ‘
O 1
(x1072) . I
it | 4 —axm-50Q
6 i R — t
F —ETTT T |
° 4 & ; AXF-5Q
o 1 e
& || AT
L 2 T
> |
- |
- Y :
an]
< (x107%) ; — x
AXF-5QBG — B32
s 1
E | L=
L I
a 1 L
0 R et
g ,| I+
] -1 :
L 8 .
T I
i
6 L
1
|
4 ,
|
i
3 2
2t == —— e
=T T - N Ep— —CYL.{5)
0 e e T oy s ——q—--81

O 2 4 6 8 10 12 14 16 18 (x10D)
MEAN PRESSURE (dynes/cm?)

Fig. 12.8. Helium permeability vs mean pressure for Poco
graphite grades AXZ-5Q, AXF-5Q, and AXF-5QBG.

applications, it is not surprising that their permeabilities
Ky, are relatively high. These were determined on
l-in.-diam disks from 0.1 to 0.5 in. thick that were
machined with specific orientations from their starting
stock. Grades HL-18, ATJS-G, HS-82, ATIJS, and
AXF-5QBG illustrate how the permeability value varies
with orientation and specimen location.

The isotropic Poco grades have been given more
attention because the AXF series have shown the most
resistance to fast neutron damage at 715°C. To more
clearly show the permeability variations from grade to
grade and within a grade, the helium permeabilities K,
are plotted vs mean pressures in Fig. 12.8 for grades
AXZ-5Q, AXM-5Q, AXF-5Q, and AXF-5QBG. The first
three grades are standard materials that represent
different ranges of bulk densities, and AXF-5QBG is an
impregnated, graphitized version of AXF-5Q. All were
fabricated from fillers having a maximum particle size
of 0.001 in. (ref. 7). In Fig. 12.8 and from Table 12.3 it
is interesting to note that grade AXZ-5Q is more than

7. Data taken from comparison chart issued by Poco Graph-
ite, Inc., Decatur, Texas.

an order of magnitude more permeable to helium than
grades AXM-5Q and AXF-5Q. The latter two_are not
much different from each other based on these few
measurements.

Data for grade AXF-5QBG for several specimens were
plotted to show the effect of the impregnation, and
that even though the material is relatively uniform, like
all grades of graphite, it can vary appreciably in
localized zones. The latter is illustrated in Fig. 12.8 by
the results obtained on specimens 831 and 832 that
were machined from the same block. The impregnation
appears to reduce the permeability of grade AXF-5Q by
one to two orders of magnitude.

12.6 REDUCTION OF GRAPHITE PERMEABILITY
BY PYROLYTIC CARBON SEALING

C.B.Pollock  W.P. Eatherly

Graphite to be used in the core of an MSBR must be
able to exclude fluoride salts and gaseous fission
products. We have been studying techniques to seal the
surface of the graphite with pyrolytic carbon. Two
techniques have been used: a vacuum-pulse impregna-
tion technique in which surface pores of the graphite
are closed by plugging them with pyrolytic carbon and
a coating procedure in which a continuous surface
coating of impermeable pyrolytic carbon is deposited
on the graphite. Both techniques have been successful
in sealing commercially available graphite suitable for
use in the core of an MSBR.

In conjunction with the fabrication program a con-
comitant irradiation testing program is being conducted
in which graphite samples sealed with pyrolytic carbon

-are subjected to MSBR conditions of neutron fluence

and temperature. A HFIR irradiation experiment con-
taining a number of coated samples and impregnated
samples was recently completed. This experiment con-
tained 10 graphite samples coated with pyrolytic
carbon and 12 graphite samples that were impregnated
with pyrolytic carbon.

Turning first to the coated samples, four of these have
been cycled twice through HFIR and now have a total
neutron dose of 2.4 X 10?2 neutrons/cm? (£ > 50
MeV) at a temperature of 700°C. Three of the samples
appeared to be intact except for some small cracks on
their ends, but independent of this, the helium perme-
abilities went from less than 1072 to greater than 1072
cm?/sec. The remaining sample was extensively cracked
over its entire surface. A second set of coated samples
has been cycled through HFIR for the first time and
received total neutron doses of up to 1.3 X 10?2
neutrons/cm? (£ > 50 MeV) at 700°C. Five of these
samples cracked badly and had helium permeabilities of
greater than 1072 cm?/sec. However, one of the
samples appeared to be unaffected by the experiment.

The only significant difference between these two sets
of coated samples is the coating thickness. The first
group had coating thicknesses ranging from 4 to 6 mils;
the second set had coating thicknesses of 2 to 3 mils.
The one sample of the second set that survived had a
coating that was 3 mils thick. Coating thickness and
microstructure are the significant variables, but further
work will be required to determine if satisfactory
coatings can be devised.

The results on the 12 pyrolytically impregnated
samples are given in Table 12.4. Six of these have been
recycled in HFIR to fluences of up to 3.7 X 10%2%; six
new samples have received only their first irradiation up
to 1.2 X 10*2. The new samples appear to be
significantly better than previous samples irradiated to
similar fluences. The pore size distribution of these
samples will be examined by mercury porosimetry. The
amount of pyrolytic carbon required for sealing can be
controlled by varying process parameters. Helium per-
meabilities of less than 107® cm?/sec have been
obtained in graphite samples whose weight increase was
less than 4% with a processing time of only 2 hr at
750°C. Graphite samples sealed in this manner should
behave more like unimpregnated base stock graphite. A
number of samples containing the minimum amount of
carbon needed for sealing has been prepared for the
next experiment in an attempt to control the neutron-
induced expansion of the graphite. :

As noted in Table 12.4, one of the samples has
increased its length by 11%, or substantially more than

Table 12.4. Impregnated samples

Permeability Permeability Fluence

Sa;nple before after (neutrons/cm?) Aé/ L
0 (cm?/sec) (cm?/sec) (X 1021 (%)

1231 23x1077 s51x107° 12.11

1208 13x107°7 22x107° 8.76

1236 22%x10° 1.0x107° 7.82

211 15x107°  62x%x1077 12.45

HR20 1x107° 23x1078 5.59

HR12 1x10® 65x10°° 11.35

1205 44x107!% 72x107° 17.06 0.7

1216 - 59x107'° 98x107° " 21.19 .22

HL32 16x10% 91x107° 20.60 0.8

1182 17x107% 15x1072 31.24 6.0

1181 5.1x107° 9.0x1072 36.97 11.0

1163 7.0x107° 29x10™* 2567 2.0

would be expected from the unimpregnated base stock
graphite at similar fluences.

Preliminary experiments have demonstrated the feasi-
bility of coating graphite with pyrolytic carbon.

Samples have been prepared in fluid bed furnaces.

designed for coating spherical fuel particles. The small
graphite samples were suspended in a bed of inert
particles that were levitated by a flow of helium. Then
the gaseous hydrocarbon was infected and cracked at
various temperatures, and the graphite sample was
coated in the same manner as the spherical particles.
Dense isotropic coatings of pyrolytic carbon tightly
bonded to the samples were obtained in this fashion;
however, this would clearly be a difficult technique to
use on large graphite tubes or cylinders.

A carbon resistance furnace has been built for coating
experiments with larger graphite samples. This furnace
is similar to conventional fluid bed coating furnaces but
will permit the investigation of depositing dense iso-
tropic pyrolytic carbon on a fixed substrate by altering
such variables as gas flow, gas mixture, and bed
condition. The furnace is complete, and samples are
being prepared for coating. '

12.7 THE MAGNIFIED TOPOGRAPHY
OF GRAPHITE SEALED WITH
PYROLYTIC CARBON

W. H. Cook

We have begun a detailed characterization of the
graphite grades sealed with pyrocarbon. The objectives
are to learn more about the pyrocarbon-sealing tech-
niques and to determine what produces a pyrocarbon
seal that has the maximum resistance to damage by fast
neutrons.

Parts of the early phases of this work are evaluations
of the various techniques available for characterization
of the pyrocarbon sealants. The scanning electron
microscope (SEM) is one of the potential tools for this
work. A preliminary examination of the surfaces of

unirradiated grade AXF specimens of graphite ma- -

chined in the shape of HFIR specimens, 0.128 in. ID by
0.400 in. OD by 0.500 in. long, has been made with an
SEM.? One specimen was as-machined, and the other
had been impregnated with pyrocarbon from butadiene
at 750°C and heat treated at 1650°C.

The examination of the as-machined specimen shown
in Fig. 12.92 suggests a tearing of the filler particles
more than a cutting operation. At this magnification of
2000X, the surfaces appear to be mosaic of irregular-
shaped, equiaxed particles. In Fig. 12.9b, at the same
magnification, the surfaces of the pyrocarbon-sealed

}a
121

PHOTO 2658-71

Fig. 12.9. Scanning electron microscope photomicrographs of the surfaces of grade AXF graphite. (a) As machined as a HFIR
irradiation specimen. (5) A like specimen that was impregnated with pyrocarbon from butadiene at 750°C and heat treated at
1650°C. 2000 . Note: The surfaces of the specimens were not coated with a film of metal to reduce charging-up effects.

specimen showed that pyrocarbon covered the exposed
filler particles so that they now had sphere-like shapes
and appeared to be a surface composed of closely
stacked, rough spherules. Since the sealing with the
pyrocarbon was designed to be an impregnation rather
than a surface coating, the latter surface appearance is
not illogical.

There are some charging-up effects observable in Fig.
12.9 that might have been eliminated or reduced by
coating the specimens with a thin film of metal prior to
their examination with the SEM. We did not do this
because such an operation would be impractical in our
future work with irradiated specimens that we periodi-
cally examine and recycle for additional radiation.

This SEM work indicates promise; however, addi-
tional work with it remains to be done in order to
evaluate it and apply it to our objectives.

12.8 REDUCTION OF PERMEABILITY
BY FLUID IMPREGNATION

W. H. Cook

It has been proposed® that unfueled salt impregnated
into the accessible pores of graphite may be a practical
way of reducing its gas permeability. Preliminary tests
with an analog impregnation with bismuth and unfueled
salt impregnation indicate that helium permeability of a
permeable graphite can be reduced to the range of 107
cm?/sec. Additional studies are required to determine
the practical limits for reducing permeabilities by this
technique and to evaluate the effectiveness of such fluid

8. R. B. Evans I1I et al., Reactor Chemistry Division.

122

impregnations for reducing fission gas penetration into
the graphite. _

In an analog test using impregnation by bismuth
(which expands on freezing) instead of molten salt, the
helium permeability was decreased two to five orders of

_magnitude, depending on the grade of graphite used.”
The lowest permeability values were in the 107°
cm?[sec range. Apparently, there were enough <0.25-
um pore entrance diameters that were not impregnated
to prevent the permeability from attaining the desired
range of <1078 cm?/sec.

Prior to the early shutdown of the MSRE, we had
planned to determine the actual effect of an unfueled
salt impregnant on graphite from the standpoints of the
reduction of gas permeabilities and gaseous fission
products penetration into the graphite in the MSRE

environment. To do this, we impregnated evacuated .

specimens of grade AXF-5Q graphite with NaF-BeF,
(66-34 mole %) under a pressure of 200 psi at 705°C.
This should have caused pore entrances down to
approximately 0.5 um in diameter to be filled. Equip-
ment limitations for the required salt temperature
prevented our using enough pressure to penetrate pore
entrance diameters down to approximately 0.25 um as
was done in the analog test with the bismuth. The
permeability values for the bismuth- and the salt-
impregnated specimens are compared with their con-
trols in Table 12.5. The higher permeability of the

9. W. H. Cook, MSR Program Monthly Progr. Rep. September
1969, MSR-69-93, p. 33.

salt-impregnated specimen over the bismuth-impreg-
nated specimen is probably due to the shrinkage that
occurs in the salt when it freezes. This is supported by
the facts that the pore entrance diameters of the
unimpregnated grades AXF and AXF-5Q are concen-
trated at 1 u, and there is a larger viscous component of
gas flow for the specimien impregnated with the frozen
salt vs the one impregnated with frozen bismuth. The
freezing of the salt apparently makes available porosity
that has relatively large effective pore diameters com-
pared with those remaining in the bismuth-impregnated
specimen. The salt in the molten state should seal
against helium flow as well as’ the frozen bismuth.
Additional studies are required to determine (1) the
practical limits for reducing gas permeabilities by this
technique and (2) the diffusion rates of fuel salt
components into the impregnant(s). The final gas
permeability values attained will determine the effec-
tiveness of such impregnations for reducing fission gas
penetration into the graphite. Xenon-135, which has a
high cross section for thermal neutrons, is the fission gas

of primary concern. Uranium (or plutonium) and |

salt-soluble fission product migration into the impreg-
nants will be the primary concerns in the diffusion rate
studies.

12.9 FUNDAMENTAL STUDIES OF RADIATION
- DAMAGE MECHANISMS IN GRAPHITE

S.M.Ohr T.8S. Noggle

Irradiation of graphite single crystals at temperatures
of 300 to 600°C in a 200-kV electron microscope leads

Table 12.5. Permeability effects on a graphite impregnated with solidified bismuth or NaF-BeF,

Permeability parameters?

Graphite Specimen

Impregnation K.. at1atm
rade 2 He
g Number Type By (cm*) Ko (cm) (cm? [sec)
AXF 1 . Hollow cylinder None 3.30x 1072 1.94 x 10~ 4.10 X 1072
1 Cylinder Bismuth 549x 1077 2.15x 1071 3.76 X 107¢
AXF-5Q 5 Disk None 1.29 x 10712 1.22x 1077 2.39 X 1072
AXF-5Q 4 Disk NaF-BeF,? 1.45 x 10715 452%x 1073

2.47x 10710

2Kye = Bo/n) DO+ % KoV,
where Ky, = permeability coefficient for helium at room temperature, 28°C (cm? /sec),
Bg = viscous permeability (cm?),
n = viscosity of gas, He (poise),
- {p) = average pressure across specimen (dynes/cm?),
K = slip coefficient (cm),
V= average molecular velocity of gas, He (cm/sec).
bThe nominal composition is NaF-BeF, (66-34 mole %).
to the formation of visible spot-type damage structure
in relatively short times. Studies of the contrast of the
defect clusters for various diffraction conditions have
shown that the defects are dislocation loops of inter-
stitial type. In dark-field micrographs as in Fig. 12.10,
the damage clusters frequently appear as both black and
white spots in roughly equal numbers.' ® This black and
white contrast is found to be quite sensitive to the
diffraction condition, that is, the deviation from the
Bragg condition. Stereoscopic study of this structure
has shown that the black and white spots are present as
several alternating layers parallel to the surface of the
specimen. The behavior of this black-white structure

10. S. M. Ohr and T. S. Noggle, MSR Program Semiannu.

Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 174-76.

with variation of the diffraction condition has been
found to be consistent with and hence attributable to
the stacking fault present in the prismatic dislocation
loop formed by the clustering of interstitial atoms.
Figure 12.11z shows the intensity of the diffracted
beam as a function of the position of a stacking fault in
a 2300-A-thick graphite crystal. The oscillation of the
intensity with depth about the perfect crystal value
(dashed line) predicts three layers of dark (black) spots
and two layers of light (white) spots as has been
observed experimentally.

Other aspects of the contrast of these defect clusters,
although qualitatively consistent with the behavior
expected of interstitial-type dislocation loops, suggest
that the strain contrast is less than would be expected
on the basis of diffraction contrast theory employing
the displacement field derived from isotropic elasticity

Fig. 12.10. (1011) Dark-field micrograph of graphite irradiated at 400°C with 175-kV electrons. Interstitial-type prismatic
dislocation loops show black and white contrast as predicted by calculations shown in Fig. 12.11.60,000x .

124

(1011) S.F.

O

to/p

d, DEPTH OF THE LOOP FROM THE TOP SURFACE

- {a)

|

f
0O 0.5

5 10 0
|$4/2 DIFFRACTED INTENSITY ‘

ORNL-DWG 74-13691

S.F. AND DISL.
I

(1011)

0.5 10

Fig. 12.11. Computed depth dependence of the diffracted intensity for interstitial-type prismatic dislocation loops in a
2300-A-thick graphite specimen. Diffraction conditions and specimen thickness correspond to Fig. 12.10. (¢) Stacking fault contrast.
(b) Stacking fault and dislocation contrast. Dashed lines indicate perfect crystal diffracted intensity, shaded areas indicate depth
regions of dark (black) loops, and unshaded areas indicate regions of light (white) loops.

theory. In order to clarify this matter, the stress and
displacement fields for a circular prismatic dislocation
loop have been derived using anisotropic elasticity
theory. The results indicate that the overall effect of
crystal anisotropy is to reduce the stress and displace-
ment fields around the loop by about an order of
magnitude, except for the component of displacement
normal to the loop plane which is increased by about a
factor of 2. These results appear reasonable when one
considers the nature of chemical bonds in graphite,
which calls for a strong covalent bonding in the basal
plane compared with a weak van der Waals-type
bonding across the basal plane. When a dislocation loop
is created between basal planes, the crystal expands
more readily along the ¢ axis giving rise to large atomic
displacements in this direction. In the basal plane,
graphite is elastically too rigid to accommodate appre-
ciable displacement.

On the basis of the anisotropic elasticity theory of
atomic displacements around a dislocation loop, calcu-
lations of image contrast in the electron microscope

* have been carried out in the IBM-360 computer. It is

found that the crystal anisotropy does influence the
intensity and apparent size of the image of a dislocation
loop. Figure 12.115 shows the calculated diffracted
intensity at the exit surface of the specimen for the
central region of a loop as a'function of the position of
the loop in a specimen for diffraction conditions similar
to those present in Fig. 12.10. The variation of the
intensity relative to the perfect crystal value is similar
to that for the stacking fault alone (Fig. 12.11a),
indicating that the stacking fault is dominant in
determining the contrast for the conditions considered
here. Experiments and calculations are currently being
made to examine more closely the agreement between
observed and calculated contrast.

is
“r

13. Hastelloy N

H. E. McCoy

The search for a chemically modified composition of
Hastelloy N with improved resistance to irradiation
damage continues. The elements of primary concern are
Ti, Nb, Hf, Zr, Si, and C. We continue to approach the
problem by initially making small laboratory melts and
then procuring 50- to 100-lb commercial melts. This
gives us some feel for the problems that will be
encountered in scaleup of these ‘alloys to production
size (10,000 Ib or greater). This work involves mechan-
ical property studies on unirradiated and irradiated
samples.

Our compatibility programs are involved with the
corrosion of Hastelloy N in several fluoride salts and in
steam. The salt of primary concern is the new proposed
coolant salt, sodium fluoroborate. Several thermal
convection and two pump loops are committed to
studying corrosion in this salt. Steam corrosion work
continues at two facilities. Hastelloy N is currently
being exposed in the unstressed and stressed conditions.

13.1 ELECTRON MICROSCOPY STUDIES
R. E.Gehlbach S.W.Cook

Our studies of the effects of alloying additions on the
microstructures in modified Hastelloy N at elevated
temperatures have concentrated in three primary areas.
The first area consists of the effects of Hf, as a sole
addition or in combination with Ti and Nb, on carbide
precipitation. These studies have included variable
concentrations of C and Si and both laboratory and
small commercial melts. We have achieved a reasonable
understanding of precipitation processes in laboratory
heats containing Hf; however, its effect in commercial
alloys is not clear.

125

The second area of interest deals with the precipita-
tion of Ni; Ti in titanium-modified Hastelloy N. Recent
observations indicate that a Ti concentration of about
2% may provide good resistance to irradiation embrit-
tlement at 700°C and higher. We are concerned whether
this amount of Ti will promote the formation of Niy Ti
at various temperatures. Precipitation of this inter-
metallic dramatically increases the strength, but causes
embrittlement.

The last area of study reported deals with preliminary
observations of strain-induced precipitation in modi-
fied alloys. Carbides which precipitate during a creep-
rupture test are different in morphology and distri-
bution from those generated during short-term tensile
tests or elevated temperature exposure in the absence of
strain. '

13.1.1 Microstructures of Hf-Modified Hastelloy N
. Laboratory Melts

The laboratory heats discussed in this report have a
base composition of Ni—12% Mo—-7% Cr—4% Fe-—-0.2%
Mn and are modified with 05% Hf, 05% Ti, and
0.6% Nb unless otherwise noted. The standard carbon
and silicon concentrations are 0.05 and <0.01%.

The microstructure characteristic of hafnium-
modified laboratory melts has been discussed previ-
ously!™ and is shown in Fig. 13.1 for an alloy

1. R. E. Gehlbach, C. E. Sessions, and S. W. Cook, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1969, ORNL-4449,
pp. 193-95. N ' '

2. R. E. Gehlbach and S. W.|Cook, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1970, ORNL-4548; pp. 131-38.

3. R. E. Gehlbach and S.'W. Cook, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 164—65.
containing 0.7% HF after aging 200 hr at 760°C. Small
particles and platelets of MC carbide, about 0.1 to 0.3
um in diameter, precipitate throughout the matrix with
the exception of a region along grain boundaries
approximately 2 um wide that is denuded of precipi-
tate. Grain boundaries contain many small MC carbides

Fig. 13.1. Typical MC carbide distribution in hafnium-
modified hastelloy N. The alloy contains 0.7% Hf and 0.06% C
and was aged 200 hr at 760°C. 5000%.

126

the same size or slightly smaller than the matrix precipi-
tates. These grain boundary carbides are closely spaced,
and there does not appear to be any significant
coarsening of the matrix carbides between 10 and 1000
hr of aging. However, a high dislocation density is
present around those precipitated during aging 10 hr at
760°C. No matrix precipitate is observed in annealed
material prior to aging.

Effect of aging temperature. The microstructure of
the reference alloy (alloy 329) in the Hf-Ti-Nb series is
shown in Fig. 13.2a after aging 1000 hr at 760°C. The
precipitation process is similar to the hafnium-modified
laboratory heat discussed above. The matrix MC car-
bides are somewhat coarser (0.3 to 0.5 um) in the more
highly modified series. In addition, a very small amount
of MC carbide precipitates in the stacking fault mor-
phology previously observed in the titanium-modified
alloys. This stacking fault precipitate (SFP) is much
finer in the alloys modified with Hf-Ti-Nb than in those
modified with Ti alone. In general, the SFP only occurs
in the interior of grains.

Aging at the lower temperature of 650°C results in
finer matrix precipitates than at 760°C. Thin platelets
are generally formed at 760°C (Fig. 13.2a), and
particles and thick platelets are formed at 650°C (Fig.
13.2b). The amount of precipitate is considerably
higher at 650°C than at 760°C, and formation of SFP is
much more prevalent at 650°C. In addition, a disper-
sion of very fine carbides (<200 A) is observed

Fig. 13.2. MC carbides in microstructure of Hastelloy N containing 0.5% Hf, 0.5% Ti, 0.6% Nb, and 0.05% C. Anncaled | hr

at 1180°C and aged 1000 hr at (a) 760°C and (b) 650°C. 5000X

throughout the matrix after aging at 650°C (not
resolvable in Fig. 13.2a). These extremely fine precipi-
tates are also present in alloys modified only with
‘hafnium after exposure at 650°C.

Effect of carbon concentration. The effect of carbon
concentration on precipitation in the Hf-Ti-Nb series is
shown in Figs. 13.2 to 13.4. The microstructures of an
alloy with 0.008% C (alloy 331) after aging 1000 hr at
760 and 650°C are shown in Fig. 13.3. The matrix
carbides are slightly smaller at 760°C (Fig. 13.32) than
those in the alloy with the nominal 0.05% C (Fig.
13.2a). but are present in a much lower concentration.
Grain boundary carbides are very numerous, however.
A high density of very fine carbides is present after
exposure at 650°C (Fig. 13.3b). With 0.014% C (alloy
332) the microstructure after aging at 760°C is similar
to that of the very low carbon alloy, but with more
carbides present.

Increasing the carbon concentration above the
standard (0.05%) level generally results in fewer but
coarser carbides. The microstructures of an alloy
containing 0.14% carbon (alloy 335) after aging at 760
and 650°C are shown in Fig. 134. Large primary
carbides of the M,C type are present at both tempera-
tures. The characteristic matrix MC carbide is virtually
absent at 760°C but is present in relatively high
concentrations at 650°C. Its occurrence is generally
limited to large-grained areas which do not contain large
quantities of M,C. The MC carbide is coarser than that

YE-10361

127

formed at 650°C in heats containing lower carbon
levels. A carbon concentration of 0.087% results in
considerably more fine matrix precipitate at each
temperature, but much primary M,C is also present.

Effect of silicon additions. Additions of silicon to
Hf-Ti-Nb-modified alloys containing the standard 0.05%
carbon results in massive silicon-rich primary MgC
carbides primarily located in grain boundaries and as
stringers. The microstructures resulting from the addi-
tion of 0.44% silicon are shown in Fig. 13.5 after aging
at 760 and 650°C. Large MgC precipitates are in the
grain boundaries in Fig. 13.5a after exposure at 760°C.
No matrix precipitation of MC-type carbides occurs at
this temperature (compare with Fig. 13.2a where the Si
content was <0.1%). At the lower aging temperature,
precipitation of all morphologies of MC discussed
previously are present in addition to the large MgC
particles in the grain boundary (Fig. 13.5b). Some fine
MC is also present in the grain boundaries after aging at
650°C.

Silicon additions of about 0.15% result in a smaller
quantity of MgC and corresponding increases in the
amount of MC precipitating in the matrix during aging
at both 650 and 760°C. Thus, silicon removes much of
the carbon from solution that would otherwise be
available for MC precipitation during aging.

In summary, MC carbides which precipitate in the
matrix of Hf-Ti-Nb-modified Hastelloy N are present in
various sizes, morphologies, and concentrations de-

YE-10360

Fig. 13.3. MC carbide distribution in Hf-Ti-Nb modified Hastelloy N containing 0.008% C. Annealed | hr at 1180°C and aged

1000 hr at (a) 760°C and (b) 650°C. 5000X.

YE-10362 » YE-10358

Fig. 13.4. Carbide distribution in high carbon (0.14%) heat of Hf-Ti-Nb modified Hastelloy N. Large carbides are M,C; small
particles are MC. Annealed 1 hr at 1180°C and aged 1000 hr at (a) 760°C and (b) 650°C. S000X.

YE-10371 ) YE-10370 -

Fig. 13.5. Effect of 0.44% silicon on microstructures of Hf-Ti-Nb modified Hastelloy N (0.05% C). Large carbides are MC;
others are MC. Annealed 1 hr at 1180°C and aged 1000 hr at (a) 760°C and (b) 650°C. 5000X.

vr

[

129

pending on the aging temperature at the standard 0.05%
carbon level. Decreasing the carbon content decreases
the size and amount of matrix MC precipitated.
Increasing the carbon level above 0.05% results in large
primary M, C carbides at the expense of fine MC.

The addition of silicon results in precipitation of large
primary silicon-rich M¢C carbides with subsequent
precipitation of MC dependent on the amount of
carbon remaining in solid solution.

Commercial alloys. We have examined the micro-

“structures of nearly all of the commercial heats of

modified Hastelloy N listed in Table 13.1 after aging at
650, 760, and in several cases 700°C. Our general

observations have been reported previously.?**> A par- -

ticular concern is that the microstructures of the heats
containing hafnium do not resemble those of laboratory
melts with similar compositions. Instead of platelets
and fine particles of MC carbides in the matrix and
grain boundaries, the microstructure consists of large
amounts of MC in jagged grain boundaries and ex-
tending into the matrix 1 to 3 um. The grain interiors

are generally quite free of precipitates. The alloy

modified only with 0.8% Hf (heat 70-796) does exhibit
some matrix MC, although it is very heterogeneously
distributed and much larger in size than that observed
in the laboratory heats. The grain boundaries are rather
typical of the other commercial heats, although less
carbide is present adjacent to the boundaries. Similar
observations were made on a commercial alloy (67-504)
containing 0.5% Hf that was produced by a different
vendor.

Material from both of these commercial heats was
remelted and fabricated at ORNL with the same
procedures used for the laboratory heats. The micro-
structures of the lab remelts after aging at 760°C are
similar to each other. The remelts differ ‘from the
straight commercial material in that grain boundary
precipitation of the reprocessed alloys strongly re-
sembles that of conventional laboratory melts: fine
particles with no precipitate adjacent to the grain

~boundaries. Figure 13.6a shows typical grain boundary

precipitate and a patch of heterogeneous matrix carbide
in the commercial alloys containing 0.8% Hf. The
laboratory remelts (Fig. 13.6b) do not have carbides
adjacent to the grain boundary. A very small amount of
matrix carbide is present in the remelts after aging
although heterogeneously distributed.

The microstructures of the remelted alloys aged for
200 hr at 650°C are both free of precipitate. This
microstructure is different from that of the commercial
and the laboratory melts, and we currently have no
explanation for this anomalous observation.

13.1.2 Ni; Ti Precipitation in Ti-Modified
Hastelloy N '

Observations that Ti concentrations above 1% result
in improved postirradiation ductility after irradiation at
temperatures above 700°C have generated renewed
interest in using Ti as the sole modifying element in
Hastelloy N. Precipitation of Ni;Ti would be unde-
sirable, as its effect is one of dramatic strengthening at

Table 13.1. Compositions of experimental alloys

* Concentration (%)

Aloy No. :
' Mo Cr Fe Mn ~Si

Ti -~ Zr Hf Nb C
70-785 123 7.0 0.16 0.30 009 11 0.012° <0.003 0.097 0.057
70-727 130 74 005 037 <0.05 21 0.011 <0.01 <0.01 0.044
70-796 125 7.5 0054 0.64 0.02 0.04 0.024 0.79 0.04 0.04
69-648 128 69 03 0.34 0.05 0.92 0.005 <0.05 1.95 0.043
69-714 130 85 0.10 035 <0.05 0.80 0.028 <0.01 1.6 0.013
70-835 125 79 068 0.60 005 071 <0.005 0.031 2.60 0.052
70-786 122 7.6 041 0.43 0.08 0.82 0.024 <0.003 0.62 0.044
69-641 139 69 03 0.35 002 1.3 0.021 0.40 <0.05 0.05
70-787 123 7.0 0.18 0.43 0.09 090 0.038 0.77 0.12 0.041
70-795 137 83 0.035 0.63 003 1.5 0.018 0.42 0.005 0.05
70-788 121 73 043 041 0.1 1.4 0.020 0.30 0.67 0.027
70-797 127 7.0 0.29 0.38 0.02 0.59 0.040 0.78 0.98 0.049
70-798 135 79 0.26 0.53 0.02. 0.71 0.012 0.28 0.94 0.036
68-688 143 171 46 0.46 0.38 001 <0.050 <0.05 <0.05 0.079
68-689 137 74 4.6 0.46 0.53 036 <0.05 <0.035 <0.05 0.081
69-344 130 74 4.0 0.56 054 077 0.019 <0.1 1.7 0.11
69-345 130 80 4.0 0.52 0.52 1.05 0.038 0.88 <0.01 0.078

YE-10395

130

YE-10396

Fig. 13.6. Microstructures of Hastelloy N containing 0.8% HF after aging 200 hr at 760°C. (a) Commercial alloy 70-796 and (b)
laboratory remelt of 70-796. Carbides are of the MC type. 5000X.

the expense of ductility. Therefore, we are investigating
the concentration of Ti required to promote the
formation of Ni;Ti at various temperatures and times,
using a series of laboratory melts containing 1.0, 2.0,

Fig. 13.7. Ni3Ti precipitation in Hastelloy N containing 2.9%
Ti. The NigTi, formed by aging 10,000 hr at 700°C, is
lenticular and exhibits fringe contrast. MC carbides in  the
stacking fault morphology are also present. Dark field micro-
graph, (111) reflection. 10,000X.

2.4, and 2.9% Ti. The specimens examined to date were
aged 10,000 hr at 700°C. Large quantities of interme-
tallic Ni;Ti are present in the 2.9% Ti alloy. very homo-
geneously distributed throughout the matrix as shown in
Fig. 13.7. The intermetallic is lenticular, appears to be
shaped somewhat like a clam shell, and is oriented
parallel to the cube planes. Although present in the
2.4% Ti alloy, the NiyTi is heterogeneously distributed
and present in much smaller quantities. Isolated parti-
cles were observed in the 2.0% Ti, but they are much
smaller than in the more highly alloyed heats.

The unstressed buttonhead from a creep specimen of
the 2% alloy was examined. The specimen had been
annealed and aged for 1000 hr at 760°C before a
47,000-psi creep rupture test at 650°C that lasted about
5200 hr. A considerable amount of Ni;Ti is present,
although poorly developed, and was probably nucleated
during the 760°C age prior to the 650°C exposure. We
will look at these alloys exposed at 650 and 760°C for
various times to define more closely the maximum Ti
concentration allowable for service in this temperature
range. The effect of irradiation on NizTi will also be
investigated.

13.1.3 Strain-Induced Precipitation
in Modified Hastelloy N

Examination of the gage sections of several creep-
rupture specimens reveals that precipitation during

elevated temperature tests is different from that which
forms in the unstressed buttonhead or in aged material.
The typical MC carbide distribution in a 1.2% Ti-
modified Hastelloy N (alloy 467-548) after a 40,000-psi
creep test lasting 450 hr at 650°C is shown in Fig. 13.8.
Precipitation occurs as very fine carbides (200 to 1000
A) in high dislocation density slip bands. These disloca-
tions are out of contrast in Fig. 13.8a. The carbides are
clearly imaged in Fig. 13.8b, a dark-field micrograph
using a carbide (220) diffraction spot. Precipitation in
the absence of the creep strain consists of carbides in
the stacking fault morphology, nucleated around re-
sidual primary carbides.

The strain-induced precipitate is not observed in
short-term tensile samples tested over a wide range of
strain rates. The dislocation structure generated in
tensile tests is characterized by relatively long, straight,
oriented dislocations with little or no tangling. The
dislocation structure of creep test specimens is mark-
edly different from that generated in tensile tests.
Although annealed specimens are used for both types of
tests, the load required for a creep test is well above the
yield point of the alloys. The application of this load to
the specimen results in very rapid plastic deformation
up to approximately 5%. The dislocation structure after
this strain-on-loading consists of bands of tangled
dislocations in [110] directions. This configuration

YE-10523

131

provides the sites for precipitation during the remainder
of the test.

A different type of strain-induced precipitation
occurs in alloys containing Nb as a major addition. This
precipitate is present as very thin sheets as shown in the
bright field and precipitate dark field micrographs (Fig.
13.9). The specimen is from a heat containing 1.95%
Nb and 09% Ti (alloy 69-648). The 63.000-psi creep
test lasted 470 hr at 650°C. As before, the initial
structure after loading consisted of bands of tangled
dislocations. Although the resulting precipitate has not
been identified, it does not appear to be the MC-type
carbide usually found in these alloys.

In reality, then, the microstructures of specimens
creep tested above the yield stress are not the same as
those which would be found in a reactor constructed of
annealed material and operating below the yield stress.
The property changes resulting from the necessity of
performing creep tests at stresses above the yield point
would be an apparent increase in yield strength and
possibly lower ductility. The extremely low creep rates
found in the Nb-Ti alloys are indicative of the higher
strength. The Nb-Ti-modified alloys having this strain-
induced precipitate are extremely strong with very
respectable ductilities, generally above 20%.

We will be investigating the extent of this strain-
induced precipitate in several of the modified alloys and

YE-10524

Fig. 13.8. Strain-induced MC carbide precipitation in 1.2% Ti-modified Hastelloy N stressed at 40,000 psi at 650°C for 450 hr.
(a) Bright field; (b) Dark field using (220) carbide diffraction spot. 12,000X.
Fig. 13.9. Strain:

(a) Bright field; (b) Dark field using precipitate diffraction spot

also the effects of this starting microstructure on
mechanical properties.

13.2 MECHANICAL PROPERTIES
OF UNIRRADIATED MODIFIED
HASTELLOY N

B.McNabb  H. E. McCoy

One of the prime requirements for nuclear reactor
materials is long-term metallurgical stability. We are
aging materials at anticipated service temperatures and
measuring the mechanical properties to determine
whether the material is stable. Some modified Hastelloy
N specimens were solution annealed (1 hr at 1177°C)
and then aged at 650 and 760°C for 1000, 3000, and
10,000 hr. These will be subsequently creep tested at
650°C and 55,000 psi or at 760°C and 20,000 psi.

The testing is partially complete, and the rupture lives
are compared in Table 13.2.

The results to date indicate that the alloys are quite
stable with only small changes in properties. The
fracture strains of all the alloys were 20% or greater,
indicating no serious reductions in ductility during
aging.

induced precipitation in 1.9% Nb—0.95% Ti-modified Hastelloy N (heat 69-648) stressed at 650°C for 475 hr.

13.3 WELDABILITY OF SEVERAL
MODIFIED COMMERCIAL ALLOYS

B.McNabb  H. E.McCoy

The double-vacuum-melted and electroslag-remelted
commercial melts of modified Hastelloy N have been
further evaluated by mechanical property tests. Some
unirradiated mechanical properties and relative welding
characteristics were previously reported.* Mechanical
property tests on transverse specimens (0.125 in. diam
X 1.125 gage length) from welds of these alloys have
been conducted.

Table 13.3 is a tabulation of the tensile properties of
these alloys at 25°C and a strain rate of 0.05 min ™' and
compares the properties of the base metal with the
welded material in the as-welded and stress-relieved
conditions (annealed 8 hr at 870°C). The base-metal
yield stresses are from 40,000 to 55,000 psi, the
ultimate tensile strengths range from 100,000 to
130,000 psi, and the fracture strains vary from 55 to

4. H. E. McCoy and B. McNabb, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 181 -88.

[

133

Table 13.2. The effect of aging on rupture life of various alloys at 650 and 760°C

1000 hr 1000 hr 3000 hr 3000 hr
at 650°C at 760°C  at 650°C at 760°C

Heat Temperature  Stress: Solution
Number? 0 (psi) anneal
69641 650 55,000 2959
760 20,000 671.3

69648 650 55,000 1759.5
760 20,000  769.5

469344 650 55,000 9127
760 20,000 231.1

469345 650 55,000 285.6
760 20,000 2598

469714 650 55,000 146.3
: 760 © 20,000 4974
470727 650 55,000 8452
760 20,000  333.8

470785 650 55,000  81.7
760 20,000  338.1

470786 650 55,000 1322
760 20,000  343.1

470787 650 55,000 4008
760 20,000  450.3

4707788 650 55,000 1251.0
760 20,000  660.0

470795 650 55,000 246.0
760 20,000  469.7

470796 650 55,000 2924
760 20,000 418.8

470797 650 55,000 817.7
760 20,000  796.6

470798 650 55,000  665.6
760 20,000 524.8

470835 650 55,000 39115
" 760 20,000  748.1

219.0 2126 162.6 193.3
488.0 376.8 399.9 497.2
703.4 956.5 1294.7 1251.7

380.4 1558.4 430.4

142.8 183.6 403.6 1529
128.2 111.7 226.7 354.7
1563.3 210.2 3394 120.7
2099 160.1 178.3 150.2

3794 301.5 269.3 393.2
361.3 413.2 403.0 362.0

1088.4 378.1 919.4 591.2
407.9 378.1 344.9 353.0
" 85.0 87.2 84.2 898
329.9 2349 2578 254.0

477.7 109.5 107.6 110.7 ’
412.0 354.5 346 .4 295.7
229.5 2548 309.7 155.1
4458 435.1 465.4 487.5
1336.2 1069.7 1164.8 792.2
686.2 758.1 478.0 481.2
283.0 151.2 2309 196.7
73020 4837 418.8 398.3
146.7 70.4 112.8 85.3
263.2 2054  435.3 293.8
546.2 4009 4854 484.2
6778  615.0 6709 706.3
333.0 239.8 258.7 214.55
580.0 6082 5244 563.8

1989.35

716.7 6024  566.1 645.4

dSee Table 13.1 for chemical compositions.

42,540 psi.

80%. The weld metal has a higher yield stress than the
base metal and a lower fracture strain, but the ultimate
tensile stresses are equivalent. The stress relief anneal of
8 hr at 870°C lowers the yield stress only slightly and
recovers only a small fraction of the base-metal duc-
tility of the modified alloys, whereas it was sufficient to
recover most of the base-metal properties in the
standard alloy. As reported previously,' a higher
temperature anneal (1 hr at 1175°C) or possibly longer
times at 870°C would be required to recover the
base-metal properties of the modified alloys. Failure
generally occurred in the weld metal of the transverse
weld specimens.

Table 13.4 shows the tensile properties of these same
alloys at 650°C. The same trends shown at 25°C are
apparent at this temperature, but the fracture strains of
the weldment specimens are lower at 650°C. A ductility
minimum in this temperature range appears to be a

characteristic of these nickel-base alloys. The base-metal
yield strengths of these alloys at 650° are in the 25,000-
to 35,000-psi range, the ultimate tensile strength range
from 70,000 to 100,000 psi, and fracture strains from
40 to 65%.

The yield strengths of the weldment specimens are
approximately double those of the base metal and are
reduced only about 10,000 psi by the 8 hr at 870°C
anneal. The fracture strains are increased a few percent
by the anneal. Failure generally occurred in the weld
metal in transverse weld specimens.

Table 13.5 compares the creep-rupture properties of
these alloys at 55,000 psi and 650°C in air. There are
large variations in the rupture lives of these alloys in
creep at 650°C, even though there are only small
differences in the yield strength of the base metal at 25
and 650°C (Tables 13.3 and 13.4). In general, the
rupture lives of the weldment specimens were much
" Table 13.3. 'i'ensile properties at 25°C of several alloys?

134

. Heat

'R

" History? Yield stress  Ultimate stress  Fracture ~ Reduction in

number (psi) (psi) strain (%) area (%)
_70-785 A 40,400 114,300 69.8 57.7
B 86,100 117,500 21.2 62.3
C 74,100 113,400 21.3 58.8
70-727 A 48,600 133,000 62.6 469
B 83,600 116,600 20.6 54.1
C 60,700 127,500 31.1 55.9
67-504 A 37,400 104,800 - 73.2 73.7
B 58,000 101,400 40.2 55.0
C 50,000 104,900 37.3 453
70-796 A 39,800 116,000 64.0 543
B 71,400 91,000 12.6 27.9
C 59,700 98,200 18.4 30.6
69-648 A 44,700 116,900 742 42.6
B 75,300 118,600 333 38.0
C 65,400 111,800 31.0 394
69-714 A 41,800 109,300 81.1 51.6
B 80,600 103,700 28.1 46.9
C 62,100 107,600 35.0 549
70-835 A 54,500 144,700 59.0 55.2
B 87,100 117,200 19.7 61.7
C 82,300 116,700 27.7 59.7
70-786 A 37,400 106,300 73.1 63.7
B 78,400 98,300 12.5 46 .4
. C 75,500 114,300 19.9 457
69-641 A 44 900 117,100 72.6 54.7
B 73,800 113,700 29.6 52.5
C 65,500 116,000 36.7 42.0
70-787 A 40,800 113,200 69.5 62.4
B - 85,800 118,00 20.4 49.9
C 66,400 93,800 13.0 24.7
70-795 A 49,700 130,800 64.3 39.7
B 87,800 121,600 20.3 39.7
C 72,200 116,500 24.9 33.9
70-788 A 42,200 114,700 66.4 63.8
B 78,500 110,900 21.2 63.0
C 69,800 113,300 29.5 59.8
70-797 A 44.000 123,400 65.4 524
B 88,300 127,500 24.9 54.0
C 65,200 111,800 249 50.8
70-798 A 42,300 121,600 66.3 49.0
B 86,000 119,800 21.6 30.2
C 75,200 115,100 19.8 59.7
68-688 A 46,100 118,700 "59.8 45.6
B 88,400 126,500 214 459
C 61,700 113,900 26.9 40.5
68-689 A 52,100 117,400 §5.7 504
B 84,500 121,600 18.6 34.1
C 54,800 114,300 315 353
69-344 A 56,500 130,400 56.9 47.2
B 80,700 125,700 343 423
C 67,900 117,200 28.5 66.4
69-345 A 48,800 126,500 . 61.3 50.6

aStrain rate of 0.05 min 1. .
bA: base metal, annealed 1 hr at 1180°C; B: weldment, as welded;

C: weldment, annealed 8 hr at 870°C.

[ LN

. V
ar

-/

135

Table 13.4. Tensile properties at 650°C of several alloys?

Heat- History® Yield stress  Ultimate stress . Fracture  Reduction in

number (psi) (psi) strain (%) area (%)
70-785 A 26,300 83,300 65.3 42.4
B 59,200 78,000 13.7 56.7
C . 49,000 73,900 13.0 537
70-727 A 33,800 94,100 64.7 41.5
B 64,900 82,900 13.9 35.7
C 43,700 72,300 18.0 35.2
67-504 A 25,600 83,000 66.9 45.1
| B 21,100 21,400 2.2 4.2
C 35,800 60,100 13.8 23.2
70-796 A 27,900 88,300 63.6 45.7
B 29,200 34,700 10.5 14.4
C 38,000 61,300 11.6 47.0
69-648 A 29,400 78,300 46.6 33.9
B 55,100 71,200 10.9 34.3
C 46,500 76,900 19.3 44.1
- 69-714 A 27,700 78,500 56.1 42.6
B 53,600 71,500 12.3 347
C 43,700 69,000 20.1 28.6

70-835 A 33,600 99,500 51.1 43.4
B 66,200 83,400 12.3 28.5
C 54,300 83,500 18.3 29.1
70-786 A 23,900 65,400 45.1 31.0
B 64,100 77,400 10.6 32.9
C 54,000 80,100 13.5 30.3
69-641 A 29,000 92,700 533 376
B 56,500 72,500 12.1 304
C 46,000 76,000 17.5 41.1
70-787 A 26,800 82,500 61.3 39.9
B 60,700 75,000 10.9 38.7
C 47,300 71,000 14.8 35.7
70-795 A 33,800 99,500 61.2 41.6
B 63,400 85,300 13.1 24.6
C 55,700 86,800 21.0 37.7
70-788 A 29,000 71,500 45.8 36.2
B 62,100 76,900 11.0 40.5
C 51,000 75,900 16.2 41.8
70-797 A 33,400 104,200 59.3 38.4
B 57,300 80,200 17.6 36.5
C 51,800 85,400 20.6 53.9
70-798 A 27,000 91,600 62.0 44 .6
B 63,900 78,500 104 36.6
C 53,300 81,600 17.5 53.6
68-688 A 30,600 85,700 414 31.9
B 67,000 89,600 13.5 28.8
C 45,500 85,400 233 29.1
68-689 A 31,200 79,200 314 21.9
B 66,000 87,300 12.5 23.1
' C 44,300 83,600 20.4 29.7
69-344 A 33,900 93,600 40.6 29.5
B 67,000 86,000 11.5 25.2
C 52,200 88,500 23.0 33.0
69-345 A 37,200 91,600 31.8

4Strain rate of 0.05 min~!." _
bA: base metal, annealed 1 hr at 1180°C; B: weldmeat, as welded; C::

weldment, annealed 8 hr at 870°C.

40.4
136

Table 13.5. Creep rupture properties of several alloys at 650°C and 55,000 psi

Heat History® Rupture  Minimum creep Fracture Reduction in
‘number life ¢hr) rate (%/hr) strain (%) area (%)
70-785 A 81.7 0.0730 53.5 43.3

B 48.0 0.028 6.1 21.9
C 52.0 0.175 19.2 60.4
70-727 A 845.2 0.0152 39.5 36.4
B 463.7 0.0046 5.1 16.3
C 189.8 0.0606 17.9 17.2
70-796 A 2924 0.0165 37.2 31.5
B 409 0.0635 7.2 . 20.5
C 6.3 1.188 18.0 45.7
21-543 A 69.0 0.038 - 24.9 26.3
B 31.7 0.094 5.7 59
C 16.8 0.53 12.5 16.2
69-648 A 1759.8 0.0005 20.5 18.9
B 9.8 0.046 5.3 17.9
C '88.8 0.052 11.6 52.8
69-714 A 146.3 0.0835 47.3 51.0
' B 455.7 0.0127 15.6 16.2
C 194 0.112 11.8 26.1
70-835 A 3911.9 0.0041 32.0 - 26.0
B 814.0 0.0029 8.02 16.8
C 557.1 0.0117 13.7 17.8
70-786 A 132.2 0.0270 34.9 44.2
B 353 0.0275 44 27.5
C 75.0 0.0848 14.4 24.3
69-641 A 265.9 0.0740 499 43.9
‘ B 346.3 0.0238 ‘13.6 19.7

ol 241 .4 0.0386 17.1 23.1
70-787 A 400.8 0.0320 49.1 47.3
B 939 0.036 9.6 35.2
: C 51.5 0.125 12.3 52.0
70-795 A 246.0 0.050 64.7 61.4
' B 144 .6 0.035 .11.9 344
C 56.8 0.14 16.9 576
70-788 A 1251.0 0.0113 42.0 35.4
B 344 .3 0.014 9.2 23.4
C 312.0 0.031 26.3 68.6
70-797 A 817.7 0.0185 48.8 45.6
B 4234 0.0193 21.8 41.1
C 158.3 0.0830 24.6 45.0
70-798 A 665.6 0.0165 55.5 46.7
B 62.7 0.0360 10.3 28.8
. C 164.7 0.0660 22.7 36.3
68-688° A 78.0 0.0400 19.6 19.0
B 41.1 0.14 10.3 20.6
C - 52.0 0.22 16.9 25.2
68-6892 A 71.9 0.0525 15.8 19.5
B 35.2 0.164 10.6 18.2
C 834 0.145 17.5 20.0
69-344 A 912.7 0.0075 24,2 26.7
B 631.1 0.0090 14.5 23.3
C 1556 0.044 13.0 21.3

2A: base metal, annealed 1 hr at 1180°C; B: weldment, as welded; C:

weldment, annealed 8 hr at 870°C.

547 000 psi.

1Y

Cam !
"i"

137

shorter than those of the base metal. The minimum
creep rate of the “‘as welded” specimen was lower than
that of the base metal, but the lower fracture strain of
the weld caused the rupture life to be reduced
considerably. After annealing 8 hr at 870°C, the
weldment specimens generally had a higher minimum
creep rate than the base metal, a shorter rupture life,
and a lower strain at fracture. Failure generally oc-
curred in the weld metal on the weldment specimens.
Rupture lives for the base metal varied from 72 to 3912
hr for tests at 55,000 psi and 650°C in air.

13.4 CREEP-RUPTURE PROPERTIES OF
HASTELLOY N MODIFIED WITH
NIOBIUM, HAFNIUM, AND TITANIUM

H. E. McCoy

" Laboratory melts having nominal compositions of
Ni—12% Cr—4% Fe—0.2% Mn-0.5% H{-0.5% Nb-—
0.5% Ti and various concentrations of C and Si were
prepared. The detailed analyses of these alloys are given
in Table 13.6. The alloys were 2-lb castings and were
fabricated to Y,-in.-diam rods. They were annealed 1 hr
at 1180°C before testing. Several of the samples were
irradiated 'in the ORR at 760°C to a peak thermal
fluence of 3 X 10%° neutrons/cm”. All of the results
reported in this section were obtained at a test
temperature of 650°C.

The first series of alloys had carbon concentrations
that ranged from 0.006 ito 0.137%. The results of creep
tests on these alloys at 650°C are shown in Figs. 13.10

and 13.11. The rupture lives show some scatter, but
they generally show the expected trend of increasing
rupture life with increasing carbon content. Alloys 330
and 331 have similar carbon contents, but the rupture
lives are different at high stresses. This may be due to
the higher oxygen content of heat 330 and its influence

‘on the carbide structure. Alloys 334 and 335 with

0.087 and 0.137% C, respectively, have the longest
rupture lives of the series, and there is no detectable
difference in the two alloys. The minimum creep rates
of these same alloys in Fig. 13.11 show a systematic
progression of Str’ength with carbon level up to 0.05%.
The data for alloys 329 (0.052% C), 334 (0.087% C),
and 335 (0.137% C) are described by a single line. Heat
333 (0.049% C) seems weaker than heat 329 (0.052%).
All of the alloys are as strong or stronger than standard
Hastelloy N. ’

The stress-rupture properties of these same alloys
after irradiation at 760°C for 1100 hr are shown in Fig.
13.12. Alloys 330 (0.006% C), 331 (0.008% C), and
332 (0.014% C) have about equivalent rupture. lives,
which slightly exceed those of standard air-melted
Hastelloy N. The other four alloys with carbon levels of
0.049 to 0.137% seem to have rupture lives that are
better than those of the lower carbon alloys. The creep
strengths (Fig. 13.13) show a similar behavior. All
alloys have creep strengths above those noted for
standard vacuum-melted material.

The fracture strains of these alloys after irradiation
(Fig. 13.14) are very good. Standard vacuum-melted
materjal has been observed to have fracture strains of
only a few tenths of a percent under these 1rrad1at10n

Table 13.6. Compositions of experimental alloys

Alloy ' Concentration (%) ‘
No. Mo Cr Fe Mn Nb Ti Hf Zr C - CSic N 0
30 . 117 7.7 3.2 0.19 062 047 0>.50 .<0.01 0.006 - 0.011 0:0022 0.031
: o K . 0.048
331 11.7 7.3 36 015 0.56 050 0.46 <0.01 0.008 <0.01 0.0016 0.0034
332 11.8 74 30  0.15 0.61 048 0.44 <0.01 '0.014 <0.01 0.0012 = 0.0060
329 11.6 7.3 4.1 0.16 0.59 048 044 <0.01 0.052 <0.01 0.0012 0.0018
333 11.8 72 - 34. 0.15 0.62 047 048 ' <0.01 = 0.049 <0.01 0.0014 0.0160
334 116 7.2 4.0 0.11 056 0.50 041 <0.01 0.087 <0.01 - 0.0012 0.0036
335 1.7 7.4 3.3 0.15 . - 0.56 049 046 <0.01 0.137 <0.01 0.0011 0.0047
336 11.7 74 31 “0.15- 0.56° 0.48 0.38 <0.01 0.053 0.14 0.0030. 0.045
S - - . : I _ . : A - 0.130
337 116 . 7.3 4.0 0.13 063 0.44 047 . <0.01 . 0.049 0.12 0.0035 0.0093
338 11.7 74 36. . 0.15 0.45 048, 0.52 <0.01 . 0.057 0.44 0.0913 . 0.0024

138

ORNL- DWG 71— {3676

70 2 o
N
I \\\ o
60 fi\\ . \\ ,\‘n
~] N N
1."\ vy v \\\ ol \4>\ <
50 T N N
"n.,.. '~‘ by L)
= \\‘_\ ST AT AT 334,335
. g ."""*- \\
o 40 "'\‘lx\\ L
g SQY
. \h-:-.\
Q 30 — * ALLOY NO. C CONTENT (%) : \.\ 330 |
= o 330 0.006 ~ i
v 33 0.008 \\
20 — \
? gzg g'g; STANDARD HASTELLOY N
Q R
o 333 0.049
10— & 334 0.087
o 335 0.137
oL LU L L |
10~ 10° 10! 102 103 10?

RUPTURE TIME (hr)

Fig. 13.10. Stress-rupture properties at 650°C of alloys modified with 0.5% Hf, 0.5% Ti, 0.5%

and test conditions,’ but the minimum fracture strain
noted for these alloys was 5%. There is a marked effect
of carbon content at lower creep rates. For alloys with
carbon contents of 0.049% or greater, the lowest
fracture strain is 13%. There is no systematic effect of
carbon content above 0.049%.

Three of the alloys contained Si additions. The
variation of the rupture lives of these alloys in the
unirradiated and irradiated conditions is shown in Fig.
13.15. Silicon does not seem to have any systematic
effect on the rupture life in the irradiated or unirradi-
ated conditions. The fracture strains are quite good for
all specimens, and there is no detectable effect of Si.

The mechanical behavior of these alloys is generally
what would be expected from the microstructural
observations made by Gehlbach. However, some
apparent conflicts exist. For example, aging at 760°C
produced a relatively coarse carbide structure in the
higher carbon alloys, whereas the mechanical properties
do not show a deterioration at the higher levels. Also,
the alloy with 0.44% Si (338) formed coarse M4C-type

5. H. E. McCoy and R. E. Gehlbach, “Influence of Irradiation
Temperature on the Creep-Rupture Properties of Hastelloy N,”
J. Nucl. Appl. Technol. 11(1), 45—-60 (May 1971).

Nb, and various amounts of C.

carbides and very little matrix precipitation during

* aging. This should result in poor postirradiation proper-

ties, but the data (Fig. 13.15) do not bear this out. Two
factors likely account for these discrepancies. First, the
precipitates that form during aging without a stress may
be different from those that form in a stressed specimen
during testing. Second, irradiation may influence the
type and distribution of precipitate that forms. Both of
these factors must be investigated further since nuclear
applications will both stress and irradiate the material.

13.5 CORROSION STUDIES
J. W. Koger

The success of a molten-salt reactor system is strongly
dependent on the compatibility of the materials of
construction with the salts comprising the primary and
secondary circuits of the reactor.

A number of factors contribute to molten-salt corro-
sion. The tendency for a reaction to occur is measured
in terms of the chemical potentials of the reactants and
the products. The reaction kinetics will be influenced
by factors such as the temperature, pressure, concen-
tration, and nobility of the salt constituents, solublllty
effects, and viscosity of the me]ts

Il “,'\
1 ys »

139

The experiments discussed in this section are intended
to assess the mass transfer characteristics of candidate
salt-alloy systems under design parameters based on the
Molten-Salt Breeder Reactor. The majority of studies
are conducted in thermal convectioh and pumped loops
constructed of Hastelloy N. The salts of interest are LiF
and BeF, based with UF, (fuel), ThF,; (blanket), and
UF, and ThF, (fertile-fissile) and a NaBF,-NaF mix-
ture«(coolant salt). '

Of the major constituents of Hastelloy N, chromium
is much more readily oxidized in fluoride salts than Fe,
Ni, or Mo. Accordingly, attack in our. systems is
normally manifested by the selective removal of chro-
mjum from the Hastelloy N. Several oxidizing reactions
may occur depending on ‘the salt composition and
impurity content, but among the most important
reactants causing oxidation are UF,, FeF,, and HF.

- Because our experimental systems are not isothermal,

temperature gradient mass transfer effects are quite
important. Conventionally, we observe weight losses in
the hotter regions of our loops and weight gains in the
colder sections. - .

The status Of the thermal convection loops in
operation is summarized in Table 13.7.

13.5.1 Fuel Salts

Loop 1255, containing an MSRE-type fuel salt with 1
mole % ThF,; and constructed of Hastelloy N, has been
terminated after 80,439 hr (9.2 years). The loop
contains insert specimens of standard Hastelloy N and a
Hastelloy N plus 2% Nb alloy. The latter alloy was
developed - for improved weldability and mechanical
properties, although similar compositions are now
under consideration because of increased resistance to
neutron damage.

This loop was operated as a “life-test” to provide
long-term data on mass transfer in a fluoride salt, air
oxidation of Hastelloy N, and corrosion properties of
various weld junctions. We elected to terminate the
experiment after a salt leak was detected under one of
the main loop heaters. Prior to the discovery of the
failure, the amperage to the loop heaters dropped by a
factor of 2. Investigation revealed that half the main
heaters were open and grounded, which suggested the
presence of salt. After removing the insulation and
heaters, a small salt leak was found under one of the
Lavite® bushings used to position the loop heaters
around the hot leg. |

6. Trade name, American Lava Company. .

ORNL-DWG 74-13677

70 oo
AT 7
A /D //6 a'/
60 pad //u// a
/’ bt ol
/ v
/éc /’ / 7 JISYES
- - |t
L/' | LA Lt
50 Pt 4 L
. | s S|
[: 207
o A
Q 40 —
O > ’
= port
o) : . 74 Al
b 30 P 'ALLOY NO.  C CONTENT %
rd
% 7 0 330 0.006
20 Bl v 33 0.008 U
'/‘ a 332 0:014
“STANDARD o 329 0.052
HASTELLOY N . 33 0.049
10 a 339 0.087 1
o 335 0.137
o N s
1074 1072 1072 107! 10° - 10"

MINIMUM CREEP RATE (%/hr)

Fig. 13.11. Creep-rupture properties at 650°C of a.lloys modified with 0.5% Hf, 0.5% Ti, and 0.5% Nb and various amounts of C.
140

ORNL-DWG 71-13678

70
D
60
STANDARD AIR A J )
MELTED \
50
\g = 1} 0 o & O
= L
z STANDARD VACUUM N
o 40 ——MELTED ' S 2 , _
S R \\ D
o ™
= " ..""'h.. v
5)) 30 \\ N\ 2 o
L - o) bl
o ™ Py
= ~ ""N-...
[¥p] ™
_ ALLOY NO. G CONTENT (%) [ L] T~
20 ® 330 0.006 TN =
| v 33 0.008 ~
o 332 0.014 ~
10 o 329 © 0.052 e
o 333 0.049
a 334 0.087
° 335 0.137
0. AR R |
107! _ 10° B o 102 103 10%

RUPTURE TIME (hr)

" Fig. 13.12. Stress-rupture properties at 650°C of alloys irradiated to a thermal fluence of 3 X 102? neutrons/cm? at 760°C. All
alloys contain 0.5% Hf, 0.5% Ti, and 0.5% Nb and various amounts of C.

ORNL-DWG 71-13679

70
i d
60
0 o h &
k "/
50 at
) o |o &1 ] //n Iy
‘g /—/<~STANDARD AiR MELTED
o 40 — a/._’ —
3 STANDARD - Jans
= VACUUM MELTED 4 ; 1 LT
’ % /’“: 0 —/
& 30 g <
o |~ //"
ju
® L~ i ALLOY NO. G CONTENT (%)
pr. |t
20 ] o 330 0.006 Al
I v 334 0.008
o 332 0.014
o 329 0.052
10 0 333 . 0.049 -+
a 334 0.087
> 335 0.137
o ' LI b 4t
10" 1073 1077 107! 10° 0’

MINIMUM CREEP RATE (%/hr)

Fig. 13.13. Creep-rupture properties at 650°C of alloys irradiated to a thermal fluence of 3 X 102? neutrons/cm? at 760°C All
alloys contain 0. 4% Hf, 0.5% Ti, and 0.5% Nb and various amounts of C,

@
1 W "

g
il IF‘* )

141

ORNL-DWG 7{-13680

28
REEREI T 1T
ALLOY NO. C CONTENT (%)
0 330 0.006
o IV 334 0.008
o 332 0.014 o A
o 329 0.052 °
0 333 0.049
A 334 0.087
20 — o 335 0.137
S
0
E 0
I 16 g il
- ‘ < 1]
$ x Pe A
a
[o]
o o] 14
o 12
04: o
U a
v 0
8 v
v o
4
0
Tl 1072 1072 107! 10°

MINIMUM CREEP RATE {%/hr)

Fig. 13.14. Postirradiation fracture strains at 650°C of alloys irradiated to a thermal fluence of 3 X 1020 neutrons/cm? at
760°C.

ORNL-DWG 71-13681

70 O——)
- ™ - CLTTIIE
N STANDARD HASTELLOY N
60 o \ | i z (UNIRRADIATED)
O5 7o iaw \1§;‘7 o ola { { j
) 32 I \L
~ >0 Sr%w 4 104 1LI!7 ézj;
2 e/
§ \"T‘fiélfeo ‘ r m ™ | 18
o} 40 O 7_01 l \\ST4ND| il 10‘.—'13
= - 039 Ny TO & +
° PN q
2 \T\Nf’% T o) TN
E 30 = Vi I/4CUUM‘ \\fi/ \
v , _\“:\.\ir £p s ‘.\\i
20 |- ALLOY NO. Si CONTENT (%) o \74/1/04% :
o 329 ~ <0.01 S~ s
. I~ 76‘4
A 336 0.14 "*»Qnr "
101-s 337 0.12 ~
¢ 338 0.44
LT ]
107! , 10° 10' 102 103 104

RUPTUYRE TIME (hr)

Fig. 13.15. Influence of various amounts of Si on the stress-tupture properties at 650°C of alloys unin'adiated'(open) and
irradiated (solid) at 760°C to a thermal fluence of 3 X 10%° neutrons/cm?2, Numbers by the points are the fracture strains.
142

The appearance of the wall near the failure is shown
in Fig. 13.16. The failure site was at the center of a
Y4 -in.-diam crater whose depth was comparable to the
thickness of the wall. The crater formed from the
outside in, and its location suggests that it was caused
by catastrophic oxidation of the tube wall associated
with the Lavite bushing which surrounded the failure
area. We have examined metallographic samples from
various sections of the loop, including the leak site, and,
except for the '5-in.-diam crater, these samples showed
only superficial corrosion (<1 mil) on either the outside
or inside surfaces of the loop. Even surfaces covered by
the leaking salt showed limited oxidation. New heaters
and insulation had been placed on the loop after 8.8
years of operation.” Based on our inspection of the
loop at that time, we believe that the failure was
induced following the heater replacement and that it
was not progressive over the life of the loop.

7. 1. W. Koger, MSR Program Semiannu. Progr. Rep. Feb. 28,
1971, ORNLA4676, p. 194,

Loop 1258, constructed of type 304L stainless steel,
has operated about eight years with the same salt as
Loop 1255. The heaters and insulation on this loop
were replaced at the same time as for loop 1255, and it
continues to operate normally.

Loop NCL-16, constructed of Hastelloy N and con-
taining removable specimens in each vertical leg, has
operated with MSBR fuel salt (Table 13 .7) for over 3.5
years. The heaviest weight loss after this period is —2.9
mg/cm?®, and the largest weight gain is +1.7 mg/cm?.
Assuming uniform loss, the maximum weight loss is

equivalent to 0.04 mil/year. The chromjum content of .

the salt has increased 550 ppm, and the iron has
decreased about 100 ppm in 30,000 hr. Titanium-
modified Hastelloy N specimens (12% Mo—7% C1—0.5%

" Ti, bal Ni) in this loop show smaller weight losses than

standard Hastelloy N specimens (16% Mo—7% Cr—5%
Fe, bal Ni) under equivalent conditions. We attribute
this difference to the absence of iron in the modified
alloy. Principal corrosion reactions acting in this loop
appear to be

Table 13.7. Status of MSR program thermal convection loops through August 31, 1971

Salt Max. Operating
Loop . . . AT .
umber Loop material . Specimens Salt type composition temp. co time
n : (mole %) - 0. (hr)
1258 Type 304L Type 304 L stainless - - Fuel LiF-BeF;-ZrF4-UF4-ThF 4 688 100 70,583
stainless steel steeld:® (70-23-5-1-1)
NCL-13A  Hastelloy N Hastelloy N; Ti-modified Coolant NaBF4-NaF (92-8) plus 607 125 24,795
Hastelloy N controls?-¢ tritium additions
NCL-14 Hastelloy N Ti-modified Hastelloy N&¢ - Coolant NaBF4-NaF (92-8) 607 150 33,370

NCL-15A  Hastelloy N
‘ Hastelloy N controls?:¢

NCL-16 Hastelloy N
» Hastelloy N controls®:¢

NCL-17 Hastelloy N
Hastelloy N controls?-¢

NCL-19A  Hastelloy N

Ti-modified Hastelloy N; Blanket
Ti-modified Hastelloy N; Fuel
Hastelloy N; Ti-modified Coolant

Hastelloy N; Ti-modified  Fertile-
Hastelloy N controls?-¢ fissile

LiF-BeF,-ThF4'(73-2-25) 677 55 26,632

LiF-BeF,-UF4 704 170 . 30,990
(65.5-34.0-0.5)

NaBF 4-NaF (92-8); plus 607 100 19,033
steam additions

LiF-BeF,-ThF4-UF, 704 170 13,419

(68-20-11.7-0.3) plus
bismuth inimolybdenum

hot finger
NCL-20 Hastelloy N Hastelloy N; Ti-modified Coolant '~ NaBF,4-NaF (92-8) 687 250 14,913
Hastelloy N controls?:¢
NCL-21 Hastelloy N Hastelloy N&:¢ MSRE fuel LiF-BeF,-Z1F 4-UF, 650 110 1,009

(65.4-29.1-5.0-0.5)

dHot leg only.
bRemovable specimens.
“Hot and cold legs.

P2
2UF4(d) + Cr(s) = 2UF3(d) + CrF,(d) .
FeF,(d) + Cr(s) = Fe(s) + CrF,(d) .

where s and d refer to crystalline solid and dissolved
state respectively.

We are now using loop NCL-16 to assist in the study
of the intergranular cracking of Hastelloy N found in
the MSRE. Specifically, we are evaluating the extent to
which the attack is related to the localization of normal
corrosion processes to grain boundaries. In any solid
solution alloy where there is a difference in nobility of
the constituents, electrochemical corrosion (oxidation-
reduction reactions) will cause the removal of the least
noble constituent, attack being preferential to the grain
boundaries. In time, given a continuing electrochemical
process, this will lead to crevices in the grain boundaries
and at defect sites in the grain surfaces. Diffusional
processes within this crevice will normally lead to its
broadening and ultimately to the formation of pits.
However, if the root of the crack is anodically polarized
and the walls cathodically polarized, the knifeline
attack will continue. Such a condition may arise if the
walls of the crevice become covered with a very noble
material (nickel or molybdenum). This covering by a
noble constituent can occur either by the noble

143

material remaining on the wall when the least noble
constituent is removed or by dissolution of all the alloy
constituents with subsequent precipitation of the more
noble constituents.

In order to test this thesis we added approximately
400 to 500 ppm FeF, to NCL-16. Chemical analysis
showed an immediate increase of 300 ppm Fe in the
salt which has significantly increased the oxidation
potential of the salt compared with past operation of
the loop.

The specimens from NCL-16 were removed after
448-hr exposure to the FeF,-doped salt and showed
weight changes typical of all our temperature gradient
mass transfer systems, that is, weight losses in the hot
section with corresponding weight gains in the cold
section. However, the weight changes obtained in the
last 448 hr were equal to those previously generated in
10,000 hr. No metallographic information is available at
this time.

NCL-21, a Hastelloy N thermal convection loop of a
somewhat different design (Fig. 13.17) with removable
specimens in each leg and containing the MSRE fuel
salt, was started and has operated for over 1000 hr. This
loop is equipped with electrochemical probes which will
measure the U**/U** ratio, and, hence, provide an
instantaneous measure of the oxidation potential of the

Y—108591

0.5 1.0
|

INCHES

Fig. 13.16. Cross section of hot leg of loop 1255 showing region where leak occurred after 9.2 years.

144

salt. Previous use of these probes had been limited to large system. This loop differs from our other thermal -
small static systems, and this experiment is intended to convection loops in that the flow up the hot leg goes
evaluate their possible use for on-stream analysis in a directly into a large surge tank. In our other loops the

PHOTO 78466

Fig. 13.17. Upper portion of Hastelloy N thermal convection loop (NCL-21) showing hot-leg surge tank (right side) into which
the electrochemical probes are placed. ¢
Fe )

v

145

salt is circulated only through tubing w}Jth is in a
harp-shaped configuration. The flow then goes out the
bottom of this tank into a crossover leg and into the
cold leg. The different design of the upper portion. of
the loop was dictated by the fact that we wanted the
electrochemical probes to see a fairly large and calm salt
surface. The electrochemical probes pass through
mechanical fittings (separated from the metal by
Teflon), which are on risers 10 in. from the surge tank
(to protect the Teflon from the heat of the salt) and
then enter the salt in the large hot-leg surge tank. There
are five access ports for the probes.

The specimens from the loop were removed, weighed,
and examined after 665-hr exposure. Weight losses and
gains, as expected in a temperature gradient mass

- transfer system, were seen in the hot and cold leg

respectively. The maximum weight loss in the hot leg
was 0.1 mg/ecm? and the maximum weight gain in the

cold leg was 0.05 mg/cm?®. As expected, the U¥*/U*

ratio increased with time, and we are correlating these
values with our weight-change data. After these initial
measurements, further steps will involve additions of
oxidants and reductants to study their influence on our
measurements. -

Details on the U**/U*" ratio as monitored by the

probes are discussed in the Analytical Chemistry section
of this report. :

13.5.2 Fertile-Fissile Salt

A fertile-fissile MSBR salt has circulated for over
13,400 hr in Hastelloy N loop NCL-19A, which has
removable specimens in each leg. The test has two
purposes: (1) to confirm the compatibility of Hastelloy
N with the salt and (2) to determine if bismuth will be
picked up by the salt and carried through the loop. The
bismuth is contained in a molybdenum vessel located in
an appendage beneath the hot leg of the loop. Assuming
uniform loss, the maximum weight loss is equivalent to
0.024 mil/year. A modified Hastelloy N alloy (Ni—12%
Mo—7% Cr—0.1% Fe—0.4% Ti—2.0% Nb) has lost less
weight than a standard alloy at the same position. The

chromium content of the salt has increased 134 ppm in .
7400 hr, and there is no detectable bismuth in the salt.

The corrosion reactlons were discussed in the NCL- 16
section.

13.5.3 B]anket Salt.

Loop NCL-15A, constructed of standard Hastelloy N
and containing removable specimens in each leg, has
operated three years with a blanket salt (LiF-BeF, with

25 ‘mole % ThF,) proposed for a two-fluid MSBR

concept. Mass transfer, as measured by the change of
chromium concentration in the salt, has been very
small. Specimens exposed to this salt have a “‘glaze” or
coating (probably a high-melting thorium compound)
that is impossible to remove without damaging the

- metal. However, metallographic studies show little, if

any, mass transfer of metallic components.

13.5.4 Coolant Salt

Loops NCL-13A and NCL-14, constructed of stand-
ard Hastelloy N and containing removable specimens in
each leg, have operated for 2.8 and 3.8 years, respec-
tively, with the fluoroborate mixture NaBF,-NaF (92-8
mole %). The maximum corrosion rate (assuming
uniform removal) at the highest temperature, 605°C,
has averaged 0.7 mil/year for both loops. Corrosion has

~ generally been selective toward chromium

FeF, (d) + Cr(s) = CrF,(d) + Fe(s) ,

but there have been short periods when gaseous
impurities entered the salt to cause general attack of the
Hastelloy N, for example, by the reaction

2HF(d) + M(s) = H, (g) + MF,(d) ,

where d, s, and g refer-to dissolved state, crystalline
solid, and gas, respectively, and M =Cr, Fe, Ni, and Mo.
The latter periods were caused by leaks in the seals of
ball valves which are located above the surge tanks.

.These valves are exposed to a mixture of He and BFj3,

gas and not to salt. We attribute the leaks both to heavy
use of the ball valves and to corrosion induced by any
air inleakage into the gas mixture. Although the overall
corrosion rate in these loops is not excessive, the rates
observed in the absence of ball valve leaks have been an
order of magnitude lower than the average rate. In
NCL-14, over the last 4450 hr, the maximum corrosion
rate has been 0.3 milfyear. Figure 13.18 shows weight
changes from specimens in NCL-13A as function of
position in the loop (temperature) and time. Note the
areas of weight gain and weight loss and the balance
points where no mass transfer occurs. ”

- Loop NCL-17, constructed of standard Hastelloy ‘N
and containing removable specimens in each leg, is
being used to determine the effect of-steam on the mass
transfer characteristics of the fluoroborate salt mixture.
After 1000 hr of normal operation, steam was injected
146

ORNL-DWG T4-8062

T 1 1

HEATED AND iNSULATED —=

d N
Y /.,-.__\
.__é;/v"_‘§§
\‘:':\\ /,/ 600
\I : N
\\/\ &;\\_ ] °%°
\ v v
\\_/—.-/\ ‘\\ \\

500

TEMPERATURE " {*C)

[
Q

} \
/ / / / THERMAL CONVECTICN LOOP NCL-13A ‘\\‘\

AN
-
N

WEIGHT CHANGE {mg /cm?2)

\
--..\
h-—"‘—---

v 958 hr \\ ‘\

D 2225 hr

® 3545 hr \ \‘
a 5339 hr 7

A 1,680 hr \
‘ : 0 14,495 hr \

® 16,843 hr

0 10 20 30 40 50 60 70 80 90 100
DISTANCE FROM HD]‘TEST PORTICN OF THE UPPER CROSSOVER {in.)
| l : Y | o | o
| | it i >
UPPER COLD LEG BEND LOWER BENOD HOT LEG
CROSSOVER CROSSOVER (VERTICAL)

(VERTICAL)

Fig. 13.18. Hastelloy N specimen weight changes exposed to NaBF 4-NaF (92-8 mole %) from NCL-13A as a function of position

and time.

into the salt,® and the loop has now operated 18,000. hr
following that injection. Figure 13.19 shows specimen

weight change as a function of position in the loop

(temperature) and time. Note the large increase in
weight change during the first 239 hr after steam
injection. Another abrupt. weight change occurred at

8. J.W. Koger, MSR Prog}am Semignnu, Prbgr. Rep. Aug. 31,
1970, ORNL4622, p. 170.

10,178 hr because of a leak in the cover gas system;
however, the overall mass transfer rate decreased
steadily between these two events.

Loop NCL-20, constructed of standard Hastelloy N
and containing removable specimens in each leg, has
operated for over 14900 hr with the fluoroborate
coolant salt at the most extreme temperature condi-
tions considered (687°C max and 438°C min) for the
MSBR secondary circuit. Forced air cooling of the cold

leg is required to obtain this AT. After 11,900 hr, the

vy

kY
0y

147

ORNL-DWG 71-8060

30 T — T T
|-—HEATED AND INSULATED —=
20 — I
' i — i
. .M =T _%
//] flé—:\\\
io 7 A \“ = —‘fi-.‘:_?\k
. ——
0 ——gf— g—t—¢ S r—r—
\ N
N L ]
v,
-10 N A b\\ ‘N ] 600
\\E\ \ \./ . 'Sy
—_— § v N L
o A _ Lt
E Al 550 &
L -20 -~ D >
g // NN /\ \\\:\fl 0§
- / T~ ] N T+ 500 E:
© _3p \ \ =
b= \ Lt
G .\ 450
— : : THERMAL CONVECTION LOOP NCL-17
T -40 |- N,
o .
-50 |/ / “e 239 hr ) - ' —%
. : o 424 hr .
4 663 hr o
-60 ‘: 12%2% :: | TIME AFTER , ; y
y agesmr [ STEAM INJECTION
o 6030 hr
s 9006 hr
-70 ‘ o 10,178 hr J: ;
~ ¢ 1054 hr BEFORE STEAM
-80 5
"-90
0 10 20 30 40 50 60 70 80 90 100
DISTANCE FROM HOTTEST PORTION OF THE UPPER CROSSOVER (in.)
| | [ el ' -
| | = { |
UPPER COLD LEG BEND LOWER BEND HOT LEG
CROSSOVER (VERTICAL) CROSSOVER (VERTICAL)

Fig. 13.19. Weight changes from Hastelloy N specimens of NCL-17 [NaBF;-NaF (92-8 mole %) salt with steam addition] asa

function of position and time.

maximum weight loss in'the hot leg was measured to be
6.2 mg/cm? (0.2 mil/year assuming uniform loss) and
was accompanied by a maximum gain in the cold leg of
+2.8 mg/cm?. Figure 13.20 shows the weight changes
of specimens in the loop as a function of time and
temperature, and Fig. 13.21 shows the overall picture

of the mass transfer in the system. The chromium

content of the salt increased by 150 ppm. Up to'11,900

hr the weight changes of the specimens and the.
chemistry changes of the salt indicated that the mass

transfer was the lowest yet attained with the fluoro-
borate mixture .in a temperature gradient system and
also confirmed the importance of salt purity on
compatibility with structural material and on operation.

Figure 13.22 shows micrographs of specimens in the
hottest and coldest positions of the loops .after
11,900-hr exposure. Microprobe analysis® of these
specimens showed no detectable concentration gradi-
ents. The deposit on the surface of the cold specimens
was rich in Fe (>25 wt %) and Ni (>64 wt %), but
showed very little Cr or Mo. ‘

Following 12,400 hr of operation, the normal 20-psia
loop ovéfpressure ‘dropped to atmospheric pressure.

‘Investigation disclosed that the safety valve on the

9. Performed by H. Mateer, T. J. Henson, and R. . Crouse of
the Metals and Ceramics Division.
148

ORNL-DWG 71-8059

10
NE NCL-20
~
o 5
£
s S|
= | 460°C COLDEST -
g 5 _figflz‘fl—:?——:"‘:'° SPECIMENS |
=4 =g
T — .
o .\?\M
T _g e T—d 660%C
o T 685°C HOTTEST
£ | SPECIMENS

l

L
S

0 2000 4000 6000 B800Q 10,000 12,000 14,000 16,000
TIME OF OPERATION (hr)

Fig. 13.20. Weight changes of Hastelloy N specimens from
NCL-20 exposed to NaBF4-NaF (92-8 mole %) as a function of
time and temperature. : :

helium regulator had failed and had released helium
from the loop. The distance from the regulator to the
loop was approximately 20 ft, and the gas line was
'/4-in. tubing. The specimens were immediately removed
from the loop and weight changes measured. The
maximum metal loss for this time period (525 hr) was
0.34 mil/year with a correspondingly low weight gain.
Thus, the mass transfer rate was not greatly accelerated.

A salt analysis did not disclose any gross changes. The
valve was replaced, and loop operation continued.

During the next 1850 hr, the maximum corrosion rate
(assuming uniform loss) was 0.7 mil/year. Part of this
increased mass transfer was probably associated with
the regulator problem, but further inspection disclosed
a leaking mechanical pressure fitting. This was repaired,
and the loop is continuing operation.

An experiment was started to evaluate the corrosion
properties of eight brazed Hastelloy N specimens in
NaBF,-NaF (92-8 mole %). The specimens are con-
tained in a nickel capsule at 607°C (1125°F). The
reason for this test is that little, if any, work has been
done to determine brazes suitable for use in the sodium
fluoroborate coolant salt. The test was intended as an
initial. screening study of brazes composed of Au-Ni,
Ag-Cu, Cu, and several types .of Ni-Cr-Fe alloys.
Specimens will be removed periodically for visual
observation and to determine weight changes.

The test has operated for over 3000 hr with all

- specimens showing a net weight gain from +3 to 5

mg/cm?. Between 1200 and 1500 hr of operation a
pressure gage between the He cylinder and the capsule
became defective. During this time some of the speci-
mens lost weight as a consequence of the more highly

ORNL-DOWG 71-80861
T T

T T T T
AlIR-
}—COOLED-‘-*‘HEATED AND INSULATED —=
SECTION
10 : . ]
/\ — 450
E\E ) /‘:Qg — AN 1°%° 5
S . | =4 == . €
o 7 N = ]
£ L= S ™ &
2 - &
< - , / © THERMAL CONVECTION LOOP NCL-20 \' N\ 600 a
Q . =
. . Lt
- -
— /]
510 : . 650
© 3784 hr _
® 6059 hr . — 700
a 8387 hr
a 11,883 hr
_20 1
0 10 20 30 . 40 50 60 70 80. 90 100
DISTANCE FROM HOTTEST PORTION OF THE UPPER CROSSOVER (in.)
a | el | |
| T . =T 1 ‘
UPPER COLD LEG BEND LOWER BEND HOT LEG
CROSSOVER (VERTICAL) CROSSOVER (VERTICAL }

Fig. 13.21. Weight changes from Hastelloy N specimens in NCL-20 exposed to NaBF,;-NaF (92-8 mole %) as a function of

position and time,

wl

iy
149

Y-107939

SO
L[]
Too05 =

Y-=107344

()

Fig. 13.22. Optical micrographs of Hastelloy N specimens from NCL-20 exposed to NaBF s-NaF, (92-8 mole %) for 11,900 hr. (a)
685°C, weight loss 6.2 mg/cm?, etched with glyceria regia; (b) 460°C, weight gain 2.8 mg/cm2, as polished.

oxidizing conditions produced by the air inleakage.
Over the 3000-hr run the iron content of the salt has
decreased 100 ppm while the chromium, nickel, and
molybdenum content has remained unchanged.

150

The above observations suggest that the corrosmn 3

reaction taking place is of the type
FeF, (d) + 2CrF, (d) ~> 2CrF5(d) + Fe(s) ,

which would produce no measurable changes in the
chromium concentration of the salt, but would account
for a decrease in the iron concentration.

13.6 ANALYSIS OF HIGH-LEVEL PROBE
FROM SUMP TANK OF PKP-1

J. W. Koger

Loop PKP-1 was operated by A. N. Smith for over a
year to test a molten-salt pump with the coolant
mixture NaBF,-NaF (92-8 mole %). We had previously
examined the BF; feed and salt level bubbler tube!®
and were asked to examine the spark plug probe
because of its unexpected discoloration.

The spark plug probe consisted of a Y-in.-diam
electrode which was installed April 8, 1968 and was
assumed to be Hastelloy N. However, metallographic
examination of the rod disclosed cracks extending 5
mils into the metal. Spectrographic analysis revealed
that the material was Haynes alloy No. 25 (50%
Co—20% Cr—10% W—10% Ni—3% Fe).

This parallels a finding in 1968 in which the specimen
support rods of thermal convection loops NCL-13 and
-14 were found to be Haynes alloy No. 25 rather than
Hastelloy N.!! That investigation revealed that the
source of the '%-in. rod was a misidentified storage
carton, and we are assuming that the rod found in the
PKP loop was from the same source.

It was shown in the thermal convection loop tests
that the Haynes alloy No. 25 was attacked much more
- than the Hastelloy N. In the case of the PKP-1 loop, the
spark plug probe comprised an infinitesimal fraction of
the loop surface, a fact that would explain the much
heavier attack compared with Hastelloy N.

10. J. W. Koger, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL4676, pp. 211-185.

- lowing overheating,

13.7 FORCED-CONVECTION LOOP
CORROSION STUDIES

W.R.Huntley .J.W.Koger H. C. Savage

Forced-convection loop MSR-FCL-1 was rebuilt fol-
'2 and the loop is being used to
evaluate the compatibility of standard Hastelloy N with
NaBF,-NaF (92-8 mole %) coolant salt at temperatures
similar to those expected in the MSBR secondary
circuit. The loop designation has been changed from
MSR-FCL-1 to MSR-FCL-1A because of changes in the
equipment and in the operating parameters. The new
velocity in the '%-in.-OD, 0.042-in.-wall tubing is
nominally 5% fps. The maximum and minimum salt
temperatures in the loop are 620 and 454°C respec-
tively. Hastelloy N corrosion test specimens are exposed
to circulating salt at 620, 548, and 454°C.

A second forced-circulation loop of improved design,

* - MSR-FCL-2, has completed shakedown tests and is in

operation. This loop will be used to study the corrosion .

and mass transfer of standard Hastelloy N in sodium
fluoroborate coolant at typical MSBR salt tempera-
tures. A new pump design for MSR-FCL-2 provides salt
velocities up to 20 fps. The loop contains three sets of
removable corrosion specimens which are exposed to

- salt at 620, 537, 454°C and salt velocities of 10 and 20

11. J. W. Koger and A. P. Litman, MSR Program Semiannu.

Progr. Rep. Aug. 31, 1968, ORNL-4344, pp. 264 —66.

fps.

13.7.1 Operation of Forced-Convection
Loop MSR-FCL-1A

We completed loop repairs and placed the loop in
operation at simulated MSBR conditions on Aug. 16,
1971. The test has operated smoothly since that time
and 'has accumulated 380 hr at design conditions as of
Aug. 31, 1971,

During the repair period, modifications were made to
the control system to prevent accidental overheating of
the loop piping. About 25 ft of Hastelloy N loop tubing
was replaced, and new corrosion specimens were in-
stalled. The specimen assembly was modified so that
specimens could be clipped rather than tack welded
into the assembly. The 27-ft-long cooler coil was
undamaged and was reused. The NaBF,-NaF (92-8
mole %) coolant salt was removed from the dump tank
and cooler coil by water flushing. This salt had been in
the system for the entire period of operation preceding
overheating (10,335 hr at design conditions). In view of

12. J. W. Koger ét al.,, MSR Program Semiannu. Progr. Rep.

Feb. 28, 1971, ORNL4676, p. 202.

ird 4
LY

LY

the changes in corrosion specimens and operating
conditions, the loop designation was changed to MSR-
FCL-1A.

The LFB pump was disassembled and repaired prior
to reinstallation in MSR-CFL-1A, and new bearings and
mechanical face seals were installed as is customary.
The pump was operated at 3000 rpm in a cold
shakedown for about 24 hr to ensure that the bearings
and seals were operating properly. An oil leakage rate of
about 4 cc/day was observed from the internal mechan-
ical seal and was considered acceptable. The LFB pump
operates at 3000 rpm in MSR-FCL-1A, which is slower
than the 5000 rpm previously used in MSR-FCL-1. It is
hoped that this reduced speed will appreciably increase
the reliability and service life of the bearings an
rotating seals. . ‘

The piping system was filled with 510°C salt from the
dump tank on Aug. 12, 1971. The salt was circulated
isothermally at 540°C for 90 hr at a flow rate of about
0.8 gpm during final system checkout. The flow rate
was then increased to the design value of 2.4 gpm
(pump speed 3000 rpm) for an additional 4 hr at
intermediate temperatures prior to reaching design
conditions on Aug. 16, 1971. The power input to the
loop at design conditions is 63 kW. This power level
represents a practical upper limit for long-term opera-
tion of the power input system because of heating
limitations of the saturable reactor.

13.7.2 Metallurgical Analysis of Forced-Convection
Loop MSR-FCL-1A

A salt sample was taken from the pump bowl on Aug.
16, 1971, after circulating the salt isothermally for 90
hr at 540°C. This initial salt analysis disclosed relatively
large amounts of chromium, iron, molybdenum, and
hydrogen. However, a later sample taken from the
pump bowl at 454°C, after the loop had operated 138
hr at design conditions, showed smaller amounts of
impurities. If any of the impurities were at saturation,
the difference in temperature at which the samples were
taken could explain the apparent discrepancy.

13.7.3 Operation of Forced-Convection
Loop MSR-FCL-2 '

Assembly and preoperational checkout of the second
molten-salt forced-convection loop MSR-FCL-2 (ref.
13) were completed during this report period. The loop

13. W. R. Huntley et al., MSR Program Semiannu. Progr.
Rep. Aug. 31, 1970, ORNL4622, pp. 176-78.

151

was filled with a specially processed sodium fluoro-
borate salt charge, placed in operation, and heat
transfer performance of the salt was measured. After a
short /period of hot operation an O-ring seal failure in
the ALPHA pump required a loop shutdown. The seal
failure was repaired and the test returned to design
conditions. This loop, constructed of Hastelloy N, will
be used to study the corrosion and mass transfer
properties of Hastelloy N in fluoroborate-type coolant-
salt systems at conditions proposed for the MSBR.
Preoperational checkout was completed in accordance
with the operating manual, and all documentation was
reviewed for compliance with the Quality Assurance °

Plan (Engineering Document Q-10566-RB-001-5-0). All

recorders and indicators of temperature, pressure, and
power input were calibrated. Heat loss measurements
were made on heater No. 2 in preparation for deter-
mining the heat transfer performance of the salt.

The salt charge for MSR-FCL-2 was processed to
improve its purity prior to transfer into the loop. The

. first routine processing was done after 25 kg of the salt

components was mixed and sealed in a vessel. The
mixed powders were heated to 149°C and evacuated
with a mechanical vacuum pump for 45 hr to remove
moisture. The salt was then heated to 500°C and 16.7
kg was transferred to a small filling pot. The salt in the
filling pot was heated under sealed-off vacuum to check
for further impurity outgassing. The pressure rose from
28.7 in. Hg vacuum at. 149°C to 26.4 in. Hg at 470°C
which indicated little outgassing above the expected
BF; vapor pressure of the salt. An equilibrium mixture
of He-BF3 was bubbled through the salt at 480°C at
about 100 cc/min for 60 hr, and the effluent gas was
passed through a cold trap at 0°C. This test was run to
check for impurity collections such as were noted
during operation of a liquid level bubbler in the PKP
loop.!* The inlet gas mixture was prepared for the
experiment by mixing dried hetium (<1 ppm H,0)
with BF; in the proper ratio to balance the vapor
pressure of the hot salt.

Impurities were found in the gas mixture after it had
bubbled through the salt. A gas rotameter' downstream
of the cold trap initially became fouled with a clear
liquid. A white film also deposited on the walls of the
glass cold trap. A total of 0.15 cc of brown liquid was
collected in the trap after 60 hr, and the materials
collected appeared similar to those noted by A. N.
Smith!* during operation of the PKP loop. Analysis

14. A. N. Smith et al., MSR Program Semiannu. Progr. Rep.
Feb. 28, 1969, ORNL4396, pp. 102-106.
was inconclusive as it only disclosed major quantities of
Na, B, and F with ppm values for many other elements.
These instances of impurity collection show that
presently available sodjum fluoroborate systems can be
expected to give acid collection and fouling problems
when devices such as liquid level bubblers are used.
From the small amount of material collected, we
concluded that our method was not practical for
removing large amounts of impurities. However, we feel
that improvements could be made on the basic process
which could provide more efficient removal.

The heat transfer performance of NaBF,-NaF (92-8
mole %) coolant salt was measured in heater section No.
2 of MSR-FCL-2 during hot shakedown operations.
Heat transfer data were obtained at Reynolds moduli
from 4200 to 48,000, salt velocities from 1.25 to 10.6
fps, heat fluxes from 43,000 to 157,000 Btu hr ™' ft ™2,
film coefficients from 230 to 2100 Btu hr ™! ft ™2 °F !,
and at salt temperatures from 454 to 616°C. The test

152

section is '4 -in.-OD tubing with a measured ID of 0.416 -

in. Resistance heating is supplied by a three-lug system
with voltage potential applied to the center lug and the
two exterior lugs at ground potential. Therefore, there
is a section at the center lug of the heated length which
has -an interrupted heat flux. The actively heated length
is 11.5 ft which gives an L/D ratio of 331. Guard
heaters are located on the heater lugs and along the
resistance-heated tube to reduce heat losses during the
heat transfer tests. :

Temperature of the bulk fluid is measured by three
thermocouples located in wells at both the inlet and
exit. Wall temperatures along the heated section are
measured by thermocouples that are wrapped 180°
circumferentially around the tubing and clamped
against the wall at about 1-ft intervals. The above-
mentioned thermocouples are all sheathed; insulated
junction, 0.040-in.-OD, Chromel/Alumel and have been
precalibrated. Four bare wire thermocouples
(0.010-in.-OD, Chromel/Alumel) are also placed on the
heated wall and read out on a potentiometer for
comparison with the sheathed thermocouples recorded
by the DEXTIR data logger. The readings of the two
types of thermocouples were in good agreement.

The heat transfer data obtained with sodium fluoro-
borate salt in MSR-FCL-2 have been put in dimension-
less form as shown in Fig. 13.23. The data are in good
agreement with the empirical correlation of Sieder and
Tate'® which is shown by the solid line. The plotted
data points were obtained with more precise thermo-

‘couple techniques than those used in earlier work! ¢ by
the writer; therefore, the data of Fig. 13.23 are
considered the best available evidence that sodium

ORNL-DWG 71-9642
5 .
L R
| L T ] io '14
_ 08, ' fig)'
Ny, = 0.027 M8 1,3 (2 \ 3
Py
» L
[T 13
HEATED SECTION L/0 RATIC = 33{ ?‘ .
T 2
5 m| = rd
SN e
2] o N
vEn_ 5 {'...‘.
7
2 ]
.
10'
103 2 5 10? 2 5 10°

REYNGCLDS NUMBER

Fig. 13.23. Heat transfer characteristics of NaBF4-NaF (92-8
mole %) flowing in-0.416-in.-ID tube.

fluoroborate does indeed perform as a typical heat
transfer fluid.

A gas leak occurred at an O-ring seal within the
ALPHA pump on July 7, 1971, as the test was being
brought to design operating conditions. At that. time
the salt had been in the loop for 510 hr and had been
circulated by the pump for 56 hr. The salt inventory
was sampled to check for impurity additions resulting
from the gas leak, and none were found. The system
was then drained to repair the pump. The gas leak
resulted from BF; attack of a BUNA-N O-ring in the
mechanical seal cartridge. The manufacturer of the
mechanical seal indicated during early engineering
discussions that Viton (a DuPont fluoroelastomer
which is more resistant to BF3 than BUNA N) would be
used in this application, but this was changed during
actual assembly of the seals. The pump has been
reassembled with new seal cartridges containing Viton
O-rings, and MSR-FCL-2 has just been placed in
operation at design conditions. Details of the ALPHA
pump O-ring seal replacement and other pump engi-
neering problems are discussed elsewhere in this report,
Sect. 3.7, MSBR Pumps. '

15. E. N. Sieder and G. E. Tate, “‘Heat Transfer and Pressure
Drop of Liquids in Tubes,” Ind. Eng. Chem. 28(12), 1429-35
(1936).

16. W. R. Huntley, MSR Program Semiannu. Progr. Rep. Feb.
28, 1969, ORNL4369, p. 254.

>

o

~
rd
&

1]

[ X

13.7.4 Metallurgical Analysis of Forced-Convection
Loop MSR-FCL-2

Corrosion test specimens were in the loop while the
heat transfer measurements were made. These speci-
mens were examined after salt exposure of 510 hr and
salt circulation with the ALPHA pump for 56 of those
hours. During the heat transfer measurements the pump
operated at speeds between 1400 and 5400 rpm, and
the specimens were exposed to temperatures between
440 and 620°C.

The weight changes were higher than expected (maxi-
mum weight loss of 1.5 mg/cm?) and showed a definite
velocity effect. The specimens exhibited expected
temperature-gradient mass transfer behavior with
weight losses in the high-temperature region, very little
change in the medium-temperature region, and weight
gains in the cold region.

During the time of the heat transfer measurements,
the chromium in the salt increased from 64 to 82 ppm,
and the iron changed from 359 to 347 ppm. The oxide
content varied from 500 to 800 ppm, and the H*
content increased from 27 to 31 ppm. Mass transfer was
not considered excessive, and the decision was made to
go to design conditions.

13.8 CORROSION OF HASTELLOY N IN STEAM
B.McNabb  H. E.McCoy

The modifications to the sample chamber at Bull Run
to accommodate tube burst tests has been_.completed
and the test facility placed back in operation. Some of
the unstressed weight gain specimens have accumulated
8000-hr exposure and are still following the same trends
reported previously.'” The Hastelloy N specimens,
both air and vacuum melted, have total weight gains of
about 0.4 mg/cm? at 538°C. The same heats of
Hastelloy N exposed to steam at 593°C in the Florida
steam plant for approximately 15,000 hr have weight
gains of about 3.5 mg/cm?.

Two of the Hastelloy N tube burst specimens installed
in the Bull Run facility failed prematurely. The highest

153

believed that these failures were caused by either flaws
in the specimens, poor machining of the gage section, or
inaccuracies in measuring the wall thickness of the
specimens. Later evidence indicates this was not the
case. Figure 13.24 is a photograph of these double-
walled tube burst specimens, with the outer tube
removed.

The 77 000 -psi specimen failed catastrophically, but
the 52,500-psi specimen developed a small crack
starting at the tip of the arrow shown at the right of the
photograph. Even though the specimens failed in a
short time, they were not removed from the facility
until 1000 hr, so the outside of the inner tube was
exposed to steam for about the same time as the
outside of the outer tubes shown at the top of the
photograph. -

These two failed specimens were replaced by speci-
mens stressed at 56,000 psi (0.014 in. wall) and 50,000
psi (0.016 in. wall). These stresses bracket the lower
stress of the failed specimens (52,500), and these
replacement specimens have been in test over 1000 hr
without failure, indicating the other specimens had
failed prematurely. These latter specimens were
machined carefully and had a superior surface finish
and appear to have uniform wall thicknesses.

Figure 13.25 is a photomicrograph of the area of
failure of the 52,500- -psi| tube burst. It appears that
stress corrosion crackmg has occurred with the cracks
starting on the ID (steam side) of the tube. Since the
other tubes stressed in the same range as this one have

" not failed, it raises the possibility that something could

have happened to this individual specimen. Either some

- contaminant was left in the tube following fabrication

stressed (77,000 psi, 0.010 in. wall thickness) specimen

‘failed in'1 hr, and the next-highest stressed (52,500 psi,

0.015 in. wall thickness) failed in 3.7 hr. These speci-
mens were stressed considerably above the yield stress
at 538°C (40,000 psi), and yielding might continue in a

thin-walled tube to produce rupture quickly. It was first.

17. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL4676, pp. 216-17.

due to poor cleaning practice or some contaminant was
introduced by the steam during operation. An investi-
gation is under way to check water quality records
during the run to determine if any unusual conditions
existed during the run. The plant had been in a period
of sustained operation for over one month before the
first specimens were inserted.

Preliminary results from studies by Hammond at
Dunedin, Fla:-showed that Hastelloy N may be suscep-
tible to stress corrosion cracking in-steam containing
sodium chloride and oxygen. Some of the test con-
ditions actually. deposited NaCl-on the material. H. W.
Pickering, F. H. Beck, and M. G. Fontana'® proposed a

18. H. W. Pickering, F. H. Beck, and M. G. Fontana, “Rapid
Intergranular Oxidation of 18-8 Stainless Steels by Oxygen and

' Dry Sodium Chloride at Elevated Temperatures,” Trans. ASM

53,793-803 (1961).
o ' 2 Y-108427
INCHES

/

HASTELLOY N TUBE BURST SPECIMEN HASTELLOY N TUBE BURST SPECIMEN
FAILED IN 1.0 hr IN STEAM AT 77,000 psi FAILED IN 3.7 hr IN STEAM AT 52,500 psi

Fig. 13.24. Hastelloy N tube burst specimens removed from the Bull Run Steam Plant atter failure. The small tube in the test
section was exposed to steam at 538°C and 3500 psi.

Y—109669 T

™ 160X

Tor

Fig. 13.25. Photomicrograph of Hastelloy N tube burst specimen. Failure occurred after exposure to steam for 3.7 hr at 52,500
psi at 538°C. Steam was on the inside of the tube.

¥ em

18]

L

(Y

155

mechanism for rapid intergranular deterioration of
austenitic stainless steels by oxygen and dry NaCl that
may be pertinent to these observations. This mechanism
was based on the formation of a nonprotective scale
containing NaCrO, from the reaction of NaCl and O,
with Cr, 03 and Cr,3C4, with reaction rates increasing
in the order listed. Since Hastelloy N contains 7%
chromium, it was decided to test it in contact with
NaCl at 593°C. An air-melted heat (5065) with 16.5%
Mo, 7.2% Cr, 3.9% Fe, 0.55% Mn, 0.6% Si, and 0.065%
C and a vacuum-melted heat (2477) with 16.3% Mo, 7%
Cr, 4.1% Fe, 0.05% Mn, 0.05% Si, and 0.06% C were
tested in air at 60,000 psi and 593°C in the uncoated
and coated conditions. Rod specimens having gage
sections 1.125 in. long X 0.125 in. diam were used for
these creep tests. The specimens were annealed 1 hr at
1177°C in argon prior to testing. The NaCl was applied
to the specimens by painting a saturated NaCl solution
on a ',-in. length of the gage section and allowing it to
dry before installing in the furnace. The coatings were
typically about 5 mils thick before testing.

Figure 13.26 shows a plot of the percent elongation
vs time for heat 5065 at 593°C and 60,000 psi in air for
the uncoated and coated specimens. The uncoated

ORNL-DWG 71-13682

250

22.5

200

17.5

15.0

12.5

ELONGATION (%}

100 O UNCOQATED

A COATED WITH NaCl

75

5.0

“A‘ o ole
\35%6.0‘0 ofo ©°°
Q 20 - 40 60 80 100 120 140 160
TIME (hr)

Fig. 13.26. Comparative creep curves of Hastelloy N (heat
5065) with and without NaCl coating stressed at 60,000 psi and
593°Cin air,

specimen had about two times the rupture life of the
coated specimen and a slightly lower creep rate. Figure
13.27 is a plot of the percent elongation vs time for
heat 2477 in the uncoated and coated conditions at
593°C and 60,000 psi in air. The first two specimens
designated anneal 000 had a mill anneal of 1 hr at
1177°C prior to machining, and the third specimen had
an additional anneal of 1 hr at 1177°C after machining
but prior to testing.

It appears that there is little effect of the NaCl on the
creep strength of Hastelloy N at these conditions, but
further testing will be required to determine the
magnitude of the effect at various stresses. The reason
for the difference in creep strengths of heats 5065 and
2477 is not presently understood, since at higher
temperatures, lsuch as 650 and 760°C, the strengths are
quite close.

Figure 13.28is a photomicrograph of the rupture of
uncoated heat 5065 after testing at 593°C and 60,000
psi in air. Figure 13.29 is a photomicrograph of the
rupture of heat 5065 coated with NaCl and tested at
593°C and 60,000 psi in air. A comparison of these
photomicrographs reveals very little if any difference
due to the presence of the NaCl.

Figure 13.30 is a photomicrogfaph of the rupture of
uncoated heat 2477 after testing at 593°C and 60,000
psi in air. Figure 13.31 is a photomicrograph of the
rupture of heat 2477 coated with NaCl and tested
under the same conditions. A comparison of these
photomicrographs and the rupture lives of these tests
(468.3! hr without and 475.9 hr with salt) shows NaCl
has little if any effect on vacuum-melted Hastelloy N at
these conditions. :

ORNL-DWG 71—13684

* T L
o TEST AS RECEIVED
{2 |-a TEST AS RECEIVED WITH SALT ® #
o TEST ANNEALED thr AT 1177°C WITH SALT ¥*
10
&
= 8
o
S
3 6
=z
o
S .
w4 L s
: - . s Iy
’ _ et A&&SCDO
> o .«6‘%9
| " Mgflfflr ®

¢ 50 100 {50 200 250 300 350 400 450 500
TIME (hr) '

Fig. 13.27. Compa:étive creep curves of Hastelloy N (heat
2477) with and without NaCl coating stressed at 60,000 psi and
593°C in air. :
156

Y—109437

Fig. 13.28. Photomicrograph of the fracture of Hastelloy N (heat 5065) stressed at 60,000 psi at 593°C without NaCl coating.

Etchant: Gly Regia.

13.9 EVALUATION OF DUPLEX TUBING
FOR USE IN STEAM GENERATORS

B.McNabb  H.E.McCoy

Duplex tubing has been proposed for steam generator
service with Incoloy 800 on the inside (steam side) and
nickel 280 (pure nickel with 0.05% Al, O for grain size
control) on the outside or salt side. A piece of the Inco
duplex tubing reported previously'? was further evalu-
ated by mechanical property tests. A 12-in. piece of the
0.750-in.-diam tubing was tensile tested in a Baldwin
Tensile Machine at 25°C and a displacement rate of
0.05 in./min. The 0.2% offset yield stress was 37,600
psi, the ultimate tensile stress was 70,780 psi, and the
elongation 50.5% in 2 in. It appears that at room

19. H. E. McCoy and B. McNabb, MSR Program Semiann.
Progr. Rep. Aug. 31, 1970, ORNL4622, pp. 181-83.

temperature the Ni 280 and the Incoloy 800 contribute
equally to the strength in the tensile test.

A tube burst specimen was tested at 538°C in an
argon atmosphere with argon internal pressure and
ruptured in 3263 hr. Calculating the stress on the entire
duplex wall thickness assuming 0.065 in. each for the
Incoloy 800 and Ni 280 would give a hoop stress of
28,720 psi. The hoop stress in the Incoloy 800,
assuming the Ni 280 did not add any strength, would be
46,000 psi. The rupture life of 3263 hr for a stress of
46,000 psi at 538°C is in good agreement with the data
in Bulletin T40 (Inco) for Incoloy 800. Therefore, it
appears that the nickel contributed very little to the
strength at this temperature.

Figure 13.32 is a photograph of the tube burst
specimen with Incoloy 800 end plugs and nickel rings
welded to the tube to reinforce the end closures. Inco
82T weld wire and GTA welding were used to make the
dissimilar metal welds. Dye penetrant was applied to
157

5
S
o S
> = 3%
8 22 B

e a e o
oo P .

g i B

Y—109439

=

0.035 INCHES
TS 100X

T

Fig. 13.29. Photomicrograph of the fracture of Hastelloy N (heat 5065) stressed at 60,000 psi at 593°C with NaCl coating.

Etchant: Gly Regia.

the gage section of the tube, and the many longitudinal
cracks in the nickel 280 can be seen clearly in the
figure. Figure 13.33 is a photomicrograph of the Ni 280
tubing showing the extent and depth of the cracking in
the outer portion of the duplex tube. Figure 13.34 isa
photomicrograph of the rupture area with the crack in
the inner Incoloy 800 tube and extending on through
the Ni 280 outer tube. The Incoloy 800 was not
cracked around the circumference as was the Ni 280,
and the interface between the two materials was
metallurgically sound.

The tubing was made by coextruding the two
materials into a tube shell and then drawing into tubing.
Since the tube burst specimen was profusely cracked
longitudinally, it was thought that prior working
direction might have influenced the cracking behavior.
Specimens from some Ni 280 sheet already on hand
were cut longitudinal and transverse to the rolling
direction and creep tested at 538°C and 20,000 psi in
argon. Figure 13.35 is a plot of the percent elongation

vs time for a pair of these tested in the as-received
condition. It appears that the longitudinal specimen is
slightly stronger, with a longer rupture life and lower
minimum creep rate. However, the fracture strains are
quite high for both specimens.

The proposed use for the tubing would have steam on
the inside (Incoloy 800 side), but it would be desirable
to know if Ni 280 would be compatible with steam in
the event of a leak. Some specimens of Ni 280 sheet
were exposed to 3500 psi steam at 538°C in the Bull
Run facility for 2000 hr. The specimens gained about
75 mg/em® in this period and appear to be almost
completely oxidized as shown in Fig. 13.36. The green
flaky oxide was easily removed, and the remaining
black oxide was easily broken in handling the specimens
with tweezers, The thickness of the specimens had
doubled because of the low density of the oxide. It
appears that Ni 280 is not compatible with steam under
these conditions.

158

Etchant: Gly Regia.

ig. 13.30. Photomicrograph of the fracture of Hastelloy N (heat 2477) stressed at 60,000 psi at 593°C without NaCl coating.

159

Y-109135

0.035 INCHES.
To 100X

Fig. 13.31. Photomicrograph of the fracture of Hastelloy N (heat 2477) stressed at 60,000 psi at 593°C with NaCl coating.
Etchant: Gly Regia.

Y-106620

INCHES

Fig. 13.32. Specimen of duplex Incoloy 800—Ni 280 tubing tested at 538°C and 46,000 psi hoop stress in the Incoloy 800. The
specimen failed in 3263 hr with 2.3% diametral strain. The longitudinal markings are due to dye penetrant that was absorbed by the
cracks.

160

Fig. 13.33. Photomicrograph of the Ni 280 from the specimen in the previous figure. As polished

BS

161

Y-106952

Fig. 13.34. Photomicrograph of the primary failure region in the tube in Fig. 13.32. Incoloy 800 is on the inside and the Ni 280
on the outside. As polished. Magnification 35X.

ORNL-DWG 71-13683

© LONGITUDINAL
+ TRANSVERSE

| o9
b

150 200 250 300 350 400 450 500
TIME (hr)

' Fig. 13.35. Comparative creep curves of Ni 280 sheet at
538°C and 20,000 psi in an argon atmosphere. Samples taken
from longitudinal and transverse orientation.
Y-106551

269

INCHES

Fig. 13.36. Samples of Ni 280 exposed to steam for 2000 hr at 538°C and 3500 psi.
Ny 3

14. Support for Chemical Processing

J. R. DiStefano

Reductive-extraction reprocessing has been proposed
as a method of removing protactinium and fission
products from an MSBR fuel. The process involves
extracting uranium, protactinium, and rare earths from
the fuel salt into molten bismuth containing dissolved
thorium and lithium as reductants. Our materials
program is aimed at selecting the most promising
material for containing these Bi-Li-Th solutions at 500
to 700°C. '

At present our efforts are concentrated in two areas.
One is the fabrication of an experimental reductive-
extraction unit from molybdenum. This unit, which
involves a packed column, will allow us to obtain
metallurgical data on the test equipment as well as
chemical engineering process data under a variety of
conditions, The second part of our program is to
evaluate the compatibility of molybdenum, TZM, tan-
talum, T-111, brazing alloys, and various grades of
graphite with several Bi-Li solutions under reprocessing
conditions.

14.1 CONSTRUCTION OF A MOLYBDENUM
REDUCTIVE-EXTRACTION TEST STAND

J. R. DiStefano

We are constructing a molybdenum system that will

H. E. McCoy

OD); gas lift pumps; freeze valve; and Y%-in.-OD X
0.020-in.-wall, ¥%-in.-OD X 0.025-in.-wall, % -in.-OD X
0.030-4in.-wall, and %-in.-OD X 0.080-in.-wall inter-
connecting lines. Molten salt and bismuth will be
countercurrently circulated through the column, sep-
arated in the disengagement sections, and returned to
the elevated head pots for gas separation, sampling,
flow measurement, and gravity flow back through the
column. The system is designed to operate isothermally
in the range 550 to 650°C.

Back-extrusion, welding, brazing, and mechanical
joining techniques have been developed to fabricate and
assemble the loop. The following is a general outline of

“the fabrication schedule:

be used for metallurgical and chemical engineering

evaluation. Details of the design of this test stand have
been reported elsewhere.!’? Basically, it consists of a
5-ft-long 1%-in.-OD packed column with enlarged (3%
in. OD) top and bottom sections; feed pots (3% in.

163

1. fabrication of containers and column,

2. machining of loop components,

3. fabrication of head pot and column subassemblies,
4. mounting of subassemblies on test stand supports,

5. joining of subassemblies and other loop components.

We have completed primary fabrication of the head
pots, enlarged end sections of the column, and the

1. W. F. Schaffer, E. L. Nicholson, and J. Roth, “Design of a
Processing Materials Test Stand and the Molybdenum Reductive
Extractive Equipment,” MSR Program Semiannu. Progr. Rep.
Aug. 31, 1970, ORNL-4622, pp. 112-13.

2. E. L. Nicholson, “Conceptual Design and Development
Program for the Molybdenum Reductive Extraction Equipment
Test Stand,” ORNL-CF-71-7 (July 1, 1971).
column itself. Loop components for subassembly fabri-
cation are presently being machined.

Evaluation of three sizes of commercially obtained
molybdenum tubing (%, in. OD X 0.020 in. wall, % in.
OD X 0.025 in. wall, and *% in. OD X 0.030 in. wall)
has revealed differences in hardness, recrystallization
behavior, and ductility. Table 14.1 shows the variation
in hardness and microstructure that occurred as a
function of heat treatment. The data for the %-in.-OD
tubing indicated it was partially recrystallized in the
as-received condition, and heating to 925°C resulted in
complete recrystallization. Contrarily, both the %- and
Y,in.-OD samples were received in a cold-worked
condition. Heat treating to 925°C did not alter the
hardness or microstructure of the Y,-in.-OD tubing, but
the '%-in.-OD sample softened and was almost com-
pletely recrystallized. :

To evaluate the ductility of the tubing as a function
of temperature, we devised a somewhat qualitative test
in which a ring sample was impact squashed a pre-
determined amount. In these tests the ring becomes
elliptical, and we have defined the amount of defor-
mation in terms of a displacement (original ring diam —
minor axis diam of ellipse) divided by the original
diameter of the ring. The data.obtained have been
summarized in Table 14.2. Note that the %-in.-OD
tubing was “‘ductile” (no cracks) at room temperature
while, under the conditions of these tests, we were
required to heat the %:-in.-OD tubing to 150 to 250°C

164

and the Y,-in.-OD tubing to 300°C before they became
ductile. We traced the behavior of the ¥%;-in.-OD tubing
to a brittle layer on the ID as indicated in Table 14.3.
Removal of 0.004 in. from the ID (0.002 in. from the
wall thickness) resulted in lowering the temperature at
which ductility was observed from 300 to 125°C; when
0.006 in. was removed from the ID the material was
ductile at room temperature. Chemical analysis in-
dicated that material from near the ID of this tubing

. contained higher oxygen and carbon concentrations

compared with bulk sample analyses. We suspect that
contamination of the tubing occurred during fabri-
cation; inadequate cleaning procedures, perhaps due to
the difficulty of getting a pickling solution into the
relatively small diameter opening, resulted in a brittle
zone near the ID. To determine if ID contamination
might also be responsible for the brittle behavior of the
%-in.-OD tubing, we removed (by etching) 0.001 to
0.002 in. from the inside. Subsequent tests then showed
good ductility at room temperature for this material
also.

The fabrication schedule used in making the %-
in.-OD tubing could also have contributed to its lack of

- ductility. We noted that the %-in.-OD tubing com-

pletely recrystallized when it was heated for 1 hr at
925°C. Since recrystallization temperature is a function
of the prior history of the material, we felt the
fabrication schedule merited investigation. Additional
¥-in.-OD tubing made by a different schedule was

Table 14.1. Hardness and microstructure of molybdenum tubing
as a function of heat treatment

fllxzbelr?; Condition  Hardness Microstructure
(OD in) DPH (1000 g)
A As received? 237 Cold worked
A 1 hr at 800°C 258 Cold worked
Y t hrat 925°C 250 Cold worked
A As received? 200 Partially recrystallized (10—15%)
% 1 hrat 800°C 2437 Partially recrystallized -
3/3 1 hr at 925°C 168 Completely recrystallized (
Y, . As received? 242 Cold worked
' 1 hr at 700°C 255 Cold worked
A 1 hr at 800°C 243 Cold worked
1/2 v 1 hr at 900°C 236 Very slightly recrystallized
A 1 hrat 925°C 206 Almost completely recrystallized (90%)

9The tubing was stress relieved for 1 hr at 870°C before delivery to us.

LY }

181y,
hiw

10

165

Table 14.2, Ductility of molybdenum tubing as a function of deformation temperature

S(fi‘e . Deformation Temperature
tubing disp!acement/tube ©0) Observation
(OD in) diameter (%)

l/4 8 25 Cracked where 1D in tension
l/4 8 100 Cracked where ID in tension
A 8 200 Cracked where ID in tension
1/4 4 25 Cracked -where ID in tension
Y 4 100 Cracked where ID in tension
g/ 4 150 Cracked where 1D in tension
A 4 175 Cracked where ID in tension
Y 4 200 Cracked where ID in tension
1/4 4 250 Cracked where ID in tension
% 4 300 No cracks

% 6.5 25 Fractured into four pieces
A 6.5 100 Cracked in three places

3/8 6.5 150 Cracked in three places

3/8 6.5 250 No cracks

3/3 3.25 25 Fractured into four pieces
3/8 3.25 100 Cracked in four places

% 3.25 150 No cracks

*% 3.25 175 No cracks

3/3 3.25 200 No cracks

1/2 10 25 No cracks

acquired, and some improvement in as-received duc-
tility was found. However, removal of 0.001 to 0.002
in. from the ID greatly improved its room-temperature
ductility, indicating that ID contamination is definitely
the most significant problem.

Table 14.3. Mechanical behavior of %-in.-OD tubing
as a function of removing incremental layers
from the ID surface

Material

removed Deformation Temperature .
displacement/tube - ‘Observation

from diameter (%) ¢0

ID (in.)
0.002 4 25 Hairline crack
0.002 4 125 Hairline crack
0.004 4 25 Hairline cracks
0.004 4 125 No cracks
0.006 4 25 No cracks
0.006 . 4 125 " No cracks
0.008 4 25 No cracks
0.008 4 125 No cracks
0.010 4 25 No cracks
0.010 4

125 No cracks

A roll-bonding technique has been developed to join
the Y-, %-, and %-in.-OD tubes to the closed-end feed
pots and upper and lower column disengagement
sections. Some of the commercially obtained tube
expanders that we are using to make these joints are
shown in Fig. 14.1. The tool fits inside the tube to be
joined such as the one shown extending into the boss of
the typical back-extruded half section in Fig. 14.1.
Expandable tube. rollers .then mechanically force the
tube against the surface of the boss to which it is to be
joined. Because of its room-temperature brittleness,
molybdenum joints are more easily made at 250°C; in
order to prevent contamination of areas that must be
subsequently joined by welding, an inert atmosphere is
required (Fig. 14.2). Leak-tight joints (<1 X 1077 atm
ccfsec) have been made under the following conditions:

. Tube size Torque
(OD in.) (in.1b)
h 10-12

% 30-35

Y, © 55-60
166

PHOTO 4990 — 74

Fig. 14.1. Tube expanders used in roll-bonding molybdenum.

These joints were also leak-tight after they were
thermally cycled ten times from room temperature to
650°C.

To further strengthen roll-bonded joints, each joint
will be sealed by a chemically vapor deposited tungsten
layer on the bismuth or salt side and will be back
brazed with the alloy Fe—15% Mo—5% Ge—4%C—1% B
on the inert gas atmosphere side. A macrophotograph
of a cross section through a typical joint and blowups
of the three types of seals are shown in Fig. 14.3.

More detailed information on the design and purpose
of the molybdenum test stand is presented in Part 4 of
this report. Progress on welding, brazing, and fabri-
cation of molybdenum components is reported in this
section.

14.2 FABRICATION DEVELOPMENT
OF MOLYBDENUM COMPONENTS

A. C. Schaffhauser ~ R. E. McDonald

We have completed development of a back-extrusion
process for fabricating 3%-in.-diam closed-end molyb-
denum vessels for head pots and disengaging sections of
the molybdenum test stand. The process has been
described previously.> A starting blank, tooling, and a
completed back extrusion are shown in Fig. 14.4. The

3. R. E. McDonald and A. C. Schaffhauser, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp.
253-54.

167

PHOTO 1991-74

Fig. 14.2. Roll-bonding molybdenum under inert atmosphere at 250°C.

tooling consists of a 3% -in.-diam plunger and a split die
with an oblong hole in the closed end to produce a boss
needed for support of inlet and exit tubes. All surfaces
of the tooling in contact with molybdenum are coated
with plasma-sprayed ZrO, to prevent chemical inter-
action.

We have fabricated all of the required eleven 37%-
in.-diam back extrusions for the molybdenum test
stand. Four of five 5-in.-long back extrusions have been
machined for the salt and bismuth head-pot assemblies,
leaving one spare. Two of three 8-in.-long back ex-
trusions are being machined for the lower disengaging
section of the packed column, also leaving one spare.
All of these back extrusions were accomplished at
1600°C with a stem load of 800 to 900 tons.

In order to make the lower disengaging section of the
column from two back extrusions of approximately
equal length, 9% ¢-in.-long half sections are required. A
12-in.-long die was procured, and three half sections
between 9 and 10 in. long have been back extruded at
1700°C. One of the half sections was made in two
steps. Under similar conditions, it did not back extrude

Y—1409360

Fig. 14.3. Section through molybdenum roll-bonded joint sealed by chemical vapor deposition and back-brazing.

the required length as did the others. It was reheated to
1700°C, and by carefully aligning it with respect to the
die it was again back extruded to increase its length
without altering the preformed boss.

Our studies have shown that back extrusion of
molybdenum depends upon the quality of the starting
blank, back-extrusion temperature, and reactivity be-
tween the tooling and molybdenum. Some of the early
starting blanks had small surface cracks, and we found
it difficult to avoid cracks in the finished product.
Initially we used a ZrO,-coated plunger and die, but the
die length was short and allowed the molybdenum to
contact the tool steel container of the press along its
OD as it was being back extruded. Interaction of the
molybdenum with the steel caused some tearing of the
OD surface. This was eliminated by use of a ZrO,-
coated long-length die as shown in Fig. 14.4. We also
found that recoating both plunger and die before each
back extrusion ensured high-quality finished products.
Since coating and tooling life is dependent upon
temperature, we tried temperatures from 1350 to
1700°C during our development period. Unfortunately,

Y=-107985

by 4

INCHES

Fig. 14.4. Molybdenum back-extrusion tooling with starting blank and as-extruded product.

we often had more than temperature as a variable and
were unable to conclude what might be the optimum
temperature. However, we did find we were able to
consistently produce good quality back extrusions at
temperatures from 1600 to 1700°C. Tensile and bend-
test specimens have been machined from back ex-
trusions made at several temperatures for mechanical
properties evaluation.

14.3 WELDING MOLYBDENUM
A.J.Moorhead

Three major types of welded joint (tube-to-header,
header-to-header, and tube-to-tube) are required for
fabrication of the molybdenum test stand for chemical
processing. Our initial work on these three joints was

described in an earlier report.* During this period we
have fabricated more complex structures which sim-
ulate actual loop components. We are also optimizing
our welding procedures and are designing and building
weld fixtures.

Asan aid to fabrication, we have begun constructing a
full-scale mockup of the molybdenum test stand.
Wooden models of the four pots were made, and short
stubs of Monel tubing were inserted in the appropriate
holes in the bosses. The tubes were previously bent to
the shapes required of the molybdenum tubes. These

4. A. J. Moorhead and T. R. Housley, MSR Program
Semiannu. Progr. Rep. Feb. 28, 197, ORNL4676, pp.
220-21.

! COUNTERBORE
2-4mils GREATER THAN
- TUBE DIAM FOR

BRAZE CLEARANCE‘—I l*—

Vgin.

|/ain.

TYPICAL
DESIGN FOR || 0.020in.
ROLL-BONDING (TYP)
AND BACK BRAZING

4"] l"“llsin.
FOR GTA WELD

ORNL-DWG 74-3745

1/|5in.

g
t
TYPICAL DESIGN FOR

"\ TUBE-TO-HEADER
EB WELD

22

ALTERNATE, FOR
EB WELD

< (7R 7 T T TTTE7D

QUTLET

Fig. 14.5. Feed pot for molybdenum chemical processing loop.

subassemblies are presently being used to check out the
rotary fixture for girth welding to ensure that the
fixture and pots are compatible. Subsequently, the pots
will be interconnected with full lengths of tubing to
evaluate the procedures and fixturing for tube-to-tube
welding and brazing in the glove box-and in the field.
‘We have developed a ‘procedure. for electron-beam
welding the 7-in.-diam tube-to-header’ type of joint.
The standard 80-mil wall thickness of the molybdenum
tubing will be machined down to 50 mils for this weld.
Therefore, 7%-in.-OD X 0.050-in.-wall tubes were
welded to 1%-in.-OD cylinders representing thé boss on
the back-extruded header. The weld joint was preheated
to about 700°C with the defocused electron beam.
Welding parameters were 140-kV accelerating potential,
10-mA beam current, 11-in./min travel speed, with the
beam defocused 15 mA. This is the final size of tubing
to be welded in developing procedures for welding the
tube-to-header type of joint.
- Two Yh-in.-diam tubular weirs were electron-beam
welded into one of the %-in.-thick molybdenum baffle

plates for the head pots. These welds, which were of the
tube-to-header type, were made by rotating the electron
beam around the joint rather than by rotating the
workpiece under the beam. The speed of beam rotation
was manually controlled. This technique is very useful
in welding parts in which the weld joint is off-centered
(as- in this case) or otherwise difficult to rotate.
Visually, both welds looked good, and no defects were
detected by fluorescent penetrant inspection.

-Three sets of back-extruded headers were successfully
joined by girth welds to form three nearly full-size feed
pots. The first set was welded using the electron-beam
process and the step joint (shown as the “alternate”
joint) in Fig. 14.5. Both headers were preheated to
above 700°C with the defocused electron beam prior to
welding; earlier work had revealed that preheating was
necessary to prevent weldment cracking, although the
minimum preheat temperature has not been de-
termined. The welding parameters were: 140-kV ac-
celerating potential, 20-mA beam current, 41.5-in./min
work travel speed, and the beam defocused 10-mA from

e

!
sharp focus. No weld defects were revealed by flu-
orescent-penetrant inspection. However, metallographic
examination revealed that weld penetration was about
45 mils rather than the desired 100 mils. We will try to
overcome this shortcoming on the next practice weld
by using a more sharply focused beam and a higher
beam current.

The other two sets of headers were joined using the
manual gas tungsten-arc process in an argon-filled
chamber. Two weld passes were required to fill the 90°

vee-groove weld joint (Fig. 14.5), using low-carbon,
low-oxygen molybdenum filler wire. No weld preheat
was required. One of these welded pots is shown in Fig.
14.6 together with the two back-extruded headers. No
defects were detected on either arc-welded feed pot
when they were fluorescent-penetrant inspected.

To more closely simulate conditions required in
fabrication of the test stand, additional tube-to-tube
welds have been made with the orbiting arc weld head.
Longer lengths of tubing were joined with the axis of

Y-106804

Fig. 14.6. Molybdenum back-extruded half sections joined by GTA welding.

the tube vertical. As we expected, problems of axial
tube misalignment after welding and weld cracking due
“to excessive restraint of the joint occurred. In order to
overcome these difficulties, we have procured longer
weld inserts to provide better internal support and a
yoke type of fixture for external support. The latter
will be left in place around the joint until after the
sleeve surrounding the weld has been attached.

"We are also in the process of optimizing our welding
procedures for each size of tubing to increase the
reproducibility of these welds. Helium-leak-tight tube-
to-tube welds have been made in all the required sizes
of tubing, but the penetration achieved on the -
in.-OD tubing has been only about one-third of the
50-mil wall thickness. The welding current required to

172

make these latter welds is relatively high (110 amp),.
and the welding speed slow (11 in: /min), resulting in"
overheating of the orbiting-arc weld head. Therefore,
we are investigating the use of a weld cycle in which the-
current is pulsed between a high current (to get- deeper;
_ penetration) and a lower current to’ permlt some’

cooling of the head.

14.4 DEVELOPMENT OF BRAZING TECHNIQUES
FOR FABRICATING THE MOLYBDENUM
TEST LOOP

N.C.Cole

The iron-base alloy, Fe—15% Mo—5% Ge—4% C—1%
B, has been selected as the filler metal for back-brazing
joints in sections of the chemical processing test loop
that will contain bismuth. Back brazing will serve two
purposes. It will provide reinforcement to the weld
zone which is brittle at room temperature, and it will
provide a seal should a leak develop through a roll bond
or a cracked weld. The alloy selected satisfies the
requirements of having a moderately low brazing
temperature (<1200°C) and adequate resistance to

bismuth at 650°C. Since specific brazing techniques for -

the variety of joints encountered have never been

developed previously, much of our work has been

devoted to basic process development and im-
provement. We have developed- several methods of
~ brazing using resistance and high-frequency induction as
the heating sources.

14.4.1 Resistance Furnace Brazing

Small thin-walled tubes have been brazed to lafge pots

3% .in. OD X 9%, in. long X Y% in. wall) in vacuum in a
resistance furnace. Into each end of the container
shown in Fig. 14.5, two or more short lengths of tubing

must be roll bonded or welded and then back brazed.
At the same time, a split ring covering the girth welds of
the pots or chambers will also be brazed. Mockups of
these subassemblies were brazed in a resistance furnace
with a 7-in. X 7-in. X 30-in.-long chamber at relatively
isothermal conditions at a pressure of 1075 torr:

To ensure proper flow of the brazing filler metal into
the joint, a joint gap of at least 0.001 in. is necessary.
However, the joint preparation for roll bonding and
welding requires a closer tolerance to produce sound
welds and proper alignment. Therefore, the portion of
the joint to be back brazed will be counter bored an
extra 0.0015 to 0.0025 in. on the diameter to a depth
of % in. We have also provided filler-metal feeder holes
by drilling into the bosses- of the large components;
these serve as reservoirs for the brazing alloy powder
and provide insurance that the filler metal does not
prematurely flow away from the joint. Otherwise, the
thin-walled tubing may reach the brazing temperature
faster than the heavy pot, and the brazing filler metal
might flow along the hotter tube (and away from the
joint) rather than into the joint.

14.4.2 Inductlon Vacuum Brazing

We plan to manually weld ‘the 1%-in.-OD X 5-ft- long
X 0.080-in.-wall packed column of the chemical proc-
essing system in a dry box filled with argon. In
addltlon the %-in. tubing that connects the bismuth
head pot to the upper disengaging section of the
column will also be welded in the dry box. Because of
the size and fragility of these large welded sub-
assemblies, we plan to induction braze the reinforcing:
sleeves over the weld before the 10-ft assembly is
removed from the dry box. The controlled atmosphere
of the dry box will also be utilized to prevent oxidation
of all portions of the assembly which will be heated
during the welding operation, N

In preliminary experiments, we learned that brazing
must be accomplished in vacuum (1075 torr) since we
experienced difficulty with arcing in helium or argon
atmospheres. To add further complication, the organic
binder, which is used to preplace the powdered filler
metal, caused arcing when it began to vaporize. As a
result, it is necessary to vaporize the binder by other
heating techniques before the induction unit is ac-
tivated or else preplace the filler metal without using
the binder. To eliminate the need for the binder, we
have machined grooves on the inside diameter of
samples of split sleeves to hold the powdered filler
metal. These grooves also allow space for the reinforced
(raised) weld bead.
We have induction sleeves over tube-to-tube welds in
the atmosphere chamber. However, actual mockup
assemblies have not been brazed but are currently being
prepared.

14.4.3 Induction Field Brazing

Many interconnecting lines of the loop will have to be
joined in the field. Tubing of the sizes %, %, 4, and
possibly % in. OD will probably be welded and the split
sleeves back brazed on site. In this regard, we have
induction field brazed sleeves on all of these sizes of
tubing (Fig. 14.7). We surrounded the part to be brazed
with a quartz tube placed inside the induction coil.
Helium gas flows over the heated area to protect the
braze from air. However, the induction coil is still
operated in air. When the induction unit is activated,
the 2-in.-long section of tubing with the split sleeve is
quickly brought to brazing temperature. Because of the
small amount of material being heated, adjacent areas

remain relatively cool, and the entire brazing cycle
takes less than 5 min.

We are currently refining our apparatus for main-
taining a protective atmosphere and are preparing to
braze more mockups of actual joint designs.

14.5 COMPATIBILITY OF MATERIALS
WITH BISMUTH

0.B.Cavin L. R. Trotter

Successful operation of a processing plant for the
extraction of fission products from the fuel salt of an
MSBR is dependent upon the compatibility of the
container material with molten Bi-Li solutions. There-
fore, we are continuing to evaluate the dissolution and
mass transport of potential structural materials in static
capsules and thermal convection loops containing bis-
muth and Bi-Li solutions with up to 2.5 wt % lithium
concentrations. The tests are being conducted at a

Y-106657

Fig. 14.7. Sleeve brazed to molybdenum tube.

- maximum temperature of 700°C, and the loops have a
AT of 95 + 5°C. For lithium concentrations up to 100
ppm, we have used quartz loops, but for higher
concentrations of lithium (<2.5 wt %) the loop is made
from the material to be tested. .

Previously we reported grain boundary cracking of
small surface grains on recrystallized molybdenum
samples.’ Since cracking was limited to these grains we
hope to circumvent this problem by removing these
grains prior to testing. This was done by electro-
polishing samples after vacuum annealing for 1 hr at
1500°C. These samples were included in loop 11, and
the results are summarized in Table 14 4.

The samples lost weight above a temperature of
650°C and gained weight at or below this temperature.
Because of difficulties in removing residual bismuth
from the specimens, we cannot be certain of the
absolute magnitude of the weight changes, although
they are relatively small. The effect of electropolishing
will be evaluated by metallographic examination of the
specimens. A maximum weight loss of 3.18 mg/cm?
was observed on one of the tensile samples and
corresponds with a uniform surface removal of 0.36

5. O. B. Cavin and L. R. Trotter, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1970, ORNL-4622, p. 189.

174 .

mil/year. The room-temperature mechanical properties
of the molybdenum were unaffected by exposure to
Bi—100 ppm Li, as indicated in Table 14.5.

Tantalum specimens exposed to Bi + 100 ppm Li in
loop 13 lost weight at all temperatures (Table 14.4).
The maximum loss (45.2 mg/cm?) corresponds to a
corrosion rate of 3.13 mils/year. This does not agree
with results obtained by Shimotake et al. in short-term
tests (100 hr) at 1000°C in which they reported no
observable corrosion.® The room-temperature me-
chanical properties of these samples were unaltered
(Table 14.5) even though there was a detectable
decrease in their dimensions. The maximum measurable

change in dimensions corresponds to a uniform surface

removal of 2.6 mils/year, which is in close agreement
with that calculated from weight-change data. |

Conversely, the T-111 alloy (Ta—8% W—2% Hf)
tested in loop 12 showed excellent resistance to
dissolution and mass transport. -‘Tensile samples all
showed slight weight losses of up to 4.45 mg/cm? (0.31
mil/year), but all the tab samples had weight increases
of up to 0.37 mg/cm? (Table 14.4). These weight losses

6. H. Shimotake, N. R. Stalica, and J. C. Hesson, “Corrosion
of Refractory Metals by Liquid Bismuth, Tin and Lead at
1000°C,” Trans. Am. Nucl Soc. 10, 141—-42 (June 1967).

Table 14.4. Weight change of materials tested in flowing bismuth? for 3000 hr

Recrystallized tabs

Tensile samples

Materi Tefnperature Weight change Surfaceb ' Temperature Weight change . Surfaceb
aterial co ~ (mgfem?) removal °0) (mg/cm?) removal

: (mil/year) {mil/year)

Mo 650¢ +0.09 660 None
(loop 11) © 670 -1.30 —0.15 690 -3.18 —0.36
"620 +0.46 640 +0.14

650¢ ) +1.09 ) 660 -0.38 —0.04
Ta ' 650 -16.29 . -1.13 660 —22.55 —1.56
(loop 13) 670 —-2.24 -0.16 690 —45.17 . =3.13
620 -38.32 ~2.66 640 —15.11 —1.05
650 —5.63 -0.39 | 660 —13.66 -0.95
T-111 650 +0.24 660 -0.28 -0.02
" (loop 12) 670 +0.37 690 —4 .45 -0.31
620 +0.15 640 -0.48 - —-0.03
650 +0.28 660 -0.75 —0.05

4Bismuth contained 100 ppm Li.
_b Assuming uniform surface removal.
¢Samples were electropolished prior to test.

LK ]

.y
el

LI

175

Table 14.5. Room-temperature mechanical properties of materials tested
in flowing bismuth® for 3000 hr :

Sample Test Fracture strain Tensile strength Yield strength®?
: temperature (°C) (% in 1.5 in.) (psi) (psi)
X 103 _ x 103
Mo-1 -~ 660 8.0 111 101
Mo-2 690 . 11.3 111 104
Mo-3 640 11.3 112 104 -
Mo-4 660 11.3 112 105
Mo As received 8.7 110 100
Ta-1 : 660 26.0 46.6 33.7
Ta-2 690 23.3 47.2 354
Ta-3 640 : 24.7 46.7 32.6
Ta-4 660 26.0 50.0 36.7
Ta As received 24.0 442 » 35.7
T111-1¢ 660 0 61.4 d
Ti11-2 690 0 51.7 d
T111-3 640 0 46.7 - d
T111-4 660 0.7 88.0 87.3
T111-5¢ 700 15.3 91.7 79.2
T111-6 As received 100 939 81.0

' 4Bj contained 100 ppm Li.
b0.2% offset.
°Ta—8% W—2% Hf alloy.
dBrittle fracture.
€Aged 3000 hr in argon atmosphere,

are of the same magnitude as those observed for
molybdenum. However, the room-temperature ductility
of the T-111 was drastically reduced, as shown in Table
14.5. This is no different from the data previously
reported for T-111 in pure bismuth.” To determine the
effects of aging T-111 at 700°C, samples were held at
this temperature for 3000 hr in an argon-filled capsule.

- No significant changes in the room-temperature me-

chanical properties were observed, and, if anything, the
ductility was slightly higher. The microstructure of the
as-received material shows a grain boundary precipitate.

Similar precipitates have been identified by Inouye® as -

HfC and by Sheffler and co-workers® as a hafnium
oxide. We are presently evaluating the extent to which
such hafnium-rich phases may interact with bismuth or
impurities dissolved in bismuth.

7. O. B, Cavin and L. R. Trotter, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1970, ORNL4622, p. 189-95. .

8. H. Inouye, private communication.

9. K. D. Sheffler, J. C. Sawyer, and E. A. Steigerwald,
“Mechanical Behavior of Tantalum Base T-111 Alloy at
Elevated Temperature,” Trans. ASM 62, 749-58 (1969).

Four grades of graphite were tested in loop 14, and
the weight-change data are shown in Table 14.6. All the
samples increased in weight except the high-density
pyrolytic sample. The available open porosity of grades
AXF-5Q and ATIJS are 15 and 13%, respectively, but
the majority of the pores in AXF-5Q have an entrance
diameter of 0.8 to 1 @, and those in ATJS have a
diameter of 3 to 5 u.1

Grafoil is a trade name for a laminated low-density
graphite foil used for gaskets. It has many large open
pores and in the uncompressed state was infiltrated by
bismuth. A graphite bolt was used to hold a second
sample of Grafoil under compression during test, but

" weight-change information was not meaningful because

the bismuth had penetrated along the threads under the
nut. Other means will be used to compare the bismuth
intrusion into the compressed and uncompressed
Grafoil. | '

The total amount of bismuth intrusion into graphite
appears to depend upon the pore entrance diameter and

10. W. H. Cook, private communication.
176

Table 14.6. Results obtained on graphite tested in flowing Bi—100 ppm Li for 3000 hr

Density Test temperature Weight change N.O rmalized
Sample . 3 : weight change
(g/ecm?) CcO (mg) (mg/cm?)
AXF-5Q-14 1.81 666 +8.4 +2.7
AXF-5Q-2¢ 1.80 , 680 +2.6 +0.8
ATIS-18 1.84 678 +36.4 +11.1
ATIS-2b 1.82 665 +20.0 +6.1
Pyrolytic-1 2.14 655 -0.6 ~0.5
Pyrolytic-2 2.12 , 650 -1.4 -1.3
Grafoil-1¢ 1.08 632 +529.4 817.7
+671.6 . 1095.8

Grafoil-2¢ 1.12 620

9Manufactured by Poco Graphite inc., Decatur, Texas.
bManufactured by Union Carbide Corp., Carbon Products Division, New York, N.Y.
€A trade name for laminated graphite foil used for gaskets and manufactured by Carbon

Products Division of UCC.

not primarily upon the total available porosity. Since
graphite generally is an inhomogeneous matérial as to
density and porosity, samples taken from the same
billet may vary considerably. This, as well as the
uncertainty in complete bismuth removal, could
account for some of the weight-change variations
observed in Table 14.6. .

An all-metal T-111 alloy loop containing T-111
tensile samples has been successfully filled with Bi—2.5
wt % Li (nominal composition) and is operating at a
maximum temperature of 700°C with a AT of 100°C.

Many of our analyses of tests recently completed are
still in progress; however, the results are encouraging.
Of the materials tested, molybdenum continues to be
the least affected by'exposure to bismuth and Bi—100
ppm Li up to temperatures of 700°C. Equal dissolution
and mass transport data were obtained on T-111 alloy,
but it became embrittled by eprsure to bismuth,
probably by an intergranular Hf-Bi reaction. If this
reaction resulting-in loss of ductility can be eliminated
by some pretreatment of the T-111 or by alloying
additions, a tantalum-base alloy may also be attractive
as container material. Graphite appears suitable if the
available open porosity can be eliminated, at least at the
surface. Tests will be done to determine the extent to
which bismuth will penetrate commercially available
graphites and to evaluate the resistance of pyrolytically
sealed graphite to bismuth and bismuth-lithjum in-
trusion. A molybdenum thermal convection loop is
being readied for operation and will contain Bi—2.5% Li
at 600 to 700°C.

14.6 CHEMICAL VAPOR DEPOSITED COATINGS

J. 1. Federer

Roll bonding of molybdenum tubes to integral bosses
attached to molybdenum vessels is being used to make
mechanical joints. Although these joints can be made
leak-tight, brazing from the outside of the vessel will
reinforce the mechanical bond. A chemical vapor

deposited coating around the joint on the inside of the -

vessel is also contemplated as an added precaution. The
coating provides an additional seal that minimizes any
compatibility problems between the braze alloy and
bismuth. Several joints have been coated for evaluation.
These were made by roll bonding 0.5-in.-OD mo-
lybdenum tubes into 0.5-in.-ID molybdenum cylinders.
The tubes extended about 4 in. or more out of the
cylinders. Four joints were coated at S00°C with
0.002-, 0.007-, 0.015-, and 0.018-in.-thick layers of
tungsten. Three other joints were coated at 900°C with
0.005- to 0.009-in.-thick layers of molybdenum. The
joints had very small but measurable helium leak rates
prior to coating (~107% std cc/sec). After coating all
the joints were ‘leak-tight as determined with a helium
leak detector having a sensitivity of 8 X 107'° std
ccfsec. '

An attempt was made to coat a Hastelloy N loop with
tungsten so that the coating could be tested in flowing

" bismuth. The loop dimensions, preliminary coating

experiments, and the method for coating the loop were

alte

ar

-
rFe)

i

described previously.'' Portions. of the coated loop

that were accessible to a borescope were visually
examined. The entire loop was radiographed. We
determined that all surfaces of the loop were coated,
but that several small cracks occurred in the coating at
the corners of two joints where tubing was welded
together at approximately a 90° angle. Also, the coating
was thinner in one cross member than elsewhere, and a

" 1,-in.-diam blister occurred in the cross member near

one of the joints mentioned above. Subsequently, we
coated the two joints and the cross member again.
Radiography revealed that the cracks on the joints were
filled, the blister was covered, and the coating thickness
in the cross member was comparable to other parts of

177

the loop; however, a new crack was found in the cross

member. This crack was probably caused by the large
difference in thermal expansion between tungsten and
Hastelloy N. The cross member was coated again, but
the crack persisted, and further attempts to coat the
loop have not been made.

Several specimens of types 304 and 430 stainless steel
were nickel plated by the electroless method so that the
adhererce of tungsten and molybdenum coatings to the
specimens could be evaluated. The plating method used

for these specimens yielded a phosphorus content of

about 2% in the nickel plate compared with about 8%
phosphorus.in specimens plated previously. One spec-
imen~ of each type steel was coated with about
0.004-in.-thick tungsten. The coating was smooth in
both cases but spalled from the type 430 specimen
during bend testing. Another pair of specimens was
coated with about 0.001- to 0.004-in.-thick mo-

11. J. 1. Federer, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL4676, pp. 231-32.

lybdenum. The coating on the type 304 specimen was
too rough for accurate thickness measurements. The
coating on the type 430 specimen was much smoother
but spalled during bend testing. Since previous ex-
periments showed that molybdenum coats smoothly on
specimens electroplated with nickel, we believe that the
electroless nickel was responsible for the poor coating
characteristics. These results and the previous obser-
vation!! that both tungsten and molybdenum coatings
are less adherent to electroless nickel containing about
8% phosphorus indicate that electroless nickel is in-
ferior to electroplated nickel for promoting adherence.

14.7 MOLYBDENUM DEPOSITION FROM MoF,
‘ J. W. Koger

We have continued to conduct experiments to op-
timize the conditions for coating stainless steel with
molybdenum by contacting the stainless steel with a
molten fluoride salt containing MoF,. Holding temper-
ature and time constant, we have determined the
concentration of MoF¢ which must be added to our salt
mixture to get a coating. Future plans involve an.
evaluation of time and temperature effects on coating
rates and coating integrity.

In one of our more successful runs, microprobe
analysis! 2 of the type 316 stainless steel capsule on
which molybdenum- was deposited showed that mo-
lybdenum ‘was deposited at the sample surface and that
grain boundaries at least 350 u into the metal were
enriched in molybdenum.

12. Performed by H. Mateer, T. J. Henson, and R. S. Crouse
of the Metals and Ceramics Division.
Part 4. Molten-Salt Processing and Preparation

L. E. McNeese

Part 4 deals with the development of processes for the
isolation of protactinium and the removal of fission
products from molten-salt breeder reactors. During this
period we continued to evaluate and develop a flow-
sheet based on fluorination—reductive extraction for
protactinium isolation and the metal transfer process
for rare-earth removal. The portion of the flowsheet
dealing with recovery and recycle of the UF4 from the
fluorinators was modified so that the .UF¢-F, gas
stream would be directly absorbed in molten salt
containing UF, and the resulting uranium fluoride
would be reduced to UF,; by contact of the salt with
" hydrogen. A method was devised for recycling fluorine,
hydrogen, and HF in the plant in order to minimize the
quantity of radioactive waste produced. A study was
completed for determining -the capital and operating
costs for a plant that processes the fuel salt from a
1000-MW(e) MSBR on a ten-day cycle. The total direct

and indirect costs for the plant were found to be $20.6.

and $15 million, respectively; thus the total plant
investment required would be $35.6 million. The
resulting fuel cycle cost was 1.12 mills/kWhr. A
0.28-power dependence of processing plant cost on
_ processing rate ‘was derived; this indicates that a
considerable saving in processing cost could be achieved
by associating a processing plant with a larger power
generation capacity than 1000 MW(e).

Studies relating to the chemistry of fuel reconstitu-
tion were initiated. It was found that gaseous UF
reacts quantitatively with UF, dissolved in LiF-BeF,-
ThF, (72-16-12 mole %) at 600°C to produce soluble

- lithium chloride, has been in operation for about nine

UFs. The molten salt containing UFS appeared to be
stable in both gold and graphite equipment. These re-
sults are in agreement with those from scouting experi-

ments carried out several years ago on fuel reconstitution.
During this report period we found that the lithium

and bismuth concentrations in LiCl in equilibrium with
lithium-bismuth solutions are higher than initially ex-
pected and increase markedly as the lithium concentra-
tion in the lithium-bismuth solution is increased. We
realized that this could have a significant effect on the
performance of the metal transfer process since lithium-
bismuth solutions having lithium concentrations of 5
and about 50 at. % are proposed for extracting rare
earths from lithium chloride in the process. Therefore, a
study of the equilibrium distribution of lithium and

bismuth between molten LiCl and liquid lithium-

bismuth solutions was initiated. Data obtained thus far
with lithium concentrations of 20 to 50 at. % are
consistent with the observed behavior of lithium in the
second metal transfer experiment (MTE-2) completed
previously. An engineering experiment involving the
metal transfer process (MTE-2B) is in operation in order
to study further the behavior of lithium in the process.
The experiment, in which a § at. % lithium-bismuth
solution is being used to extract rare earths from

weeks. To date, the results indicate that the equilibrium

~concentration of lithium in lithium chloride above a §

178

at. % lithium-bismuth solution is less than 0.3 wt ppm.
It is not believed that the metal transfer process will be
affected significantly by the unexpected behavior of

vy

d e
o

lithium and bismuth in the process. The design of the
third engineering experiment (MTE-3) for development
of the process was completed. Most of the equipment
was fabricated, and the main process vessels are now
being installed. The experiment will use salt flow rates
that are 1% of the estimated flow rates required for
processing a 1000-MW(e) reactor.

Our work on contactor development was continued
successfully during this report period. Mass transfer
experiments were carried out in which the rate of
transfer of zirconium from molten salt to bismuth was
measured in a 24-in.-long, 0.82-in.-ID column packed
with Y-in. molybdenum Raschig rings. The results
indicate HTU values of 1 to 4 ft, with the lowest values
being given by data in which we have the most
confidence. These results indicate that packed column
contactors can be used successfully in MSBR processing
systems. We have initiated studies on mechanically
agitated salt-metal contactors as an alternative to
packed columns. This type of contactor is of particular
interest for use in the metal transfer process since
designs can be envisioned in which the bismuth would
be a near-isothermal internally circulated captive phase.
The hydrodynamic performance of a mechanically
agitated contactor was investigated using mercury and
water to simulate bismuth and molten salt. These data,

179

along with data from the literature, indicate that
satisfactory mass transfer performance may be achieva-
ble in this contactor without dispersing either the salt
or the metal phase. Contactors of this design may be.
easier to ‘fabricate than packed columns, and the
problem of entrainment of bismuth in salt to be
returned to the reactor should be eliminated if the
phases are not dispersed.

We have continued studies of oxide precipitation as
an alternative to the fluorination—reductive-extraction
method for isolating protactinium and for subsequently
removing uranium from MSBR fuel salt. Protactinium
pentoxide was selectively precipitated from LiF-BeF,-
ThF, (72-16-12 mole %) that contained up to 0.25
mole % UF, by sparging the salt with HF-H, O-Ar gas
mixtures at 600°C. Recent results on the equilibrium
quotient for the precipitation reaction are in agreement
with earlier results and are consistent with earlier
indications that Pa, 05 can be precipitated from MSBR
fuel salt as a pure or nearly pure solid phase. Installa-
tion of equipment was completed. in preparation for
engineering studies of the precipitation of UO,-ThO,
solid solutions from molten fluoride salts. Two experi-
ments in which as much as 40% of the uranium initially
in the salt was precipitated were carried out success-
fully. Results of these experiments are encouraging.

| 15. Flowsheet Analysis

A study was completed in which the capital and
operating costs were determined for a plant that
processes the fuel salt from a 1000-MW(e) MSBR on a
ten-day cycle. The plant is based on the reference
flowsheet that uses fluorination, reductive extraction,
and metal transfer. Capital and operating costs were
calculated for two methods for obtaining and disposing
of hydrogen, HF, and fluorine in a processing plant.
Calculations were made of the thermal power that will
be produced in an MSBR processing plant by the decay
of noble-metal fission products as a function of the
residence time of these materials in the primary reactor
system and of the fraction of these materials accom-
panying the fuel salt to the processing plant. The
long-term hazard of high-level radioactive wastes pro-
duced by MSBRs was considered. A survey was made of
published data on the availability of and future demand
for natural resources required for an MSBR power
economy. |

15.1 COST ESTIMATE FOR AN MSBR
PROCESSING PLANT

W. L. Carter E. L. Nicholson
L. E. McNeese

We have completed a study to determine the capital’
and operating costs for a plant that processes the fuel
salt from a 1000-MW(e) MSBR on a ten-day cycle. The
processing plant is based on the previously described
reference flowsheet!? that uses fluorination for ura-
nium removal, reductive extraction for protactinium
removal, and the metal transfer process for rare-earth
removal. The most recent version of the flowsheet,

1. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb.
28, 1970, ORNL-4548, pp. 277-88.

2. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL-4676, pp. 234-38.

180

ORNL-DWG 71-78694A

s‘!

Hp, TO PURGE COLUMNS _ Hyp (95%) 5% -
- - — Il AI,05 SORBER . || CHARCOAL SORBER VENT TO STACK
(10 scfm)
: (SeFg—TeFg ) : (Kr—Xe) ‘
Hp MAKEUP ‘ , '
H,+HF FROM HYDROFLUORINATORS ' !
: AND PURGE COLUMNS DRAWOFF TO WASTE ,
CAUSTIC - EVAPORATION AND LITHIUM ADDITION .
SCRUB SALT MINE STORAGE (539 mole/day) .
, H, +HF H, _| _ _ -
L (14.9 scfm) (10.5 scm) - ' - BISMUTH
SALT MAKEUP B : KOH (HBr-HI) | _ , RE 2+ U ke
BeF,(74.8g mole/day) | HF {2 gal/ min) -Accuggbfi?su | . (06gal/day) ‘ f
ThF, (100.3 g mole/day) ~ DISTILLATION IN BI } 7 |
UFs KOH ' — o — : | | |
REDUCTION ' RESERVOIR | | }
He /;-I e R A | |
|,3 CARRIER SALT DISCARD ' : [
.Léj (LiF—BeF2 —Tth,) . 24 I I
o | | | (2.38 gal/day) - '| fSTRR'IEPPER I Bi-Li~Re®* DRAWOFF |
'3 - e—t Bi-50 at. % Li-REZ2? | (509 mole Li/day) f.
— H ' ) ————-—‘-——,——-—'————"L——————(0—6——| ————————— —"'l
BISMUTH ‘g 2 l (0.17 gal/min) gal/day) |
REMOVA L BlE ' T = : i
el - R '
()] ]
oW FLUORINE |
o J l HE CELL ' FISSION PRODUCT | ! Lt THIUM ADDITION | |
& ':1 | - ! l TRANSFER | S {65 g mole/day) I
5|3 RARE EARTH | L | TO LiCl "o o
E EXTRACTION | Y | | l ' ” BISMUTH .
SALT CLEANUP AND i l HF TO - ’ - (0.66 golay) RE TSI oo T T T T T T T Lo
U3+ Ju4+ - | HYDROF LUORINATORS I I f (0.66 gal/day) ACCUMULATION | (5.7 gal /day) ! I
, ' : lN BISMUT
‘ VALENCE & b gy — g J_ X i Ha-HF TO | : I
ADJUSTMENT ?.:% _BISMUTH_(RES*—RE®Y ) [ ] ] | : RECYCLE : I i
6 | | Le—mp—=2WTL R TRE T A T S I R [ PO ———
)  .— - T _ (12 4 gu{/mln) . . . I . : 2-- ) ' 4 ) | BI_Ll_RE3+ DRAWOFF I I |
FUEL SALT RETURN | ! ] : ‘ | (50g mole Li/day) ! )
TO REACTOR l | RE3t (5.7 gal/day) i ||
{0.87gal/min) ! I UFe BRODUCT I ! STRIPPER I 'j | I :
| | " REMOVAL ' | : o { : !
SECONDARY ' ' |
| FLUORINATION (165g U/day) | | N | HYDROF LUORINATION |
- _ [ | LLicl (REZ*-RE | | ||
SALT FROM REACTOR . 1 ey | , | (32.3gal/min) l : : :
LiF-BeFa—ThFa—UF,. 30-min ] ¢ __ Bi-5at. % Li-Re3* | I ||
T1.7-16.0-12.0-0.3 mole % SALT Pa | —————— S
© 87' l/ . ') - HOLDUP EXTRACTION (Bgul/mln) | : [
. a min
k » TANK : _rullx_!V | :
‘ |
HyHF TO | :
PRIMARY 2 I
FLUORINATION|. RECYCLE Po STORAGE AND DECAY | FISSION PRODUCT AND : |
LiF-ThRy=ZrFs-PaF, , | LITHIUM ADDITION. WASTE SALT ACCUMULATION |
BISMUTH (71.0—26.0-2.8—0.23mo|e %) ) L (371 mole /day) (3.14 gal /day) | L
(Li-Th-Zr-U-Pa) 150-175 ft lr- - « I [ | ‘
— — — ‘ |
(0.43 gal /min) — -
t / BATCHWISE DISCARD( | | !
| (2513 /220 days) i | I l
PRIMARY STREAMS 2 1
: , | HYDROFLUORINATION |
LiF-BeF,—ThF, SALT { 220-day | ULTIMATE DISPOSAL IN ||
—— e =m == BISMUTH | Pa DECAY SALT MINE AFTER Lo
—— e i C | ’ | I ~9-yr DECAY PERIOD |
e R | | - )
| (0.78gal/ min) i I ]
|
| ————— — — I ___________________________________________ —’l I

Fig. 15.1. Fluorination—reductive extraction—metal transfer process flowsheet for processing MSBR fuel salt. Flow rates are shown for processing a 1000-MW(e) reactor on a ten-day cycle with an assumed uranium removal efficiency in the primary

fluorinator of 95%.

L]

.7

>

LI

.y

181

shown in Fig. 15.1, includes two significant improve-
ments over the previous version of the flowsheet: (1) an
improved method for recovering UF4 from the fluorina-
tor off-gas and (2) a method for recycling fluorine,
hydrogen, and HF in the processing plant in order to
minimize the quantity of radioactive waste generated.

The behavior of fission products and other com-
ponents of the fuel salt in the processing plant is
summarized in Table 15.1, which shows only the
dominant removal method for each of several groups of

fuel salt constituents. The removal times for the noble

gases (50 sec) and noble and seminoble metals (2.4 hr)
are short as compared with the ten-day processing
cycle, and only small amounts of these materials are
removed by the processing plant. In the previous
flowsheet,! the fluorinator off-gas (primarily a mixture
of fluorine and UF) passed through NaF sorption beds
and cold traps operated at —40°C for UFg recovery.
The UF; was then combined with the processed fuel

carrier salt for return to the reactor. Several difficulties
were encountered with this approach, for example: (1)
the uranium inventory in the UFg collection system
would be appreciable (3 to 10% of the uranium
inventory in the reactor); (2) beds of granular heat-
emitting solids (NaF) would have to be filled and
discharged remotely; (3) the discharged NaF would
require subsequent treatment in order to ensure ac-
ceptably low uranium losses; and (4) collection of UFy
in the cold traps might be complicated by heat

produced by the decay of fission products that are
desorbed from the NaF beds with the UF. Initially in
the present study, consideration was given to recycling
the unused fluorine after UF, collection; however,
several problems were encountered with this mode of
operation. The two major problems were: (1) fission
product iodine accumulated to appreciable concentra-
tions in the fluorine recirculation system and repre-
sented a significant heat source (210 kW as shown in

Table 15.1. Methods for removing fission products and fuel salt constituents
in an MSBR processing plant '

Group : Components Removal time Primary removal operation
Noble gases K, Xe 50 sec Sparging with inert gas in reactor
fuel circuit
Seminoble metals Zn, Ga, Ge, As, Se 2.4 hr Plating out on surfaces in reactor
vessel and heat exchangers
Noble metals Nb, Mo, Tc, Ru, Rh, 24 hr Plating out on surfaces in reactor
Ag, Cd, In, 8Sn, Sb, . vessel and heat exchangers
Te

Trivalent rare earths Y, La, Ce, Pr, Nd,
Pm, Gd, Tb, Dy, Ho,

Er (note: Yttrium

Varies for different
nuclides; effective
removal time = 25

Reductive extraction into Bi-Li alloy
followed by metal transfer via LiCl
into Bi—$5 at. % Li solution

is not a rare earth days
. but behaves similarly.) .
Divalent rare earths Sm, Eu, Sr, Ba Varies for different

and alkaline earths

nuclides; effective

removal time = 25 days

Alkali metals Rb, Cs 10 days
Halogens Br, I 10 days
Uranium 2?’3U,234U,235U, 10 days
236 U, 237
Zirconium Zr, 233p, 10 days
and protactinium
Corrosion products Ni, Fe, Cr 10 days
Carrier salt Li, Be, Th ~15 years

Reductive extraction into Bi-Li alloy
followed by metal transfer via LiCl
into Bi—S§ at. % Li solution

Reductive extraction into Bi-Li alloy
followed by accumulation in LiCl

Volatilization in primary fluorinator
followed by accumulation in KOH solution
in gas recycle system

Volatilization in primary fluorinator;
returned to carrier salt and recycled
to reactor

Reductive extraction into Bi-Li alloy

- followed by hydrofluorination into Pa
decay salt

Reductive extraction with Bi-Li
alloy followed by hydrofluorination
into Pa decay salt :

Salt discard

Sect. 15.3); and (2) maintenance on the sealed dia-
phragm-type fluorine compressors is known to be
higher than desired. The alternative of disposing of the
unused fluorine by absorption in aqueous KOH solu-
tions was unattractive because of the large volume of
radioactive waste (mostly KF) that was generated.

In the current flowsheet, difficulties associated with
UF¢ collection and recycle or disposal of unused
fluorine were avoided by sending the F,-UFs gas
mixture directly to the UF¢ reduction step, where both
the F, and UFg¢ are absorbed in a sufficient quantity of
molten salt containing UF, to produce an average
uranium valence between 4+ and 5+. The resulting
~uranium fluoride is reduced to UF, by contact of the
salt mixture with hydrogen. The reconstituted fuel salt
is then returned to the reactor after a final cleanup
operation. The HF produced in the UF, reduction
operation is sent to a remotely operated fluorine plant,
where F, and H, are generated for recycle to the
processing plant. As discussed in Sect. 15.2, the cost for
recycling the fluorine and hydrogen via a fluorine
generation plant is only 0.024 mill/kWhr as compared
with a total cost of 0.11 mill/kWhr associated with
absorbing the HF in an aqueous KOH solution and
disposing of the fesulting radioactive wastes. :

Cost estimate. In preparing the cost estimate, prelimi-
nary designs for all major process equipment items were
completed, and the equipment cost was estimated using
unit costs for fabricating the various structural shapes.
The general design criteria assumed that a material
having a cost-equal to that of Hastelloy N would be
used for vessels containing only molten salt and that
molybdenum would be used for vessels that contained
 bismuth. The cost of fabricated molybdenum equip-
ment was assumed to be $200/lb; a value of this
magnitude was used because of the difficulty of
fabricating molybdenum. The costs of piping, instru-
mentation, and certain auxiliary equipment items were
estimated as appropriate fractions of the costs of
fabricated equipment items.

A summary of the capital costs for the processing
plant is shown in Table 15.2. The total direct and
indirect costs are $20.6 and $15 million, respectively;
thus the total plant investment is $35.6 million.

Fuel cycle cost. A breakdown of the fuel cycle cost is
given in Table’ 15.3 for a -1000-MW(e) MSBR that is
processed on a ten-day cycle (a processing rate of 0.88
gpm) by fluorination—reductive extraction—metal
“transfer. The inventory charges on fissile materials and
salt in the reactor are those given for the reference
MSBR design.®> The net fuel cycle cost (1.12 mills/
kWhr) is composed largely of fixed charges on the

182

Table 15.2. Capital costs for a processing plant for a
1000-MW(e) MSBR based on the fluorination—reductive
extraction—metal transfer flowsheet
Reactor fuel volume = 1683 ft>
Processing cycle time = ten days

Thousands
of dollars
Installed molybdenum process equipment 4,579
Installed molybdenum pumps 245
Installed molybdenum heat exchangers 90
Installed molybdenum piping : ' 1,474
Installed Hastelloy N, stainless steel, 3,129
and nickel equipment
Installed auxiliary equipment : 2,486
Process piping (other than molybdenum 2,342
piping)

Process instrumentation 2,711
Cell electrical connections . 494
Thermal insulation 588
Radiation monitoring . 150
Sampling stations 1,275
Fluorine plant : 1,005

Total direct cost 20,568
Construction overhead® 4,114
Engineering and inSpectionb 3,790
Taxes and insurance® . 864
Contingencyd : 2,836

Subtotal 32,172
Interest during construction® 3,442

Total plant investment 35,614

430% of total direct cost.

b5.3% of total direct cost, plus a $2.7 million “novel” design
premium,

€4.2% of total direct cost.

920% of total direct cost, exclusive of the cost of molyb-
denum components.

®Money borrowed at 8%/year for a three-year penod

processing plant (0.7 mill/kWhr), inventory charges on
salt and fissile materials in the reactor and 2*?Pa in the
processing plant (0.42 mill/kWhr), and processing plant
operating charges (0.083 mill/kWhr); credit is taken for
the excess fissile material produced (0.089 mill/kWhr).

Variation of the capital cost of the processing plant
with plant throughput. The variation of the capital cost
of the processing plant with plant throughput was
estimated by determining the capital cost for a plant

3. R. C. Robertson (ed.), Conceptual Design Study of a
Single-Fluid Molten Salt Breeder Reactor, ORNL-4541 (June
1971), pp. 180-81..

*u

ol

- >

>
183

Table 15.3. Fuel cycle cost for a 1000-MW(e) MSBR

Processing cycle time = ten days
Plant factor = 80%

Mills/kWhr
Fixed charges on processing plant at 13.7%/year 0.6962
Inventory charges at 13.2%/year
LiF-BeF,-ThF 4 in reactor 0.0608
2337+ 235y + 233pa in reactor 0.3281
LiF-BeF,-ThF, in processing plant 0.0073
233y 4+ 235y +233pyip processing plant 0.0286
LiCl in processing plant 0.0006
Bi in processing plant 0.0040
0.4294
Operating charges
Li metal makeup 0.0171
BeF,-ThF4 makeup 0.0209
HF makeup 0.0001
H, makeup 0.0002
KOH makeup 0.00003
Waste disposal 0.0164
Payroll 0.0285
' 0.0832
Gross fuel cycle cost 1.2088
Production credit (3.27%/year fuel yield). -0.0894
Net fuel cycle cost 1.1194

-p

based on a 3.3-day processing cycle (a processing rate of
2.9 gpm). The total plant investment for this case was
$48.5 million. The resulting cost data are shown in Fig.
15.2, where a power-type dependence of plant cost on
throughput is assumed. The resulting cost-vs-throughput
relationship is:

. R 0.28
C=Co|— ,
O(Ro)

where
C = cost- of processing plant,
R = processing rate,
Co = cost of plant for a ten-day processing cycle,

R, = processing rate for a ten-day processing cycle.

We believe that this relationship can be used for
processing cycle times of about 3 to 37 days. However,
for process cycle times longer than 37 days(a proc-

essing rate of 0.24 gpm), it is believed that a plant cost
of $25 million should be used (see Fig. 15.2).

The cost information shown in Fig. 15.2 indicates
that a significant saving in processing.cost can be
achieved by using a processing plant with a larger power
generation capacity than 1000 MW(e). Although this is
undoubtedly true, the processing costs associated with a
higher plant throughput cannot be obtained directly
from Fig. 15.2 since a larger amount of radioactivity
would be present in the processing plant.

ORNL-DWG 71-9543A

100
ey
w 50
o) L~
° _—
g SLOPEE)—.‘?_B’/——
(] e
| //
=
=
fl<‘- 20
[&]
10
[oR] 0.2 0.5 1 2 5

SALT PROCESSING RATE (gal/min)

Fig. 15.2. Variation of capital cost for a fluorination—reduc-
tive extraction—metal transfer processing plant with salt
throughput for a 1000-MW(e) MSBR.

15.2 COMPARISON OF COSTS FOR ALTERNATE
OFF-GAS TREATMENT METHODS FOR AN
MSBR PROCESSING PLANT

W. L. Carter

Capital and operating costs associated with two
methods (see Fig. 15.3) for obtaining and disposing of
HF, H,, and F, in a plant were determined. In the first
method, HF, H,, and F, are purchased, and the H, and
HF that appear in the plant off-gas are subsequently

. disposed of; in the second, the HF and H, from the

plant off-gas are collected, and the HF is electrolyzed to
produce H, and F,, which are recycled to the
processing plant. The once-through gas cycle results in
the purchase of a large amount of F,, H,, and HF and
in the generation of a significant volume of radioactive
wastes. The second method, however, results in the
purchase of only minor amounts of HF and H, and
produces only a small quantity of radioactive waste.
The principal fission product activity in the process
off-gas is iodine, although small amounts of bromine,
krypton, xenon, and volatile noble and seminoble metal
fluorides are also present.

Once-through gas cycle. As shown in Fig. 15.3,
hydrogen and HF are collected from the plant off-gas
streams resulting from UF4 reduction, hydrofluorina-
tion of bismuth, and hydrogen sparging. The combined
HF-H, stream enters a caustic scrubber, where it is
contacted with an aqueous KOH solution for removal
of HF, HI, and HBr. It was assumed that 135 ft® of 10
M KOH solution would be recirculated through the
scrubber for 100 hr and that the KOH concentration
would be reduced to 0.5 M during this period. The heat
generation rate in the scrubber solution after 100 hr
would be about 153 kW, and the solution would be
held for an additional 400 hr for fission product decay

184

before the solution is evaporated to produce a solid
waste residue. Evaporation is carried out in the waste
container, which has a diameter of 2 ft and a length of
10 ft. This is the largest waste container presently
considered for interment in a salt mine. The aqueous
condensate produced by evaporation of the waste is
recycled to produce additional 10 M KOH solution.

The hydrogen stream leaving the scrubber is assumed
to contain low concentrations of SeFg and TeF4. The
gas is passed through a dryer, an Al,0; bed for
sorption of SeFs and TeF4, and a charcoal bed for
retention of noble gases. About 10.5 scfm of hydrogen
is discharged to the stack.

This off-gas treatment method results in the produc-
tion of 489 kg of solid waste per day and requires that

ORNL~-DWG 71-9553A

H ]
2.f orver | - A,05 | ~{crarcoaL}— Hz, T
Hp, FROM SPARGING o feROH_
«KOH MAKEUP |22
Hz -HF FROM : KOHI 1
HYDROFLUORINATORSl | scrusser kon | .
- RESERVOIR
Hp- HF
FUEL SALT } | KOH-KF-KI 1.
gl
+ URy UFs-UF, WASTE CAN
REDUCTION EVAPORATE
TO SOLID
H
FUEL SALT 2 RESIDUE
Fa-UFg ONCE- THROUGH GAS CYCLE
Hz FROM SPARGING Ho- HF H ‘TO
2” 2 .
HARCOAL
Hp -HF FROM m STACK
HYDROFLUORINATORS § ¢ Hz RECYCLE KOH
| . N oH TO SPARGING |
CONDENSER CONDENSER | (o he o KOH <OH Ho0
l l MAKEUP :
Ho- HF Hy| | KOH
HF - RESERVOIR
b0 KOH-KF -KI
HF RECYCLE e kAN
TO. HYDRO- WASTE CAN
FLUORINATORS EVAPORATE
- TO SOLID
FUEL SALT |——| : HF HF _| FLUORINE RESIDUE
+ UF, UF, -UF, . | DISTILLATION PLANT L1
REDUCTION H |
- | 2 :
FUEL SALT F, RECYCLE TO
FLUORINATORS
Fo-UFg

GAS RECYCLE SYSTEM

Fig. 15.3. Alternate methods for treating off-gases in an MSBR processing plant.

da
L

105 waste cans be stored annually in a salt mine.
Approximately 66 kg of fluorine, 102 kg of HF, and 41
kg of hydrogen would be required daily. '

Recycle of H, and F, via electrolysis. In this off-gas
treatment method, hydrogen and HF are collected from
the various off-gas streams in the process. The mixture
is compressed to a pressure of about 2 atm and chilled
to a temperature of —40°C to condense the bulk of the
HF. The liquid HF is distilled at essentially total reflux
to separate volatile fission product fluorides (principally
HI, HBr, SeFg, and TeF¢). About 70 kg of HF is
electrolyzed per day in a remotely operated fluorine
plant in order to produce the hydrogen and fluorine
that are required for plant operation. The remaining HF
is recycled to the plant, where it is used principally for
hydrofluorination of bismuth streams containing dis-
solved metals.

The hydrogen stream leaving the HF condensation
and distillation operations will contain most of the
fission product activity in the initial H,-HF stream and
a small amount of HF. This gas stream is contacted with
an aqueous KOH solution to remove HF, HI, and HBr.
It is assumed that 20 ft3 of 10 M KOH solution will be
recirculated through the scrubber for 34 days. At the

185

end of this period, the KOH concentration will have
been reduced to 0.5 M, and the steady-state heat
generation rate in the solution will be 210 kW. The
solution would be held for 45 days for fission product
decay before it is evaporated to produce a solid waste.
The solid waste production rate is only 9.7 kg/day, and
only 2.1 containers having a diameter of 2 ft and a
length of 10 ft would be filled annually.

The hydrogen stream leaving the caustic scrubber is
dried and recycled to the processing plant. About 5%
(0.5 scfm) of the hydrogen is discarded through an
Al, 03 bed (for removal of SeFg and TeFg) and a
charcoal bed (for retention of noble gases). This off-gas
treatment system requires the purchase of only 1.4 kg
of hydrogen and 13.4 kg of HF per day.

Comparison of costs for off-gas treatment. Costs
associated with the operation of the two off-gas
treatment methods were determined by calculating the
capital equipment and operating charges for each.
Results of these calculations are summarized in Table
15.4 for a 1000-MW(e) MSBR that is processed on a
ten-day cycle. The contribution to the fuel cycle cost
for the once-through gas cycle is about 0.11 mill/kWhr,
while the cost for the treatment system resulting in

Table 15.4. Fuel cycle cost contribution for two methods of treating
process gases in an MSBR processing plant

Reactor power = 1000 MW(e)

Plant factor = 80%

Fuel cycle cost (mills/kWhr)

Once-through gas cycle

Gas recycle

Waste disposal charges
Waste containers
Carriers
Shipping
Salt mine storage

Chemicals and interest on capital
equipment

KOH tanks

Fluorine plant

Fluorine

Hydrogen

Hydrogen fluoride

Potassium hydroxide

HF distillation equipment

Total

0.0485 0.0009
0.0022 0.0004
0.0050 0.0001
0.0045 0.0001
0.0602 0.0015
0.0124 0.0013 -
: 0.0196
0.0304
0.0055 0.0002
0.0026 0.0005
0.0038 0.0001
_ 0.0007
0.0547 _ 0.0224
0.11 0.024

almost complete recycle is only 0.024 mill/kWhr. The
higher cost for the once-through gas cycle is due
primarily to the cost of disposing of a relatively large
quantity of solid waste and to the purchase of fluorine.
The most significant cost for the gas recycle method is
amortization of the remote fluorine plant, which has a
direct cost of $1.005 million. In making the cost
estimate, capital costs were amortized at 13.7%/year.
Salt mine storage charges® were assumed to be $300 per
waste container, which is the minimum interment
charge. Based on this comparison, the gas recycle
method was selected for use in the MSBR fuel
reprocessing plant.

15.3 EFFECT OF NOBLE-METAL AND HALOGEN
REMOVAL TIMES ON THE HEAT GENERATION
RATE IN AN MSBR PROCESSING PLANT

M.J.Bell ~ L. E. McNeese

The thermal power that will be generated in an MSBR
processing plant by decay of the noble-metal fission
products will depend on the residence time- of these
elements in the primary reactor system and the frac-
tions of these elements accompanying the fuel salt to
the processing plant. We have performed a series of
calculations in which the noble metals were assumed to
be transferred from the fuel salt, on a 50-sec cycle, into
a holdup volume having a residence time ranging from 1
sec to 1 year. The holdup volume was assumed to be in
contact with fuel salt so that materials produced by the
decay of noble metals could return to the fuel salt if
these materials were not noble metals. The model is
sufficiently general that the holdup volume could
represent noble metals associated with circulating gas
bubbles, a stagnant film of noble metals in a pump
tank, or noble metals deposited on graphite and metal
surfaces. The residual thermal power of the noble
metals leaving the holdup volume is shown in Fig. 15.4
as a function of residence time in the holdup volume.
The residual thermal power is defined as the integral,
over all time, of the instantaneous heat generation rate
for all isotopes leaving the holdup volume. The noble
metals were divided into two groups: those that form
stable volatile fluorides (As, Se, Nb, Mo, T¢, Ru, Sb,
and Te) and those that do not (Rh, Pd, and Ag).
Elements in the first group are assumed to be com-
pletely removed from the salt in the primary fluorina-

4. Staff of the Oak Ridge National Laboratory, Siting of Fuel
Reprocessing Plants and Waste Management Facilities, ORNL-
4451 (July 1970), pp. 6-44—6-47.

186

tor. The elements in the second group will remain in the
salt and will be extracted into bismuth in the protac-
tinium isolation system. If the removal of 1% of the
noble metals from the holdup volume occurs with a
residence time of 1 min, the heat generation in the
processing plant will be increased by 200 kW, with the
bulk of the heat being generated in the UF4 recovery
system. About the same heat generation rate would
result if 10% of the noble metals reached the processing
plant after a one-day holdup or if 20% reached the
plant after a-ten-day holdup.

The heat generated by the halogens (Br and I) in the
processing plant will depend upon both the halogen
removal time and the noble-metal removal time since
halogens are produced by the decay of some noble
metals. If the noble metals are removed from the fuel
salt on a very short cycle, their daughters are prevented
from entering the fuel salt and thus will not reach the
processing plant. The heat generation rate resulting
from the decay of halogen fission products is shown in
Fig. 15.5. An effective noble-metal removal time of 0.1
day -is believed to be reasonable for the present

‘flowsheet. This would result in a heat generation rate of

210 kW for the halogens for a halogen removal time of
ten days or 520 kW for a removal time of three days.
For a noble-metal residence time in the holdup volume
as long as one year, the halogen thermal power would
be increased by only 60% over that for a 0.1-day

_residence time.

15.4 LONG-TERM DISPOSAL
OF MSBR WASTES

M. J. Bell

A recent study® has been made of the long-term
hazard associated with high-level radioactive wastes
produced by nuclear reactors operating on the enriched
235y, uranium-23°Pu, and thorium-233U fuel cycles.
In this investigation, the reference MSBR was assumed
to be typical of the thorium-233U fuel cycle. It was
found that the ingestion hazards represented by the
high-level wastes produced by the three fuel cycles are
quite similar after comparable periods of decay. These
high-level wastes were assumed to include all fission
products produced by the fuel cycle and, in the case of
the 225U and uranium-23°Pu fuel cycles, to include

5. M. J. Bell and R. S. Dillon, The Long Term Hazard of
Radioactive Wastes Produced by the Enriched Uranium, Pu-
238y and 233U .Thorium Fuel Cycles, ORNL-TM-3548 (in
press).

a1
0

187

ORNL-DWG 71-2845A

10 ==
< =
= !‘{_“ {
P OTAL
w ]
g y R NOBLE METALS WITH ’
g 0 2 STABLE VOLATILE FLUORIDES
I = =

P —l i
= P~ 3 -
1] o
ba e N
= 4° N ™
h .
+
@ ™~ Ry
) NOBLE METALS WHICH DO NOT iy N \
a5 FORM STABLE VOLATILE FLUORIDES—] by N
= e —H
T T ]
(1] LLLRLLRL 1 [
-} 1 minute 1 hou iday ~| [} 10 daoys {] 1 year T{{]|
m N :. " : N paghy L u“ ™ I 1; N
g | i N
2 - L H T N
$0° " 10" 162 10° 10% 10° 108 ol 108

HOLD-UP TIME (sec)

F
system.

e

ORNL-DWG 71— 2923A

05 AN

\\ \INOBLE METAL REMOVAL

TIME 10 d
N \ e
™
\\
0-2 \\\

HALOGEN THERMAL POWER (MW)

041 N \\
~
~ 04 dcy\\
N
q N
0.05
\\
Qsec
0.02
\

0.01 —
1 2 5 10 20 50 100
HALOGEN REMOVAL TIME (days) )

Fig. 15.5. Effect of halogen and noble-metal removal times
on the thermal power of the halogen fission products.

0.5% of the uranium and plutonium and 100% of other
transuranium isotopes and their daughters present in
the spent fuel at the time of chemical processing (150
days after discharge for the 233U fuel and 90 days after
discharge for the 23Pu fuel). In the case of the MSBR
it was assumed that the thorium in the waste salt would
not be recovered, that 0.5% of the reactor uranium
inventory would be discarded over the 30-year life of

g. 15.4. Residual thermal power of noble metals leaving the holdup volume as a function of holdup time in the primary reactor

the plant, and that all of the transuranium isotopes and
their daughters would be present in the waste. The
removal times for the transuranium elements were
assumed to be ten days except in the case of neptu-
nium, which was assumed to have a removal time of 16
years. ' .

The volume of water required to dilute the three
types of high-level wastes to the radioactivity concen-
tration guide values for continuous ingestion in unre-
stricted areas is shown in Table 15.5 for wastes of
various ages. Also shown for comparison are the
volumes of water required to dilute comparable quanti-
ties of (1) uranium ore tailings and (2) material
containing uranium and thorium at concentrations
equal to their respective average concentrations in the
earth’s crust. The quantities of these materials used in
the comparison were assumed to be those that would
occupy the same volume as-the high-level waste and
associated salt and shale in the proposed Federal salt
mine repository. Wastes aged 300 years would present
an ingestion hazard similar to that from uranium ore
tailings, and wastes aged 10,000 years would present a
hazard about one order of magnitude greater than that
associated with naturally occurring uranium and tho-
rium at their average concentrations in the earth’s crust.

The study identified two potential problems relating
to the MSBR fuel cycle, as envisioned at the present
time. As a result of the inefficient use of relatively
inexpensive thorium in the present MSBR concept, the
mass of wastes produced per MW(e) by an MSBR is
substantially greater than the masses of waste produced
by the other reactor types. Also, MSBR wastes aged one
188

-

Table 15.5. Volume of water (in m>) potentially contaminated to radioactivity concentration
guide values for ingestion in unrestricted areas (10CFR20, Table 11, column 2) by
solidified high-level waste resulting from 33,000 MWd of exposure of enriched
2?’SU, 239Pa, and 233U fuels

2.0 x 10®* m® = volume of water required to reduce ingestion hazard of the corresponding amount of
uranium ore (0.17% U).

1.1 X 107 m® = volume of water that results from dissolving the salt required to store waste equivalent
to 33,000 MWd exposure to a terminal concentration of 500 ppm.

1.07 x 10® m> = approximate volume of water required to reduce ingestion hazard potential of the
corresponding amount of earth containing naturally occurring uranium plus thorium
in equilibrium with their daughters at the average concentration in the earth’s crust.

Age of waste 2355 239p,b 2334¢
(years)
30 1.26x 10! ®%sp 7.22x 10'° ®%sp) 2.11 x 10*! (®%sr)
100 2.24 x 10*° (®%sp) 1.32 x 10'° (®*°sp) 3.75x 10'° (°%sn)
300  2.00x 10% (*%n) 393x 10° **'am)  3.76 x 10° °%8r)

1,000 155x 107 **'Aam)  1.09x 10% ®*'Am)  4.72x 10° *?*Ra and 228Ra)
3,000  653x10°C*°Am) 224 x 107 ***Am)  5.06 x 10® *?’Ra and 2?®Ra) -
10,000  4.26 x 10% *3°Pu) 1.26 x 107 ***Am)  9.34 x 10% (*2%Ra)
30,000  2.44 x 10% (23°Ppu) 6.88 x 10% (2*°Pu) 2.2 % 107 (*2Ra)
100,000  2.14x 10° **°Ra) 574 x 10° *?®Ra) - 4.45x 107 (**°Ra)
300,000  2.38 x 10° (*?°Ra) 5.81 x 10® (326Ra) 4.68 x 107 (*26Ra)

1,000,000  1.58 x 10% (*?°1) 2.03 x 10% (*2°p) 9.42 X 10° (*25Ra)
3,000,000  9.93 x 10° (*?°1) 9.53 x 10° (12°]) 1.80 X 10° (228 Ra)
©10,000,000  5.28 x 10° (**°1) 4.88 x 10° (*2°1) 1.56 X 10° (328Ra)
30,000,000 257 % 10° (*2°1) 2.38 x 10° (12°)) 1.20 x 10% (228Ra)

dReference PWR fueled with 3.3% enriched uranium, operated at a specific power of 30 MW per
metric ton of heavy metal charged to reactor. Processing losses of 1/2% of uranium and plutonium to
waste are assumed. ' )

BAI reference oxide LMFBR mixed core and blankets fueled with LWR discharge plutonium and
diffusion plant tails. Average specific power of blend is 58.2 MW per metric ton of heavy meta]
charged to reactor. Processmg losses of é% of uranium and plutonium to waste are assumed.

CReference MSBR with continuous protactinium isolation on a ten-day cycle and rare-earth
removal by the metal transfer process. Thorium is discarded on a 4200-day cycle, and l/2% of the

uranium inventory in the reactor is assumed to be lost to waste over a 30-year plant life.

million years or more have a greater ingestion hazard
associated with them than other waste types, princi-
pally as the result of radioactivity from 232 Th daugh-
ters. Both of these problems can be alleviated by
making more efficient use of thorium in the fuel cycle,
which is, at present, utilized with an efficiency of only
13.7%. We are investigating processing schemes that can
increase the thorium utilization to greater than 90% and
thus eliminate these problems.

The study also revealed that a greater ingestion hazard
is associated with MSBR wastes during the period
30,000 to 1 million years as the result of the presence

of 22®Ra, a daughter of 233Pu. The isotope 23%Pu
exists in MSBR wastes in substantial concentrations as a
result of the long removal time for neptunium and the
short removal time for plutonium in the present
processing scheme. The amount of 2*#Pu in the MSBR

wastes could be reduced by removing neptunium more -

efficiently or by use of a processing scheme that would
allow plutonium to remain in the fuel salt and be
consumed by neutron capture. Both of these possi-
bilities are being considered as improvements to the
present processing flowsheet. :

N
189

15.5 AVAILABILITYIOF NATURAL
RESOURCES REQUIRED FOR
MOLTEN-SALT BREEDER REACTORS

M.J. Bell

A study has been made of published data on the
availability of and future demand for natural resources
of special importance to the MSBR program.® Materials
considered in the survey included the - constituents of
coolant and fuel salts, Hastelloy N, and materials
required for construction and operation of a chemical
processing plant. Table 15.6 presents the cumulative
demand for a number of materials through the year
2000, compares the requirements of the assumed MSBR
economy with those of the rest of the world, and
compares the world demand with world reserves and
estimated world resources of these materials. The

6. M. J. Bell, The Availability of Natural Resources for
Molten-Salt Breeder R eactors, ORNL-TM-3563 (in press).

known world reserves of beryllium, fluorine, and
bismuth are being rapidly depleted by non-MSBR uses
and are expected to be exhausted before the end of the
century. Ample beryllium resources (in the form of
bertrandite) and fluorine (in the form of phosphate
rock) exist; however, an improved mining technology
will be required to make recovery of these materials
economical. Bismuth resources are limited by the
demand for lead, copper, and zinc ores from which
bismuth is recovered as a by-product. In order to satisfy
the cumulative world demand through the year 2000, it
is necessary that new base metal ores be developed and
that the technology for recovering bismuth during the
refining of copper and zinc ores be improved. .

Ores from which thoria can be recovered for $10/1b
will be available well into the next century. Large
reserves of higher-priced thoria are also available. MSR
demands for all materials, with the exception of
thorium and hafnium, comprise only a small fraction of
the world demand for these materials and will not
require the development of new industries for the sole
purpose of sustaining an MSBR economy.
190

 J
Table 15.6. Estimated world primary demand for MSBR materials over the period 1968 —2000
. . . Cumulative
MSBR cumulative World cumulative Cumulative
MSBR demand world demand
Element demand for the demand for the as percentage of world demand as percentage of
€ period 1985—-2000  period 1968—2000° P g as percentage of p & ¢
world demand b  world resources” plus
(tons) (tons) world reserves
. world reserves
Fuel salt Li 7.3 x 103 2.7-3.3 x 10° 2-3 3644 d
Be® 2.1x 103 3.0-4.3 x 10° 5-17 250-360 1.9-2.7
W 4.0 x 104 2.3-8.4 x 10° 498 '4.5-16 1.7-6.3
ph 4.2 x 10* 1.2-1.4 x 10® 0.03-0.04 500-600 180-210
Hastelloy N Ni 2.5 % 10° 2.4-29 x 107 0.8-1.0 32-39 d
Mo 3.9 x 10° 4.2-5.0 x 10° 0.8-0.9 78-93 d
Cr 2.3 % 104 0.9-1.1 x 108 0.02 11-14 d
Fe 6.5x 103 1.7-2.1 x 10'° <107 17-21 3-4
Mn 6.5 X 10° 3.9-4.6 x 108 <107 53-63 2-3 -
Ti 1.6 x 103 1.9-44 X 106 0.04—0.08 1.2-29 d .
Nb 3.2 x 102 2.7-4.0 X 10° 0.08-0.12 2.7-4.0 d .
_ Hf 3.2 x 102 1.9-2.9x 103 11-17 0.6—0.9 d °
Other Bi 52x% 103 1.4-1.8 x 10° 2.9-3.7 140—180 78—100 .
B 1.5 x 10* 1.4-1.8 x 107 0.08-0.1 d

20-25

2U.S. Bureau of Mines, Mineral Facts and Problems, Bulletin 650, 1970 ed., U.S. Govt. Printing Office.

bReserves are known materials that may or may not be completely explored but may be quantitatively estimated; considered to
be economically exploijtable at the time of the estimate.

©Resources are materials other than reserves that may be ultimately exploitable; these include undiscovered but geologically
predicted deposits similar to present reserves as well as known deposits whose exploitation awaits more favorable economic or
technologic conditions.

dQuantitative estimates of world resources of these elements are not available.

€Beryllium resources include deposits containing at least 1% equivalent beryl (0.1% BeO).

fThorium reserves and potential resources recoverable at $10 per pound of ThO,.

&Ratio based on high range of forecast world cumulative demand.

#Fuorine reserves and resources of ores containing at least 35% CaF;, or equivalent value in fluorspar and metallic sulfides.
R

16. Processihg Chemistry

b

L. M. Ferris

Studies relating to the metal transfer process'*? for
the removal of rare-earth and other fission products
from MSBR fuel salt were continued. Included in this
work were measurements of the equilibrium distri-
bution of lithium and bismuth between liquid Li-Bi
alloys and molten LiCl, and of the mutual solubilities of

" thorium and selected rare earths in liquid bismuth. The

equilibrium precipitation of Pa, Os from MSBR fuel
salt by sparging with HF-H, O-H,-Ar gas mixtures was
also studied. The effect of temperature on the solubility
of Pa, Qs in fuel salt that was also saturated with ThO,

" was determined. Studies relating to the chemistry of

fuel reconstitution were begun. The initial phase.of this
work involves a study of the reaction of gaseous UFg
with UF, dissolved in MSBR fuel salt.

16.1 DISTRIBUTION OF LITHIUM AND BISMUTH
BETWEEN LIQUID LITHIUM-THORIUM-BISMUTH
ALLOYS AND MOLTEN LiCl

L.M. Ferris J. F.Land

In the metal transfer process,'’? rare earths and the
attendant small amount of thorium would be stripped
from the LiCl acceptor salt into lithium-bismuth so-
lutions having lithium concentrations of 5 to 50 at. %.

1. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb.
28, 1970, ORNL-4548, p. 277. ’

2. D. E. Ferguson and Staff, Chem. Technol Div. Annu.
Progr. Rep. Mar. 31, 1971, ORNL-4682, p. 2.

Work done at Argonne National Laboratory® with
LiCl-LiF solutions indicated that high lithium and
bismuth concentrations (up to 0.1 mole %) might be
expected in LiCl that is in equilibrium with lithium-
bismuth solutions in which the lithium concentration is
5 to 50 at. %. Concentrations of this magnitude could
have a significant effect on the performance of the
metal transfer process. Consequently, we have initiated
a study of the equilibrium distribution of lithium and
bismuth between molten LiCl and liquid lithium-
bismuth solutions. The results of this study should aid

.in optimizing operating conditions for the metal

191

transfer process.

In this study, LiCl containing less than 0.05 mole %
LiOH was equilibrated with various lithium-bismuth
solutions at 650°C. Thorium was added to the system
in some cases to minimize the oxide and hydroxide
concentrations in the salt. The apparatus and general
procedure have been described elsewhere.* The equi-
librium lithium concentration in the salt phase was
generally determined by removing a salt sample with a
stainless steel sampler, hydrolyzing the sample-in water,
and measuring the amount of hydrogen evolved by a
gaséchromatographic procedure. Samples for bismuth
analysis were taken with quartz samplers. When the

3. E. ). Cairns et ‘al., Galvanic Cells with Fused-Salt Elec-
trolytes, ANL-7316 (November 1967), p. 119.

4. L. M. Fenis, J. C. Mailen, J. J. Lawrance, F. J. Smith, and
E. D. Nogueira, J. Inorg. Nucl. Chem. 32, 2019 (1970).
bismuth concentration in the salt was greater than
about 100 wt ppm, a colorimetric analytical method
was used. An inverse-polarographic method was used
for the determination of bismuth at lower concen-
trations. ' .

Data obtained thus far at 650°C are summarized in
Table 16.1. As can be seen in the table, both the
equilibrium lithium and bismuth concentrations in the
salt phase increased markedly as the lithium concen-
tration in the metal phase was increased. This is also
shown quite clearly in Fig. 16.1, which is a log-log plot
of lithium concentration in the salt phase vs lithium
concentration in the metal phase. In those cases where
both lithium and bismuth analyses of the salt were
made, the Li/Bi atom ratio in the dissolved species was
about 3. Thus, the dramatic increase in the equilibrium
concentrations of lithium and bismuth in the salt with
increasing lithium concentration in the metal phase
appears to be related to the simultaneous increase in
concentration in the metal phase of a saltlike species,
such as Li;Bi, which has a high solubility in LiCl. The
data given in Table 16.1 indicate that the presence of
thorium in the metal phase had no measurable effect on
the distribution of lithium and bismuth between the
two phases. Measurements have also been made of the
solubilities of pure bismuth and of Li;Bi in molten
LiCl. The preliminary results indicate that the solubility
of bismuth is about 0.1 wt ppm at 650°C, whereas the
solubility of Li,Bi is about 0.3 mole %.

The magnitude of the equilibrium lithium and
bismuth concentrations in LiCl determined in this
investigation is consistent with the limited amount of

192

previous work done at Argonne National Laboratory>-®
with LiCl-LiF (75-25 mole %) as the salt phase and solid
Li;Bi or liquid Li-Bi (50-50 at. %) as the metal phase.
We do not agree, however, that the Li/Bi atom ratio in
the species dissolved in the salt is markedly lower than

.3 when the metal phase is a lithium-bismuth solution

rather than solid LizBi. The present study also yielded
values for the solubility of pure bismuth in LiCl that
were substantially lower than those obtained by the
ANL workers® with the molten LiCI-LiF salt phase.
Extrapolation of the results of this study indicates
that the equilibrium concentrations of both lithium and
bismuth will be less than 1 wt ppm in LiCl that is in
equilibrium with Li-Bi (595 at. %), which is the
composition of the alloy proposed for stripping the
trivalent rare earths and thorium from the LiCl in the
metal transfer process.'>? Lithium and bismuth dis-
solved in the LiCl at these low concentrations should
not produce a- deleterious effect on the performance of
this part of the metal transfer process. However, the
present results indicate that the lithium and bismuth
concentrations in the LiCl in contact with the divalent
rare-earth strip solution, Li-Bi (50-50 at. %), would be
about 0.3 and 0.1 mole % respectively. These concen-
trations are sufficiently high that a significant amount
of lithium and bismuth (as much as 370 and 1110
moles/day, respectively) will transfer from the lithium-
bismuth solution used to remove divalent rare earths

5. M. S. Foster, C. E. Crouthamel, D. M. Gruen, and R. L.
McBeth, J. Phys. Chem. 68(4), 980 (1964).

Table 16.1. Equilibrium di'stribution of lithium and bismuth between molten LiCl and liguid Li-Th-Bi solutions at 650°C

Concentration in metal phase Lithium concentration in Bismuth concentration in Li/Bi
Expt. Lithium " Thorium salt phase salt phase mole ratio
wt % at. % (ppm) wt ppm " mole ppm wt ppm mole ppm in salt
8la 1.09 24.9 0 218 126 + 50 186 + 20 38+4 34+14
79 1.20 _ 26.8 0 3027 184 £ 42 225 + 175 46 + 15 40=x1.6
76 a 1.85 36.2 750 110 £ 50 673 + 300 '
74 a 1.90 36.8 0 176 £ 15 1080 + 90
78 2.00 38.1 0 108 + 16 662 + 96 1016 + 20 206 3 3.2+0.5
74 b 2.09 39.1 0 210 + 40 1280 + 225
75 2.16 399 7500 220 + 40 1340 = 225
76 b 2.26 41.0 1300 298 + 50 1820 + 325
76 ¢’ 2.52 43.8 1800 430 £ 75 2620 = 460
‘ 76 d 2.90 47.4 2500 525t 60 3210 £ 340
82 3.00 48.2 0 4260 + 460 864 + 93

R 3

4y

193

ORNL-DWG 71-9539A

5000
A
/
®
2000 !
!
1000

| I
1]
1/

LITHIUM CONCENTRATION IN LiCl (mole ppm)

200

j

100

LITHIUM CONCENTRATION
IN Li-Bi ALLOY (at. %)

Fig. 16.1. Equilibrium distribution of lithium between mol-
ten LiCl and lithium-bismuth solutions at 650°C.

from LiCl. It is not believed that transfer of these
amounts of lithium and bismuth will cause difficulties
with the present flowsheet, however, since the lithium
will be available for the extraction of uranium and
protactinium.

10 20 30 40 50

16.2 MUTUAL SOLUBILITIES OF THORIUM
AND RARE EARTHS IN LIQUID BISMUTH

F.J.Smith  C.T. Thompson
J.F. Land

Lithium-bismuth solutions having lithium concen-
trations of 5 to 50 at. % will be used to strip thorium
and rare earths from the LiCl acceptor salt in the metal
transfer process.'*> We have been determining the
solubilities, both individual and mutual, of thorium and
selected rare earths in lithium-bismuth solutions to help
define optimum process conditions. Previously,® we
studied the precipitation of thorium bismuthide from
solutions containing rare-earth metals and observed
partitioning of lanthanum (initial concentration, ~1000
ppm) between the liquid and solid phase(s) at tempera-
tures below about 500°C. The available data were
interpreted in terms of coprecipitation of lanthanum
and thorium bismuthides. Results obtained in our
current investigations of the “mutual interactions of
thorium and rare earths in lithium-bismuth solutions
have led us to reevaluate our original work. Data
obtained recently have confirmed the earlier obser-
vations but cannot be explained in terms of a co-
precipitation model. The general behavior observed is
consistent with the formation of a compound con-
taining a rare earth and thorium (possibly a bis-
muthide). The general reaction for the formation of
such a compound would be:

Thygjy +x Loggjy + Bigjiqy = ThLn, Biy, (o15q) >

in which Ln denotes a lanthanide metal. A mole
fraction solubility product can be written as

Ksp =NThNpo *Ngi - . (1)

Under- the experimental conditions employed, the Ln
and Th concentrations were low; therefore, the mole
fraction of bismuth was close to unity and the mole
fraction solubility product can be closely approximated
by Ksp = NypNp,*- The variation of KSp with
temperature would be expected to follow the usual
relationship

log Ky, = log (NppNpn*) = A +B/T . (2)

6. F.J. Smith and C. T. Thompson, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1969, ORNL-4449, p. 217.
194

In each of a set of experiments, 300 g of bismuth
containing the equivalent of about 2000 ppm of rare
_ earth (either La or Nd) and about 4000 ppm of thorium
was equilibrated for at least 8 hr at several temperatures
between 750 and 300°C. At the end of each equili-
bration period, a filtered sample of the liquid phase was
removed for analysis. The experiments were initiated by
heating the system to about 750°C, where both the.
thorium and the rare-earth metal were soluble. The
temperature of the system was then progressively
lowered to about 300°C; finally, the temperature was
returned, in steps, to 750°C. The results from one of
our recent experiments (LA-3), along with those from a
previous® experiment (LA-2), are shown in Fig. 16.2.
Agreement among the results of these two experiments
is obvious. As the temperature of the system was
lowered, the thorium concentration in the liquid
remained at its original value until about 650°C was
teached. Below 650°C, the thorium concentration in
the liquid decreased regularly with decreasing temper-
ature (Fig. 16.2). The equilibrium thorium concen-

ORNL-OWG 7{-78684
TEMPERATURE (°C)

4 700 650 600 550 500 450 400 350
el i s ! ] q —
——INIiTIAL CONCENTRATIONS — |
e Lo, EXPT LA-2 ——
5 o La, EXPT LA-3 _
e i a Th, EXPT LA-2 ]
\ _ A Th, EXPT LA-3
. ]
_Q..._‘L}g S
_ 2 S =|8 - E]A}'A‘
£ \ N
o Lo L NA g
¥ 3¢ b N~ e
~ 40 N Y
-4 AN
<} N d
§ AW kN
a 5 o\
i ey
z AN -8-\ 5
g \D\ A \ i
= N2 ~N
. = &
o 2 L3
= 4 » .
5]
10 L]
N
AN
\\
° AN
4 N\g
N\
z a
10 T 12 13 14 15 16 17
10,000/, °K)

Fig. 16.2. Mutual solubilities of lanthanum and thorium in
liquid bismuth solutions. Excess solid thorium bismuthide was
present, fixing the thorium concentration in solution,

trations in each experiment are in excellent agreement

with the published? values for the solubility of thorium

in liquid bismuth. Similarly, as the temperature was
lowered, the lanthanum concentration in the liquid
remained at its original value until some temperature
below 550°C was reached; then it decreased regularly

with decreasing temperature (Fig. 16.2). The lanthanum

behavior in each experiment was identical. The equi-
librium lanthanum concentrations in the liquid were
much lower than the reported solubility values.®>* On
the basis of the lanthanum results, we postulate the
formation® of a La- and Th-containing solid phase that
has a very low solubility in liquid bismuth. The results
obtained for thorium indicate that there was sufficient
thorium present in the system to satisfy the solubility
product of the compound containing the rare earth and
thorium and to saturate the system with thorium
bismuthide. In this case, the maximum thorium concen-
tration in solution at a given temperature is determined
by the solubility of thorium in bismuth. The variation
of thorium.solubility in bismuth with temperature has
been shown’ to follow the usual relationship

logNpy =A'+BT. 3)

* Substitution of Eq. (3) into Eq. (2) yields

xlog Ny, =A" +B"IT,

indicating that a plot of the 10garithm of the lanthanum
concentration in the bismuth solution vs 1/7 should be
linear. As seen in Fig. 16.2, lanthanum behaved as

" predicted. It should be noted that a value of x cannot

be determined from data obtained in the manner
outlined above if thorium bismuthide is present as a
second solid phase. '

An experiment (Nd-6) with neodymium as the rare
earth was conducted in the same manner as experiments
LA-2 and LA-3, and the results obtained were similar to
those obtained with lanthanum. In addition, an approx-
imate value of x = 1 was obtained in an experiment
(E-70) conducted entirely at 640°C. In this experiment,
neodymium was added to a thorium-bismuth solution,
and the equilibrium concentrations in solution were
determined after each addition. From Eq. (1), a plot of

7. C. E. Schilling and L. M. Ferris, J. Less-Common Metals
20, 155 (1970).

8. V.I. Kober et al., Russ. J. Phys. Chem. 42, 360 (1968).

9. D. G. Schweitzer and J. R. Weeks, Trans. ASM 54, 185
(1961).

., .
L3

195

log Ny, vs log Ny4q was expected to be linear with a
slope of —x. Our data covered only. a limited range of
concentrations, but the slope of the plot indicated x to
be about 1. A third experiment (Nd-7) was conducted
in an attempt to measure the solubility of the Nd- and
Th-containing solid in the absence of excess thorium
bismuthide. The solution initially contained about 1 wt
% Nd and about 3500 ppm of Th. The temperature of
the system was varied between 500 and 700°C, and
samples of the liquid phase were removed at selected
temperatures. Analyses of these samples showed that
both the thorium and neodymium concentrations in the
liquid were much lower than the solubilities of the
respective bismuthides. From these results, we assumed
that only one solid phase, a Nd- and Th-containing
compound, was present. Assuming x = 1, we calculated
the mole fraction solubility product Ksp = NNaNTh»
using data from the three experiments involving neo-
dymium. The resulting log K, vs 1/T plot is shown in

Fig. 16.3. As seen, the data yield the expected linear
plot. It should be noted that agreement among all the
data cannot be obtained if values of x other than 1 are

ORNL-DWG 71-7867A
TEMPERATURE (°C)

-a 700 650 600 550 500 . 450 400
10 T T T H i = H
EXPT. Nd~6. SOLUTION ALSO SATURATED _|
\ . WITH THORIUM BISMUTHIDE, —
5 s ====[FEXTRAPQLATION OF DATA FROM Nd-6. —
\\ ® EXPT. Nd-7. NO THORIUM BISMUTHIDE —
N & PRESENT ]
‘\ A EXPT. E-70 TEMPERATURE CONSTANT AT
T \® 640 °C. AVERAGE OF 2 POINTS WITH
2 \ VARYING Th/Nd RATIO IN SOLUTION
N
\
-5
_© ary
r )
Z -
Z g \
-4 AJ
I AY
E \
5 e
s 2 A
& A\
x ‘ \
&8 \\
A}
5 \\
g \\
-7 .
10
10 k! 12 13 14 15 16
‘ 10,000/7 °K)

Fig. 16.3. Temperature dependence of the solubility product
for the compound ThNdB1 in liquid bismuth.

used. The equation for the line-in Fig. 163 is log
Kop(Na-Th) = 3233 ~ 7285/T(°K). Assuming that x is
also 1 for the La- and Th-containing solid phase, we

_.obtain, from Fig. 16.2, log Ksp(La Th)y = 2410 —

6480/ T(°K).

In reference to the metal transfer process,'»? the K,
values for both lanthanum and neodymium indicate
that the desired thorium and rare-earth concentrations
should be attainable in the lithium-bismuth strip so-
lutions without precipitation of a rare-earth- and
thorium-containing compound if the temperature of the
trivalent rare-earth strip solution Li-Bi (595 at. %) is
kept at 600°C or higher.

Using the data and interpretation given above, we
considered a scheme for removing rare earths from
MSBR fuel salt that involved reductive extraction
followed by selective precipitation of thorium bis-
muthide. Rare-earth fission products would be ex-
tracted from fuel salt that is free of uranium and
protactinium into bismuth that contains lithium and
thorium as reductants. If the extraction were conducted
at 600°C, the thorium concentration in the bismuth
leaving the extraction unit would be about 1800 wt
ppm for the reference case of a 1000-MW(e) MSBR and
a 25-day rare-earth removal cycle. The corresponding
rare-earth concentration would be less than 107 mole
fraction. Equilibrium cooling of the bismuth to 350°C
would precipitate thorium bismuthide and decrease the
thorium concentration in solution to about 35 wt ppm.
No rare earths would precipitate because their concen-
trations in solution would be too low to satisfy the K
for the compound containing the rare earth and
thorium. For example, the K., for the La-containing
solid at 350°C is about 1078, With a thorium concen-
tration of 35 wt ppm (Npp = 2.9 X 107°), the
lanthanum concentration in solution could be as high as
3 X 107 mole fraction (about 240 wt ppm) without
precipitation of a La-containing solid. The supernatant
solution resulting from the precipitation of thorium
bismuthide would be removed, and the dissolved species
(Li, Th, and rare earths) would be stripped, either by
hydrofluorination or hydrochlorination, into a suitable
salt phase for disposal. The clean bismuth would then
be used to dissolve the thorium bismuthide, more
thorium and lithium would be added, and the resulting
solution would be recycled to the reductive extraction
unit. The thorium decontamination factor for the
conditions cited above would be about 100 but could
be increased to about 1000 if the extraction were
conducted at 750°C and the resulting bismuth solution
were cooled to 300°C. The decontamination factor is
not affected markedly by increasing the rare-earth
removal cycle to at least 100 days. The chief dis-
advantage of the above scheme is the economic penalty
involved in the large lithium and thorium losses, about
30 and 15 kg/day, respectively, for the reference case.

16.3 OXIDE PRECIPITATION STUDIES
0O.K. Tallent F.J. Smith

Studies in support of the development of oxide
precipitation processes for isolating protactinium and
uranium from MSBR fuel salt have been continued.
Previous work'® has shown that addition of oxide to a
salt solution containing Pa®* results in precipitation of
pure, or nearly pure, Pa,0s and that Pa,Os can be
precipitated in preference to UQ,.'''? We are in-
vestigating a process in which salt from the reactor
would be treated with the appropriate HF-H, O-H, gas
mixture to convert practically all of the protactinium to
Pa®* and to precipitate a large fraction of the protac-
tinium as Pa,Os without precipitating uranium oxide.
We ‘have also determined the effect of temperature on
the solubility of Pa,O; in MSBR fuel salt that was
saturated with both ThO, and NiO. These measure-
ments aid in defining the lowest protactinium concen-
trations attainable. at a given temperature without
changing the uranium or thorium concentrations in the
salt.

Protactinium has been systematically precipitated
from molten LiF-BeF,-ThF,-UF, solutions at about
600°C by equilibrating the salt with various HF-H; O-Ar
and HF-H,O-H,-Ar gas mixtures. The data obtained
with HF-H, O-Ar gas mixtures were considered in terms
of the equilibrium

PaFs(d) + °4H,;0(g) = "4Pa, O5(s) + SHF(g), (4)
for which the equilibrium quotient at a given tempera-
ture can be written as

5

0, = (5)

pHgOS/zNPaFS

In the above expressions, (d), (g), (s), &, and p denote
_dissolved species, gas, solid, mole fraction, and partial

10. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR
Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622, p.
92,

11. J. C. Mailen, L. M. Ferris, and J. F. Land, MSR Program
Semiannu. Progr. Rep. Aug. 31, 1970, ORNL-4622 p. 208.

12. J. C. Mailen and L. M. Ferns, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 245.

196

pressure respectively. If the ratio py o/pyy is fixed at
some value A, Eq. (5) can be written, in logarithmic
form, as

log Np,p, =2.5logpyp —logQ, —2.5log 4. (6)
For a given value of A4, a log-log plot of protactinium
concentration in the salt vs HF partial pressure should
be linear with a slope of 2.5. If the gas present at
equilibrium contains hydrogen in addition to HF and

H, O, tetrapositive protactinium is produced according
to the reaction

@)

PaFs(d) + ', Hy(g) = PaF,(d) + HF(g),
for which the equilibrium quotient is:
 Npap,PHF
0, = 12 (8)
‘NPanpH'Z '
The total protactinium concen;[rati_on in the salt is
NPa =NPaFa, +NPaF5 )

When Pa, O is the solid phase at equilibrium, Npap, i
defined by Eq. (5). Substitution of Eqs. (5) and (9) 1nto
Eq. (8) thus yields:

Q2 =NPaQ1AS/2pHF-3/2pH2—1/2

-1/

— PHFPH, 2. (10)

Two experiments designed to yield values of Q, at
600°C have been completed. Each experiment was
initiated by loading 200 g of LiF-BeF,-ThF, (72-16-12
mole %), a weighed amount of 2*3U oxide, the
equivalent of about 100 wt ppm of 23'Pa, and 1 mCi
of 233Pa tracer into a nickel vessel. After the system
had been heated to 600°C, the salt was sparged for two
days with HF-H, (50-50 mole %) to dissolve the
uranium and protactinium in the salt. After residual HF
and H, had been removed by sparging with argon,
filtered samples of the salt were taken for analysis. By
analysis, the initial uranium concentration in the salt
was 0.288 + 0.028 wt % (about 0.08 mole %) in
experiment | and was 0.93 + 0.03 wt % (about 0.25
mole %) in experiment Il. The salt was then sparged
with either an HF-H,O-Ar or an HF-H,O-H,-Ar gas
mixture. A standardized aqueous HF-H, O solution
(H, O/HF mole ratio = 3) was fed at a controlled rate to
a heated vaporizer, using a Sage syringe pump. with a
~

197

Teflon syringe and piston. The vaporizer was a heated
cylindrically shaped 500-ml Monel vessel that had a
gold-plated cup in the bottom. The HF-H,O solution
was fed into the cup, where it was vaporized and then

was carried into the nickel reaction vessel by a stream .

of argon or an Ar-H, mixture. The argon and hydrogen
flow rates were controlled using calibrated DP cells. In

-all cases, the salt was sampled without measurably

changing the composition of the gas phase. In prelim-
inary experiments, in which samples of the salt were
removed periodically, it was shown that equilibrium
was attained in less than 16 hr. All data reported below
were obtained by allowing at least 16 hr for equili-
bration. Each sample was first analyzed for protac-
tinium by gamma spectrometry and then was submitted
to the Analytical Chemistry Division for dissolution and
analyses for 2*1Pa and 222U by the alpha-pulse-height
method. :

Data obtained at 600°C in the two experiments
outlined above are summarized in Table 16.2. As seen,
when HF-H,O-Ar gas mixtures were used, the equi-
librium protactinium concentration in the salt varied

regularly with the partial pressure of HF in the gas
mixture. A log-log plot of the protactinium concen-
tration vs pyy is shown in Fig. 16.4. The protactinium
concentrations used in this plot were those obtained by
gamma spectrometry, since this analytical method gave
what appeared to be the most accurate and self-
consistent results. Increasing the UF,. concentration
from 0.08 to 0.25 mole % had no measurable effect on
the equilibrium protactinium concentrations. The slope
of the line in Fig. 16.4 is 2.5; we interpret this as
confirmation that the equilibrium involved was that
given by Eq. (4). The equation of the line is log (wt
ppm Pa) = 2.5 log pyg + 4.870, from which we derive
log NPaF5 = 2.5 log pyp — 1.6925. Since 4 = 3, we
obtain @, = 3.2 at 600°C.

The uranium analyses of the salt, despite their
considerable variation, support the contention that
protactinium oxide was precipitated as a pure solid
phase. However, we do not yet have sufficient data to
define accurately the lowest protactinium concen-
trations attainable under given conditions without
precipitating uranium and thorium. oxides. Our data

Table 16.2. Equilibrium protactinium concentrations obtained by spa'rgi_ng LiF-BeF2—ThF4"(7 2-16-12 _inole %)
containing UF4 with HF-H,0-Ar or HF-H,O-H,-Ar gas mixtures at 600°C ’

Pa concentration

in salt (wt ppm) UrantluT
. P D concentration
Experiment (alglrg) (a?rfi) By gamma By alpl}a- in salt Q2
spectrometry pulse-height (Wt %)
analysis

I 0.019 0 5.5 9.8
1 0.024 0 7.2
| 0.030 -0 13.0
1 0.031 0 134 10.7 0.241
I 0.040 0 194 15.9- 0.292
| 0.042 0 33.6 25.7 0.259
I 0.043 0 36.0 31.2 0.174
I 0.043 0 264 20.2 0.199
I 0.047 0 37.5 36.0 0.907
I 0.050 0 44.3
1 0.052 0 36.2 59.6 .. 0.263
I 0.052 0 41.6 43.9 0.262
II 0.055 0 66.1 71.0 0.920
II 0.058 0 - 84.0 88.0 0.937
I 0.063 0 73.0 71.5 0.334
I 0.065 0 72.0 58.5 0.332
II 0.065 0 64.9 86.0 0977
I 0.070 0 69.0 58.5 0.282
I 0.0212 0.101 10.4 0.08
I 10.0227 0.0843 26.4 0.28
| 0.0340 0.140 49.5 0.19
| 0.154 62.0 0.20

0.0375

198

indicate that at 600°C the protactinium concentration
can probably be reduced to about 5 wt ppm without
the attendant precipitation of uranium oxide.

When hydrogen was present in the gas mixture, the
equilibrium protactinium concentration at a given value
of pyp was higher than that obtained in the absence of
hydrogen (Table 16.2). We attribute this behavior to
the formation of Pa%*, according to Eq. (7). The values
of @, given in Table 16.2 were calculated, using Eq.
(10}, from the data obtained with hydrogen present in
the gas mixture. Our average value of @, = 0.2 at 600°C
is in good agreement with the value of about 0.5
obtained by extrapolation of values given by
Bamberger, Ross, and.Baes."?-'4

The solubility of Pa, O in LiF-BeF,-ThF, (72-16-12
mole %) that was also saturated with ThO, and NiO
was determined over the temperature range 560 to
690°C. The equilibrium involved was assumed to be

% ThF 4(d) + 4 Pa, O5(s) = PaF(d) + % ThO,(s), (11)

for which

Np,r,

Qy=—+. (12)
Npyg, '

The apparatus and general technique have been de-
scribed elsewhere.* Initially, 100 g of salt containing
about 10.5 wt ppm of 23! Pa and about 1 mCi of 23Pa
tracer was hydrofluorinated at 650°C in a molybdenum
crucible to dissolve the- protactinium in the salt. After
stripping residual HF and H, from the system with pure
argon, sufficient ThO, (~1.2 g) was added to saturate
the melt with ThO,. About 0.4 g of NiO (an oxidant)
was also added to ensure that the protactinium was in
the 5+ oxidation state.!® After an initial period of six
days, duplicate filtered samples of the salt were taken at
each of several temperatures. At least 20 hr was allowed
for the attainment of equilibrium at each temperature.
The samples were analyzed for protactinium by
counting the gamma rays emitted by the 233Pa. Where
possible, the samples were also analyzed for >2!Pa by

13. C. E. Bamberger, R. G. Ross, and C. F. Baes, J1., Reactor
Chem. Div. Annu. Progr. Rep. May 31, 1971, ORNL4717, p.
8.

14. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr.,, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p.
120.

ORNL-DWG 71— 9538A

100 T I

|8 EXPT 1Ny, =0.08 4/
= | a EXPT II;Nyp, = 0.25 %
g. UFq ‘/'A
(o}
2 50 : /
5 .
- A
n eo,le
z °
=
o °
I—-
& /
[ e

20
> /"
L
Q
>
o
s ]
p=
2 0 //
=
O /
= 7
o
T /
a
o /
5 y
0.01 0.02 0.05 o R
Pur {atm)

Fig. 16.4. Equilibrium protactinium concentrations obtained
by sparging LiF-BeF,-ThF, (72-16-12 mole %) containing UF4
with various HF-H,O-Ar gas mixtures at 600°C. The H,O/HF
mole ratio was 3. The equation for the line is: log (wt ppm Pa)
= 2.5 log pyg + 4.870.

the alpha-pulse-height method. The agreement among
the analyses was generally good.

The data obtained are summarized in Table 16.3 and
are shown graphically ‘in Fig. 16.5 as a plot of the
logarithm of the protactinium concentration vs 1/7.
The plot includes our previously reported'? values. The
least-squares equation for the line is log (wt ppm Pa®")
= 19.471 — 17890/T(°K), which is equivalent to log Q3
= 14.06 — 17890/T(°K); the standard deviation of log
Q5 is 0.3. At temperatures above 650°C, our values of
Q5 are in reasonable agreement with those determined
by Ross, Bamberger, and Baes;'® however, our values
are much lower than theirs at temperatures below
650°C.

Values of @5 can also be estimated by combining the
quotients for the following two equilibria:

Ya UFa(d) + ', Pa; O5(s) =PaFs(d) + % UO,(ss) , (13)
~

UF,4(d) + ThO,(ss) = UQ,(ss) + ThF,4(d) .  (14)

In these expressions, (ss) denotes ThO,-UQO, solid
solution. It is easily seen that log @3 =log @,3 — % log
(@1 4. Utilizing the values of @, 3 determined by Mailen
and Ferris'? at temperatures above 600°C and the
values of Q,, reported by Bamberger and Baes,!® we
calculate values of @3 in the temperature range 600 to
750°C that are within a factor of 3 of the values
obtained by our direct measurements. Both Q,; and
Q.4 were evaluated at NUF4(d) = 0.0022 for the
comparison.

Table 16.3. Solubility of Pa;Os in LiF-BeF;-ThE,
(72-16-12 mole %) at ThO, saturation of the salt

Protactinium

Temperature concentration

O in salt 10° 05
{wt ppm)

562 0.014 0.06
610 - 0.17 0.66
632 0.32 | 1.2
642 0.48 1.9
655 3.75 . 15
660 ' 0.63 2.5
675 4.95 . : 19
690 4.60 - 18

16.4 CHEMISTRY OF FUEL RECONSTITUTION
M. R. Bennett

The current flowsheet? for the processing of MSBR
fuel involves removal of the uranium from the salt as
UF, by fluorination prior to isolation of protactinium

by reductive extraction and removal of rare earths by -

the metal transfer process. The proposed fuel re-
constitution method consists in contacting, in the
presence of hydrogen, the purified UF with carrier salt
that contains some dissolved UF,. The expected
sequence of reactions, which could occur almost simul-
taneously, is ' :

UF¢(g) + UF4(d) = 2UF5(d), (15)

2UF(d) + H,(g) = 2UF,(d) + 2HF(g) . (16)

15. C. E. Bamberger and C. F. Baes, J1., J. Nucl. Mater. 35, - -

177 (1970).

199

ORNL- DWG 71-9540A
TEMPERATURE (°C)
750 700 650 600 550

100

I I T I
1 ' ! ' ]

log (wt ppm Po®*)= 19,471 -17890/7 (°K) _

.5
P6°" CONCENTRATION IN SALT (wt ppm)

h Y
LY
\
®
N
T\
o\\
.
04 \
—— @ THIS WORK Y
[ & MAILEN AND FERRIS'2 \\
N
0.01
9.5 10.0 10.5 1.0 1.5 12.0
10,000/ oy, :

Fig. 16.5. Solubility of Pa,0s in molten LiF-BeF,-ThF,
(72-16-12 mole %) that is also saturated with ThO, and NiO.

Some time ago' ¢>!7 this concept was tested by adding
gaseous UF4 at 600°C to LiF-ZrF, (5149 mole %) that
initially contained about 1 wt % UF,. The salt, which
was held in a nickel vessel, was sparged with hydrogen

.after the addition of UF,. All of the UF4 added to the

system was absorbed by the salt; however, the uranium
was found to be almost entirely in the 4+ oxidation
state even before hydrogen was admitted to the system.
It was postulated that UFs; was initially formed
according to reaction (15) and that the UFs was very

16. L. E. McNeese and C. D. Scott, Reconstitution of MSR
Fuel by Reducing UFg Gas to UF4 in a Molten Salt,
ORNL-TM-1051 (Mar. 11, 1965).

17. L. E. McNeese, Unit Operations Section Quarterly Progr.
Rep. April—June 1965, ORNL-3868, p. 43.
rapidly reduced by the nickel container according to

- the reaction

2UF5(d) + Ni(s) = 2UF4(d) + NiF,(d) . (17)

We have begun a more comprehensive study of the
chemistry involved in the fuel reconstitution step. Qur
initial goals were to establish the validity of the
stoichiometry of reaction (15) and to find a suitable
material for containing molten-salt solutions in which
uranium is present in oxidation states higher than 4+.
The experimental equipment is shown schematically
in Fig. 16.6. The UF, generation system consisted of a
1-Ib UF¢ cylinder fitted with the appropriate control
valves and gages and a calibrated DP cell for the
accurate control of UFg flow rates. The entire
generation system was wrapped with resistance heaters
and thermal insulation so that it could be maintained at
65 to 70°C. The nickel reaction vessel, which was
similar to the vessels used in- our related studies,® was
equipped with %;-in.-diam ports for a thermowell and
sparge tube and a % -in.-diam ball-valve sample port.
The general experimental procedure used was as
follows: First, the UF, generation system was re-
peatedly pressurized and purged with UFg4 at about 15
" psig in order to remove argon from the lines and valves.
Excess UF¢ was vented to NaF trap No. 1 (Fig. 16.6).

- | DP CELL
@ ar > —| 65-70°C

A predetermined amount of UF4 was then metered into
purified molten salt contained in a crucible inside the
nickel reaction vessel. The UF, flow rates generally
ranged from 0.1 to 0.25 g/min. Argon was admitted at a

low flow rate at point B (Fig. 16.6) at the same time

that UF, was flowing into the reaction vessel. This
argon flow was maintained after the UF, introduction
period until the thermocouple and sparge tube had been
retracted to positions above the salt level. The system
was then left under argon for the desired period of time
before the salt was sampled. Samples were taken by
inserting a cold graphite rod through the ball valve into
the salt and withdrawing it rapidly. The sample was
hydrolyzed in phosphoric acid, and the resultant
solution was analyzed for total uranium and US*. We
assumed that uranium present in the salt in an
oxidation state other than 4+ was present as UF;, and
thus on hydrolysis the following reaction occurred:

2US = U™ + US* . - (18)

Obviously, when all the uranium in the salt is present in
the 5+ state, the U%* concentration is one-half the total
uranium concentration.

Our first series of experiments was initiated by
loading 2.9 g of UF; and 197.1 g of LiF-BeF,-ThF,

(72-16-12 mole %) into a high-density, high-purity .

ORNL-DWG 71-9542A

Lo

CAPILLARY
TUBE

THERMOCOUPLE ——)

\Z %
%

BALL-VALVE
SAMPLE PORT

A
! H%l |

NaF TRAP NO. 2

{

___NICKEL REACTION

i SPARGE TUBE

=
M

| - ——MOLTEN SALT

)

Fig. 16.6. Schematic diagram of apparatus used in the study of the reaction of gaseous UF¢ with UF, dissolved in a molten

fluoride salt,

1

1Y

i
graphite crucible, which was then sealed in the nickel
reaction vessel under argon and heated to 600°C. When
molten, the salt produced a pool about 2 in. deep. The
salt was then sparged with HF-H, (50-50 mole %) for
about two days to ensure dissolution of the UF, and to
remove oxide from the system. The sparge tube used
for the hydrofluorination was fabricated of a relatively
low-density spectrographic-grade graphite. Analysis of
the salt after hydrofluorination gave the expected 1.1
wt % uranium. With the system at 600°C, we attempted
to introduce UF, gas into the system through the
graphite sparge tube, which extended to within % in. of
the bottom of the reaction vessel. However, the UF
was quantitatively reduced to UF, at the inlet of the
tube, even though the temperature was 200°C or lower
at this point. Analysis of the salt after this attempt
showed that the uranium concentration in the salt was
still 1.1 wt %.

Prior to a second test, the graphite sparge tube was
replaced by one made of high-fired Al,O5. Then, five
times the amount of UF4 necessary to convert the
uranium to UFg was fed to the system. Following the
addition, the Al, 0, sparge tube was retracted to a
position above the liquid salt, and the system was
maintained under an argon atmosphere for several hours
before the salt was sampled. Analyses showed that all of
the uranium added to the system was present in the salt
(the uranium concentration had increased from 1.1 to
5.1 wt %), but little, if any, UFs was produced. The
Al, 05 sparge tube had been severely corroded.

The Al, O sparge tube was replaced by one made of
copper, and the amount of UF¢ required to convert all

201

of the UF, in the salt to UFs was introduced. The

copper sparge tube was then retracted so that it was not
in contact with the molten salt, and the system was
again maintained under argon for several hours before
the salt was sampled. Analyses of the samples showed
that all of the uranium introduced as UF¢ had been
absorbed in the salt; the total uranium concentration
was now 9.63 wt %. About 60% of the uranium was
present as UFs, and the Cu/UF, ratio was about 1. The
copper sparge tube was badly corroded. Either the
copper had reacted with UF¢ according to the reaction

UF4(g) + Cu(s) = UF,(d) + CuF,(d) (19)

during the 0.5-hr period in which UF4 was being added
to the system, or it was corroded according to the
following sequence of reactions:

UF(g) + UF,4(d) = 2UF5(d) , (15)

2UF(d) + Cu(s) = CuF,(d) + 2UF,(d) . (20)

Despite the corrosion of the copper sparge tube, the
results of this test strongly indicated that the graphite
crucible was inert, at least for a short time, to UF;
dissolved in a molten salt.

Even more convincing evidence for the stability of
graphite to dissolved UFs was obtained in the final test
in this series. The salt from the prior test involving the
copper sparge tube was left at 600°C under argon for
about two days while we awaited analyses of the
samples. Once these analyses were available, the amount
of UF¢ necessary to convert the 40% of the uranium
calculated to be present as UF,; to UF; was admitted to
the system through a gold sparge tube. The gold was
unaffected during the 0.5-hr period required for the
addition of the UF¢. After the gold sparge tube had
been retracted, the salt was left under argon for about 6
hr before it was sampled. Analyses confirmed not only
that all of the uranium added as UFg4 ‘had been
absorbed (the total uranium concentration was 12.8 wt
% compared with the expected 12.5%) but also that the
uranium was present in the salt, within analytical
uncertainty, as UFs (the U concentration in the acid
was equivalent to 6.57% U** in the salt). These results
strongly indicate that the UFs concentration in the salt
had not been reduced in the two-day period between
the tests with the copper and gold sparge tubes,
providing additional confirmation that graphite is stable
to UF; dissolved in molten LiF-BeF,-ThF, (72-16-12
mole %). These results also support the stoichiometry
of reaction (15) and suggest that gold would be a
suitable material for the containment of dissolved UFs.
Consequently, we are planning an experiment in gold
apparatus not only to evaluate the long-term stability of
gold to dissolved UFs but also to obtain further
confirmation of the stoichiometry of reaction (15).
17. Engineering Development of Processing Operations

L. E. McNeese

Studies related to the development of a number of
processing operations were continued during this report
period. Additional information on® the behavior of
lithium during metal transfer experiment MTE-2 was
developed. The data are consistent with recently ob-
tained data on the concentration of lithium in LiCl that
is in equilibrium with lithium-bismuth solutions. A new
experiment (MTE-2B) is under way to study further the
transfer of lithium from a Li-Bi solution containing
lithium at the concentration proposed for extracting
trivalent rare earths from lithium chloride (5 at. %).
Studies on mechanically agitated salt-metal contactors,
which are being developed as an alternative to packed
columns, were also continued during this period. The
design of the third engineering experiment for develop-
ment of the metal transfer process (MTE-3) has been
completed, and most of the equipment has been
fabricated. Installation of the completed equipment
items is in progress. Mass transfer experiments were
begun in which the rate of transfer of zirconium from
molten salt to bismuth was measured during the
countercurrent contact of salt and bismuth in a packed
column. Equations were developed for predicting the
rate of heat generation in a frozen-wall fluorinator that
uses radio-frequency induction heating. Calculations
were carried out for predicting the performance of
continuous fluorinators. The installation of equipment
for engineering studies on uranium oxide precipitation
was completed, and two experiments were carried out.
Work on the design of a processing materials test stand
and the molybdenum reductive extraction equipment

was continued. Studies of a salt-bismuth interface
detector for use in salt-metal contactors were also
continued.

17.1 LITHIUM TRANSFER DURING METAL
TRANSFER EXPERIMENT MTE-2

E.L. Youngblood L.E.McNeese

Additional analytical results were obtained for the
concentration of lithium in the Li-Bi solution used for
extracting rare earths from lithium chloride in metal
transfer experiment MTE-2, which was completed
previously. During the experiment,! the lithium -con-
centration decreased from an initial value of 0.35 to
0.18 mole fraction after about 570 liters of lithium
chloride had been contacted with the Li-Bi solution (see
Fig. 17.1). Only a small fraction of this decrease can be
accounted for by the reaction of rare earths with
lithium. The major portion of the decrease is believed

~ to be associated with the circulation of the lithium

202

chloride since little or no decrease occurred during
periods in which LiCl was not circulated. Recently, it
has been determined that the equilibrium concentra-
tions of lithium and bismuth in lithium chloride in
contact with lithium-bismuth solutions are appreciable

1. E. L. Youngblood and L. E. McNeese, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp.
249-53.
L B

ORNL-DWG 71-9546A
0.40

0.35 feen—
[ W
-
0.30 | hd
’ 3
L ]
..

PREDICTED Li
CONCENTRATION

Li IN Li-Bi SOLUTION (mole fraction)

Py [ ]
0.25 AN »
\..\
*
‘;""‘--_ -
0.20 » ";--
H
045 ;
o] 100 200 300 400 500 600

VOLUME OF LiClI PUMPED (liters)

Fig. 17.1. Variation of lithium concentration in the Li-Bi

203

solution used for removing rare earths from LiCl during metal

transfer experiment MTE-2.

and that they increase markedly as the lithium concen--
tration in the metal phase is increased (see Sect. 1'5).

A mathematical analysis has been carried out using
the recently obtained data on the equilibrium lithium
concentration in lithium chloride in contact with Li-Bi

where

‘Vgi.Li = volume of the bismuth-lithium solution,
~ moles,

XLiBi) = concentration of lithium in lithium-
bismuth solution, mole fraction,

t = time, min;
f = fractional approach to equilibrium ob-
tained at the LiCl-Li-Bi interface,

F =lithium chloride circulation rate, moles/
min,
XLicLicl) = concentration of lithium in lithium chlo-
ride in equilibrium with Li-Bi solution,

mole fraction,
R, , = rate at which lithium is removed from the

Li-Bi solution by reaction with LaCls,
moles/min.

In the above relation, it has been assumed that
essentially all of the lithium was removed from the

lithium chloride by extraction into the main bismuth

solutions. These data can be represented by the

relation:
XLicLicy) = 0-2278 X ;¢piy* ' (1)

where

XLi(Lic1) = concentration of lithium in lithium chlo-
ride, mole fraction,

Xyipiy = concentration of lithium in lithium-
bismuth solution, mole fraction.

In experiment MTE-2, lithium would have been re-
moved from the lithium-bismuth solution as a result of
(1) circulation of the LiCl and (2) extraction of
materials into the lithium-bismuth solution. During the
period of interest (which constitutes most of the
experiment), only lanthanum was present in sufficiently
high concentrations to react with an appreciable
amount of lithium. Thus a material balance on lithium
in the Li-Bisolution can be written as follows:

d .
~VBi-Li 3; XLi(Bi) =fFxpicLicn T RiLa (2)

pool in the experiment. Equation (1) was substituted
into Eq. (2), and the resulting equation was solved
numerically for various values of fin order to determine
the value that best represented the data. The curve
shown in Fig. 17.1, which resulted from an f value of
0.37, is in good agreement with the measured lithium
concentration in the Li-Bi solution during the experi-
ment, and the value of fis in reasonable agreement with
the value of 0.2 reported earlier, based on the rate of
accumulation of lanthanum and neodymium in the
Li-Bi solution. It should be observed that essentially no
transfer of lithium occurred until after about 50 liters
of lithium chloride had been circulated; at this time,
operation of a gas sparge tube in the Li-Bi container was
initiated. Similar behavior was observed in determining
the rate of accumulation of lanthanum and neodymium
in the Li-Bi solution during this period in that essen-
tially no accumulation was observed until the sparge
tube was activated. The agreement of the measured
lithium concentration with the predicted values was less
satisfactory after a total of 400 liters of lithium
chloride had been circulated because, by this time, a
hoie had developed in the Li-Bi container. This hole
caused poor contact of part of the solution with the
lithium chloride during the remainder of the experi-
ment. Thus the measured lithium concentrations would
be expected to be lower than the predicted values, as is
observed in Fig. 17.1.
The agreement between the predicted and observed
behavior of lithium in experiment MTE-2 provides good
confirmation of the recently measured data on the
equilibrium concentration of lithium in lithium chloride
in contact with a lithium-bismuth solution having a high
lithium concentration.

17.2 OPERATION OF METAL TRANSFER
EXPERIMENT MTE-2B

E. L. Youngblood L. E. McNeese

As discussed in the previous section, a much larger
decrease than expected was observed in the lithium
concentration in the bismuth into which rare earths
were extracted from the lithium chloride in metal
transfer experiment MTE-2. A new experiment, desig-
nated MTE-2B, is now under way to study further the
transfer of lithium from a.lithium-bismuth solution
containing lithium at the concentration proposed for
extracting trivalent rare earths from lithium chloride
(ie., 5 at. %).

The experimental equipment is shown schematically
in Fig. 17.2. All components that contact salt and

204

bismuth are fabricated of carbon steel. The main vessel
is constructed of 6-in.-diam sched 4Q pipe and is divided
into two compartments by a partition that extends to
within % in. of the bottom of the vessel. The two
compartments are interconnected by a pool of bismuth
containing reductant. One compartment contains fluo-
ride salt (67-33 mole % LiF-BeF,) to which were added
11 mCi of '*7"NdF; and sufficient ThF, to produce a
concentration of 0.19 mole %. The other compartment
contains lithium chloride. A separate, electrically insu-
lated vessel containing a 5 at. % Li-Bi solution is
connected to the lithium -chloride compartment with a
Y4-in.-diam sched 80 pipe. During operation, molten
lithium chloride is circulated between the Li-Bi vessel
and the compartmented vessel by pressurizing the Li-Bi
container with argon. Gas-lift sparge tubes are used to
improve the contact of the salt and metal phases. The
experiment is being operated at 645°C.

The quantities of materials used in the experiment are
shown in Table 17.1. These materials were purified for
removal of oxides prior to being used in the experi-
ment. The two bismuth phases were sparged with H, at
650°C for 12 hr and were charged to the carbon-steel

ORNL~DWG 71-6199A

ARGON INLET AND VENT ~ T

WY/ 2

6-in. CARBON STEEL PIPE —4

LEVEL
ELECTRODES

ARGON INLET
/AND VENT

%@%m

24 ~in.

4-in. CARBON
STEEL PIPE

3-in. CARBCN
STEEL PIPE

5at. % LiIN Bi

LiF - BeFp - ThF4
166.8-33-0.19 mole %)

BISMUTH WITH 0.2 at % Li

ALUMINA INSULATORS

Fig. 17.2. Schematic diagram of equipment used for metal transfer experiment MTE-2B,
-

205

Table 17.1. Materials used in metal transfer experiment MTE-2B

Volume Moles
(cm?) (8)

Fluoride salt ’ 794 47

(LiF-BeF,-ThF 4,

66.8-33-0.19 mole %

plus 11 mCi of 147Nd

as 'tracgr)
Bismuth (containing 1130 52

about 50 ppm of Li

and 200 ppm of Th)
LiCl : 1338 47
Li-Bi (5 at. % Li) 187 9

vessel, which had also been contacted with hydrogen at
650°C for 12 hr. The lithium chloride was contacted
with bismuth that had been saturated with thorium at
650°C, and the fluoride salt was obtained in purified
form from the Reactor Chemistry Division.

The experiment is designed in a manner such that
data on lithium transfer can be obtained by the
following independent methods:

1. Direct determination of lithium in the Li-Bi solution
used for extraction of rare earths from the lithium
chloride. '

2. Direct determination of the lithium and thorium
concentrations in LiCl in equilibrium with the Li-Bi
solution. -

3. Determination of the rate at which lithium is
transferred from the Li-Bi solution to the main
bismuth pool, as indicated by changes in the
distribution ratios for thorium and '*’Nd. The
composition of the fluoride salt and the relative
volummes of fluoride salt and bismuth were chosen
such that the maximum thorium concentration that
can be attained in the bismuth is only one-half the
solubility of thorium in bismuth. This will prevent
the bismuth phase from becoming saturated with
thorium. If saturation occurs, the. thorium and
neodymium distribution ratios will not be sensitive
to transfer of lithium into the main bismuth pool.

4. Measurement of the voltage that is developed when
the two bismuth phases containing lithium are
connected by the lithium chloride. It was antici-
pated that the developed voltage could be interpre-
ted in terms of a concentration cell involving
bismuth phases that contain lithium at different
concentrations, ‘

Data obtained thus far are summarized in Table 17.2.
During the first two weeks of the experiment, decreases
were observed in the concentrations of lithium in the
Li-Bi solution and in the main bismuth pool. Simulta-
neous decreases were noted in the distribution coeffi-
cients for thorium and neodymium, whereas an increase
was noted in the emf measured between the two
bismuth phases. We believe the decreases in the lithium
concentrations resulted from reacticn of lithium with
impurities in the system. Two weeks after the beginning
of the experiment, 8.5 g of thorium was added to the
Th-Bi solution in order to increase the concentration of
reductant to the desired value. At this time, the
measured emf between the two bismuth phases de-
creased to near the expected value of about 250 mV,
and the distribution values for thorium and neodymium
increased 'to about the expected values. During the
subsequent operating period, the lithium concentration
in the Li-Bi solution decreased slightly (from 4.40 to
4.0 mole %), as shown in Fig. 17.3. During this period,
the emf measured between the two bismuth phases
increased steadily from about 250 to about 276 mV.
Calculated values for the lithium concentration in the
Th-Bi solution, based on the measured emf values and
the concentration of lithium in the Li-Bi solution, are
observed to be in reasonable agreement with lithium
concentrations determined by analysis. Data on the
variations of the thorium and neodymium distribution
ratios during the experiment are shown in Fig. 17.4.
Calculated results based on emf measurements and on
lithium analyses -of the Th-Bi solution are shown for
comparison with experimentally determined values.
Although the scatter in the experimental data is
appreciable, the values appear to lie closer to the results
that are based on emf measurements than to those
based on lithium analyses.

Several conclusions can be drawn from these data.
The observed decrease in the lithium concentration in
the Li-Bi solution represents an equilibrium lithium
concentration in the LiCl of less than 0.3 wt ppm;
however, transfer of this amount of lithium from the
Li-Bi solution to the Th-Bi solution would have
increased the concentration of reductant in the Th-Bi
solution by about 10% and would have also increased
the neodymium and thorium distribution ratios by
about 33 and 46% respectively. In addition, a decrease
in measured emf of about 5% would have been
observed. The data appear to be consistent with a slight
loss of reductant from both the Li-Bi and the Th-Bi
solutions by reaction with impurities in the system. It is
believed that further operation of the experiment will
Table 17.2. Summary of data from metal transfer experiment MTE-2B

Measured distribution ratios

) Volume Li concentration in:
Time Purfipmg of LiCl Measured LiBi Main Bi Thorium Neodymium Neodymium Inventories (%)
(days) Time pumped emf solution I between between between Thori Neodvmium Comments
(hr) dorgy - (mV) poo Bi and Bi and Bi and onum ocy
(liters) (mole %) (mole %) . . : ' ‘
_ fluoride fluoride LiCi

0 0 0 2544 4.934 0.196¢ 0.15 41 100

7 0 0 332 4.88 0.00003 0.006 106
12 0 0 345 433 0.02 0.003 <0.00004 <0.0003 150 69
14 0 0. ‘ Added 8.5 ¢

of thorium
. to Th-Bi

15 0 0 250 0.07 2.6 50 solution
18 0 0 253 4.33 0.169 0.10 0.16 2.5 ' 86 20
20 7.3 11.6 255 4.60 0.157 0.10 0.15 1.4 87 11
22 21.0 36.6 255 4.05 0.18 1.7 10
25 27.7 46.9 253 4.60 0.30 0.08" 0.16 : 1.2 96 6
27 41.8 69.9 253 4.60 0.151 0.11 0.18 1.8 84 8
32 62.4 103.7 260 4.05 0.157 = 0.06 0.15 3.5 91 10
34 81.6 131.9 259 4.60 0.139 0.07 0.15 1.7 76 10
39 160.2 244.6 267 4.33 0.151 0.06 0.12 1.5 78 5
41 173.4 265.0 262 4.60 0.160 0.05 0.06 0.6 75 5
46 194.3 293.3 264 4.05 0.111 0.04 0.20 04 85 4
53 . 280.7 428.0 273 4.05 0.142 0.03 0.09 1.2 76 11
60 384.7 591.1 276 4.08 0.0994 0.06 0.4 13

2Calculated value.

90¢C
ORNL-DWG 71-9548RA

10
4
) 5 . .ok
et o_| o °® Li IN Li-Bi

— ¥ 1 B — Fy
P0 b n 4
[
©
E
~ 2
.
-
o
=
2 1
1)
2
s
= 0.5
o
«c
-
= —a0-
w
z

Q.2 ~ i i-
8. — : Li IN Bi-Th FROM emf
3 .

0.1 ° o}

Li IN Bi-Th —
BY ANALYSIS —
|
0.05 l 1
0 100 200 300 400 500 600

VOLUME OF LiCl CIRCULATED

Fig. 17.3. Variation of lithium concentration in Li-Bi and
Th-Bi solutions during experiment MTE-2B.

show lower equilibrium lithium concentrations in the
LiCl in contact with the 5 at. % Li-Bi solution than the
present limit of about 0.3 wt ppm of lithium.

17.3 DEVELOPMENT OF MECHANICALLY
AGITATED SALT-METAL CONTACTORS

H.O.Weeren L. E. McNeese

. Mechanically agitated salt-metal contactors are being
developed as an alternative to packed columns for
MSBR processing systems. This type of contactor is of
particular interest for use in the metal transfer process
since designs can be envisioned in which the bismuth in
the contactor would be a near‘isothermal, internally
circulated captive phase. It is believed that contactors
of this design may be easier to fabricate than packed
column contactors. The proposed contactor, shown
schematically in Fig. 17.5, consists of a vessel that is
divided into two compartments by a central partition
that does not extend to the bottom of the vessel. A seal
between the two compartments is provided by a pool of
bismuth that.will contain reductant. In-operation, fuel
carrier salt containing rare earths will flow. through one.

207

ORNL-DWG 71-9547RA

5
L
[ ]
2 g Nd BETWEEN
~_e LiCl AND Bi
o \'\ FROM emf
.\ ®
. B
] ¢ \\
2 0.5 Nd BETWEEN
< LiCl AND Bi FROM —®
P Li ANALYSIS
=
o
5 0.2
o
[hed
s
L o Th BETWEEN
(] -
0.1 oo “— FLUORIDE AND
—O0~~4c——/—Bi FROM emf
0
S~
©-0% N i
\
\\\ \_\
_ Th BETWEEN ] T~
0.02 FLUORIDE AND Bi ]
FROM Li ANALYSIS ~—~——
0.01

0 100 200 300 400 500 600
: VOLUME OF LiCl CIRCULATED

Fig. 17.4. Variation of Nd and Th distribution ratios during
experiment MTE-2B.

compartment, and lithium chloride will be circulated
through the other compartment. An agitator having a.
paddle in each phase will be .placed in each  of the
compartments. The paddles will be located at a consid-
erable distance away from the-interface, and the phases
will be agitated as vigorously as possible without
actually causing dispersion of one phase in the other.
The hydrodynamic performance of this type of
contactor has been investigated using water and mer-
cury to simulate molten salt and bismuth in order to
determine favorable operating conditions. Test of sev-
eral sizes of contactors with different agitator arrange-
ments established that the common factor that Jlimits
the agitator speed is entrainment of water in the
mercury circulating between the two halves of the
contactor. Entrainment of water was found to begin at
a definite agitator speed; below the limiting speed, no
entrainment was observed. The limiting agitator speed
was found to be relatively independent of the size and
ORNL-DWG 74-9544A

fl o

FLUORIDE SALT ———{

= LiCl

+— PARTITION BETWEEN
CELL HALVES

| - BISMUTH CONTAINING
REDUCTANT

— 4-BLADED PADDLE

GAP FOR
BISMUTH FLOW

\IIL;lI

Fig. 17.5. Mechanically agitated salt-metal contactor for metal transfer process experiments,

shape of the contactor vessel but was strongly depend-
ent on agitator diameter, as shown in Fig. 17.6.
Consideration of these-data, along with data from the
literature,2 on mass transfer coefficients in this type of
contactor suggest that optimum mass transfer perform-
ance will be obtained by use of the largest possible
agitator diameter. This will also result in the lowest
possible agitator speed, which should facilitate main-
taining a gas-tight seal on the agitator shaft.

It was found that the distance between the lower edge
of the partition and the bottom of the contactor vessel
could be increased from Y, to ¥ in. without appreci-
ably affecting the limiting agitator speed. However, the
limiting agitator speed was reduced as the separation
distance was increased above % in. Considerably higher
limiting agitator speeds could be obtained when the
agitator blades were canted at a 45° angle instead of
being vertical and the direction of rotation was such

that the agitator lifted the mercury phase toward the .

2. 1. B. Lewis, Chem. Eng. Sci. 3,248-59 (1954).

mercury-water interface. It is believed that the entrain-
ment of water in the mercury phase results from flow

. of water down the agitator shaft to a point below the

water-mercury interface, where it is then dispersed into
the mercury phase. The limiting agitator speed with
canted blades was decreased significantly when the
contactor contained baffles or other items such as
thermowells or sampling tubes. In all cases, however, a
higher limiting agitator speed was obtained with canted
blades than with vertical ones.

A test was also carried out to observe the effect on
contactor performance of the difference in densities of
the two phases. The test was made in a heated
contactor with Cerrolow-105 (42.9-21.7-18.3-8.0-5.0-
4.0 wt % Bi-Pb-In-Sn-Cd-Hg) and water. The metal
phase has a density of 8.1 g/cm® and a liquidus
temperature of about 38°C. At the operating tempera-
ture of 60°C, the limiting agitator speed was nearly
identical to that obtained with mercury and water in
the same contactor. Thus, it appears that the perform-
ance of the contactor will not depend significantly
upon the difference in the densities of molten salt and
bismuth. )

~
-

<
209

ORNL-DWG 71-9541A

o CIRCULAR CONTACTOR,
5V in. DIAM
4 CIRCULAR CONTACTOR, 10 in. DIAM

@ RECTANGULAR CONTACTOR,
12%24 in.

BLADE DIAMETER {in.]

20 50 100 200

LIMITING AGITATOR SPEED (rom)

Fig. 17.6. Correlation of limiting agitator speed with agitator
diameter in several contactors.

The rate at which mercury circulated between the
two halves of a 10-in.-diam contactor having a 3-in.-
diam agitator was determined by measuring the rate of
return of the liquids to a common temperature on
resumption of mercury flow after one side of the cell
had been chilled and the other side of the cell had been
heated. The mercury flow rate was found to be about
15 liters/min and depended, to some extent, on
propeller configuration. A bismuth circulation rate of
0.5 liter/min or greater is desired in the contactor for
metal transfer experiment MTE-3, which uses a 10-in.-
diam contactor. Thus, it appears that circulation of the
bismuth phase will be more than adequate.

17.4 DESIGN OF THE THIRD METAL
TRANSFER EXPERIMENT

E. L. Nicholson ~ W. F. Schaffer, Jr.
L. E. McNeese E. L. Youngblood
H. 0. Weeren

The design of the third engineering experiment
MTE-3 for development of the metal transfer process
for removing rare earths from MSBR fuel carrier salt has
been completed. Most of the equipment has been
fabricated, and the main process vessels are now being
installed. This experiment will use salt flow rates that
are 1% of the estimated flow rates required for
processing a 1000-MW(e) reactor. The planned experi-
ment and equipment have been described previously.

3. E. L. Nicholson et al., MSR Program Semiannu. Progr
Rep. Feb. 28, 1971, ORNL-4676, pp. 245-55.

The three carbon steel vessels required for the
experiment are shown in Fig. 17.7. The fluoride salt
reservoir is on the left, the compartmented salt-metal
contactor is in the center, and the rare-earth stripper
vessel is on the right. The fluoride salt pump, which has
been fabricated, will be inserted in the large flange on
the fluoride salt reservoir. Both the pump discharge line
to the salt-metal contactor and the return stream from
the contactor pass through the upper horizontal
diam pipe between the two vessels. The lower pipe is a
structural member. The agitators for promoting contact
of the salt and bismuth will be mounted on the two
flanged nozzles on the salt-metal contactor vessel (only
one nozzle can be seen in this view) and on the vertical
flanged nozzle on the rare-earth stripper. The lithium
chloride will be pumped back and forth between the
salt-metal contactor and the rare-earth stripper vessel by
varying the gas pressure in the stripper vessel. Figure
17.8 shows a detailed view of the top of the rare-earth
stripper vessel with the agitator shaft seal housing and
shaft cooler installed in the agitator nozzle. The vessel is

PHOTO 182874

Fig. 17.7. Equipment for metal transfer experiment MTE-3,

210

PHOTO 183074

Fig. 17.8. Top of rare-earth stripper vessel showing the nozzles.

electrically insulated from the other equipment to
minimize the rate of transfer of lithium between the
two bismuth pools (which have different lithium
concentrations) in the experiment. The LiCl transfer
line is constructed of %-in. sched 80 pipe and is
insulated from the rare-earth stripper vessel by the
water-cooled, Teflon-gasketed flange assembly shown
on the right side of the figure. The portion of the
carbon-steel pipe located adjacent to the cooled flanges
is heated by a nickel-jacketed copper sleeve on which a
Calrod heater will be mounted. The method used for
mounting the heater will facilitate replacement of the
heater in the event that this becomes necessary. The
heater will be wound in the grooves in the enlarged
threaded part of the nickel-clad copper sleeve. This
figure also shows the flame-sprayed, oxidation-resistant

nickel aluminide coating that was applied to the
carbon-steel parts of the vessel.

We are still testing the agitator drive unit and the
nonlubricated shaft seal assembly, which is water-
cooled and buffered with inert gas. The assembly
contains two standard graphite-impregnated Teflon
Bal-Seals (product of Bal-Seal Engineering Co.). Figure
17.9 shows the test equipment prior to completion of
the piping and installation of the electrical heaters and
thermal insulation. Two seals were tested initially for
820 hr at shaft speeds of 150 to 750 rpm while salt and
bismuth were agitated at 650°C in the 3-in.-diam test
vessel. The seals were buffered with argon at 16 psig.
After an initial run-in period, the seal leak rate was
about 8.6 cm® of gas per hour. After 700 hr of

operation, the leak rate increased rapidly up to the time

<
a m PHOTO 0455-71A
e B

VARIABLE
SPEED MOTO

AGITATOR BEARING
AND SHAFT SEAL
ASSEMBLY

3

Fig. 17.9. Equipment for testing the agitator drive and the shaft seal assembly.
212

the test was terminated. Gas leakage was principally to
the outside rather than into the vessel. On completion
of the test, examination of the upper seal ring showed
signs of excessive wear. Therefore, both seals were
replaced- with a new flange-mounting type of seal that
had light-duty seal expander springs and a thicker cross
section. The seal holder was also modified to facilitate
self-centering of the seals on the shaft. Subsequent tests
were made at room temperature, since the previous
tests showed that the shaft and seal cooling system
maintained the critical parts at room temperature with
salt and bismuth in the vessel at 650°C. After prelimi-
nary testing, the seals were replaced with ones having
medium-duty expander springs. These seals have been
operated for 200 hr at 125 rpm to date; the seal leak
rate is only 0.13 cm?/hr, which is quite satisfactory.
The vapor-deposited tungsten coating on the interior of
the test vessel and the sprayed stainless steel undercoat
beneath the nickel aluminide protective coat on the
outside will be evaluated after tests of the seal are
completed. No evidence of stable emulsions of bismuth
in salt was found in the first test.

17.5 REDUCTIVE EXTRACTION

ENGINEERING STUDIES
B. A. Hannaford W.M. Woods
D. D. Sood L. E. McNeese

‘We have begun mass transfer experiments in which
the rate of transfer of zirconium from molten salt to
bismuth is measured by adding ®?ZrQ, tracer to the
salt phase prior to contacting the salt with bismuth
containing reductant in a packed column. Three of the
experiments (ZTR-2, -3, and -5) resulted in the transfer
of 15 to 30% of the °7Zr present in the salt and gave
measured HTU values that ranged from 4 to 1 ft. Other
experiments (ZTR-1, 4, and -6) were carried out under
conditions where no transfer of tracer between phases
was expected to occur. '

Preparation for mass transfer experiments using * ’Zr
tracer. The new salt feed tank and newly installed lines
were treated with hydrogen for 13 hr at 600°C to
reduce surface oxides in preparation for the ° 7 Zr tracer
experiments. Thorium reductant was added to the
bismuth in the treatment vessel, and about 18 liters of
salt (72-16-12 mole % LiF-BeF,-ThF,) was. charged to

the treatment vessel to replace salt that was removed.

when the original salt tank was replaced. The salt and
bismuth were equilibrated for 20 hr and were then fed
through the system to complete the removal of surface
oxides not reduced by the earlier hydrogen treatment.

The measured pressure drop across the column when
only salt was flowing (~70 ml/min) was about 2 in.
H, O, which is the same value that was measured soon
after the column was installed. Thus the column
characteristics apparently have not changed as 4 result
of runs made to date.

A small amount of Zircaloy-2 (5.3 g) was dissolved in
the bismuth to bring the zirconium inventory in the

‘system to about 10 g. The steel draft tube in the

treatment vessel was replaced with a molybdenum tube
of similar dimensions in preparation for H,-HF treat-
ment of the bismuth and salt. Following a 20-hr
treatment with 30% HF in H,, the bismuth and salt
were sparged with argon for 20 hr. A 48-g charge of
thorium metal was suspended in a perforated steel
basket in the bismuth phase in the treatment vessel in
order to add reductant to the bismuth. After 25 g of
thorium had dissolved, a 240-g batch of 1.75 wt %
lithium-bismuth was added to the vessel in order to
make a rapid, final adjustment of the reductant
inventory to an expected value of about 1 g-equiv.
Following a 22-hr equilibration period, the salt and
about one-half of the bismuth were transferred to the
feed tanks. Transfer of all of the bismuth was prevented
by a leak in the transfer line at the weld joining the
molybdenum tube to the mild-steel tube inside the
treatment vessel.

A 6.7-mg quantity of °®ZrO, that had been irradi-
ated for 12 hr (to produce about 8 mCi of ®7Zr) was
transferred to a steel capsule to facilitate its addition to
the salt phase. Analyses of samples taken periodically
from the salt feed tank after the capsule had been
immersed in the salt indicated that the tracer was
completely distributed throughout the salt after 2 hr.

Mass transfer runs using ® 7 Zr tracer. When the ° "Zr

activity in the seven sets of samples taken during

experiment ZTR-1 was counted, it became obvious that
no measurable transfer of ®7Zr tracer had occurred.
The lack of transfer was due to an unexpectedly low
distribution coefficient for zirconium; postrun samples
from the treatment vessel showed that D, was only
0.023 instead of the expected value of 1.0.

An additional 1 g-equiv of lithium, in the form of
lithium-bismuth alloy, was added 'to the treatment
vessel in preparation for experiment ZTR-2. Following
a 105-hr equilibration period, a new charge of ®7Zr
tracer (about 10 mCi) was added to the treatment
vessel. Bismuth and salt samples taken at various times
showed that, when the tracer addition capsule was
removed (after being immersed for 80 min in the salt),
the concentration of tracer in each phase was within
about 90% of the equilibrium tracer concentration that
o

~

<

was observed after -a 32-hr period. The equilibrium
distribution- coefficient for zirconium, determined by
®7Zr counting, was 0.32. Since the addition of reduc-
tant should have produced a D,  of about 5, it is
believed that reductant was consumed by side reactions
with materials such as HF, FeF,, or air (a contaminant
in the argon cover gas). The same explanation would
account for very low D, observed in ZTR-1. Subse-
quent samples demonstrated that some loss of reduc-
tant continued to occur as the mass transfer experi-

‘ments were performed.

The results of the three experiments that yielded mass
transfer data are summarized in Table ~17.3. The
calculated HTU value is quite dependent on the
zirconium distribution ratio (D,,) and the *7Zr con-
centration in the bismuth leaving the column. Since the
quality of the data improved from run to run, the
calculated HTU values are not directly comparable.

Samples of bismuth exiting from the column during
the second °7Zr tracer experiment, ZTR-2, indicated
that about 30% of the tracer had transferred to the salt;
however, analyses of salt samples indicated that only
15% of the tracer had transferred. The postrun zirco-
nium distribution ratio (determined from: the °7Zr
activity) was found to be 0.034. This is in excellent
agreement with the value of 0.038 * 0.020, which was

- calculated from postrun uranium . distribution data.

Since the observed -°7Zr transfer could not have
occurred with such a low zirconium distribution coef-
ficient, it is apparent that some loss of reductant must
have occurred between the packed column and the
bismuth and salt receivers.

In order to increase.D; ' before the next experiment,
Li-Bi alloy containing 0.38 g-equiv of lithium was added
to the treatment vessel; the addition of ®7Zr tracer
indicated a D, of 0.25. An additional 0.75 g-equiv of

213

lithium was then added to the treatment vessel, thereby
increasing the D,_ to 1.90. Since each of the additions
of reductant produced an effect corresponding to only
about two-thirds the amount of Li-Bi alloy added, it is
possible that the lithium in the alloy was partially
oxidized before it was added to the treatment vessel.

~ After the °7Zr tracer that was used to monitor the
adjustment of D, had decayed, additional ® 7 Zr tracer
was added to the salt feed tank in preparation for
experiment ZTR-3. About midway of the experiment,
the presence of a frozen-salt plug in the flowing salt
sampler caused a brief interruption since it halted the
sampling operation (Fig. 17.10). With the exception of
samples that reflected the effect of this interruption,
the scatter in ®7Zr counting results was satisfactorily
low. The fraction of tracer transferred from the salt was
0.30 and the calculated HTU was 2 ft, based on a D,
of 1.9. The postrun zirconium distribution coefficient,
measured after a 2-hr equilibration of the salt and
bismuth in the treatment vessel, was 0.77.

In order to remove suspected oxidants from the
system, the bismuth and the salt were countercurrently
contacted in the column about 20 hr prior to experi-
ment ZTR-5. Tracer was then added to the salt feed
tank, and run ZTR-5 was carried out with each phase
having a nominal flow rate of 150 ml/min. Counting
results for five of seven bismuth-samples (Fig. 17.11)
were within 7% of their average. However, the
counting rates for the two remaining samples were 2
and 170 times the average of the other samples.
Following a 3-hr equilibration period in the treatment
vessel, both phases were returned to the feed tanks, and
run ZTR-6 was carried out. Counting results for salt
samples were comparable with results obtained during
run ZTR-5. However, the tracer concentrations in
bismuth samples were generally a factor of 2 to 3 times

Table 17.3. Summary of mass transfer results from expériments with 97 Zr tracer

" Salt (72-16-12 mole % LiF-BeF,-ThF4) and bismuth contacted
at 600°C in an 0.82-in.-ID by 24-in.-long packed column

Bismuth Salt Flow Zirconium Fraction
flow rate, flow rate, rate distribution § 7 HTU
Run . .. of 721
) V Vg ratio, coefficient, (ft)
B L transferred
(ml/min)  (ml/min) ~ Vg/Vg Dq.
ZTR-2 232 70 3.3 0.32 0.154 - ~4
ZTR-3 93 168 0.55 1.90 0.30 , 2
ZTR-5 . 147 158 . 093 0.24 0.14 1

%y alue based on material balance of the salt.
214

ORNL-DWG 71-8563RA

SALT FEED CONC. ' h

___________ gl \
N AP
/'\\\\V//_-_\

/ B!/SMUTH ./

0 10 20 30 40 50 60 70
RUN TIME (min)

972¢ ACTIVITY (dis min~' m)™!
n
N
/
|
1

S
(S
[ ]

Fig. 17.10. Zirconium-97 activity in flowing salt and flowing
bismuth samples obtained in run ZTR-3. Salt flow rate, 168
ml/min; bismuth flow rate, 93 ml/min. Dg = 1.9.

o5 ORNL-DWG 7!-8567RA
] I —
- ZTR-5 : 1 ZTR-6 } {
= T3 . SALT — .
';‘5 51 ’G;r'm /
T: - > \/ TS5m i 1/} \ i T5
E I N 14"
" /‘\
=)
: [N
.
=
s i / "\
[ BISMUTH
(8]
4 40° Al . J b
! 0-..5’ TSe
T5
o) I ?TZ m . TE: .}‘
o 40 80 120 © 40 80 120

TIME (min} TIME (min}

Fig. 17.11. Results from consecutive ° ’ Zr tracer experiments
ZTR-5 and ZTR-6. °7ZrQ, tracer was added to the salt feed
tank preceding ZTR-5; tracer was equilibrated in T5 preceding

ZTR-6. Feed tank designations:  T1 (bismuth), T3 (salt). -

Receiver tank designations: T2 (bismuth), T4 (salt), Treatment
vessel, TS. . .

higher than in the previous experiment. Suspected
contamination of the bismuth samples with salt was
ruled out by visual inspection of several of the samples.
A review of the sampling procedure indicated that small
amounts of air might possibly be entering the flowing

stream samplers. Such a contaminant could have re-
sulted in the accumulation of °7Zr, as ® 7Zr0,, at the
bismuth surface in the sampler; subsequently, some of
this material might have been drawn into the sample
capsule since the samples were taken from near the
bismuth surface.

We are presently replacing several bismuth and salt
transfer lines in which leaks have occurred. We are
preparing for additional ®7Zr tracer mass transfer
experiments that will use an improved technique for
obtaining bismuth and salt samples.

17.6 DEVELOPMENT OF A FROZEN-WALL
FLUORINATOR: INDUCTION HEATING
EXPERIMENTS

J. R. Hightower, Jr.

An experiment to demonstrate protection against
corrosion in a continuous fluorinator by use of a frozen
wall will use high-frequency induction heating to serve
as a heat source in the molten salt. Heat generation
rates, were measured with four induction coil designs in
a system that used 31 wt % HNO; as a substitute for
molten salt. With these measured heat generation rates,
correction factors could be computed -for use in
equations for calculating the amount of heat that is
induced in similar, but idealized, geometries. The
equations were then used to design equipment for a
molten-salt induction heating experiment for testing (1)
the operation of the proposed heating system and (2)
the means by which power leads are introduced into the
fluorinator. The results of measurements with the nitric
acid system and the design of equipment for the
molten-salt induction heating test are summarized in
the remainder of this section. _

Results of experiments with a simulated fluorinator.
The equipment for measuring heat generation rates in a

system that uses 31 wt % HNO; as a substitute for -

molten salt has been described previously.* Twenty-
nine runs were made to determine heat generation rates
in the acid, in the pipe surrounding the acid, and in
each of four induction coils. The length of each
induction coil assembly was 5 ft, and each coil assembly
was made of a number of smaller coil sections con-
nected electrically in parallel. Each small section had an
inside diameter of 5.6 in. The characteristics for each of
the coil assemblies are given in Table 17.4.

4. J. R. Hightower, Jr., MSRP Semiannu. Progr. Rep. Feb,’

28, 1971, ORNL-4676, pp. 262—64.

-~
215

Table 17.4. Characteristics of coils tested in continuous fluorinator simulation

) No.  Diameter ™M No.  Winding
% i of of of of of
Coil Material small .
small conductor . small adjacent
. . section . T
sections (in.) . sections sections
. - (in.)
I Monel 17 Y 3 6%  Opposing
I Stainless steel: 18 % 3 6 Assisting
mn Stainless steel 18 3/8 3 6 Opposing
v Copper 10 1/4 4 11 3/4 Opposing

The measured heat™ generation rates were used to where
calculate correction factors for idealized design equa- _
tions that had been derived previously. The design P, = heat generated in pipe, W,
equations (which define the correction factors) are liot>1, N, and L are as defined for Eq. (1),
given below. The equation for heat generation within '

- i - g, = specific conductivity of pipe material, Q 'm!,
the fluid zone inside the coil is: p =P y PIP

ap, = inside radius of pipe, m, -
| lio)\2 [ o pp = (2nfgyu,)~ 112,
P=k |0.3818 <fi> (i) LJ LD P i —
- o N py/ & - kp = correction factor, dimensionless,
‘/ where 1, = magnetic permeability of pipe material, N A™2.
e :
7 _ This equation is valid for (ap/pp) > 10. The equation
14 Py = heat generation rate, W, for the heat generated in the induction coil is:
n = average turns per meter over length of coil ‘
assembly, L BN - fro.\ 1/?
- Po=Ki—+ \Z7) Lo L, (3)
N, = number of small coil sections, N, d.l, o
a =radius of fluid zone, m, o
where

L = length of coil assembly, m,

pr = (2nfg) =112, |
g; = specific conductivity of fluid, @ 'ni’!,

I, L, and N, are as defined in Eqs. (1) and (2),

P = heat generation rate in coil, W,

u; = magnetic permeability of fluid, N A2, b = inside diameter of coil, in.,

f = frequency, Hz, d, = conductor diameter, in.,

[. . = total rms coil current. A I, = length of a small section of coil, in.,
ftot » £

k = correction factor, dimensionless. Ny - m_lmber of turns in a small coil section,

. . : . . ) o, = specific resistivity of coil material, £ 2+ cm,
Equation (1) is an approximate equation and is valid for c %P y : H3

(a/p)) < 1.4. The equation for the heat generation in - - f1 = frequency, kHz,
the pipe surrounding the coil is: ' ' ‘ K, = proportionality constant having dimensions de-
) ~ fined by Eq. (3). '

: 2 N - The effect of bubbles in the nitric acid on heat
nliot ap\ L . ) . ) .

* P, =k, | T ~ ) = (2) generation rate was investigated with coils III and IV.

: s Pp/ &p Seven runs with coil III and two runs with coil IV were
made with air flow rates up to 2.16 scfm, which
produced bubble volume fractions in the acid as high as
18%. In the range of bubble volume fractions examined,
the value of k, defined by Eq. (1), varied approximately
linearly with bubble volume fraction. The effect can be
expressed by the fol]owmg relation:

k=k, (1 —1.079), . (4)

where
k, = the value of k¥ with no bubbles present,
€ = bubble volume fraction.

Values for k,, k,, [see Eq. (2)] L and K, [see Eq. (3)]
for the four work coils tested are given in Table 17.5.

Table 17.5. Comection factors for heat
generation rate equations

216

decrease in the axial component of the field. It is the
axial component that provides the proper eddy currents
for heat generation.

Although any one of the four coils could be used to
generate heat in the proposed fluorinator, coil IV would
require the lowest. coil current to perform the required
heating and, for this reason, would be the most
desirable. Calculations have shown that, for a 5-in.-diam
molten-salt zone, a 5.56-in.-ID coil made from %-in.
nickel tubing (using a coil IV design in which each small
section” has 9% turns over a 3¥%-in. length, with a

2Y,-in. space between small sections), and a 6% ¢-in.-ID

Coil k,, (no bubbles) ‘ kp K,?

X 1073

1 0.130 0.624 1.915
1l 0.089 " 0.447 1.755

11 0.178 0.623 1.885

v 0.216 0.586 2.15

%The dimensions of K, are defined by Eq. (3).

The values for all constants for coil 1l are smaller than
the corresponding constants for the other coils. This is
apparently the result of winding all of the small coil
" sections in the same direction, since other character-
istics are similar to those of coils I, III, and IV. Coil II
requires the largest current to produce the required heat
generation rate in the molten salt but might have a high
efficiency for heating the salt if the molten diameter is
sufficiently large.

The value for k, was largest for coil IV. This is
probably due to the difference in spacing between the
small coil sections. Coil IV had 4-in.-long small coil
sections with ‘11%, turns per section, and the coil
sections were separated by a space of 2 in. The smaller
sections in the other coils were placed close together so
that the coil turns were spaced evenly over the length of
the acid column; the total number of turns was about
the same for each coil. In the case of coils I and II, this
closer spacing compressed the magnetic field between
each of the small coil sections, thereby effecting a

nickel fluorinator vessel, an efficiency of heating the
salt (with no bubbles) of about 49% would be achieved
with a total current of less than 140 A. With coil III,
the efficiency of heating the salt would be about 58%,
but a coil current of 267 A would be required.
Molten-salt induction heating test. The expenmental
results obtained with the simulated fluorinator indicate
that sufficient heat can be produced in the molten salt
by induction heating. However, before a fluorinator
experiment is designed, a number of factors require

further study. These include: (1) verifying the predicted-

coil performance, (2) demonstrating a means for intro-
ducing the rf power leads into the fluorinator vessel,
and (3) checking the general operability of a system
that contains molten salt. Equipment has been designed
and installed for inductively heating molten salt in a
vessel similar in design to that expected for the
fluorinator experiment.

The test vessel, essentially a short version of the
proposed frozen-wall fluorinator, is a 5-ft-long by
6%, ¢-in.-ID nickel column having two conduits for the
induction coil electrical leads that extend along the

vessel wall from the top to within 1 ft of the bottom of

the vessel. The conduits are made from 1-in. sched 40
nickel pipe. The gas inlet nozzle near the bottom of the
column is made from 21/2 -in. sched 40 nickel pipe and
enters at a 45° angle to the axis of the test vessel. The
test section in which the frozen-salt layer will be
formed is 3 ft long and begins 1 ft from the bottom of
the vessel. Figure 17.12 shows the installed vessel with
Calrod heaters and thermocouples in place. The induc-
tion coil to be used for this test is of a design similar to
coil IV. The coil, shown in Fig. 17.13, consists of six
small coil sections connected electrically in parallel to
leads that are made of %-in. nickel tubing. Each small

section is 3%, in. long and 5% in. ID and consists of

9', turns of %-in. nickel tubing. Adjacent coil sections
are wound in opposite directions and are separated by a
distance of 2, in. The leads to the small coil sections
fit into the conduits on the vessel wall and allow the

R
217

coil to be located very close to the wall. Quartz
insulators on the %-in. leads and on the inside walls of
the vessel prevent the coil from making electrical
contact with the vessel when frozen salt is not present.

17.7 PREDICTED PERFORMANCE
OF CONTINUOUS FLUORINATORS

J.S.Watson L. E. McNeese

Thus far, most of the flowsheets considered for use in
processing molten-salt reactor fuel salt require fluorina-
tion of molten salt for removal of uranium at one or
more points in the flowsheet. These include: (1) re-
moval of trace quantities of uranium from relatively
small salt streams prior to discard, (2)removal of
uranium from a captive salt volume in which 233Pa is
accumulated and held for decay to 233U, (3) removal
of most of the uranium from relatively large fuel salt
streams prior to isolation of protactinium and removal
of rare earths, and (4) nearly quantitative removal of
uranium from a salt stream containing 2**Pa in order
to produce isotopically pure 22?U. Not all of these
operations would require continuous fluorinators. In
fact, the use of batch fluorinators has definite advan-
tages in certain cases. However, as the quantities of salt
and uranium to be handled increase, the use of
continuous fluorinators becomes mandatory in order to
avoid undesirably large inventory charges on uranium
and molten salt, as well as the detrimental increase in
reactor doubling time that is associated with an
increased fissile inventory.

Although the literature contains many references to
the removal of uranium from molten salt by batch
fluorination, information on continuous fluorinators,
particularly on fluorinators capable of handling salt
flow rates on the order of 100 ft*/day, is rather meager.
A mathematical analysis for predicting continuous
fluorinator performance was completed, and calcula-
tions were made for fluorinator operating conditions of
interest for MSBR processing.

Mathematical analysis of open-column continuous
fluorinators. Consider a differential height of a con-
tinuous fluorinator in which fluorine and molten salt
containing uranium are in countercurrent flow. If the
rate of removal of uranium from the salt is assumed to
be first order with respect to the uranium concentration
in the salt, a material balance on the differential salt
volume yields the relation:

DI- v _ke=o, )

PHOTO 116971

Fig. 17.13. Induction coil for molten-salt induction heating test.

where
D = axial dispersion coefficient, cm?/sec,
C = concentration of uranium in salt, moles/cm®,

X = position in column measured from top of col-
umn, cm,

V = superficial salt velocity, cm/sec,

k = reaction rate constant, sec ™ .

The terms in Eq. (1) represent the transfer of uranium
in the salt by axial diffusion, the transfer of uranium in
the salt by convection, and the removal of uranium
from the salt by reaction with fluorine respectively. The
assumption of a first-order reaction does not imply a
particular rate-limiting reaction mechanism; however, it
is consistent with the assumption that the rate-limiting
step is diffusion of uranium in the salt to the gas-liquid
interface. In this case, the first-order expression would
imply that the concentration in the salt at the interface
is negligible in comparison with the uranium concen-
tration in the salt at points a short distance from the
interface.

The boundary conditions chosen for use with Eq. (1)
assume that the diffusive flux across the fluorinator
boundaries is negligible: at X = 0 (top of fluorinator),

dac

Vv
g =B Crwa =) @

and at X =/ (bottom of fluorinator),

dc

=l .59 3
o). 2P ®)

where
Creeq = concentration of uranium in salt fed to the
fluorinator,
Cy+ = concentration of uranium in salt at the top of
the fluorinator.

Note that Cy, is not equal to Cg,,4 since there is a
discontinuity in uranium concentration in the salt at
the top of the column where the salt enters.

The solution of Eq. (1) with the. stated boundary
conditions yields the following expression for..the ratio
of the uranium concentration in salt leaving the column
(at X =) to the concentration in the feed salt:

a (1/2“7 +l> £V TTE7-1/2)

Creed '.\/1 +4n 2
-1

(A2 D) anevTE | g
W1+dn 2/ ’

where
kD
n=——_,
;- VL
D

" Application of Eq. (4) to the design and evaluation of
continuous fluorinators requires information on the
rate constant k and the axial dispersion coefficient D.
Values of the dispersion coefficient were obtained from
correlations resulting from studies in which air and
aqueous solutions were contacted countercurrently in
open bubble columns, as indicated below. A value. for
the rate constant was calculated using data obtained
previously during a study of the performance of a
1-in.-diam continuous fluorinator.

Dispersion coefficient values. Studies in which air and.
aqueous solutions were contacted countercurrently in
open bubble columns having diameters of 1, 1.5, 2, 3,
and 6 in. were carried out previously. The viscosity of
the aqueous solutions was varied between 1 and 15 cP
by the use of water-glycerol mixtures. The viscosity of
molten salt under conditions of interest is about 15 cP.
The effect of the surface tension of the aqueous phase
was determined by varying the surface tension from 68
to 27 dynes/cm, using water-isopropanol mixtures. The
surface tension of molten salt is much higher than these
values; however, the data allow estimates of the effect
of surface tension on the dispersion coefficient. Data
from these studies were correlated® by use of dimen-
sionless groups in the following manner: for low gas
flow rates (bubbly flow), .

5. A. K. Padia, G. E. Marion, and R. H. McCue, Axial Mixing
in Open Bubble Columns, MIT Report CEPS-X-121 (in prepara-
tion),

219

Np'e =-18'0NR80.33 N, 0435 NSu-o.ovs p00475 ,
and at high gas flow rates (slug flow),
- 0. . -0.
NPe - 0'46NRe ! NArO ! NSu 0-38

>

where -

V
Np, = —~£_& = Peclet number,
' D

pd. V .
Npe = I7e £ = Reynolds number,
My :
dc IU .
Ng, = —= Suratman number,
Ky
d.>p’g
Ny, = % = Archimedes’ number,
My ‘

d,. = column diameter,

V, = superficial gas velocity,

D = dispersion coefficient,
p; = density of the liquid,
uy = viscosity of the liquid,
o = surface tension of the liquid,
£ = acceleration of gravity,

n = number of gas inlets in disperser.

These relations are believed to allow accurate and
reliable - estimates of axial dispersion in continuous
fluorinators for the range of diameters and physical
properties of interest.

. Evaluation of fluorination reaction rate constant. The
fluorination reaction rate constant was evaluated from
Eq. (4) using previously obtained data on the perform-

ance of a l-in.-diam open-column continuous fluori-

nator® and the axial dispersion coefficient data dis-
cussed above. The fluorinator performance data were
obtained in studies with two different uranium concen-
trations in the salt feed (0.35 and 0.13 mole %) at
600°C and one uranium concentration (0.35 mole %) at
525°C. Three data points corresponding to different
salt flow rates were obtained for each set of tempera-
tures and inlet compositions. The fluorine flow rate was
different for each data point; however, according to Eq.

6. L. E. M-cNeese, Engineering Development Studies for
Molten-Salt Reactor Processing No. 6, ORNL-TM-3141 (in
preparation). . :
220

(1), which defines the present model, the fluorine flow
rate only affects the results by changing the axial
dispersion coefficient. The data obtained at 525°C were
used to evaluate the rate constant &, since the operating
temperature of the proposed frozen-wall fluorinators
will be 10 to 15°C above the salt liquidus temperature
of 505°C. Application of Eq. (4) to the three data
points obtained at 525°C produced the results shown in
Table 17.6. The scatter in the calculated values for & is
acceptably small, and no trend with salt or fluorine
flow rates is observed. Although these data do not
confirm the validity of the model, they do not
contradict it.

Predicted performance of open-column continuous
fluorinators. The performance of large, open-column
continuous fluorinators was estimated from Eq. (4)
using the previously discussed estimates of the reaction
rate constant ¥ and the axial dispersion coefficient D.
The required fluorinator height is shown in Figs. 17.14
and 17.15 for a range of salt flow rates for two
fractional uranium removal values (0.95 and 0.99). The
uranium concentration in the inlet salt was assumed to
be 0.0033 mole fraction in all cases, and the fluorine
flow rate was assumed to be equal to 1.5 times the
stoichiometric requirement. These results are encourag-
ing since they suggest that single fluorination vessels of
moderate size will suffice for removing uranium from
MSBR fuel salt prior to the isolation of protactinium
by reductive extraction. The reference flowsheet for
isolating protactinium by fluorination—reductive ex-
traction requires fluorination of fuel salt at the rate of
170 ft3/day, which is equivalent to a ten-day processing
cycle. A 6-in.-diam fluorinator having a height of 10.2
ft will be required for a uranium removal efficiency of
95%, and an 8-in.-diam fluorinator having a height of
17.8 ft will be required for a uranium removal
efficiency of 99%.

17.8 ENGINEERING STUDIES
OF URANIUM OXIDE PRECIPITATION

M.J.Bell D.D.Sood
L. E. McNeese

The fabrication and installation of equipment’ for a
single-stage experiment to study the precipitation of
UQ,-ThQ, solid solutions from molten fluoride salts
have been completed. In the two experiments carried
out thus far, the major equipment items operated
satisfactorily. Several minor difficulties that were en-
countered, including the freezing of salt and the
condensation of water in several lines in the system,
have been alleviated. A third experiment is in progress.

In the first experiment (OP-1), a gas stream con-
taining 15% water—85% argon was fed to the precipita-
tor vessel at the rate of 0.5 scfh for a period of 4 hr
with the salt at 600°C. The gas flow was interrupted
periodically during the run, and filtered salt samples
were obtained. The precipitate and the salt were
allowed to equilibrate for 64 hr, after which a sample of
the oxide phase was obtained for analysis by x-ray
diffraction. The salt was then slowly transferred from
the precipitator vessel to the treatment vessel in order
to leave most of the precipitate in the precipitator.
Subsequently, the salt was hydrofluorinated in the
treatment vessel and sampled again in order to deter-
mine the amount of precipitated uranium that had
transferred to the treatment vessel. Finally, the salt was
returned to the precipitator vessel, where it was
hydrofluorinated for 12 hr using a 20% HF—80% H,

gas mixture at the rate of 3 scfh in order to redissolve

7. M. ]. Bell and L. E. McNeese, MSR Program Semiannu.
Progr. Rep. Feb. 28, [971, ORNL-4676, p. 267.

Table 17.6. Evaluation of fluorination reaction rate constant

Salt supe.rfic1al - C) F, flow rate Dlsg_)e.rsnon Rate constant, k&
velocity z (cm3/sec) coefficient, D (sec™1)
(cm/sec) feed (cm?/sec) ,

0.0625 0:0257 6.8 17.6 0.00805
0.0445 0.0096 5.0 14.6 0.01033
. 10.6 0.00886

0.0225 0.00457 3.82

Average: 0.00908

oy
l‘ po

221

ORNL-DWG 71-10788A
C T T ] 1T T T7T1 T T T T 17 FTTT
I URANIUM REMOVAL EFFICIENCY: 95%
CINLET URANIUM CONCENTRATION: 0.0033 mole fraction

100

50 - FLUORINE FEED RATE:150% OF STOICHIOMETRIC 7
= /://
— g
- A //

T 20 7| AT .
g Pl
I FLUORINATOR DIAMETER | A 4]
% 0 L SRS
Q 1 -
= i g s
= A7 7
S s dnpayd
2 A |\AL
w
. / ALY
2 4 //'.Oin.
%
Bin /
1 /
10! 2 5 102 2 5 103 2 5

SALT FLOW RATE {ft3/day)

Fig. 17.14. Variation of calculated fluorinator height with
salt flow rate and fluorinator diameter for a uranium removal
efficiency of 95%.

ORNL—~DWG 74—10793A

500

200 (URANIUM REMOCVAL EFFICIENCY: 99%
INLET URANIUM CONCENTRATION; 0.0033 mole fraction
FLUORINE FEED RATE: 150 % OF STO!CHIOMETRIC

= 100 s
: .
A4
'—
e
w L —/ T
xI
. L
&S 20
e FLUORINATOR | || ] =
3 DIAMETER ,,;:;/
Z 10 l— 6 in_Jt2 T
S i &
™ L8 in. LA
o g [0 [/
10 in. 4
12 in,
2
M
1ot 2 5 10° 2 5 10° 2 5

SALT FLOW RATE (ft%/day)

Fig. 17.15. Variation of calculated fluorinator height with
salt flow rate and fluorinator diameter for a uranium removal
efficiency of 99%.

the uranium and thorium oxides. Salt samples were
withdrawn at intervals during this hydrofluorination
period. . | -

Analysis of samples taken during the precipitation
phase of the experiment indicated that 17% of the

uranium initjally present in the salt was precipitated
during the run. However, the analyses show consider-
able scatter, and the uranium concentration in the salt
does not decrease monotonically with time. The analy-
ses indicate a water utilization of about 25%, although
there is some uncertainty in this value because of
problems encountered with the condensation of water
in gas lines during the experiment. Analysis of the oxide
phase collected at the end of the precipitation period
revealed that a UQO,-ThO, solid solution having a UQO,
concentration of 0.94 mole fraction was formed, and
that 5 to 15% of the precipitate was ThO,. The
concentration of BeO in the precipitate was below the
limit of detection (< 1 wt %). The presence of ThO,
and the fact that the uranjum content of the solid
solution was slightly lower than the value expected at
equilibrium (0.96 mole fraction UO,) indicate that
equilibrium between the salt and the precipitate was
not obtained. The uranium concentration in the de-
canted salt after hydrofluorination was found to be the
same as that obtained for the filtered .alt sample taken
after the precipitate had been allowed to settle for 64
hr. This indicates that it may be possible to separate the
precipitate from the salt with relatively simple equip-
ment.

In the second experiment (OP-2), the salt temperature
was again held at 600°C, and the composition of the
inlet gas mixture was 15% water—85% argon, as before.
In this experiment, the salt wis first contacted with the
gas mixture at a rate of 0.5 scfh for 9 hr; then the gas
flow rate was increased to 1.5 scfh for an additional
5-hr period. The total precipitation time (14 hr)
extended over a period of about ten days, during which
the run was interrupted several times in order to modify
gas lines and add heaters and thermal insulation to a
number of lines where condensation of water was
suspected. The uranium concentration observed in
filtered salt samples obtained during the run and the
expected uranium concentration based on the amount
of HF evolved during the run are shown in Fig. 17.16.
The discontinuity in the curve based on HF evolution at
6.7 hr resulted from the fact that condensate was
observed in an off:gas line between periods of precipita-
tion. The condensate was collected and titrated; then
the resulting correction was made. Additional heaters
were placed on this line, as well as other lines, before
the precipitation phase of the experiment was con-

‘tinued. The greater rate of HF evolution observed after

these changes had been made suggests that the conden-
sation problem has been eliminated.
Salt samples taken early: in the experiment (shortly

. after the precipitation had been discontinued, as well as
222

ORNL-DWG 71-9549A

1.2
1.0 b0
'--.._______..-‘ o
;5 ."‘----.fi
3 08 L<< caLcuLATED FROM
— ~\_HF EVOLUTION
5 ~
~
by ~
73] N "\
2 0.6 = ~
= . . \.\
=
> S
Z 04 5 - ,
5 SETTLING TIME
o2 L o { hr
: ® 16 hr
a 168 hr
0 ' _
0 2 4 6 8 10 12 14

PRECIPITATION TIME (hr)

" Fig. 17.16. History of uranium ooncenttatiofl in salt samples
during second uranium oxide precipitation run (OP-2). Salt
temperature was 600°C. .

after an extended equilibration period) were found to
have essentially the same uranium concentration (i.e.,
within the range of analytical error). However, analyses

of samples taken later in the experiment showed large

variations between samples taken shortly after precipi-
tation had been discontinued and those taken after a
long equilibration-time. This behavior is not currently
understood. Several samples of the oxide phase ob-
tained during the run may be helpful in interpreting the
results. : |

17.9 DESIGN OF A PROCESSING MATERIALS
TEST STAND AND THE MOLYBDENUM
REDUCTIVE EXTRACTION EQUIPMENT

E. L. Nicholson ~ W. F. Schaffer, Jr.
J. Roth

A report describing the conceptual design and fabrica-
tion development program for the molybdenum reduc-
tive extraction equipment test stand was completed.
The conceptual design for the system has been de-
scribed previously;®™!® fabrication development details
are reported in Sects. 14.1-14.4.

The detailed designs of the extraction column and the
salt and.bismuth head pots have been completed, and

fabrication has begun. Conceptual design sketches for -

the equipment supports in the containment vessel,
machined tees, and the freeze valve have been com-

pleted. Preliminary drawings for the molybdenum
piping arrangement inside the containment vessel have
also been completed; however, further work on these
drawings will be deferred until the piping can be
analyzed for thermal expansion stresses and a full-size
mockup of the system can be completed to investigate
fabrication and field assembly problems. The overall
height of the molybdenum components of the loop
(from the underside of the containment vessel flange to
the lowest point) is about 17 ft. Both the salt and the
bismuth gas-lift pumps are designed to operate with a
minimum submergence of 50%, which should provide
the maximum flow rate desired for either phase (1.1
liters/min). '

Final tests were completed with the revised design of
the full-scale transparent plastic mockup of the bismuth
head pot and the top portion of the extraction column.
Mercury and water were used to simulate molten
bismuth and salt. Problems with internal flow restric-
tions and gas venting noted previously were solved by
the revised design. Entrainment of liquid in the exit gas
was reduced to very low levels by routing the gas
through a settling chamber and Raschig-ring packing in

the top of the head pot.

The design of the probe for determining the position
of the salt-bismuth interface in the disengagement
section below the extraction column was completed, -
and a prototype probe has been built for testing. Work
associated with the development of liquid-level probes
is discussed in the following section.

.17.10 DEVELOPMENT OF A BISMUTH-SALT
INTERFACE DETECTOR

H. O. Weeren  E. L. Nicholson
C.V.Dodd C.C.Lu
J. Roth

An eddy current type of detector!’ is being devel-
oped to allow detection and control of the bismuth-salt
interface in salt-metal extraction columns or mechani-
cally agitated salt-metal contactors. The probe consists
of a ceramic form on which bifilar primary and

8. M. W. Rosenthal et al., MSR Program Semiannu. Progr.
Rep. Feb. 28, 1970, ORNL-4548, pp. 289-300.

9. M. W. Rosenthal et al., MSR Program Semiannu. Progr.
Rep. Aug. 31, 1970, ORNL-4622, pp. 212—13.

10. M. W. Rosenthal et al., MSK Program Semiannu. Progr.

" Rep. Feb. 28, 1971, ORNL-4676, pp. 267—68.

11. M. W. Rosenthal et al., MSR Program Semiannu. Progr.
Rep. Feb. 28, 1970, ORNL-4548, pp. 300-301.

Bl

L
)

Pt
=

secondary coils are wound. Contact of the coils with
molten salt or bismuth is prevented by enclosing the
coils in a molybdenum tube. In operation, a high-
frequency alternating current is passed through the
primary coil and it induces a current in the secondary
coil. The induced current is dependent on the conduc-
tivities of the materials located adjacent to the primary
and secondary coils; since the conductivities of salt and
bismuth are quite different, the induced current reflects
the presence or absence of bismuth.. The principal
problem associated with this type of detector stems
from the low permeability of molybdenum, the fabrica-
tion material of the protective sheath. Two approachés
for obtaining an output from the detector are being
pursued. The first is based on measuring changes in the
magnitude of the induced current; the second is based
on measuring the phase shift that occurs between the
voltage imposed on the primary coil and that which is
induced in the secondary coil.

Initial tests in which measurements were made of
variations in the induced current caused by changes in
the bismuth interface position showed that the probe

223

these calculations revealed the manner in which the

was quite sensitive to drift in the electronic circuit, to

minor variations in line voltage, and to changes in the
temperature of the electronic components. To circum-
vent this problem, an improved electronic circuit, which
is relatively insensitive. to amplifier drift and to changes
in the temperature of the electronic components, was
devised. Difficulties caused by-line voltage variations
were avoided .by use of a constant-voltage power
supply. A full-scale response of about 1.5 mV was
obtained at the optimum frequency for the primary coil
(35 kHz) for a 9-in. change in the bismuth interface
position at 600°C.

During this reporting period, calculations were made
for predicting the performance of interface detectors.
These calculations were based on either measurements
of the magnitude of the induced current or the shift in
phase between the primary and the scc'ondary voltages.
Optimum operating conditions were defined, and a
prototype probe for use in the molybdenum reductive
extraction column was fabricated. Tests for confirming
the predicted performance of the probe were carried
out at room temperature. Equipment for final testing of
the probe at 600°C with molten bismuth has been
installed. Details of this work are discussed in the
remainder of this section.

Calculations for determining probe performance and
optimum operating conditions. Preliminary calculations
were performed using a two-conductor model consisting
of a bifilar coil inside a molybdenum tube that was

surrounded with either liquid bismuth or air. Results of

magnitude of the signal in the secondary coil and the
difference in phase between the signals in the primary
and secondary coils varied as tube size, tube wall
thickness, and frequency of the signal impressed on the

.primary coil were changed. The results indicated that an

adequate phase shift (a change of approximately 11°

from the condition in which no bismuth was present to

the condition in which the tube was submerged in
bismuth) should be obtained with a molybdenum tube
having an outside diameter of 0.785 in. and a wall
thickness of 0.030 in. Additional calculations were then
carried out with a three-conductor model, which
allowed consideration of the effect of the container in
which the bismuth and the probe are located. It was
determined that the phase shift technique would have
about the same sensitivity at any frequency in the range
of 20 to 34 kHz. The maximum change in magnitude of
the voltage induced in the secondary coil was about
30% as the submergence in bismuth increased from 0 to
100%; the optimum operating frequency was 8 to 12
kHz. ‘

After we had determined the conditions under which
maximum sensitivity could be achieved for the two
measurement techniques, additional calculations were
made to find the conditions that would minimize the
change in probe output resulting from a variation in
probe temperature from 600 to 700°C. The following
effects were considered:

1. The calculated thermal expansion of the molyb-
denum tube is approximately 0.05%; the phase shift
“due to this effect was calculated to be less than
0.01°, which is considered negligible.

2. The resistivity of the bismuth increases from 145.7
Q-cm to 151.3 £-cm, and the resistivity of the
molybdenum increases from 17.7 £-cm to 21 -cm.
‘While these changes produced a significant phase
shift over most of the frequency range, it was found
that a minimum change of about 0.1° occurred at
'25.2 kHz when no bismuth was present and at 30.2
kHz when the probe was completely submerged in
bismuth. On this basis, it was decided that the probe

* would be- least sensitive to these effects at a
t_"req&ency of . 27.7 kHz with the salt-bismuth inter-
face at the midpoint of the detector.

3. The dc resistance of the primary and secondary coils
increases from 69.4 to 79.2 Q in the temperature
interval considered; however, by matching the im-
pedance of the primary and secondary circuit.with
the cable capacitance, it was possible to essentially
eliminate this effect (a phase shift of less than
0.001°).

. Phase shift changes of less than 0.01° were calcu-
lated for changes of 1% in the operating frequency,
the amplitude of the signal impressed on the primary
coil, the output impedance of the primary signal, the
input impedance of the secondary signal, or the
cable capacitance.

Design of prototype interface detector for the molyb-
denum reductive extraction system. The results ob-
tained from the calculations discussed above were used
to design a prototype bismuth-salt interface detector

224

e ¥ e ey o

for use in the molybdenum reductive extraction system.
The probe, shown in Fig. 17.17, consists of a high-
density 99.8%-pure alumina core tube (0.720 in. OD)
on which bifilar 0.010-in.-diam platinum wire coils are
wound in 0.022-in.-wide grooves having a depth of
0.020 in. The active length of the probe is 13.5 in.; the
primary and secondary coils contain about 189 turns
each. The probe was designed for insertion in a
0.785-in.-OD, 0.725-in.-ID molybdenum tube on which
a molybdenum end cap was welded by an electron
beam technique (see Fig. 17.17). The lower end of the
probe is fitted with an adapter that fastens the probe to
the probe mounting boss and holds the high-
temperature electrical connector plug for the instru-

PHOTO 1922-71

Fig. 17.17. Prototype interface detector probe before assembly. The molybdenum sheath tube and bifilar-wound primary and

secondary coils are shown.
ment cable. The lower end of the probe tube was
brazed into a carbon-steel adapter to allow insertion of
the probe in a carbon-steel test vessel. When the probe
tube is installed in the molybdenum reductive extrac-
tion system, it will be welded, via an electron beam
technique, into the disengaging section below the
column without using an adapter.

Room-temperature measurements for confirming pre-
dicted probe performance. Experimental measurements
have been made at room temperature to verify the
predicted performance of the probe. These measure-
ments were made using the coil and molybdenum tube
described above. Inconel disks 1 in. thick with a
diameter of 4 in. were stacked on the probe to simulate
increases in the bismuth level. Difficulties were en-
countered initially with use of the cable proposed for
the molybdenum reductive extraction facility because
of shorting that occurred when the cable was flexed
slightly. Therefore, a cable consisting of two single-
conductor shielded wires was used as a substitute, and
phase shift measurements were carried out at fre-
quencies of 10 and 20 kHz. The measured results at 10
kHz were linear and reproducible to within 0.01°. The
sensitivity was 0.6°/in. of Inconel level, which is in
satisfactory agreement with the calculated value of
0.53°/in. The sensitivity at 20 kHz was considerably
less (0.3°/in.) than at 10 kHz due to the high
conductivity of molybdenum at room temperature. The
results obtained with a frequency of 20 kHz were as
reproducible as those at 10 kHz but were not quite as
linear. Surrounding the probe with a 2.5-in. sched 40
ferromagnetic steel pipe caused a decrease in sensitivity
of 0.02°/in. at 10 kHz and room temperature, although
a somewhat greater loss in sensitivity would be observed
at 650°C. The random noise in the measurements at
both 10 and 20 kHz was 0.015° due to the wide
frequency response of both the driving and the re-
ceiving amplifiers. This noise can be easily eliminated
by limiting the bandpass of cither amplifier.

Carbon-steel probe test vessel. A carbon-steel vessel
(Fig. 17.18) has been fabricated for use in tests of the
prototype probe described above. The enlarged upper
part of the vessel, which is the bismuth reservoir, is
arranged in such a way that known amounts of molten
bismuth can be added to, or removed from, the probe
test chamber that is located below the reservoir. The
probe extends upward from the bottom of the test
chamber and will be secured in the test vessel by
welding the carbon-steel adapter to the bottom of the

225

test chamber, The enlarged lower section, which simu-
lates the interior of the high-temperature containment
vessel for the molybdenum reductive extraction equip-
ment, will be filled with an inert gas. Space is provided
for the 13-ft length of high-temperature electrical cable
(maximum rated service temperature, 2000°F) and for
the nozzle that duplicates the electrical and service
penetration on the containment vessel flange. The
probe will be installed in the test vessel as soon as
difficulties with shorting in the high-temperature elec-
trical cable and connector plug assembly have been
corrected.

Fig. 17.18. Carbon-steel vessel for testing prototype bismuth-
salt interface detector.

18. Continuous Salt Purification

R. B. Lindauer

Studies were carried out to determine the cause of the
occasional high iron concentrations reported for
samples obtained during the previously reported experi-
ments' on the reduction of iron fluoride in molten salt
by countercurrent contact with hydrogen in a packed
column. Possible causes for the high iron concentration

are: (1) bypassing of suspended iron particles around

the poorly sealed sampler filter and (2) contamination
of salt samples with iron from tools used during
removal of salt from the nickel sampler. The presence
of suspended iron particles in the filtered salt appears to
be ruled out by low iron concentrations (average, 68

ppm) reported for four large (4-g), unfiltered samples-

taken from the system. Also, leaching of nickel speci-
mens that had been exposed to the salt in the system
with aqueous HCI showed only a negligible amount of
iron deposited on the specimens. Therefore, it is
believed that the samples are being contaminated during
the removal of salt from the nickel samplers. In order to
minimize such contamination, the samplers were rede-
signed to obtain a larger (4-g) sample and were
constructed of copper, which will facilitate removal of
the salt. The larger salt sample should also provide
improved sensitivity in the analysis for iron at low
concentrations. .

Experiments ‘for measuring the rate at which iron
fluoride is reduced during the countercurrent contact of

1. R. B. Lindauer, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL4676, pp. 269-70.

226

molten salt with hydrogen were continued after the
iron fluoride concentration in the salt (72-14.4-13.6
mole % LiF-BeF,-ThF,) was increased to about 1000
ppm - (which should be sufficiently high to allow
accurate iron analysis). The pressure drop across the
column increased significantly due to a partial restric-
tion and limited the gas flow rate to about 20% of that
used previously with a comparable salt flow rate. Five
iron fluoride reduction runs were carried out at the
reduced gas flow rate; however, the extent of iron
fluoride reduction averaged only 6% per run (see Table
18.1) rather than 27%, as had been observed with
higher gas flow rates. With the lower gas flow rates, the
HF concentration in the exit gas stream was about 60%
of the equilibrium value, based on the concentration of
iron in the exit salt stream, instead of about 27%, which
has been noted with higher gas flow rates. During run
15, a hydrogen flow rate of only 300 ¢m?/min was
used, and the extent of iron fluoride reduction was
about 100% of that which would have been obtained if
the salt and gas streams had been brought to equilib-
rium at the operating temperature of 700°C. The
average mass transfer coefficient measured in this run
was only 0.002 ft/hr as compared with about 0 018
ft/hr measured in previous runs.

Following these runs, the column was removed and
cut into sections in order to determine the cause of the
restriction. Most of the Raschig-ring packing in the
column contained salt that had not drained from the
packing. In addition, a considerable amount of salt
could be seen between rings in the upper part of the

RTRY

irg
227

Table 18.1. Data from iron fluoride reduction runs

Iron

Reduction . fowrate  Salt flowrate  concentrationin  Mass uransfer
i 3 /mi st (ppm)
T Giters/min) (e /min) PP s
Feed  Product
12 26 50 o1l 885 0.0007
13 35 80 885 762 0.0056
14 38 78 762 740 0.0011
15 03¢ 87 740 705 0.0020
16 47 7 705 661 0.0028

10% hydrogen in helium.

column, and a black deposit was noted on the rings in  packing (Fig. 18.2). Several of the salt globules from
this area (see Fig. 18.1). In contrast, very little material the upper part of the column remained white after
appeared to be deposited on the packing in the lower  exposure to moist air; however, some of the globules
part of the column; also, less salt was noted in the  became rust-colored after exposure. Subsequent analy-

PHOTO 142671

‘I
»

<

v Fig. 18.1. View of Raschig-ring packing near top of packed column, showing metallic deposits and salt that did not drain from
the packing.

ses of the salt for iron indicated the expected concen-
tration of iron (640 ppm) for the material that did not
change color, while two samples that became rust-
colored in appearance contained 890 and 2055 ppm of
iron. It appears that some of the salt globules contained
dispersed iron particles which oxidized on exposure to
air. Analysis of the black deposit in the upper part of
the column (after the material had been exposed to air)
indicated a nickel content of 49.2 wt % and an iron
content of 3 wt %. Analyses of packing removed at
12-in. intervals along the column showed little differ-
ence in the amount of deposited iron; the deposition on
the packing had an average thickness of 0.05 mil. The
observed differences in the manner in which nickel and
iron deposited on the packing are quite reasonable since
NiF, is much more easily reduced by hydrogen than
FeF,. Hence one would expect the nickel to deposit in
the upper part of the column and the iron to deposit

228

almost uniformly throughout the column. The total
amount of iron contained on the packing, based on the
average iron deposition, is about 17 g. The amount of
iron believed to have been reduced in the system thus
far is about 24 g. Therefore, it appears that most of the
iron is deposited on the column packing as the FeF, is
reduced.

A column of slightly modified design was fabricated
and installed. The cross-sectional area of the liquid
deentrainment section at the top of the column was
increased by 75% and was packed with a 5-in. depth of
Raschig rings. The wall thickness of the ;-in. Raschig-
ring packing in the column was decreased from ¢ to
Y4, in., which should increase the throughput and
lengthen the operating time before flow restriction due
to reduced metals necessitates removal of iron from the
packing.

PHOTO (425-7¢

Fig. 18.2. View of Raschig-ring packing near bottom of packed column.

-

4k

OAK RIDGE NATIONAL LABORATORY

MOLTEN-SALT REACTOR PROGRAM

AUGUST 31,1971

M. W, ROSENTHAL, DIRECTOR R
R.B. BRIGGS, ASSOCIATE DIRECTOR D
P.N. HAUBENREICH, ASSOCIATE DIRECTOR R

e e

=

Tt

PART TIME ON MSRP
DUAL CAPACITY
ON ASSIGNMENT FROM TVA

GUEST SCIENTIST FROM COMBUSTION ENGINEERING

GUEST SCIENTIST FROM INDIA
GUEST SCIENTIST FROM TAIWAN
CO-0F

H. . BEATTY,* BUDGET B
DESIGN STUDI
TUDIES COMPONENTS & SYSTEMS DEVELOPMENT INSTRUMENTATION & CONTROLS PHYSICS MATERIALS MSBR PROCESSING DEVELOPMENT CHEMISTRY
E.S. BETTIS R . . . 1. R.WEIR, JR. M&C L E.McNEESE*" oT
DUNLAP SCOTT R 5.J.DITTO | M. R -E. .
W1 LUNDIN® n &C A.M.PERRY W.R. GRIMES RG
.CHEMICAL DEVELOPMENT
I I MSR PHYSICS ANALYSIS HASTELLOY N STUDIES
. i L. M. FERRIS cT
” R ENGEL A H. E. McCOY MAC
SBR OESIGN STUDIES PUMP DEVELOPMENT PROCESS INSTRUMENTATION R.E. GEHLBACH MAC M. R. BENNETT er REACTOR CHEMISTRY
L. G. ALEXANDER A . . J.W. KOGER"* M&C E.J.SMITH et
C.E.BETTIS® GE A.G.GRINDELL R R.L.MCORE 1&C G. M. SLAUGHTER" MAC 0. K. TALLENT cr C.F.BAES, JR." RC
R.B. KORSMEYER® R P.G. HERNDON® 1&C MSR PHYSICS METHODS B. McNABB MBC 1 F.LAND cT C.F.BAMBERGER RC
H. A.McLAIN R L.V. WILSON'* R 0. L. SMITH R W, R. HUNTLEY A C.T. THOMPSON T E.G.HOHLMANN" RC
R, C. ROBERTSON R H.C. YOUNG® R H. C. SAVAGE R . 5.CANTOR AC
g W. H. DUCKWORTH* R - ENGINEERING DEVELOPMENT E L COMPERE RC
J.R. TALLACKSON 1&C i
R. P. WICHNER R L.E. McNEESE*" cT -5 KIRSLIS RC
L V. WILSON-* R MSBR CORE DESIGN GRAPHITE STUDIES R.G. ROSS RC
L e . NUCLEAR INSTRUMENTATION AND WP EATHERLY M. 3 BELL o 1 H.SHAFFER RC
L. CONTROLS ANALYSIS H. F. BAUMAN® R .8, CAVIN wac B A HANNAFORD a W, JENNINGS, JR. Ao
COMPONENTS & SYSTEMS J.L. ANDERSON"® 1&C W, H. COOK M&C 4. A. HIGHTOWER, JR. cT W.P. TEICHERT RC
L R AMNES® e ©.W.KEE o1 R.M. WALLER RC
DUNLAP SCOTT** A W.H. SIDES™ 1&C A. L. HAMNER
A MSBR EXPERIMENTAL PHYSICS C.R. KENNEDY M&C R.B. LINDAUER oT
MSBE DESIGN STUDIES C.J.CLAFFEY R O . BURKE 1ac G. L. RAGAN R R.W. McCLUNG* M&C 4. ROTH! cr .
J.L. CROWLEY A L 1P MOORE" MBC - or PHYSICAL AND INORGANIC CHEMISTRY
' C.E.BETTIS GE C.H. GABBARD R & B POLLOCK MEC o P TUNGH or 1 BRAUNSTEIN® ac
C.W. COLLINS A G. M. WINN R 1S WATSON® or PRt re
W. K. FURLONG"* R R.H. GUYMON** A H. 0. WEEREN cT F. A DOSS RO
H. A McLAIN R R. €. HELMS R CHEMICAL PROGESSING MATERIALS E. L. YOUNGBLOOD cr L. 0. GILPATRICK RC
2 R MAWHERTER” R R.d. KEDL R J R.DISTEFANO® W, M. WOODS cT B.F. HITCH i
W, TERRY , A A. 1. KRAKOVIAK R HE Mccov** 1 BEAMS or : .
D, WILSON™*" R . REID R FUNDAMENTAL RADIATION DAMAGE v. L FOWLER or - P.MALINAUSKAS RC
H. A. McLAIN R IN GRAPHITE N.C.COLE" M&C s J.D. REDMAN RC
A N.SMITH R J.1, FEDERER® M&C R. 0.PAYNE . cr D. M. RICHARDSON RC
£.C.ZAPP’ D. K. HOLMES® 33 J.W.KOGER™* M&C T.O. ROGERS cy G.D. ROBBINS RC
o i R. E. McDONALD" M&C 4.5. TAYLOR cT R. A. STREHLOW RC
HEAT TRANSFER AND PHYSICAL PROPERTIES A 1. MOOREHEAD" M&C L.M.TOTH /C
T.S,NOGGLE" S5 - M&C C. F. WEAVER RC
M. W. HOFFMAN® R ‘ TN ® A. C. SCHAFFHAUSER PHROCESS DESIGN
CONSULTANT TECHNICAL SUPPORT E. L. NICHOLSON cT
J.J. KEYES® R . R
LW, COOKE* R F.N.PEEBLES uT M, 0. ALLEN M&C W._L,CARTER CcT
¢ T.5.KRESS" R - N. M. ATCHLEY?® M&C W_F.SCHAFFER,JR. cT
: . . E.R.BOYD" MBC ANALYTICAL CHEMISTRY DEVELOPMENT
S.J. CLAIBORNE R -R. : e
5 B.J.SUTTON" A S.W. coox” JC.WHITE® AC
C.E.DUNN M&C
MSRE SURVEILLANCE J.C. FELTNER MEC
- J.W.GEER M&C RESEARCH AND DEVELOPMENT
R.H. GUYMON""" R ME&C
j‘ :; f.r;fHTH MBC A 5. MEYER® AC
C.W. HOUGK™ MBC R.F. APPLE AC
. J.M.DALE AC
F.P. JEFFERS M&C N ot
H. KEATING® MaC -P.
E. J. LAWRENCE Mac
W. ). MASON® M&C ANALYSES
R.S. PULLIAM" m&g LT CORBIN® AC
L.G.RARDON" & G. GOLDBERG AC
W.H SMITH, JR. Mac C.E.LAMB" AC
H.R.TINCH® M&C €. K. TALBOTT* AC
L.R. TROTTER M&C E.|.WYATT® AC
B. C. WILLIAMS MEC
. M
AC ANALYTICAL CHEMISTRY DIVISION " F'VV:(');';JMH%RU':S, e
8 BUDGET AND PROGRAM PLANNING OFFICE dd
© CHEMISTRY DIVISION
CT CHEMICAL TECHNOLOGY DIVISION SPECIAL ASSISTANCE
D OIRECTORS DIVISION X .
GE GENERAL ENGINEERING DIVISION M. A.BREDIG . c
I&C INSTRUMENTATION AND CONTROLS DIVISION H. R BRONSTEIN c
M&C METALS AND CERAMICS DIVISION A. 8. DWORKIN c
R REACTOR DAVISION A
. RC REACTOR CHEMISTRY DIVISION
- 88 S0LID STATE DIVISION
UT UNIVERSITY OF TENNESSEE

T

S b e ey $ e