ORNL-4832.txt

R i

3 yy5k 0360209 2 (:11-1

MOLTEN-SALT REACTOR
PROGRAM

Semiannual Progress Report
Period Ending August 31,1972

This document has been reviewed and is determined to be
APPROVED FOR PUBLIC RELEASE.

NameTitle: Leesa Laymance, ORNL TIO
Date: 712712017

OAK RIDGE NATIONAL LABORATORY
CENTRAL RESEARCH LIBRARY
DOCUMENT COLLECTION

LIBRARY LOAN COPY
DO NOT TRANSFER TO ANOTHER PERSON
If you wish someone else to see this
document, send in name with document
and the library will arrange a loan.

OAK RIDGE NATIONAL 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.

ORNL-4832
UC-80—Reactor Technology

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT
- FOR PERIOD ENDING AUGUST 31, 1972

M. W. Rosenthél, Program Director ‘
R. B. Briggs, Associate Director
P. N. Haubenreich, Associate Director_

MARCH 1973

RGY SYSTEMS LIBRAR

OAK RIDGE NATIONAL LABORATORY
MARTIN MARIETTA E

wonozzswonsnon | IR

L

U.S. ATOMIC ENERGY COMMISSION 3 445k p3L0109

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
ORNL-4449
ORNL4548
ORNL4622
ORNL4676
ORNL-4728
‘ORNL-4782

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
Period Ending August 31, 1971
Period Ending February 29, 1972
Contents

INtrodUCtiON .« « . . e e e e e e e e e e vii

UMY . .ottt et e e e e e e ix

1. DESIGN ....... ... ... ... ... ... ... it e et e e e e 2
1.1 Lead-Cooled MSBRS . . ... ... . e 2

1.2 MSBR Industrial Design Study ......... PP 5
1.2.] DrainTank ............... D 5

1.2.2 Physics Calculations . ............ e e e e et 10

1.2.3 Chemical Processing . ... ... .. .. i e 11

1.3 Xenon Behavior in the MSBR Fuel Salt System .................. ... .. ... . 11

1.4 Hybrid Computer Simulation of the MSBR . ... ... ... .. ... . . .. . . 13

2. REACTOR PHYSICS ... ... ittt e e e e 14
2.1 Analysis of HTLTR-MSR Lattice Experiments . . ........... ... . . .. 14

2.2 Nuclear Performance of Lead-Cooled Molten-Salt Reactors ... ........ ... ... ... ... ....... 14

2.3 Plutonium Use in Molten-Salt Reactors ........... ... . it 16
2.3.1 Fuel for Molten-Salt Converter Reactors. .. .. ... ... ... ... .. . . ... 16

2.3.2 Start-up of an MSBR on Plutonium Fuel . ..... .. ... .. .. ... . ... ... . . 22

2.3.3 ROD Code Modifications .. .......... .ottt e et e et ciee e 25

3. SYSTEMS AND COMPONENTS DEVELOPMENT ..... ... ... ... . ... ... ... .. ..., 26
3.1 Gaseous Fission Product Removal . .. ... ... . ... .. . .. .. 26
3.1.1 Bubble Separator and Bubble Generator ......... e e e 26

3.1.2 Bubble Formation and Coalescence Test . ........... ... .. i, 29

3.1.3 Mass Transfer to Circulating Bubbles .................. L e PR 29

3.2 Gas Systems Technology Facility .. ........... ... ... ... ... ... ..... e 30

3.3 Molten-Salt Steam Generator Industrial Program -. . .. ... ... ... ... . ... . . ... . . 30

3.4 Coolant-Salt Technology Facility (CSTF) ................ e e e 31

35 SaltPumps ... ... e e DI 32
3.5.1 Salt Pumps for MSRP Technology Facilities . . ......... ... . ... ... ... ... ........... 32

3.5.2 ALPHA Pump . ... e e 33

4. HEAT TRANSFER AND PHYSICAL PROPERTIES .......... e 35
4.1 Heat Transfer ........ e e e e e 35

4.2 Thermal Conductivity ... .... ... i e e e e 36

1ii
iv

PART 2. CHEMISTRY

5. BEHAVIOR OF HYDROGEN AND ITSISOTOPES .. ... ... . e 38
5.1 The Solubility of Hydrogen in Molten Salt . ............ ... . . .. .. . . . .. 38
5.2 Hydrogen Permeation Through Metals ... ... .. ... .. .. . . . . . 39
5.3 The Chemisorption of Tritium on Graphite ........... ... ... .. . i, 42
6. FLUOROBORATE CHEMISTRY .. ... ... e e e e e et 43
6.1 Solubility of BF; in Molten LiF-BeF,-ThF, ... ... ... . . . 43
6.2 Solubility of BF; in MSBR Primary Salt Containing 8 Mole % NaF:
Some Safety Considerations . ....... R 43
6.3 Corrosion of Hastelloy N by Fluoroborate Melts .......... ... .. ... ... .. . 00 iinnenn... 44
7. BEHAVIOR OF SIMULATED FISSION PRODUCTS .. ... .. e e 47
7.1 Effects of Selected Fission Productson Metals . .. ....... ... .. ... .. .. ..., 47
7.2 Stability and Volatility of Metal Tellurides . .............cuiui it 48
8. DEVELOPMENT AND EVALUATION OF ANALYTICAL METHODS . ... ... ... ... ... 50
8.1 In-Line Chemical Analysis of Molten Fluoride Salt Streams ............ ... ... .. ... ....... 50
8.2 Voltammetric Studies of Protonated Species in Molten NaBF,-NaF
(92-8MOIE ) . . oo ottt e 50
83 Infrared Spectral Studies of NaBF,-NaF ... ............. I 52
8.4 Studies of Protonsin Molten NaBF,; . ... ... ... .. . . . 52
8.5 Analysisof Coolant Cover Gas . . ... ... . it e 53
8.6 Spectral Studies of Molten Salts . .. ... .. ... e 54
9. OTHER MOLTEN-SALT RESEARCH . . ... .. .. . e 56
9.1 The Disproportionation Equilibrium of UF; Solutions in Graphite .......................... 56
9.2 EMF Studies of Oxide Equilibria in Molten Fluorides . ........... ... ... ... ... .......... 57
9.3 The Lithium Fluoride—Tetrafluoroborate Phase Diagram ............... ... uuueerouo. .. 58
9.4 Chronopotentiometry Based on Diffusion of Mobile Nonelectroactive Species ................. 59
PART 3. MATERIALS DEVELOPMENT
10. INTERGRANULAR CRACKING OF STRUCTURAL MATERIALS EXPOSED
TO FUEL SALT ... e et e e e e e e e e e e 63
10.1 Cracking of Samples Electroplated with Tellurium .. ........ ...ttt i, .63
10.2  Intergranular Cracking of Several Alloys When
Exposed to Tellurium Vapor . ... .. 63
10.3 Mechanical Properties of Structural Materials Containing
Strontium and Tellurium ........... e e e e e e e e e e 69
10.4 The Isothermal Diffusion of ! 27 Te Tracer in Hastelloy N,
Ni-200 and Type 304L Stainless Steel Specimens . .. ... ... o, 70
10.5 Tube-Burst Experiments . . .. ...t e 71

10.6 Strain Cycle Experiments . . ... ... ... . . 76
11.

12.

10.7 Identification of Reaction Products from Telluriurri—Structural-Material Interactions .. .......... 79
10.7.1 Knudsen Cell Reactions . ... ... ... ... ittt i e, 81
10.7.2 Possible Telluridesin Hastelloy N . . ... ... o i et 82
10.7.3 Tellurides from Knudsen Cell Reactions . .. ...........ciouiireiiereennennennnns 83
10.7.4 Tellurides from Other Reaction Methods .......... ... ... ... ... . 0 i, 83
10.7.5 SUMIMAIY .+ ottt it et e e e e e e e e e e e e e e et e e e e e e e 85
10.8 Auger Electron Spectroscopy of Intergrahular Fracture Surfaces of Nickel
and Hastelloy N Exposed to Tellurium Vapor at 700°C ... ...... ... 86
T0.8.1 Method . .. ... e e e e e 87
10.8.2 NicKel . . ottt e e e e e e 88
10.8.3 Hastelloy N ..o e et e i e e 88
10.9 Design of an In-Reactor Experiment to Study Fission Product .
Effects on Metals . ... .. .. . e e 90
10.9.1 Design Criteria . ...... ... [ 90
10.9.2 Configuration .. .. ... .. .. . .t e et e 91
10.9.3 Thermal Considerations . ......... ... e 91
10.9.4 Containment . ... ... .. et 93
10.9.5 Fuel Salt Chemistry ... ... .. i i e e et 93
GRAPHITE STUDIES . . .. e e e e e ettt et e 95
11.1 The Irradiation Behavior of Graphite . ........ .. ... .. .. . . . . 95
11.2 Procurement and Characterization of Various Grades of Graphite ... ... ... ... ... ... ..... 100
11.3 Characterization of ORNL Graphites . .......... .. . i i 101
11.4 Graphite Fabrication ........ ... .. . . . i e 103
11.5 Thermal Property Testing . . . ... ... i i e et eeeae e 106
11.6 Reduction of Helium Permeability of Graphite by Pyrolytic
Carbon Sealing . .. ... .. e e e e e 107
11.7 TIrradiation of Pyrocarbons . ... ... .. .. e e e 111
11.8 Examination of Unirradiated and Irradiated Pyrocarbon Strips with
the Scanning Electron Microscope (SEM) . . .. ... .. . e 111
11.9 Texture Determinations . . .. .. ... .. . i et e e 113
HASTELLOY N e e e e e e e e e e e e i e 117
12.1 Mechanical Properties of Several Commercial Heats of Modified
Hastelloy N ..o e 117
12.2 lrradiation of Hastelloy Ninthe HFIR .. ... ... ... . . ... . i 120
12.3  Salt Corrosion Studies .. ... .. i e i 124
123.1 FuelSalt........ ... ... ... ..... e 126
12.3.2 Fertile-Fissile Salt . . ....................... PP 126
12.3.3 Blanket Salt .. ...t e 127
12.3.4 Coolant Salt . ...ttt e et 127
12.3.5 Corrosion of Type 304L Stainless Steel and Hastelloy N
by Mixtures of Boron Trifluoride, Air,and Argon .......... .. ... ... .. ... .. . ... 132
12.3.6 Oxide Additions to Sodium Fluoroborate . ......... ... ... .. .. ... .. .. ... ... 134
12.4 Forced-Convection Loop Corrosion Studies . ........ ... .. .. . i 135
12.4.1 Operation of Forced-Convection Loop MSR-FCL-1A ......... ... ... ... ... ... .. .. 135

12.4.2 Corrosion Results from Forced-Convection Loop MSR-FCL-1A ...................... 135
13.

14.

15.

16.

17.

vi

12.4.3 Operation of Forced-Convection Loop MSR-FCL-2 . ..................... e 136
12.4.4 Corrosion Results from Forced-Convection Loop MSR-FCL-2 .. . ..................... 137
12.5 Corrosion of Hastelloy N in Steam ... ... ... ... .. i 137
SUPPORT FOR CHEMICAL PROCESSING . ... ... ... i 147
13.1 Construction of a Molybdenum Reductive-Extraction Test Stand ... ........................ 147
13.2 Welding of Molybdenum ... ... .. . . 147
13.3  Development of Brazing Techniques for Fabricating the Molybdenum
Test LO0D . . oo 149
13.4 Compatibility of Materials with Bismuth and Bismuth-Lithium Solutions ..................... 151
13.4.1 Tantalum ALlOys . . .. ..ot 152
13.4.2 Molybdenum AllOyS . . .. ... o 152
13.4.3 Graphite . ... e 152
PART 4. FUEL PROCESSING FOR MOLTEN-SALT REACTORS
FLOWSHEET ANALY SIS ... e e 159
14.1 Multiregion Code for MSBR Processing Plant Calculations .. .. .......... ... .. ... ......... 159
PROCESSING CHEMISTRY . ......................... A 160
15.1 Equilibria in Fused Salt-Liquid Alloy Systems .. .. ... ... . ... . 0t aenan.. 160
15.2  Solubility of Thorium in Liquid Li-Pb Alloys . .. .. ... ... .. . . . 162
15.3  Reductive Extraction of Titanium and Cesium . ... ........... ... .. ... . ... e .. 163
15.4 Chemistry of Fuel Reconstitution . ........ ... ... .. . . i 164
15.5 Protactinium Oxide Precipitation Studies. . ... ... .. .. . . . 165
ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS ... ... ... ... ... 168
16.1 Operation of Metal Transfer Experiment MTE-3 ... ... .. ... ... ... . . ... ... . .0 ... 168
16.2 Design of the Metal Transfer Process Facility ... ... e e e 171
16.3 Development of Mechanically Agitated Salt-Metal Contactors . ............ouvuuneeuenenn... 171
16.4 Reductive-Extraction Engineering Studies . . ... .. .. .. . ... ... 173
16.5 Reductive-Extraction Processing Facility Development . ....... ... ... . ... ... ... . ... ..... 175
16.6 Design of a Processing Materials Test Stand and the Molybdenum
- Reductive-Extraction Equipment . ... ... ... .. . 177
16.7 Removal of Bismuth from MSBR Fuel Salt ... ....... .. .. .. .. . . . . . ... 178
16.8 Frozen-Wall Fluorinator Development .. ... ... ... . ... .. . . . . ... 180
16.9 Uranium Oxide Precipitation Studies ... ........ ...ttt 182

16.10 Development of a Bismuth-Salt Interface Detector .. ............ovtiinnen ... 183

CONTINUOUS SALT PURIFICATION .. ... e e e e 185
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
233UF, dissolved in a salt that is a mixture of LiF and
BeF,, but 22°U or plutonium 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% ZiF,, 29% BeF,, and 65% " LiF; the
uranium was about 33% 2°°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

vii

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 radicactive 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 uranium
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 223U was
added to the original carrier salt, and in October 1968
the MSRE became the world’s first reactor to operate
on 233U. 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
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 five 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 development of
processes that use liquid bismuth to isolate protac-
tinium and remove rare earths from a salt that also
contains thorium. Qur studies showed that a one-fluid
breeder based on these processes can have fuel utiliza-
tion 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.

viii

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.

In the congressional authorization report on the
AEC’s programs for FY-1973, the Joint Committee on
Atomic Energy recommended that the molten-salt
reactor be reappraised so that a decision could be made
about its continuation and the level of funding appro-
priate for it. Consequently, we undertook a thorough
review of molten-salt technology to provide informa-
tion for an appraisal. A considerable effort during the
reporting period covered. by this progress report was
devoted to the preparation of ORNL-4812, “The
Development Status of Molten-Salt Breeder Reactors,”
a 416-page report that contains the results of our
review.
Summary

PART 1. MSBR DESIGN AND DEVELOPMENT
1. Design

Molten-salt reactors cooled internally by direct con-
tact with a stream of lead were investigated briefly to
see if fuel inventories could be reduced substantially.
Concepts in which the lead was mingled with the salt
throughout the core showed little promise because of
excessive neutron captures in 2®7Pb and high inventory
[2.2 kg/MW(e)]. The doubling time was estimated to be
31 years. A concept wherein the lead was confined to a
peripheral region surrounding the main core showed
slightly better performance with a doubling time of 26
years. This performance was deemed to be insufficient
to justify the development of materials to contain lead.

Current efforts in the MSBR Industrial Design Study
by Ebasco and its subcontractors are aimed at demon-
strating the feasibility of the conceptual design reported
in Task 1. This includes the preparation of CSDDs,
trade-off and parametric studies, independent physics
calculations, transient analyses, drain tank design, proc-
ess plant engineering, and a plant cost estimate.
Computations indicated that the intermediate heat
exchanger design concept recommended in Task I could
withstand a scram transient, Parametric studies show
that a wide range of drain tank designs will accommo-
date 50 MW and maintain safe temperatures. The
independent physics calculations are being tested
against the HTLTR-MSBR experiments. The ORNL
chemical processing flowsheet has been reduced to a
practical engineering design,

A digital computer program MSRXEP has been
written describing in detail the '2°Xe behavior
throughout the MSBR fuel salt system. The intent of
this effort is to confirm the results of the previous
simplified calculations or to make improvements where
required. For the trial case having 44 bubbles per cubic
centimeter of salt and a gas separator efficiency of 90%,
the 133 Xe poison fraction was calculated to be 0.0038,
and the average bubble diameter and void fraction in

ix

the fuel salt loop were calculated to be 0.065 ¢cm and
0.0055 respectively. :

A hybrid computer simulation model of the reference
1000-MW(e) MSBR and results of some test simulations
were described in ORNL-TM-3767.

2. Reactor Physics

Further analysis of MSR lattice experiments at
temperatures to 1000°C gave calculated values of k_
that agree remarkably well with experimental values.

A version of the ROD computer code capable of
calculating the time-dependent behavior of fluid-fuel
reactors with changing fuel composition and neutron
energy spectrum was made operational.

Studies of various fueling schemes for batch-processed
molten-salt converter reactors indicated lower power
costs when LWR-produced plutonium at $9.90/kg
fissile is used instead of enriched uranium. Either
plutonium or enriched uranium could be used to start
up a molten-salt breeder reactor, with little difference
in the economics. Availability of uranium from other
molten-salt reactors increases the range of fueling
options, but would have little effect on attainable
performance of the converter reactors studied. For
start-up of a breeder reactor, use of plutonium involves
some penalty in lifetime-average breeding ratio, pri-
marily because the present chemical processing flow
sheet makes it necessary to defer processing while
plutonium is in the reactor.

3. Systems and Components Development

The detailed designs of the bubble separator and
bubble generator for the gas system technology facility
(GSTF) were. completed, and the drawings were re-
leased for fabrication. The velocity and pressure distri-
butions in the separator vortex were measured under
various flow conditions. A procedure was developed for
predicting the pressure distribution in the bubble
generator at various liquid and gas flow rates.
The bubble formation and coalescence tests on
2LiF-BeF; and 72LiF-16BeF,-12ThF, MSR fuel salt
were delayed because of a mechanical failure of the
shaker drive and clouding of the quartz furnace liner.
The equipment has been repaired, and new capsules of
salt have been loaded in preparation for resuming the
tests,

The test section in the facility for measuring the mass
transfer coefficients to gas bubbles suspended in flow-
ing aqueous solutions was changed to a 1% -in.-diam
conduit to obtain data for comparison with the original
results obtained with a 2-in.-diam conduit. The system
was recalibrated, and preliminary tests were made to
validate supporting information on bubble size distribu-
tion and void fraction originally developed for the 2-in.
conduit. A set of mass transfer data was obtained with a
25% mixture of glycerine and water, and the data were
partially reduced. The system was shut down for
periodic maintenance on the bubble separator.

‘An analytical expression was proposed to relate the
bubble size produced in the bubble generator to the
flow rate, dimensions, and fluid properties. The rela-
tionship agreed with data from the bubble generator in
the mass transfer facility but did not agree with
preliminary results for the proposed MSBR bubble
generator. Possible differences in the mechanism of
generation between the two are therefore being ex-
plored.

Work on the GSTF was resumed in July after three
months’ suspension for budgetary reasons. The me-
chanical and electrical design approached completion,
and some procurement was started. A preliminary
system design description (PSDD) and the system
design description (SDD) were issued in March. The
master copy of the SDD is being used as the primary
design control document and is being continuously
updated.

Foster Wheeler continued with the conceptual design
study of a molten-salt steam generator for use with
molten-salt reactors. Foster Wheeler subcontracted with
Gulf General Atomic for portions of the study concern-
ing tube rupture analysis and dynamic stability. The
Task I steam generator design, conforming to the
conditions of the MSBR reference steam cycle, is
nearing completion.

Construction and installation of the coolant-salt
technology facility (CSTF) was completed in August,
and check-out was started on the control circuitry and
equipment. The piping of the loop was heated to 950°F
and purged with dry helium to remove most of the
moisture from the loop surfaces. Salt circulation is
scheduled for September.

The spare rotary assembly for the MSRE coolant-salt
pump was reconditioned and installed into the CSTF.
The Mark II fuel salt pump is being reconditioned for
the GSTF. The ALPHA pump has accumulated 5800 hr
of operation in the MSR-FCL-2 facility, but was
operated only intermittently due to an oil leak at the
pump shaft and other operational problems related to
the test facility.

4. Heat Transfer and Physiéal Properties

Heat transfer. Final analysis of heat-transfer data fora
proposed MSBR fuel salt (LiF-BeF,-ThF4-UF,; 67.5-
20.0-12.0-0.5 mole %) has resulted in three correlations
covering the Reynolds modulus range from 400 to
28,000. The average deviation of the data from these
correlations is at most 6.6%. However, the data for
Reynolds modulus greater than 5000 average 13%
below predictions based on the correlations of Hausen
and Sieder-Tate for transitional and fully developed
turbulent flow. The data for Reynolds modulus less
than 1000 are only 1.6% above the Sieder-Tate correla-
tion for laminar flow.

Thermal conductivity. A detailed analysis of uncer-
tainties affecting the measurement of thermal conduc-
tivity using the variable-gap apparatus had resulted in a
correlation of standard and maximum error with the
magnitude of the thermal conductivity. Narrow error
bands include the effects of radiation on the uncer-
tainty. It is concluded that the maximum error limits
for the apparatus include the deviations of measured
conductivity from published values using H, O, Hg, Ar,
He, and HTS (KNO;3-NaNQ,-NaNO;; 44-49-7 mole %)
for 93% of the measurements.

PART 2. CHEMISTRY
5. Behavior of Hydrogen and Its Isotopes

Hydrogen and helium solubilities in Li BeF4 have
been determined at 500, 600, and 700°C, and deute-
rium solubility values for the same salt have been
obtained at 600 and 700°C. The two hydrogen isotopes
yield approximately equal solubility results within the
mutual limits of experimental error. Some data have
been obtained which suggest that pretreatment of
the heat exchanger components with hydrogen might
retard tritium transport through these members.

Careful measurements with “unoxidized’ metals con-
tinue to show that permeability of hydrogen and its
isotopes varies with the square root of hydrogen partial
pressure to values as low as our experimental techniques
will permit. Nickel, whose oxide should be trivial over
the entire range of our experiments, follows the half-
power dependence precisely over the range 750 to 8§ X
107* torr. Several alloys, whose oxides cannot be com-
pletely avoided in our experiments, show only minor
variations from the half-power relationship.

Decreases in hydrogen permeability by oxide films
(such as might be expected on the steam side of steam
generator tubing) on 18-8 stainless steels and on
Hastelloy N appear relatively nonrewarding. However, a
film of Cr,0; obtained by 900°C oxidation of a
chromium film electrodeposited on nickel tubing has
markedly decreased the hydrogen permeability and has
produced a situation in which the permeability depends
nearly upon the first power of hydrogen partial
pressure. Very preliminary data also suggest that the
film readily formed by wet oxidation of Incoloy 800
affords a useful reduction of hydrogen permeability.

Upon exposure at elevated temperatures to tritium (at
1 to 10 ppm by volume in helium) graphite readily
sorbs appreciable amounts of this hydrogen isotope.
The highest loading observed to date, on a lampblack
graphite prepared at ORNL and pretreated to increase
its surface area to about 2 m?/g, corresponded to 5 X
10'3 tritium atoms per square centimeter of surface;
several percent of the surface carbon atoms appear to
have bonded tritium atoms, Rinsing with water and with
alcohol removes less than 1% of the sorbed tritium.
Studies continue to determine whether this phenome-
non can be used to assist in containment and manage-
ment of tritium in an MSBR.

6. Fluoroborate Chemistry

Measurements of the solubility of BF, in LiF-BeF,-
ThF4 continue to show that, at constant temperature,
BF; solubility appears to vary with the activity of LiF.
Solubilities of BF; were also measured in a melt
containing 8 mole % NaF in MSBR fuel solvent to aid in
assessing the consequences of coolant leaking into fuel;
the measurements were reasonably consistent with an
exploratory experiment involving a mixture of MSBR
coolant and fuel.

Measurements of the equilibria involving NaNiF; and
NaFeF; in molten NaF-NaBF, have been repeated in
order to establish the composition of the gas phase. The
results obtained yielded improved values for the free
energies of formation of the double fluorides and for
the free energies of reaction (corrosion) of Hastelloy N
constituents with HF and molten NaF.

7. Behavior of Simulated Fission Products

Comparative evaluations of the effects of tellurium on
Hastelloy N and other alloys of interest to the

Xi

Molten-Salt Reactor Program have been provided as
part of a program to identify those fission products
capable of producing superficial grain boundary crack-
ing in alloys proposed for construction of a molten-salt
breeder reactor. Metal tensile specimens representing 14
different alloys and about 24 minor modifications of
Hastelloy N have been exposed to tellurium by a vapor
deposition technique. The results show marked selec-
tivity of the corrosion process on the various alloys
tested.

Stability and volatility of pertinent metal tellurides
are under study by mass spectrometric techniques
assisted by x-ray-diffraction analysis and by direct
observation of chemical reactions. CrTe, which shows
no evidence of volatility, decomposition, or reaction
with Ni at 750°C, appears to be the most stable
telluride of the major constituents of Hastelloy N. The
most stable telluride of nickel appears to be NizTe,; it
slowly decomposes at 800°C under vacuum to form Nij
and tellurium vapor. Both MoTe, and FeTe are less
stable than Ni;Te, .

Both NiTe, and Ni;Te, have been shown to react at
elevated temperatures with CoF;. Products of the
reaction vary with reaction temperature and starting
material but include TeF,, TeF,, NiF,, and CoF,.
These reactions are generally similar to that previously
observed between CoF; and Te, but the observed
partial pressures of TeF, and TeF4 at corresponding
temperatures are lower when the nickel tellurides are
fluorinated.

8. Development and Evaluation of Analytical Methods

We have prepared a variety of voltammetric electrodes
for use in the in-line analysis of molten NaBF, eutectic
in the Coolant Salt Technology Facility. Our original
choice of platinum electrodes proved unsatisfactory
because of a decrease in the cathodic limits of the melt
at higher temperatures. Presently we plan to install
electrodes of iridium, pyrolytic graphite, gold, copper,
and evacuated palladium to effect measurement of iron,
chromium, and an active proton species in the coolant
sait. ‘

The role of hydrogen in the coolant salt has been
shown to be more complex than originally thought. It
now appears that both melts and salts contain
NaBF;0H as a relatively stable species. In the melts,
however, studies indicate that a small fraction of the
hydrogen is present as an electroactive species (possibly
HF, HBF,, HBF;0OH) which is in relatively rapid
equilibrium with a vapor species. In addition, the frozen
salt may contain adsorbed water that is partially
converted to NaBF;OH on melting. Tests with chro-
mium metal have shown that its reaction with the active
proton species is quite rapid, whereas BF;0H™ is
relatively inert. Also it appears that Cr(II) can be
present when NaBF, melts are contacted with chro-
mium metal.

The original infrared pellet calibration data for proton
determination (as BF30OH™) seem confirmed through
standard additions of NaBF;OH to NaBF, salts which
were then melted under isothermal conditions. A liquid
condensate collected from the vapor above NaBF,
melts was found by NMR analysis to contain about
equal quantities of nonionic hydrogen and fluorine. No
evidence for isotopic exchange was seen when hydrogen
isotopes were diffused into NaBF,; melts in a silica cell,
but chemical reactions which formed BF;OH™ (or
BF;0D ") did occur.

An improved Karl Fischer titration technique is being
developed for the calibration measurement of hydrol-
ysis products in coolant off-gas streams. A newly
designed automatic coulometric titrator and a rotating
polarized cathode are used to obtain improved precision
in the measurement of microgram quantities of water.
We are also assembling an apparatus to measure
hydrogen in coolant off-gas after its diffusion through a
palladium membrane.

The spectrum and chemistry of Ce®* and Te? have
been studied in fluoride melts. The analytical usefulness
of the Ce(IV)-Ce(III) reaction has also been considered.
Further spectral studies of Cu?* in LiF-BeF, have been
made, leading to a determination of the solubility of
CuO in that melt when contained in a SiO, cell.
Spectral studies have confirmed the existence of dis-
solved Li3zBi in molten LiCl or LiBr and have inferred
the nonexistence of a corresponding lead pulmbide in
molten LiCl.

9. Other Molten Salt Research

Equilibrium quotients (Q) have been measured for the
reaction

3UF, + UC, = 4UF, + 2C

in dilute molten salt solutions of LiF-BeF,;. The
stability of UF; as measured by these Q values is
strongly influenced by both temperature and solvent
composition. Higher temperatures and larger mole
fractions of BeF, favor greater equilibrium concentra-
tions of UF;.

Emf studies have been initiated of oxide equilibria in
MSBR solvent salt (LiF-BeF,-ThF,, 72-16-12 mole %).
These studies will provide improved oxide solubility
data and, in addition, will confirm activity coefficient

xil

estimates for this salt system. Primary efforts to date
have been concerned with design and assembly of
equipment and with improving the LaF; reference
electrode in order to achieve stability and reproduci-
bility.

The LiF-LiBF, phase diagram was determined. A
comparison with the other alkali fluoride—tetrafluoro-
borate systems shows an increasing positive deviation
from ideality with decreasing cation size.

A beryllium metal anode in molten NaF-BeF, (75
mole % BeF,) shows a reproducible and well-defined
chronopotentiometric transition time. A quantitative
interpretation of such chronopotentiograms, based on
interdiffusion of BeF, formed at the anode surface and
the BeF,-NaF mixture, has been developed.

PART 3. MATERIALS DEVELOPMENT

10. Intergranular Cracking of Structural
Materials Exposed to Fuel Salt

Several alloys have been vapor- and electroplated with
various amounts of Te. Some of the alloys are resistant
to intergranular cracking, and some form shallower
cracks than standard Hastelloy N. Mechanical property
tests on melts of standard Hastelloy N containing
additions of Te and Sr show that Te is detrimental and
that Sr has no effect. The grain boundary and volume
diffusion parameters for Te at 650 and 760°C in
Hastelloy N, Ni, and type 304L stainless steel were
measured. The depth of penetration was greatest for Ni
and least for type 304L stainless steel. Tube-burst tests
on samples of Hastelloy N, Inconel 600, and type 304L
stainless steel plated with Te showed that intergranular
cracks formed in the Hastelloy N and Inconel 600, but
not in the type 304L stainless steel. Similar behavior
was noted in samples of these materials that were strain
cycled.

The reaction products formed on samples exposed to
Te in several experiments have been examined. Several
tellurides have been identified. Thin sheet samples of Ni
and Hastelloy N were embrittled by exposure to Te,
fractured, and the fracture surfaces analyzed by Auger
electron spectrometry. The brittle fracture surfaces
were enriched in Te.

An in-reactor fuel experiment has been designed and
is being fabricated to evaluate the effects of fission
products on intergranular crack formation in Hastelloy
N, Inconel 601, and type 304H stainless steel.

11. Graphite Studies

Irradiation results are now available on experimental
graphites fabricated at ORNL. First-generation graph-
ites employing raw coke directly have been irradiated to
3.5 X 10?? neutrons/cm? (£ > 50 keV) and are more
stable than the reference graphite (Great Lakes grade
H-337) and approach the behavior of Poco grade AXF.
Second-generation graphites based on modified raw
coke have seen up to 2 X 1022 neutrons/cm? and
appear to be superior to the earlier materials. Graphite
fabrication studies continue on a limited scale, and a
third generation utilizing pressure baking is now being
developed.

A few graphites from commercial vendors continue to
be received and are evaluated for their interest in
relating radiation damage to microstructure. Graphites
are also being evaluated for their application as struc-
tural materials for fuel processing.

The major effort continues to be development of
coatings for xenon exclusion from the core graphite.
Process parameters have been identified to produce the
desired high-density isotropic pyrolytic carbon, and a
set is now under irradiation in the HFIR. Preliminary
results on these samples after irradiation to 1 X 10?2
neutrons/cm? should be available in December. A
number of free-standing pyrolytic strips have also been
examined after irradiation to 3.3 X 10%? neutrons/cm?
and confirm earlier results at low fluences that the
high-density isotropic carbons derived from propene are
dimensionally stable. Conversely, an anisotropic meth-
ane-derived coating expanded over 500% in the pre-
ferred c-axis direction but maintained structural integ-
rity!

12. Hastelloy N

Three commercial 100-1b heats of Hastelloy N modi-
fied with 2% Ti were evaluated. The weldability and un-
irradiated mechanical properties are superior to those of
standard Hastelloy N. Two of the alloys were irradiated
at 650 and 760°C to a thermal-neutron fluence of 3 X
102° neutrons/cm? and were found to have acceptable
postirradiation mechanical properties. Commercial al-
loys modified with 0.3 and 0.7% Hf had unirradiated
mechanical properties superior to those of standard
Hastelloy N. However, autogenous welds in the alloy
containing 0.7% Hf cracked. Several types of Hastelloy
N were irradiated to a thermal-neutron fluence of 1.6 X
1022 neutrons/cm?® at 650°C and found to be brittle in
postirradiation creep tests.

Corrosion tests in oxidizing fuel salts have exhibited
high corrosion rates, but no intergranular cracking. A
tellurium-plated Hastelloy N sample was placed in the
hottest region of a fuel salt loop, and the Te was
transferred to a cooler region of the loop. Several

Xiii

natural circulation loops and two pumped loops show
that low corrosion rates can be obtained in pure sodium
fluoroborate, but oxidizing impurities increase the
corrosion rate. Compatibility tests involving stainless
steel and Hastelloy N exposed to fuel and coolant salts
in gas environments of Ar, air, and BF; showed that
Hastelloy N is more resistant to attack than stainless
steel. Air and BF; increased the amount of attack of
both metals. A static test of Hastelloy N in sodium
fluoroborate with an addition of 350 ppm oxide
showed that the presence of the oxide did not result in
increased corrosion.

Hastelloy N specimens exposed to steam at 538°C for
13,000 hr were oxidized at a metal consumption rate of
less than 0.25 mil/year. The oxide was primarily NiO
with some MoO,, a spinel, and Cr, O;. Several stressed
specimens are in test, and the fractures thus far indicate
that the rupture life in steam may be somewhat shorter
than in Ar.

13. Support for Chemical Processing

Construction of the molybdenum reductive-extrac-
tion test stand continued. Three steps have been
completed: (1) fabrication of pots and column, (2)
machining of all subassembly components, and (3)
construction of unjoined mockup. from molybdenum
components. Concurrently, the feed pot and column
subassemblies are being fabricated. The final step will
then consist of interconnecting the subassemblies by
field welding.

Final procedures were determined for all tube-tube
and tube-tee welds. Brazing techniques were developed
for each of the types of configuration that are required.
Induction and portable resistance heating methods of
brazing were successfully used. Several commercial
brazing filler metals were investigated for use in
portions of the system not exposed to bismuth.

Studies of the compatibility of potential structural
materials with bismuth and bismuth-lithium solutions at
600 to 700°C were continued in static capsule and
thermal convection loop tests.

A T-111 thermal convection loop test completed
3000 hr in Bi—2.5 wt % Li at a maximum temperature
of 697°C. The maximum weight loss measured was 2.73
mg/cm?, and, though the samples showed slight oxygen
increases, they were ductile. A molybdenum thermal
convection loop circulating Bi—2.5 wt % Li at a
maximum temperature of 696°C has operated for 5500
hr and will be continued to accumulate long-term test
data.
Samples of graphite were attacked by lithium when

xiv

exposed for 1000 hr at 700°C in a graphite crucible. No -

metallic lithium was found in the crucible, but the walls
of the crucible moved inward, and the samples also
exhibited weight gains. X-ray-diffraction results on the
samples showed the presence of Li; C, and an expanded
graphite lattice. However, no chemical reaction has
been noted between graphite and Bi-Li solutions con-
taining up to 3 wt % Li.

PART 4. FUEL PROCESSING FOR
MOLTEN-SALT REACTORS

14. Flowsheet Analysis

Work was continued on the development of a
computer program that can be used for calculating
steady-state concentrations and heat generation rates in
an MSBR processing plant. The behavior of 687
nuclides in as many as 250 regions can be considered;
however, the use of only about 70 regions has been
adequate for representing processing plant flowsheets
considered thus far. Modifications made in the com-
"puter program during this report period facilitate its use
and allow an improved representation of equipment
items. The modified program has been used successfully
for evaluating a number of flowsheets that are based on
fluorination—reductive-extraction—metal transfer but
differ appreciably from the reference flowsheet.

15. Processing Chemistry

Studies related to the development of the metal
transfer process were continued. The following results
were achieved:

1. The solubility of Li;Bi in molten LiCl was deter-

mined over the temperature range 650 to 800°C.

Some data were obtained on the equilibrium distri-
bution of lithium and bismuth between liquid Li-Bi
alloys and molten LiBr or LiF.

3. Measurements of the equilibrium distribution of
lithium and lead between liquid Li-Pb alloys and

molten LiCl were made at 650°C.

The solubility of thorium in selected Li-Pb alloys
was determined over the temperature range 400 to
700°C.

In addition, the reductive extraction of titanium and
cesium from molten MSBR fuel salt into liquid bismuth
was studied at 600°C. Studies of the chemistry of fuel
reconstitution were continued. Spectrophotometric evi-
dence showed conclusively that UFs is the product of

the reaction of gaseous UF4 with UF, dissolved in
MSBR fuel salt. Further progress was made in the study
of the equilibrium precipitation of protactinium from
LiF-BeF,-ThF, (72-16-12 mole %) by sparging the salt
with HF-H, O-Ar gas mixtures. Preliminary values for
the equilibrium quotients for the reaction PaFs(d) +
% H,0(g) = Y%Pa,0s(s) + SHF(g) can be represented
by log Q) = 12.74 — 10,880/T(°K) in the temperature
range 535 to 650°C.,

16. Engineering Development of
Processing Operations

Studies related to a number of processing opera-
tions were continued during this report period. The
purification and charging of the salt and metal phases
to metal transfer experiment MTE-3 were com-
pleted, and operation of the experiment was initi-
ated. The equipment used in the experiment con-
sists of a fluoride salt reservoir containing about 30
liters of fluoride salt (72-16-i2 mole % LiF-BeF,-
ThF,), a salt-metal contactor, and a rare-earth stripper
containing 4.6 liters of 4.9 mole % Li-Bi solution. The
10-in.-diam contactor is divided into two compart-
ments that are interconnected by a pool of bismuth
containing reductant. During operation of the experi-
ment, salt is circulated from the fluoride salt reservoir
to one side of the contactor, and LiCl is circulated from
the rare-earth stripper to the other side of the con-
tactor. Mechanical agitators are used in the contactor to
contact the phases without causing dispersion of either
phase. Mass transfer coefficients were measured for
228Ra during the first two runs; the resulting values
were in good agreement with a literature correlation
that is based on mass transfer between aqueous and
organic phases in the type of contactor used for the
present experiment. Quantities of LaF; (510 g) and
154EuF, (14 mCi) were then added to the fluoride salt.
The mass transfer coefficient values measured for
europium and lanthanum during four additional runs
were lower than predicted; the values for lanthanum
were considerably lower than expected, while those for
europium were only about 10% of those expected. It is
believed that these low values are caused by the
presence of an oxide layer at one or more of the
salt-bismuth interfaces in the system. We believe that
the rate of transfer of europium and lanthanum can be
increased considerably either by removing the inter-
facial material or by increasing the agitator speeds
beyond the 100- to 200-rpm range employed thus far.

Additional progress was made on the design of the
metal transfer process facility, in which the fourth
metal transfer experiment will be carried out. This
experiment will use salt and bismuth flow rates that are
5 to 10% of those expected for processing a 1000-
MW(e) MSBR. The three-stage salt-metal contactor will
be fabricated from graphite. During this report period,
“discussions concerning the design and fabrication of the
contactor were held with a graphite manufacturer.
Minor design changes suggested by the manufacturer
were effected, and the preliminary quotation for
fabrication of the completed assembly appears to be
satisfactory.

Studies involving the measurement of mass transfer
rates in simulated mechanically agitated salt-metal
contactors were pursued further. Since the important
physical properties of molten salt-bismuth systems
(viscosities, densities, and density difference) are quite
similar to those of the mercury-water system, investiga-
tions carried out with contactors using mercury and
water will allow examination of the hydrodynamic and
mass transfer characteristics of this type of contactor
under highly desirable conditions for experimentation.
Work during this report period was aimed at finding
materials that distribute between mercury and aqueous
phases in a manner suitable for mass transfer studies.
The extraction of lead from an aqueous phase into
mercury containing zinc at low concentrations, accord-
ing to the reaction

Pb(NO3), [H, O] + Zn[Hg]

- Pb[Hg] + Zn(NO;), [H,0] ,

was found to be quite satisfactory. Results obtained in
a series of experiments performed with a 6-in.-diam
contactor indicate that the individual mass transfer
coefficients in a water-mercury system (and probably in
a salt-bismuth system) are predicted reasonably well by
an existing correlation that is based on mass transfer in
aqueous-organic systems.

Three mass transfer experiments were performed in
the mild-steel reductive-extraction system. In these
experiments, the rates of transfer of ®’Zr and **7U
from molten salt to bismuth were measured by adding
the respective tracers to the salt phase prior to
contacting the salt with bismuth containing reductant
in a 0.82-in.-ID by 2-ftlong packed column. The
fractions of the materials transferred ranged from 36 to
80% at extraction factors ranging from 1.3 to 6.3. The
height of an overall transfer unit (HTU) based on the
bismuth phase ranged from 4.7 to 7.3 ft. A correlation
was developed which shows that the HTU value is
constant and equal to about 4.3 ft for extraction
factors near unity.

Xv

Construction of the reductive-extraction process fa-
cility will require joints in molybdenum tubing, some of
which must be made in the field. During this report
period, the feasibility of flaring commercially available
molybdenum tubing in Y-, ¥%-, and Y%-in. sizes was
demonstrated. Since molybdenum-on-molybdenum
threads tend to gall and since it is difficult to tap
threads in a molybdenum nut, we have considered
designs in which the flare fittings have molybdenum
bodies, but the nut and ferrule are made of materials
other than molybdenum. However, because of the very
low coefficient of thermal expansion for molybdenum,
almost any other potential material for the nut and
ferrule would expand more than molybdenum; thus
such a fitting would loosen upon being heated. A means
was devised for fully compensating a fitting for both
axial and radial differential thermal expansion. Test
fittings in Y-, %-, and %-in. sizes that had a 45° flare
and were compensated for axial expansion were fabri-
cated. Although one of these fittings (¥ in.) remained
leak-tight (as shown by a helium leak test) during
thermal cycling from 300 to 600°C, it was concluded
that compensation for both axial and radial thermal
expansion is required. A set of fittings that have a 37°
flare and are fully compensated for differential thermal
expansion is currently being fabricated.

The design of the molybdenum reductive-extraction
equipment and the processing materials test stand is
essentially complete. A trial assembly of the compo-
nents, piping, and support system was made to check
the design drawings; no problems were found. The
remaining design work involves the electrical wiring for
the containment vessel and transfer line heaters, the
instrumentation and control panels, and the services.
The assembly frame for the molybdenum equipment
was test loaded and was found to perform satisfactorily.
Procedures and checklists for moving the completed
molybdenum system from the assembly area to the
operating area were prepared, and a transfer of the
frame and equipment mockup was made successfully.

Several types of samplers were used to take portions
of fluoride salt from various engineering experiments
for bismuth analyses. The reported concentrations of
bismuth in the salt ranged from less than 0.1 to greater
than 1000 ppm. The results indicate that several of the
samples were contaminated during their passage
through the sample port and that an improved sampler
design will be required in order to avoid contamination.
Samples withdrawn from the fluoride salt surge tank in
metal transfer experiment MTE-3 have shown low
bismuth concentrations consistently (0.45 to 1.7 ppm).

A series of experiments on autoresistance heating of
molten salt was performed in a simulated fluorinator
that was protected from corrosion by means of a frozen
wall. Results of these experiments indicate that an
electrically insulting layer of frozen salt can be formed
reliably if the temperature of the fluorinator wall is
held below that of the salt solidus. The measured values
for the thickness of the layer agreed well with values
calculated from heat transfer considerations and from
measurements of the electrical resistance of the molten
salt in the simulated fluorinator. Operational stability,
desired specific heat generation rates, and adequate heat
removal rates by natural convection and radiation were
demonstrated for operation of a frozen-wall fluorina-
tion experiment,

The first series of experiments in the uranium oxide
precipitation facility has been completed. On disas-
sembly, the precipitator vessel was found to contain a
large quantity of undissolved oxide which appears to
have resulted from inadequate agitation of the oxide
during the dissolution step. Examination revealed negli-
gible corrosion of the equipment by HF attack or
agglomeration of oxide on the surfaces. Thus oxide
precipitation continues to appear to be an attractive
alternative to fluorination for removing uranium from
protactinium-free MSBR fuel salt.

An eddy-current-type detector is being developed to
allow detection and control of the bismuth-salt inter-
face in salt-metal extraction columns or mechanically
agitated salt-metal contactors. The.probe consists of a
ceramic form on which bifilar primary and secondary

xvi

coils are wound. A high-frequency alternating current is
passed through the primary coil, and a current that is
dependent on the conductivities of materials adjacent
to the coils is induced in the secondary coil. A probe
assembly was tested at 550 and at 700°C using the
amplitude measurement technique. At each tempera-
ture, the probe readings were linear and reproducible
for bismuth levels between 4 and 12 in. Although the
temperature compensation circuit did not function as
expected, the probe is entirely usable at any tempera-
ture for which a calibration curve is available. With a
constant temperature and bismuth level, a variation of
about 5% in the probe reading was noted over a five-day
period. The reason for this variation is not known.

17. Continuous Salt Purification

The system was modified to provide for the continu-
ous circulation of a small volume of salt (about 4 liters)
through the packed-column gas-salt contactor. The new
equipment consists of a check valve pump that has
tungsten ball checks and an orifice head pot for
measuring the salt flow rate, Initial tests with the
system showed that the equipment for measuring the
salt flow rate performed quite satisfactorily; however, it
will be necessary to modify the pump to permit
operation with salt containing small quantities of
particulate.
Part 1.

R. B. Briggs

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) plant. This design study uses the
ORNL design for background and is to incorporate the
experience and the viewpoint of industry. Task 1 of this
study, the selection of a reference concept, has been

completed and described in a report.2 We also do brief

studies of new or advanced concepts such as the
lead-cooled reactor that is discussed in this report.

One could not, however, propose to build a 1000-
MW(e) plant in the near future, so we have done some
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 Re-
actor Experiment, would incorporate a minimum of
fuel reprocessing, and would have the purpose of

1. Molten-Salt Reactor Program Staff, Conceptual Design
Study of a Single-Fluid Molten-Salt Breeder Reactor, ORNL-
4541 (June 1971).

2. Ebasco Services Incorporated, 100-MW(el Molten-Salt
Breeder Reactor Conceptual Design Study — Final Report —
Task 1.

3. I. R. McWherter, Molten Salt Breeder Experiment Design
Bases, ORNL-TM-3177 (November 1970).

4. E. S. Bettis, L. G. Alexander, and H. L. Watts, Design
Studies of a Molten-Salt Reactor Demonstration Plant, ORNL-
TM-3832 (June 1972). '

MSBR Design and Development

P. N. Haubenreich

demonstrating the practicality of a molten-salt reactor
for use by a utility to produce electricity. The present
AEC program does not include construction of an
MSBE or MSDR, and studies of these reactors are
inactive.

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 en-
riched *3°U for fueling molten-salt reactors. Some
studies related to safety are in progress preliminary to a
comprehensive review of safety based on the ORNL
reference design of a 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. A gas
system technology facility is being built for use in
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. A coolant system technology facility has been
completed for use in studies of equipment, processes,
and chemistry of sodium fluoroborate for the sec-
ondary 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 has been undertaken to provide the
technology for designing and building reliable steam
generators for molten-salt reactors. As the first step in
this program, Foster Wheeler Corporation is preparing
the conceptual design of a steam generator for large

molten-salt reactors.
1. Design

E. S. Bettis

‘1.1 LEAD-COOLED MSBRs

Report TM-3832 on the 300-MW(e) Molten-Salt
Demonstration Reactor (MSDR) was issued. The MSDR
study involved the most detailed design of a large
single-fluid molten-salt reactor that has been done to
date, and we believe that it represents a practical
reactor.

Having finished the demonstration reactor, we de-
cided to take a more serious look at an advanced
reactor concept originally proposed about ten years
ago. This concept involves the use of lead as a direct
contact coolant for the molten salt. Lead is immiscible
with fluoride salts and is chemically compatible with
the fuel salt. Because of the difference in density
between the salt and lead, the two liquids can be
separated by gravity. When the lead is pumped into the
salt, it transfers momentum to the salt, and so it can be
used to circulate as well as extract heat from the fuel
salt. The lead must be cooled by another coolant loop
(or secondary salt loop) which is then used to make
steam.

Although the idea has received cursory attention
during the past ten years, no serious design study was
undertaken. We therefore began an evaluation of this
idea to see if it had any practical potential. A major
incentive for using direct contact cooling is that the
power density in the heat extraction system promises to
be high and so should reduce the amount of salt in the
primary circuit external to the reactor. A

These two advantages are quite attractive; however,
rather serious difficulties are associated with the use of
lead so that, for the concept to warrant serious
consideration, the lead-cooled reactor must show
superior performance, particularly in reducing primary
salt inventory.

The first serious liability of the lead system is that it

must be contained in a refractory metal (probably

. molybdenum). Successful lining of large vessels with
molybdenum and fabrication of molybdenum heat
exchangers, pumps, and piping probably can be accom-
plished, but they are certain to be expensive and to
require considerable development,

Also, effecting a complete separation of salt from the
lead is difficult, and some undetermined amount of salt
will always circulate with the lead. An extensive
experimental program would be required to determine
the extent of entrainment, For our study we only made

estimates of the gross disengagement and the amount of
salt in the salt-lead interfacial region at the bottom of
the reactor.

A third disadvantage of the direct contact lead
cooling is that the lead and salt flows are cocurrent.
This means, of course, that the maximum lead tempera-
ture is the same as the minimum salt temperature; thus
the reactor maximum salt temperature tends to be high
for this system. A further temperature disadvantage is
that the low lead temperature must be kept above the
melting point of the salt to prevent entrained salt from
freezing out in the lead. These factors limit the At
taken in the lead, and the low A¢ in combination with
the low specific heat requires that the volumetric flow
of lead be large.

We decided that the complexity of the lead piping
around the top of the reactor vessel would make it
impractical to replace the core graphite. Therefore, we
designed for a 30-year core life, which meant that the
reactors would be quite large. We studied three dif-
ferent types of reactors, the difference between them
being the way in which the lead was disposed in the
reactor vessel.

All three reactors were similar in basic arrangement.

. Each reactor had a top reflector similar to that of the

MSDR except that the reflector was cooled by lead.
Each also had.a radial reflector consisting of graphite
spheres cooled by lead. The lead pool beneath the core
served as the bottom axial reflector. The core graphite
in each case floated on the lead pool and pushed against
the top reflector. The core was firmly compressed by
the bouyant action of the spheres in the radial reflector.

None of the reactor types really proved to be worth
pursuing, so we will not describe them in detail. Rather,
we will describe the different core designs we tried and
will show where they failed.

The first core was very similar to the core of the
MSDR except that a %-in.-wall, 6-in.-OD graphite tube
was put into the center of each graphite core cell. A
typical cell is shown in Fig. 1.1. It is 13'4 in.2, made up
of four corner posts, some slabs of graphite with flow
passages between them, and the graphite downcomer
tube in the center. A lead plenum is located directly
below the top reflector. Directly under the lead plenum
is a gas space where xenon can be purged from the salt.
A layer of salt completely covers the core which bears
against the bottom of the lead plenum through graphite

_ studs on the core cover plates.
.

A 4-in.-diam nozzle introduces lead from the plenum
into the center of each downcomer. The lead velocity
leaving this nozzle is 32 fps, and the salt enters the
annulus around the nozzle with a velocity of 8.3 fps.
The pressure in the lead plenum to produce this flow is
80 psi. Mixing of the lead and salt results in a pressure
rise of 63 psi which, with the static pressure of the
mixed liquids, produces a pressure of 178 psi at the
bottom of the core. This pressure is dissipated in
forcing the salt up through the slots in the cells to the
top of the core, and circulation is thus maintained in
the core.

In this design, the lead enters the top of the reactor at
a temperature of 1000°F and leaves at a temperature of
1100°F. The temperature of the salt is 1200°F at the
top of the core and, of course, is the same (1100°F) as
the lead at the bottom. Thus both lead and salt have a

At of only 100°F. A lead flow of 900 cfs and a salt
flow of 300 cfs is required. For the flow area in the
core and with the center power density, this requires
the mixture to flow at 18 fps in the center channels.

This core had to be abandoned for two reasons. The
high velocity of the jets leaving the core at the bottom
of the reactor carried too much salt into the lead pool
at the bottom. It was not possible to get an exact figure
for this entrainment, but it amounted to at least 600
ft*. This minimum figure (it probably is much higher)
rendered the performance of the reactor unattractive.

In addition, we found that the lead in the interior of
the core was a greater poison than we had anticipated,
and the nuclear performance suffered. This is discussed
in Chap. 2.

The second design was an attempt to correct the two
faults of the first. Again, the same reactor, reflector,

ORNL—DWG 72-13700

INCHES

Fig. 1.1. Plan view of graphite core cell for lead-cooled MSBR.
and general configuration were used, but the cell design
was completely changed. This was done to decrease the

entrainment at the bottom and also to decrease the _

poison of the lead in the core.

The cell is still 134 in.2, but it is constructed of four
extrusions which, when assembled, form a square with
an 11%-in.-diam hole in the center. Inside this hole are
two concentric graphite cylinders, each having a wall
thickness of 17, ¢ in. In the center of the inner cylinder
is a 3-in.-diam rod. This cell provides two annuli of
%-in. thickness and an outer annulus of -in. thick-
ness. The two inner annuli provide riser channels for the
salt, while the outer annulus provides the downcomer
and contains 1.5 volumes of lead per volume of salt.
Figure 1.2 shows both a plan and elevation of this cell
arrangement.

At 8-ft intervals in the outer annulus, there are
turning vanes which impart a rotational component to
the mixture of lead and salt. This rotation generates a
1-g centrifugal field which separates the lead from the
salt by the time the mixed stream reaches the bottom

/ |
r

[ | E—
.‘0'0

SR

Q)
(XA
0.0 000 5
RS

()
X2

)
&

(P
*

X )
5%

P22
x5
%

*

5
SR
&

ol
X

>
9,
>

TS
S8
ot led
Podetele
3%

e

(D

odole
%

RIS
0:09:0:0:0.0
SEAEOEEL

of the core. The salt collects in a plenum (approxi-
mately 2 in. deep) and flows back up through the two
annular risers in each cell.

In order to reduce the neutron absorptions in the lead
in this core, the ratio of lead to salt is made 1.5:1,
necessitating a 200°F At in the lead and thus raising the
lead temperature to 1200°F and the salt temperature to
1300°F. From a nuclear standpoint, this core is
superior to the first core, but the performance is still
mediocre. The breeding ratio is about 1.05, the dou-
bling time is 31 years (full-power years with compound
interest), and the specific inventory is about 2.2
kg/MW(e).

Although the salt disengagement volume is reduced to
a tolerable amount by the centrifugal separation, a
hydrodynamic problem is introduced. In order to pump
the salt around the core circuit, a lead jet velocity of 60
fps is required. The corresponding lead plenum pressure
to attain this velocity is 267 psi, a value which we
consider to be excessive because of the stresses imposed
on the reactor vessel.

ORNL-DWG 72-13704

v
o
X
o)

5

e
! o,
flo%e%e L35
B2 I Al I
B2 SALT AND i Nl L
PO GAS SPACE —= o
ORI \ 554
050505 I - %8
: .:0:0. ;.... L E A D §| X 4 % : :
o9, o.o. S XRKRIF FLOW TUBE : Q4 XX
995 1 %% 0 |
%, % e%e%e%
RS | Yoted ‘ ]
SR e Jode%
: 153 peted
K ToTeds: f
o « KX 2%
y e RS =
1’ K2 X KX X
ol oo [ %%
: oge00, 5] SRR
P p 1958,
= LS 33 1] |
. %e% Setetetet
% I [ (R 0308 ]
a ¥l K] ;
1" R¢5d Potele . ]
n 5] 55 1
” (5 L
SN .z : La9a?, @ L
Leap FLow | | \ saLT FLow
SALT AND
LEAD FLOW

Fig. 1.2. Plan and elevation of graphite core cell for lead-cooled MSBR.
A final core design was studied in which the lead
downcomers were removed from the core and placed
peripherally around it. The core cells are exactly like
those of the demonstration reactor, They are squares
made up of corner posts, inside which are located
graphite slabs. The salt flows up through the interstitial
spaces in these cells, across the top of the core under
the reflector, and down in the lead mixing cells around
the core.

These peripheral cells are squares 13, in. on a side
with an 11-in.-diam hole in the center. Inside this hole
is a 6-in.-diam rod leaving a 2"} -in. annulus down which
lead and salt flow. This lead-salt-graphite region serves
as a kind of blanket and liquid-liquid direct contact
heat exchanger. By removing the lead from the center
of the core, its poison effect is greatly reduced. Outside
the blanket region is a reflector consisting of lead-
cooled graphite spheres.

Lead with a velocity of 25 fps is introduced into an
annulus through an 8-in.-OD nozzle centered around
the central graphite cylinder. The lead jet entrains salt,
and the mixed liquid flows with a velocity of about 14
fps down the annulus. Again, turning vanes generate a
centrifugal field to separate the lead from the salt. A
lead plenum pressure of 50 psi is required to impart the
necessary velocity and establish circulation through the
core.

This design may have possibilities. From a nuclear
standpoint, it appears to be rather attractive. Again,
details are reported in Chap. 2. There is, however, a
question which has not been resolved. It is very difficult
to determine how flow is established across the core at
the top and bottom. So far, we have been unable to
calculate the depth of either plenum, and it is quite
possible that these plenums will become so deep as to
contain a prohibitive amount of salt.

It should be pointed out that all these studies have
been concerned with thermal reactors. It appears now
that lead cooling does not improve the performance of
thermal reactors' sufficiently to justify attempting to
solve the difficult materials problems associated with
the use of lead. We therefore propose to summarize the
results of the studies in a report and to do no further
work on these concepts. Lead cooling may have some
merit for intermediate or fast-spectrum molten-salt
reactors, and studies of such reactors might be worth-
while at some future time.

1.2 MSBR INDUSTRIAL DESIGN STUDY
M. 1. Lundin

The Task I Report! was delivered to ORNL early in
the report period. The report described a conceptuat

design for a 1000-MW(e) Molten-Salt Breeder Reactor
plant, the salient features of which were described in
the last semiannual report.2 Task II was started and is
currently in progress. This effort is defined as follows:

1. demonstrate the feasibility of the conceptual design
presented in Task I;

2. prepare CSDDs for the design;

3. conduct trade-off and parametric studies to optimize
the design;

4. perform the benchmark physics calculations;

5. prepare a cost estimate for the recommended design.

Work is active in each of these five areas.

In the design engineering area, where practicality of
the design must be demonstrated, work is currently in
progress on the drain tank, the transient analysis of the
plant, handling of the reactor vessel top head and of the
reflector graphite, structural support of the primary
system, and design of a reactor cell structural support
system. The present arrangement is shown in Fig. 1.3.
The primary system concept has been studied, and a
decision has been made to utilize a lined system to
protect the plant from thermal shock. This cooled liner
also increases the design allowable strength of the
reactor vessel metal to 11,000 psi.

Babcock & Wilcox completed a transient analysis of
the intermediate heat exchanger. The scram transient
specified resulted in a drop in fuel salt inlet temperature
of 250°F in two stages: 200°F in 10 sec followed by a
50°F drop in 4 sec. The recommended Task I design
withstood this transient. This design utilizes a sine wave
tube with a nonremovable tubesheet but having access
to the tubesheet for tube plugging (see Fig. 1.4).

1.2.1 Drain Tank

A parametric study is under way to identify and
quantify the important design parameters and to permit
the selection of a credible design. The design base is a
hypothetical steady-state condition in which the fuel
salt produces 50 MW while at a uniform temperature of
1300°F. A tentative design based on this condition will
be examined under normal steady-state operation and
under transients. These conditions may provide bases
for refining the design.

The conceptual design uses bayonet tubes mounted in
the tank head for cooling the tank contents. For

1. 1000-MWfe) Molten Salt Breeder Reactor Conceptual
Design Study — Final Report — Task I, Ebasco Services, Inc.

2. MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL-4782.
!

ORNL - DWG 72-13702

74
/// ’

y

7 NG
||'
Il.;n,'g lm

i\

TAKEN FROM EBASCO DWG NO. SK-{8635

Fig. 1.3. Design study of MSBR cell.

simplicity, reliability, and ease of maintenance, it is
desirable to have as few thimbles as practical. To
achieve this, the present Ebasco concept utilizes duplex
tube thimbles having a mechanical bond. In this
concept heat transfer is limited primarily by the salt
film coefficient rather than by radiation as in the
ORNL concept. Thus, a greater heat flux and far fewer
thimbles are possible in the Ebasco concept.

Heat transfer characteristics of fuel salt to cooling
thimbles were examined. Figure 1.5 shows the calcu-
lated film coefficient as a function of salt and metal
temperature based on a free convection model.3 It can
be seen that a coefficient of about 200 Btu hr™! ft™2
°F7' can be expected if the cooling surfaces are

maintained at 1100°F and if the salt is allowed to reach
1300°F. Being cooler than the bulk, fuel salt near the
thimble surface flows downward, becoming turbulent
very near the free surface. Because the flow is turbulent
over almost the entire thimble length, the film coeffi-
cient is constant over the entire length. The thickness of
the turbulent boundary layer was calculated (Fig. 1.6)
based on a vertical flat plate model.# The results suggest
that it may exceed 1 in. near the bottom of a long
thimble. The same model was used to estimate the

3. M. Jacob, Heat Transfer, Vol. I, Wiley, 1949, p. 529.
4, E. R. G. Eckert and T. W. Jackson, NACA Technical Notes
2207, October 1950.
A, FILM COEFFICIENT (Btu hr! ft72eF 1)

PRIMARY

-— —
r SALT INLET

5ft

ORNL-DWG 72-13703

ACCESS HOLE

SECONDARY

N
J———”—E~ UPPER END GUIDES

CENTER CYLINDER GAS OR
H STAGNANT COCLANT SALT

! 7C00 TUBES (0.375in. OD

: 0.259in. ID 0.625in. PITCH)

Rie— BUNDLE (59in. OD)

I~ 7 saLT QUTLET ""—]

26 ft 1 i
|
-
IMPACT FLOW BAFFLE
- I SECONDARY
SALT INLET
S ft
TAKEN FROM
BABCOCK & WILCOX REPORT
PRIMARY SALT
OUTLET
Fig. 1.4. Fixed sine-wave tube IHX with tubesheet access.
ORNL-DWG 72- 13705
1.5
<
2
ul BULK SALT
z TEMPERATURE 1200°F
Q
G e
ORNL- DWG 72-13704
400 o
I ! >
Gr, i - Gr, (To—Ta)
£=0.091 (=5 Pr)") WHERE —%5 =g B —25* - 1500°F
>
300 METAL SURFACE E
TEMPERATURE / 3
Moot / 8 /
- @ 0.5 -—
200 / -
Z
/ 1200 °F g
2
d &
100 2
/ «©
o % 5 10 15
1200 1300 1400 1500

Fig. 1.5. Film coefficient for fuel salt in drain tank.

T, BULK SALT TEMPERATURE (°F)

x, DISTANCE FROM TOP OF THIMBLE (ft)

Fig. 1.6. Turbulent boundary layer thickness on a vertical flat
surface resulting from natural convection.
temperature distribution (Fig. 1.7) and the velocity
distribution (Figs. 1.8 and 1.9) in the boundary layer. It
can be seen that most of the temperature rise occurs
within half the boundary layer thickness, and the salt
velocity can approach several feet per second. These
results indicate that natural convection provides a great
deal of mixing, so a uniform bulk salt temperature can
be assumed for design purposes.

Based on the film coefficient shown in Fig. 1.5, the’

heat transfer surface required to accommodate 50 MW
was determined (Fig. 1.10). If the cooling surfaces
operate at 1100°F (average) and the bulk salt is

ORNL-DWG 72-13706

1.0
-
O
a4 5
o =
4 w
<1 @
< W
o L /
& /
QO D
& w 05
w o
E P
N o
|\9 w
\ a
E =
~ W
3
- [l
0
0 0.5 1.0

¥/'8, FRACTION OF BOUNDARY LAYER

Fig. 1.7. 'Temperature distribution in fuel salt near thimble
wall.

ORNL— DWG 72~ 13707
10 I

T = 1500 °F
y
_ 1300°F

{200 °F_—

\s\\\

0 5 10 15
x,DISTANCE FROM LiQUID SURFACE (ft)

Fig. 1.8. TI', a constant used with the velocity distribution in
Fig. 1.5.

permitted to approach 1300°F, it can be seen that an
area of 4200 ft* is required. The heat transfer area
together with the fixed heat load determine the heat
flux and, hence, the temperature drop between the
thimble surface and the flowing NaK. Figure 1.11
shows calculated NaK temperatures for various choices
of salt and salt temperatures. For 1300°F salt and

ORNL-DWG 72-13708

0.5

Vy /F, RELATIVE SALT VELOCITY

O \
0 0.5 1.0
y/S, FRACTION OF BOUNDARY LAYER

Fig. 1.9. Relative salt velocity distribution in turbulent
boundary layer.

ORNL-DWG 72-13709

15,000 ,
— g
A= RN, - 1100)
¢g=50 MW =17 x 108 Biu/hr

o A7) 1S GIVEN IN FIG. 1.5
w
Q 10,000
[T
@
]
v
@
wi
.
(%)
=
<I
@
i 5000
5 N
[TV ]
I .
‘f \

0

1200 1300 1400 1500

7., BULK FUEL SALT TEMPERATURE (°F)

Fig. 1.10. Heat transfer area required for thimble surface
temperature of 1100°F.
[ 1]

"ORNL-DWG 72-13710

1200 l

METAL SURFACE TEMPERATURE 7m

i
; — 1 1050°
1000 ,//1000"!-'
\ \

& 800 ~J
wl \
x
-
’_
<J
W
W 600
=
Wl
—
b
2
% 400 — %zUT(Tm“TNuK)
Ut~ 280 Bty hr=! f172 °F !
4 TAKEN FROM FIGURE 1.10
200
0
1200 1300 1400 1500

T+ BULK SALT TEMPERATURE (°F)

Fig. 1.11. NaK temperature required to maintain a specific
surface temperature for various bulk salt temperatures.

1100°F metal the NaK temperature would have to be
about 960°F. This result is based on bare thimbles, that
is, no fins. However, if vertical fins are put on the
thimbles so that the number of thimbles can be
reduced, the heat flux will be greater and the NaK
temperature will have to be substantially lower.

Because heat transfer through the thimbles is limited
primarily by the salt film coefficient, fins greatly
improve the heat-removal capacity of a thimble. A
standard analysis of fin efficiency shows that a fin
height of ' in. is near optimum. Calculations show that
l4-in.-thick fins spaced -4 in. apart increase the
effective heat-removal capacity of a thimble by approxi-
mately 50%.

The temperature rise of the NaK as it flows down and
up a thimble was determined by constructing two heat
balances on a thimble segment; one for downflow and
one for upflow. Based on a uniform heat flux Q, the
solutions to the two differential equations are presented
and plotted in Fig. 1.12. It can be seen that an
approximately uniform thimble temperature can be

achieved by choosing design parameters such that the
NaK temperature at the bottom is equal to that of the
upflow at the top (i.e., at the salt-free surface). Equal
temperatures (top and bottom) can be achieved by
choosing design parameters such that

2nr, U, NHAT/Q = 2,

where AT is the overall NaK temperature rise (see
nomenclature). Assuming that the heat-rejection system
does not impose additional constraints on the drain
tank design, the problem is now reduced to one of
choosing design parameters which satisfy all the pre-
vious relationships. It is desired, however, for the
number of thimbles &V and their height H to be as low as
practical and for AT to be as large as practical. The
latter is to provide adequate driving head with a
minimum elevation differential between heat source
and sink. Thus, in the previous equation the down-
comer perimeter 27r, and the overall heat transfer
coefficient U, between NaK upflow and downflow are
the only parameters which can be arbitrarily adjusted to
satisfy the equality. Current efforts are directed toward
determining the practical limits of each parameter so
that a definitive choice can be made.

Nomenclature
h = salt film heat transfer coefficient (Btu hr™' ft™ °F ')
x = distance down a thimble from the salt-free surface (ft)
Gr, = x3g BT — Tm)/vz, Grashof number
£ = acceleration of gravity (32.2 fpsz)
8 = salt volumetric expansion coefficient (°F—l )
v = salt viscosity (Ib ft ! hr 1)

Tw = bulk salt temperature

T,,, = thimble surface temperature
Pr = Prandtl number
5 = turbulent boundary thickness (in.)
¥ = horizontal distance into salt from thimble surface
v, = vertical salt velocity (fps)
" = proportionality constant (fps)
T(x) = Nak temperature (°F)
T; = cold NaK temperature at salt-free surface (°R)
To = hot NaK temperature at salt-free surface (°F)
AT =Ty - T;
. NN = number of cooling thimbles
H = heated length of cooling thimbles (ft)
2nr p = perimeter of downcomer in thimble (ft)

Up = overall heat transfer coefficient from hot to cold Nak
(Btuhr™' ft 2 °F 1)
10

ORNL - DWG 72—-1371

e | | | |
Tolx)-T, x AT (x 2
A ML TR, S rUNH———(_ - =)
To-T, H T TR T\ T o2
1.4 — (DOWNFLOW) 7
W ————-\
®
14 s —
w12 é s ~ 2mr el
D - e ’
= - - N
nq: o - \\ 2.22
w //’
=
w 1.0 2.00
(-
h'd
o
= 70
2 {.
< 0.8
o
'_
| &)
<
14
w
- 0.6
-
!
,\O
™~
S~ 0.4
x
I~
///
T.{x)- T AT ( x xe
0.2 L = UNH—(———-—)——‘
4 oT AT \F T 52
4 (UPFLOW)
. | | |
0 0.2 0.4 0.6 0.8 1.0

x/H ,FRACTIONAL DISTANCE FROM SALT SURFAGE

Fig. 1.12. Relative NaK temperature as a function of position in thimble.

Ug = overall heat transfer coefficient to NaK excluding salt
film coefficient

Q = total emergency dump heat load (50 MW)

1.2.2 Physics Calculations

As part of the industrial MSBR design study, Ebasco
is performing physics calculations to provide an inde-
pendent check on the breeding ratio reportedS by
ORNL for the reference design. To attain a high level of
independence Ebasco is using different sources of data,
different computer codes, and different mathematical
treatment from that used by ORNL. Specifically,

S. Conceptual Design Study“of a Single Fluid Molten-Salt
Breeder Reactor, ORNL-4541 (June 1971).

Ebasco is using ENDF/B data, whereas ORNL used the
XSDRN library; Ebasco is using the 3D Monte Carlo
code ESP for generating multigroup cross sections,
whereas ORNL used the 1D Sn code XSDRN,; finally,
Ebasco is using the multigroup, multidimensional neu-
tron diffusion code CITATION, whereas ORNL used
ROD, a multigroup neutron diffusion code which
synthesizes two dimensions via a buckling iteration
procedure. This system of codes offers not only a high
level of independence but rigorous mathematical treat-
ment as well. Because it does not have the capability to
search for the equilibrium salt composition, it is
inappropriate for routine MSR calculations. But for a
reactor having a well-defined composition, this limita-
tion does not constitute a handicap.
1.2.3 Chemical Processing

Continental Oil Company has developed engineered
process diagrams and mechanical arrangement drawings
for the chemical processing flowsheet supplied by
ORNL. This work represents the reduction of the
ORNL flowsheet to a practical engineering design in
which attention has been given to the details of cooling,
heating, shielding, construction, remote maintenance,
safety, recovery from accidents and process interrup-
tions, and replacement of major process components. A
systematic attempt has been made to include all
components and controls that would allow for normal
operation. Further, an attempt has been made to utilize
the materials that have the lowest possible cost for the
service conditions. Specifically, graphite, nickel, and
Hastelloy have been utilized in addition to molyb-
denum.

The process, equipment, and plant layout are de-
scribed in the Task 1 report! for the flowsheet shown
in Fig. 1.13. Continental Oil Company is presently
preparing conceptual systems design descriptions for
the plant.

1.3 XENON BEHAVIOR IN THE MSBR
FUEL SALT SYSTEM

H. A.McLain L.W.Gilley T.C. Tucker

The digital computer program MSRXEP (Molten-Salt
Reactor XEnon Poisoning) has been written describing
in detail the ' 3 Xe behavior throughout the MSBR fuel
salt system. This program incorporates the BUBBLE
program reported previously® describing in detail the
inert purge gas behavior throughout the MSBR fuel salt
system. The intent of this effort is to either confirm the
results of the simplified calculations of Kedl? describing
the noble gas behavior in the MSBR or to make
improvements where required.

For the trial case using this program with the
following assumptions:

Coated graphite

Number of bubbles: 44 bubbles per cubic centimeter
of salt

Total helium dissolved in salt and in gas bubbles: 1.0
X 107® moles/cm?>

Gas separator efficiency: 90%

6. MSR Program Semiann. Progr. Rep. Feb. 29, 1972,
ORNL4782.

7. MSR Program Semiann. Progr. Rep. Aug. 31, 1968
ORNL-4344, pp. 712—74.

11

the following results were calculated:
Poison fraction: 0.0038
Average void fraction: 0.0055
Average bubble diameter: 0.0646 cm

Average bubble mass transfer coefficient: 0.0166
cm/sec.

The following mass 135 fission product decay chain
along with the indicated half-lives were assumed for the
MSRXEP program:

lJSmxe
(15.6 min)
ISSSb - IZi.'n'l-e - 135[ Fd ]35Cs - I35Ba
(2.0sec) (29.0sec) (6.71 hr) (3.0 X 10® years) (stable)
135Ke £
(9.21 hr)

We assumed for the purposes of these calculations that
the antimony, tellurium, and iodine are “‘dissolved” in
the salt and do not migrate to the bubbles, graphite, or
the metal walls. We reasoned that the half-lives of the
antimony and tellurium are sufficiently low that the
amount of these materials leaving the fuel salt other
than by radioactive decay is insignificant. All chemical
evidence so far indicates that the iodine remains
dissolved in the fuel salt and does not migrate into the
graphite. The xenons have a very low solubility in the
salt and tend to migrate into the graphite and the gas
bubbles in the salt. We assumed that the cesium is stable
because of its very long half-life. The yields of '*3Sb
and ' % Xe during the fissioning process have not been
established with complete confidence and depend on
the fissile material being used. For the initial calcula-
tions we assumed that these yields were 5.05 and
1.11% respectively. Also, there is some question about
the branching ratio of the '3°1 decaying to '*°™Xe
and '35 Xe. We assumed that 30% of the ' >®1I decays to
135m¥e in the initial calculations.

Material balances in the form of first-order linear
differential equations were written for each of the
isotopes present in the fuel salt indicated in the decay
chain shown above. Similar material balances were
written for the xenons that had migrated into the gas
bubbles circulating with the fuel salt. These first-order
differential equations were then solved simultaneously
throughout the fuel salt loop by an iterative process on
the digital computer until a unique solution was
obtained.

As was done by Kedl,7 we assumed that the rate of
migration of the xenons within the graphite and the
pyrolytic coating was determined by the Knudsen
diffusion of these gases within the graphite. We assumed
BeFa AND ThFq4

ORNL-DWG 72-13712

SALT MAKE-UP Ha-HF Ha—HF
264 2004
.88 .88 l
T 08 0.8 5o o.88
REACTOR SALT R-101 W-102 - W-103 ' R-104 R-406 R-107 F-108 | REACTOR SALT
RS REACTOR SALT 0.68 Po SALT RE saLT | Do109 UFg REACTOR SALT s SALT
FROM REACTOR FLUORINATOR EXTRACTOR EXTRACTOR | SaLT REDUCTION PURIFICATION | SALT FILTER | TO REACTOR
DRAIN TANK Hz-HF ‘ / SALT DRAIN TANK
x 8i i l Bi REACTOR
71 |8 1 Lo it DAY
a3
Foo|¥ P (02) (o.2) | -
UFs } 12.39) 2 2
REDUCTION C0:41¢ 1 min /day
Hz ' Licl W-201
RE - LEGEND:
TRANSFER
EXTRACTOR > ) FLOW, gom
0.1 Bi-Li [/ FLOW, scth
MIXTUR
Bi ofBOM () mole % LITHILUM
H
§ 5 ) PERIODIC FLOW. 5 gpm
Bi-Li MIXTURE Bi-Li MIXTURE 15 miny/5 days FOR 1Smin EVERY 'S doys
FROM FROM
HOLD TANK ) HOLD TANK
LiCl
(Cs0) ' (5 ) -
Smin/day e 15 min/day
N Fa
X-204 w-202 - w-203 x-205 wl:’egz 133 g/day OF UFg
= RE 2 RE 2 ‘ I RE 3 RE 3 o] " ——>
HEAT EXCHANGER EXTRACTOR Lic EXTRACTOR HEAT EXCHANGER YFefa BED TO PRODUCT CYLINDER
(50 ) & & ‘
Bi 1.1 hr /doy
XD 2§ -
5 min/d Hp=HF
min/day Bi
(50 (5 15 min/day |
m @ 24 hr /68 doys
Hz-HF [ et TN J I
[
_ 1 min/day } 0.68 Pa SALT A\ ‘ N .
]
; P RE SALT
B N o SALT 5-303 | -~ - - D-402 R-403
R-301 R-302 / R-301 RE SALT RE SALT
Pa-RE 2 Pa SALT : Pa — RE 3 Y 2:4 ) oLB-0p a3tk
HYDROFLUORINATOR |ug— FLUCRINATOR Hy-HF DECAY TANK Y0.38 ) HYDROFLUORINATOR TANK FLUBRINATOR
[ 2 / { — [. o RE SALT \
Bi @ " 15 min /day l RE SALT 10 n,
1 24 hr /68 days STORAGE
HF RECYCLE Bi
F2 & R~ ,
T L0 e Po SALT e 60 min /68 days 60 min/68 days
REDUCTION
RECYCLE Bi F,
TO HOLD
TANK
0.68 TAKEN FROM CONOCO DWG
Hy SD4700-42-1-C

Fig. 1.13. Fuel salt chemical processing plant block flow diagram.

4!
that the migration of '35™Xe and of '*%Xe was
independent of each other and that the decay of
135Mye to ! 3% Xe in the graphite did not influence the
rate of migration of '*%Xe from the fuel salt to the
graphite. These isotopes lose their identity in the
graphite either by radioactive decay or by reacting with
the neutrons in the reactor.

When the unique solution of the material balances for
these isotopes is obtained, the distribution of these
isotopes throughout the fuel salt loop is then known.

From the distribution of ! ** Xe within the reactor core,
the poison fraction due to this isotope, the ratio of the
neutrons absorbed in '**Xe to those absorbed in the
fissile material, can be determined. We assume that all
of the '3°™Xe that migrates into the graphite was
decayed to !'3°Xe when we calculate the poison

13

fraction. This should give a high value for the poison
fraction in the reactor.

1.4 HYBRID COMPUTER SIMULATION
OF THE MSBR

O. W. Burke

A hybrid computer simulation model of the reference
1000-MW(e) MSBR was developed. The model simu-
lates the plant from the nuclear reactor through the
steam throttle at the turbine. The simulation model is
being used to determine the dynamic characteristics of
the plant as well as to discover the problems associated
with the control of the plant. The model and results of
some test simulations are described in ORNL-TM-3767,
Hybrid Computer Simulation of the MSBR.
14

2. Reactor Physics
A. M. Perry

2.1 ANALYSIS OF HTLTR-MSR
LATTICE EXPERIMENTS

G.L.Ragan O. L. Smith

We have completed the calculations of k__ vs tempera-
ture for the lattice used in these experiments,' using
the cross-section library and calculational techniques
described in the last semiannual report.2 Results of the
calculations and a comparison with experiment are
given in Table 2.1. While the calculated and measured
values compare very favorably, it should be noted that
we do not claim such high accuracy for our calcula-
tions, since changes larger than the differences noted in
Table 2.1 would be produced by variations in cross
sections well within our present range of uncertainty.

Calculations relating to the other reactivity worth
measurements are well along, but are not yet complete.
These include reactivity measurements of three types:
(1) fuel salt constituents such as Li, Be, and F,
individually measured in a centrally located sample
position; (2) simulated control rods; and (3) variations
in lattice cell geometry and density.

2.2 NUCLEAR PERFORMANCE OF LEAD-COOLED
MOLTEN-SALT REACTORS
J. R. Engel

Some survey calculations have been performed to
estimate the nuclear performance of large graphite-
moderated molten-salt reactors cooled internally by

1. E. P. Lippincott, Megsurement of Physics Parameters for a
MSBR Lattice in the HTLTR, BNWL-1633 (January 1972).

2. MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL4782, pp. 18-21.

molten lead. Because of the anticipated complexity of
the core structure in such reactors, we considered only
systems in which the core was large enough (power
density low enough) to allow a 30-year graphite life. In
addition, it was presumed that continuous chemical
processing for fission product removal and protactinium
isolation would be available. Various combinations of
channel dimensions, lead:salt ratios, and thorium
concentrations were examined. The objective was to
determine if the performance potential of such reactors
is sufficiently superior to that of externally cooled
pumped-salt systems to justify the very substantial
effort required to develop lead cooling (see also Sect.
1.1.). Although the calculations were much less exten-
sive than would be required for a detailed design, they
did provide a basis for comparison. While breeding
would be possible in these particular thermal lead-
cooled systems with (in some cases) a somewhat lower
fissile inventory, it was concluded that the improve-
ment in nuclear performance relative to comparably
sized reactors with external cooling was at best marginal
and therefore insufficient to overcome the technolog-
ical disadvantages of the concept.

Calculations were made for each of the lead-cooling
concepts described briefly in Sect. 1.1. For the cases
involving lead flow through the entire core, we made
complete core calculations for start-up conditions with
233U as the only fissile nuclide. The XSDRN? program
was used to produce effective neutron cross sections
that were appropriate for each salt mixture and core
geometry in 27 energy groups from the XSDRN

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

Table 2.1. Comparison of measured and calculated k.. for HTLTR-MSBR lattice?

km

20°C 300°C

Ak., (300 = 1000°C)

627°C 1000°C

1.0291 £ 0.0012
1.0316

0.9976 + 0.0023

1.0127 £ 0.0010
1.0152

0.9975 £ 0.0010

k.. (measured)

k.. (calculated)

k. measured b
calculated

1.0065 + 0.0010
1.0069

0.9996 + 0.0010

1.0037 £ 0.0012
1.0038

—0.0090 £ 0.0016
0.0114

0.9999 + 0.0012 0.79 £0.15

2., for bare critical core of given composition.

bQuoted errors include only experimental errors; uncertainties in calculated quantities are undoubtedly greater than these (see

text).
Table 2.2 Nuclear performance of molten-salt reactors with internal lead cooling throughout the core

15

Downcomer ‘l;uel mixtulie Cc?re Sl;;s;;;;ac Breeding ratio Lead poison
Type et Mo oo ey e bes  fton

6 in. cylindrical 3:1 72/16/12 6,300 1.53 1.04 0.99 a

6 in. cylindrical 3:1 72/16/12 10,700 1.17 1.00 0.95 a

7/8 in. annular 3:1 72/16/12 10,700 1.16 0.97 0.92 0.15
% in. annular 3:1 71/9/20 7,500 1.66 1.05 1.00 0.10
% 6 in. annular 1.5:1 71/9/20 8,700 1.54 1.08 1.03 0.06
% in. annular 1.5:1 71/9/20 5,400 2.24 1.10 1.05 0.06

2Not computed.

123-group cross-section library.* The resultant cross
sections were used in a 27-group one-dimensional
diffusion calculation (also with XSDRN) to determine
the start-up breeding performance. The reactor was
represented in spherical geometry with a 2.5-ft-thick
reflector and constant composition in the entire core
region. We made no attempt to optimize the cores in
terms of neutron damage to the graphite by varying the
fuel fraction across the core, since it appeared that the
small changes in breeding performance associated with
this exercise would not affect the conclusions regarding
these reactors. Although the calculations did not
explicitly include the buildup of higher isotopes of
uranium or the poisoning due to the equilibrium
concentrations of fission products, we estimated that
the reduction in breeding ratio from these effects would
be 0.05 to 0.06. The results of the computations are
summarized in Table 2.2.

The first calculations were made for an array of core
cells with a 6-in.-diam cylindrical downcomer of lead
and salt in the middle of each cell. A fuel salt
containing 12 mole % ThF, was assumed, and the lead:
salt ratio in the downcomers was 3:1. Although this
concept was rejected on the basis that the salt inventory
would be excessive because of salt carry-under in the
disengagement region below the core, the nuclear
performance was also unacceptable because positive
breeding gains were not attained. Since it appeared that
excessive lumping of the thorium in the downcomer
was adversely affecting the breeding, the concept was
modified to provide an annular lead-salt downcomer in
each cell. (This also reduced the problem of salt
carry-under below the core.) However, the nuclear

4, J. L. Lucius, J. D. Jenkins, and R. Q. Wright, The INDEX
Data System: An Index of Nuclear Datq Libraries Available at
ORNL, ORNL-TM-3334 (March 1971).

performance was still unacceptable with only 12 mole
% ThF, in the salt and the 3:1 ratio of lead:salt in the
downcomers. The next step was to increase the thorium
concentration to see if thorium could be made to
compete more effectively with the parasitic neutron
absorbers. A concentration of 20 mole % was selected
even though we realized that the liquidus temperature
(>500°C) might be unacceptably high for practical
designs. As shown in the fourth case in Table 2.2, the
higher thorium concentration was beneficial, but the
nuclear performance was still marginal.

The detrimental effect of the lead in the core is
clearly shown by the lead poison fraction (absorptions
in lead as a fraction of absorptions in fissile material),
which varied from 0.1 to 0.15 depending on the
thorium concentration. In an effort to reduce the
neutron losses to lead, some computations were made
for a lead:salt ratio of only 1.5:1 in the downcomers.
This also led to improved performance, but the net
result was no better than for a reactor with salt
circulating outside the reactor vessel.’

In a final effort to minimize the poisoning effect of
the lead, a core was considered in which the lead was
confined to a region of downcomer cells surrounding a
core that contained only salt and graphite. It was not
obvious that the fluid mechanics of such a core could
be made acceptable, but it did represent the best
nuclear performance we could expect with internal lead
cooling. Calculations of the type described above
suggested that an effective breeding ratio near 1.1 could
be attained with a 1.5:1 lead:salt ratio in the down-
comers and 20 mole % ThF,. To estimate the potential
at lower thorium concentrations with better accounting
for fission products and higher uranium isotopes, we

5. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970,
ORNL4622, pp. 26-28.
reexamined an earlier calculation® for a pumped salt
reactor. A first-order perturbation calculation was
applied to a core with 14 mole % thorium in the fuel to
estimate the loss in breeding ratio due to lead in the
outer core region and the reduction in fissile inventory
associated with elimination of the external salt loop.
This calculation indicated that the breeding-ratio pen-
alty would be about 0.02, while the inventory would be
reduced by about 15%. These estimates were in
essential agreement with the performance’ estimated
from the calculation at 20 mole % thorium. Since none
of the above-described results indicates any clearly
superior performance for lead-cooled thermal reactor
systems, this study has been discontinued.

2.3 PLUTONIUM USE IN
MOLTEN-SALT REACTORS

H. F. Bauman

The plutonium that will become available from
light-water reactors in increasing quantities over the
next several decades could be utilized in molten-salt
reactors with a minimum of fuel preparation facilities.
One use for LWR plutonium could be as fuel for
molten-salt converter reactors (MSCR’s) that have
periodic batch processing. The other is as the start-up
fuel for molten-salt breeder reactors (MSBR’s) with
on-line processing. Over the past two years, we have
made some limited studies of the nuclear performance
of molten-salt reactors using plutonium in various ways
and have calculated the economics of these uses. In the
course of these studies we established the need for a
computer program that could calculate the time-
dependent behavior of a fluid-fueled reactor, including
the effects of significant changes in the neutron
spectrum caused by changes in the fuel composition,®
and we developed the HISTRY-2 option of the ROD
code to meet this need.” In the current six-month
period, the HISTRY-2 option has been made opera-
tional, and the plutonium studies have reached a
culmination.

Our results indicate that plutonium is 4 desirable fuel
for MSCR’s: if assigned a value of 9.9 $/kg fissile,
plutonium gives somewhat lower fuel cycle costs in an
MSR than does enriched uranium. Start-up of an MSBR
on plutonium is feasible but appears to offer little or no
cost advantage. The studies and results are described in
detail in the following sections.

6. MSR Program Semiannu. Progr, Rep. Aug. 31, 1971,
ORNL4728, pp. 21-25.

7. MSR Program Semignnu. Progr. Rep. Feb. 29, 1972,
ORNL4782, pp. 23-25.

16

2.3.1 Fuel for Molten-Salt
Converter Reactors

Molten-salt converter reactors are basically similar to
molten-salt breeders. In the MSCR, however, the salt
may be processed only at intervals of several years
instead of being processed continuously as in the
MSBR. The losses due to fission products and protac-
tinium in a batch-processed MSCR reduce the breeding
ratio below 1.0, but even so most of the fissions over
the lifetime of the reactor are in bred 233U,

In the simple batch process assumed in our studies,
after six equivalent full-power years (efpy) of operation
the uranium is recovered by the fluoride volatility
process, and the remaining salt, containing fission
products and any plutonium that may be in it, is
discarded. The cycle is then repeated. (It is presently
uneconomical to recover plutonium from MSR salt.)
For our MSCR studies we used a simplified reactor
model consisting of a spherical graphite core with a
0.12 salt volume fraction surrounded by a 78.4-cm
reflector with a 0.01 salt fraction. (Earlier studies had
shown that the spherical model adequately represents
the actual cylindrical core.) The core diameter was
adjusted in each case to give a peak damage flux
equivalent to a graphite life of 30 years in a
2250-MW(1t) plant operating at a 0.8 load factor.

The initial fissile loading and subsequent feed for an
MSCR could be enriched uranium, plutonium, or 233U
from other reactors. Lower fissile loadings are required
with plutonium because of its larger cross sections. The
presence of much *#°Pu tends to harden the neutron
spectrum and increase the critical loading. However,
this effect can be offset by reducing the concentration
of the main fertile material, thorium, to maintain a
well-thermalized spectrum.® The conversion ratio that
is approached asymptotically in an MSCR depends
mainly on the fertile loading and only slightly on the
choice of fissile feed. The lower initial fissile loading
and the lower assigned value for LWR plutonium tend
to reduce fuel-cycle costs relative to those for uranium-
fueled MSCR’s. Although the plutonium cannot be
recovered economically from an MSCR, discard of large
amounts can be avoided by switching to uranium feed
for the last two or three years of each cycle, allowing
most of the plutonium to burn out. The uranium
recovered in the batch process may be used in the next
cycle, or it may be sold.

We have examined the performance of molten-salt
converter reactors with various combinations of initial

8. Ibid, pp. 24—25.
17

Table 2.3. Feed compositions assumed in MSR studies

Plutonium atom fraction

Description Key

239Pu 240Pu 241Pu 242Pu
First-cycle LWR plutonium? Pu(l) 0.60 0.24 0.12 0.04
Second-cycle LWR plutonium? Pu(2)- 0.40 0.32 0.18 0.10

Uranjum atom fraction

233y 234y ?35U 236y 238y,
Enriched uranium U(e) 0.93 0.07
Recycled MSCR uranium U 0.78 0.178 0.032 0.01
Equilibrium MSBR uranium U(g) 0.664 0.225 0.058 0.053

2Composition typical of light-water reactor discharge after one or two cycles.

Table 2.4. Composition and liquidus temperature
of fuel carrier salts used in MSCR
and MSBR calculations

Composition (mole %) Liquidus
temperature?

ThF4 BeF, LiF 0)
MSCR salt

14 17 69 495

12 20 68 485

10 . 23 67 475

8 27 65 465

6 30 64 450

4 32 64 440

3 32 65 440

0 33 67 460
MSBR salt

8 20 72 485

10 18 72 490

12 16 72 500

AR. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor
Materials, ORNL-2548, p. 80 (November 1959). .

fissile loading, primary feed (first part of each cycle),
and secondary feed (latter part of each cycle).
Plutonium and enriched uranium feed. A series of
cases was run in which plutonium discharged from an
LWR after one cycle was the primary feed and enriched
uranium the secondary feed. The feed compositions are
given in Table 2.3. (The key abbreviations shown in this
table are used subsequently to identify the feed
compositions.) Based on previous work, we selected a
reactor lifetime consisting of four cycles, each 6 efpy in
length with feed switched at 4 efpy. In this study with
the salt fraction held fixed, the thorium concentration in
the carrier salt is the dominant parameter determining
the moderator ratio, fissile inventory, and the conver-

sion ratio of the reactor. The fuel carrier salt composi-
tions used are shown in Table 2.4. The salt composition
is specified for a given reactor by giving the thorium
concentration. _

The results of six cases, arranged in descending order
of thorium concentration, are shown in Table 2.5.
Based on the plutonium fissile loading study reported in
the last semiannual report,® in cases involving higher
thorium concentrations, we used a lower thorium
concentration in the first cycle than in subsequent
cycles to reduce the initial fissile loading. The results
show that thorium concentration has little effect on the
core diameter required for a 30-year graphite life. With
decreasing thorium concentration, the fissile inventory,
as indicated by the initial loading, decreased, but the
conversion ratio also decreased and the lifetime fissile
requirements increased. These trends are reflected in
the fuel cost breakdowns, which show the fissile
inventory cost decreasing and the burnup cost in-
creasing. The result is a broad minimum in the fuel cost
at a relatively low thorium concentration, in the range 8
to 4 mole %, as shown by cases A57, A58, and A62. Of
these three cases, A57, with an 8% thorium concen-
tration for all but the first cycle, has the lowest lifetime
fissile requirement and may be considered a near-
optimum case. Because of a high conversion ratio (near
0.90), the fissile requirement (defined as the fissile
material purchased over the reactor lifetime less that
recovered at the end of life) for this case is less than
3000 kg for the entire lifetime.

The important nuclide inventories for case A57 are
plotted as a function of time in Fig. 2.1. (To avoid
overcrowding in this figure and those following, not all
nuclides are shown for every cycle.) Relatively large
concentrations of plutonium are required in the first
cycle, before the 233U has built in. Some additional
plutonium is required to start the second cycle, as the
thorium concentration was raised from 6 to 8 mole %,
but little plutonium is required in the last two cycles,
when the reactor is nearly self-sustaining. For the last
two cycles, more fissile uranium is available from the
previous cycle than is required for criticality at the
beginning of the cycle. In these calculations, the
uranium in excess of the cycle start-up requirement is
held and used as feed for as long as it lasts. Plutonium
feed is then used until the end of 4 efpy. This accounts
for the zero plutonium concentration at the beginning of
the last two cycles. In each cycle, the >*°Pu concen-
tration begins to decline abruptly, and the *?°U
concentration begins to rise as the feed is switched from
plutonium to enriched uranium at the end of the fourth
year.

Plutonium and recycled uranium feed. The cases just
described require the purchase of enriched uranium in
addition to plutonium. In a sense, enriched uranium is

18

interchangeable with uranium bred in the reactor, since
both are recoverable at the end of every cycle. Any or
all of this recovered uranium could be withheld (instead
of being loaded at the start of the next cycle) to be
used in place of enriched uranium in the latter part of
the cycle. Of course, any uranium used in the first cycle
would have to come from an ‘external supply, but this
requirement could be repaid later out of the uranium
recovered at the end of the cycles. To explore the
advantages of such an arrangement, we have calculated
several cases in which exactly enough uranium is
withheld from an MSCR to supply all the uranium used
in this reactor at other points in its lifetime. In physical
terms, we have in effect established a “borrow pile”; in
economic terms, we have established a market in
recycled MSCR uranium. When recycled uranium is
available in this manner, it can be used not only as feed
during the last years of each cycle but also in the initial

Table 2.5. Lifetime-averaged performance of MSCR designs using plutonium
feed for various thorium concentrations

Lifetime: four 6-efpy cycles

Feed:? Primary, plutonium, first cycle, Pu(l)
Secondary, 235U, 93% enriched, U(e)
Switch from primary to secondary at
4 efpy in each cycle.

Case identification A56.1 AS52.3
Thorium concentration,? mole %
Cycle 1 10 8
Cycles 2 to 4 14 12
Core diameter, cm 1064 1064
Fissile material balance, kg
Initial loading, plutonium 1265 813
Purchases
Plutonium 3720 3496
235y 230 561
Discard, plutonium 176 107
Recovery at end of life
233 2923 2665
235y 230 223
Net fissile requirement® 797 1170
Conversion ratio,d lifetime av 0.982 0.963
Fuel costs,® mills/kWhr
Fissile inventory 0.635 0.564
Salt inventory 0.069 0.065
Fissile burnup —0.012 0.015
Salt replacement 0.140 0.134
Total fuel costs 0.83 0.78

A37.23 AS57 A58 A62
6 6 6 4

10 8 6 4
1062 1064 1064 1060
582 586 586 390
3285 3196 3374 4417
1161 1610 2050 3103
71 76 77 59
2154 1672 1288 845
318 372 409 461
1973 2763 3728 6231
0.927 0.893 0.851 0.752
0.478 0.395 0.318 0.219
0.062 0.063 0.063 0.060
0.067 0.116 0.172 0.306
0.128 0.123 0.118 0.112
0.73 0.70 0.67 0.70

aRefer to Table 2.3 for feed compositions.
bRefer to MSCR salt compositions, Table 2.4.
€Assuming discarded plutonium is not recoverable.

9Nuclear conversion ratio, not considering plutonium discard or fissile processing loss.

€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 yearvl, compounded quarterly, and inventory charge rate = (0.132 year-'. Values of
11.9 $/g 233U, 13.8 $/g 222U, and 9.9 $/g fissile plutonium were assumed.
19

ORNL-DWG 72-14144

| Pu(1) fa—Ute)—~ | Pull) f—U(e)—=f e—Pul)—=—ute)—] Fe—Pu( 1) —f=—U(e)—=]
2000 T T 1 - T
~80,000 kg Th ~ 110,000 kg Th ~{10,000 kg Th -~110,000 kg Th
{6 mole% ThF,) (8 mole % ThFy) {8 mole % ThF,) {8 mole % ThF,)
Il
! 2335 _, 233
233p,, , 233 Pat+ U 23%pg 1+ 2y
1600
= 1200 / INVENTORIES
= 23py B3I A
bl
[ /
g /
=
L
>
£ 800 23—
| v
234
/ ] Pl
239, a4 /
200 240p,, 242p, % s
24 2V e '235 235 236,
234
s = 20y eaip, 22py [T~ e S
3 236, Pu /1 236y | ] 7 235
— =y 239 2%y
o ] — Pu e
O 2 ) 6 6 8 10 12 12 14 16 18 18 20 22 24

OPERATING TIME {equivalent full-power years}

Fig. 2.1. Principal nuclide lifetime inventories for MSCR case A57, with. first-cycle plutonium and enriched uranium feed.

loading to avoid a high plutonium concentration and an
excessively hard neutron spectrum. We investigated two
repayment schemes. In one the borrowed uranium is
repaid immediately at the start of the second cycle. In
the other, the borrowed uranium is repaid in equal
installments at the start of the remaining three cycles.

These possibilities are illustrated in the seven cases
shown in Table 2.6. In every case, the feed is switched
to recycled uranium from the stockpile at the end of
the third efpy of each cycle. This uranium is returned to
the stockpile at the end of the cycle.

The first six cases in Table 2.6 form a series in which
the thorium concentration was held fixed at 10 mole %.
In the first case, the reactor was started up entirely on
plutonium. In the next two cases, about 500 and 1000
kg of recycled uranium were borrowed and used in the
initial loading, and the borrowed uranium was repaid at
the end of the first cycle. In the next case, the reactor
was also started up with 1000 kg of borrowed recycled
uranium, but this was repaid over the remaining three
cycles. The next case was started with about 1500 kg of
borrowed uranium, repaid over three cycles. In the
sixth case, A55.3, the amount of uranium borrowed
initially was chosen so that, when it was repaid in three
installments, the plutonium loading would be approxi-
mately equal in all four cycles.

The last case was run with a 6 mole % thorium
concentration, for comparison with the minimum fuel-
cost case from the previous series. As in case A55.3, the
amount of uranium supplied initially was adjusted to

give approximately the same plutonium concentration
in each cycle.

Among the six cases with the same thorium concen-
tration, there are only minor differences in lifetime
performance despite the large differences in the com-
position of the initial fissile loadings. The most visible
effect is that using more plutonium early in the lifetime
lowers slightly the fuel cost. However, there may be
advantages for a more uniform loading over the reactor
lifetime that do not appear in the data. For example,
the zoning of the core to reduce the peak damage flux
is probably more effective when the cycles are more
nearly uniform.

It would appear from these results that the reactor
designer has a wide latitude in the choice of feed modes
without greatly affecting the reactor performance. This
is further illustrated by comparison of case A60.1 with
A58 in Table 2.5. In case A58, the first loading was all
plutonium, and the switch feed was enriched uranium.
The conversion ratios and fuel costs for these two cases,
however, differ by less than 1%.

The inventories of the important nuclides for case
A60.1 are plotted as a function of time in Fig. 2.2. The
four cycles are very nearly alike. The major difference
from cycle to cycle is the increase in inventory of
234y, 2354, and 23°U, which do not reach equilib-
rium in one reactor lifetime.

Start-up with enriched uranium or recycled plu-
tonium. In all the foregoing cases, we assumed that
initial fissile loading was either plutonium discharged
20

Table 2.6. Lifetime-averaged performance of MSCR designs with plutonium
and borrowed recycled uranium feed

Lifetime: four 6-efpy cycles
Feed:? Primary, Pu(1)
Secondary, U(r)
Switch at 3 efpy in each cycle

Case identification A6l AS53.4 AS54.1 AS59 AS50.3 AS55.3 A60.1
Thorium concentration,? mole % (all cycles) 10 10 10 10 10 10 6
Number of uranium return cycles 1 1 3 3 3 3
Core diameter, cm 1062 1064 1062 1062 1062 1062 1066

Fissile material balance, kg
Initial loading, by cycle

Plutonium
1 1154 813 568 - 568 290 241 240
2 94 311 530 163 226 241 254
3 0 0 16 129 215 233 247
4 0 0 0 125 213 229 245
Uranium
1 0 495 991 991 1479 1569 758
2 1805 1412 1061 1729 1636 1611 774
3 2068 2060 2022 1842 1689 1654 812
4 2125 2119 2115 1884 1713 1674 829
Purchases, plutonium 3614 3603 3546 3529 3521 3518 4712
Discard, plutonium 120 99 84 79 76 76 33
Recovery at end of life
2066 2067 2059 2059 2050 2048 970
235y _ 195 191 188 178 168 166 109
Net fissile requirement® 1353 1346 1298 1293 1302 1303 3634
Conversion ratio,? lifetime av 0.955 0.953 0.954 0.953 0.953 0.952 0.857
Fuel costs,® mills/KWhr
Fissile inventory 0.509 0.518 0.528 0.535 0.545 0.546 0.336
Salt inventory 0.070 0.070 0.070 0.070 0.070 0.070 0.064
Fissile burnup 0.029 0.029 0.027 0.027 0.028 0.028 0.157
Salt replacement 0.131 0.131 0.131 0.131 0.131 0.131 0.119
Total fuel costs 0.739 0.748 0.755 0.762 0.773 0.774 0.676

" Note: Refer to explanatory notes following Table 2.5.

ORNL-DWG 72-14145

——rpum I utr) o feePuth) } ufr) L f——putn | utr) | | P 1)U () ——]
1800 T T T T T T I i
~B0,000 kgTh ~80,000 kg Th ~ 80,000 kg Th ~B0,000 kg Th
(6 mole % ThFy) {6 mole % ThFy) {6 mole % ThE,) (6mate % Th,)
1600 233, 233 233p, 4233, 233, , 233 233p, 4 233
1400 / / /
1200 " / / /
<z /
% 1000 / / /
x / / / /
>
4 800 [7 7 4 y
z 234 234
600 234y s //
234, 23%p,, / / 239p, ]
/ 239 1 /
400 —aze, —1 Pu
Pu 24|Pu - S— o4t 240 24|p
240p,, / 2azp, 2adp,, ,29lp, 242p, 240p, Pu 2a2p, Py u 242y,
— U S—
200 235, 236 %zw 235 /& /235y S 236y
— e~ — fi o - ZBGU} / ] 235U/
O e
[¢] 2 4 6 6 8 10 1212 14 16 18 18 20 22 24

OPERATING TIME ({equivalent full-power years)

Fig. 2.2, Principal nuclide lifetime inventories for MSCR case A60.1, with first-cycle plutonium and recycle uranium feed.
from an LWR after one cycle, Pu(1), or uranium
borrowed from other MSCR’s, U(r). Other possibilities
for the initial loading and primary feed are enriched
uranium, U(e), or plutonium that has been recycled in
an LWR, Pu(2). The series of cases shown in Table 2.7
compares the use of the three different materials: U(e),
Pu(1), and Pu(2). (Refer to Table 2.3 for the composi-
tions assumed for these materials.)

We selected similar initial conditions for the cases
with the various feed materials; no attempt was made to
optimize the conditions for each feed material. The first
two cases with enriched uranium feed are to be
compared with the third case with first-cycle plutonium
feed. Enriched uranium was the secondary feed in all
the plutonium feed cases. Cases A41.1 and A37.23
appear most nearly alike, having the same thorium
concentration in each cycle. However, the plutonium
feed case has additional fertile material in the form of
249Py, and may be better compared with case A42,
with 10 mole % thorium in the first cycle. The

21

difference in first-cycle thorium concentration affects
the conversion ratio; of the two uranium feed cases, one
has a higher and the other a lower conversion ratio than
‘the plutonium feed case. Both uranium feed cases have
significantly higher fuel costs than the plutonium feed
case.

We can make two comparisons of first- and second-
cycle plutonium as feed materials. Cases A37.23 and
A51.12 each have 6 mole % thorium in the first cycle
and 10 mole % in subsequent cycles. Cases A58 and
A63 have 6 mole % thorium in all cycles. The

" calculations show very little difference in the perfor-
mance obtained with first- and second-cycle plutonium
feed. Second-cycle plutonium appears to have a slight
edge, giving fuel cycle costs that are lower by 0.01 to
0.02 mill/kWhr.

None of the cases discussed thus far take into account
the neutron reactions of the transplutonium nuclides.
G. Alesii studied the buildup of nuclides in the chain
from 2%2Pu through 24°Cm as a function of time at

Table 2.7. Lifetime-averaged performance of typical MSCR designs
with various feed materials ‘

Lifetime: four 6-efpy cvcles
Switch to secondary feed, where
applicable, end of efpy: 4

Case identification A42 Ad1.1
Feed?
Primary U(e) U(e)
Secondary None None
Thorium concentration,® mole %

Cycle 1 10 6
Cycles 2 to 4 10 10
Core diameter, cm 1026 1020

Fissile material balance, kg
Initial loading 2369 1453
Purchases
Plutonium 0 0
Uranium 3810 4569
Discard, plutonium 18 20
Recovery at end of life
233y 1985 1978
235y 374 386
Net fissile requirement€ 1451 2215
Conversion ratio,? lifetime av 0.939 0.907
Fuel costs,? mills/kWhr
Fissile inventory 0.555 0.518
Salt inventory 0.064 0.058
Fissile burnup 0.070 0.113
Salt replacement 0.122 0.118
Total fuel costs 0.811 0.807

A37.23 A51.12 A58 A63 Ab4
trans-Pu

Pu(l) Pu(2) Pu(l) Pu(?) Pu(2)
U(e) U(e) U(e) U(e) U(e)
6 6 6 6 6

10 10 6 6 6
1062 1062 1062 1062 1062
582 700 586 700 709
3285 3123 3374 3190 3263
1161 1187 2050 2089 2198
71 84 77 91 104
2154 2155 1288 1286 1287
318 316 409 414 423
1973 1839 3728 3580 3751
0.927 0.931 0.851 0.857 0.851
0.478 0.471 0.318 0.311 0.322
0.062 0.062 0.063 0.063 0.063
0.067 0.061 0.172 0.166 0.175
0.128 0.128 0.118 0.118 0.118
0.734 0.721 0.676 0.658 0.678

Note: Refer to explanatory notes following Table 2.5,
several flux levels.® This study showed that the trans-
plutonium chain is always a net absorber of neutrons,
although for very long times fissions in *#*Cm and the
241Pu produced by way of curium returned nearly as
many neutrons as are absorbed. For a typical MSR flux
over a six-year cycle, the net absorptions in the
transplutonium were about 0.4 of the absorptions in
242py. In case A64, we simulated the effect of the

transplutonium chain by allowing 0.4 additional absorp- .

tion per neutron absorbed in 2*2Pu. As compared with
case A63, the conversion ratio was reduced by about
1% and the fuel cost increased by about 0.02 mill/
kWhr. Since this effect is proportional to the amount of
242Py in the reactor, we would expect about half this
effect for first-cycle plutonium feed, where the 2*2Pu
concentration is about half. We conclude from these
calculations that the effect of the transplutonium
nuclides, while not entirely negligible, is small enough
that explicit representation of these nuclides in the
ROD-HISTRY calculation is not necessary.

The inventories of the principal nuclides in case A64
are given as a function of time in Fig. 2.3. The pattern
is not greatly different from that of the first-cycle
plutonium feed case given in Fig. 2.1, except that the
higher plutonium nuclides, particularly 2% °Pu, are more
prominent.

To sum up this study of the use of plutonium in
MSCR’s, we conclude that plutonium, enriched ura-
nium, and recycled uranium are essentially interchange-
able as feed materials. Because of the high conversion

9. G. Alesii, Oak Ridge Associated Universities participant
from

the University of Arizona, internal communication

22

ratio in these reactors, their performance is dominated
by 233U no matter what feed material is used or at
what point in the reactor lifetime it is supplied. Either
first- or second-cycle plutonium has about a 0.1
mill/kWhr cost advantage over enriched uranium, due to
a lower assigned value for the plutonium feed.

2.3.2 Start-up of an MSBR
on Plutonium Fuel

Although molten-salt breeder reactors will eventually
produce excess 233U, only small quantities of 233U
may be available at the time that the first generation of
MSBR’s are brought on the line. To provide the initial
fissile inventory for these reactors, we must use the
available fuels: enriched uranium or recycled pluto-
nium. Both are chemically stable in molten-salt systems,
and from a nuclear standpoint they are essentially
interchangeable. As shown in the MSCR study reported
above, plutonium has the advantage that the initial
loading is only about half that required for uranium in a
well-thermalized spectrum. On the other hand, pluto-
nium start-up of an MSBR would be inconvenient for
two reasons. First, in the present flow sheet for
chemical processing, plutonium is extracted from the
fuel salt and is not recovered in the MSBR protactinium
process. Second, recycled plutonium carries enough
fertile material along with it, in the form of 2*°Pu, to
require reducing the usual thorium concentration of the
fuel salt. Neither of these disadvantages applies to the
start-up of enriched uranium. However, if supply or
economic conditions should strongly favor the use of
plutonium over enriched uranium, the MSBR could be

(1970). started up and operated for several years without
ORNL-DWG 72-14146
00 b= Pu(2)———f=—U(e}—] b—Pu(2)—f—ute)— e——Pu(2) (e} —] le—Pu(2)——f=—ute)—]
~80,000 kg Th ~80,000 kqTh ~80,000 kgTh ~80,000 kgTh
1400 (6 mole % ThFy) (6 mole % ThFy) (6mole % ThFy) (6mote % ThFy)

o
s
O

1200 zsspfizV 7@?23%\

233pg 4233y

|
233pgy 4233y

INVENTORY (kg}
®
Q
o

234
600 Y
235
400 u
200 242p,
: 240Pu
)
0 2 4 66 8 10 12 12 14 16 18 18 20 22 24

OPERATING TIME {equivalent full-power years)

Fig. 2.3. Principal nuclide lifetime inventories for MSCR case A64, with second-cycle plutonium and enriched uranium feed.
processing, as in the first cycle of some of the MSCR
cases.

We have made a few exploratory calculations on the
start-up of an MSBR using this scheme. We began by
making an equilibrium calculation of the reference
MSBR, using the same spherical core model that was to
be used for the HISTRY calculations. The resuiting
equivalent calculation, case SCC219, gave results in
good agreement with an earlier calculation using a
cylindrical approximation (case CC207, reported pre-
viously 7).

It was necessary to make the HISTRY calculation in
two stages, the first with plutonium feed and no
processing and the second with continuous processing.

23

For the first stage, we picked somewhat arbitrarily a

duration of 39 equivalent full-power months (efpm).
This coincides with the planned graphite life in the
reference MSBR. It is not necessary that the two should
coincide, however, since the preliminary feed cycle may
be ended at any time by starting continuous processing.
Plutonium was used for the initial fissile inventory and
as feed for the first three months. For the last 36
months, the feed was switched to equilibrium uranium,
of the composition given in Table 2.3. For the first trial
calculation, we reduced the thorium concentration
from the 12 mole % normal in the MSBR to 10%. This
gave an initial critical loading of 1050 kg of fissile
plutonium. This was a relatively high loading for

Table 2.8. MSBR start-up with plutonium and
equilibrium recycle uranium feed

Initial Approach to Lifetime rer
yer s total or Equilibrium
cycle equilibrium
average
Case identification P3A.1 P3iB SCC219
Processing Batch Continuous Continuous
Thorium concentration,? mole % 8 12 12
Time, efpm 0-39 39-279 0-279
Feed?
Primary (first 3 months) Pu(l) U(q) None
Secondary (last 36 months) U(q)
Fissile material balance, kg
Initial loading
Plutonium 528 0
Uranium, from initial cycle 0 1099
Uranium, feed 0 223
Total 528 1322
Purchases, gross
Plutonium 896 0 896
Uranium 722 318 1040
Total 1618 318 1936
Production, Uranium 0 1493 1493
Recovery at end of life, Uranium (1 099)d' 1366 1366 1366
Discard, Plutonium 6 0 6
Net fissile requirement® 520 —1442F -923
Conversion ratio® 0.857 1.079 1.048 1.076
Fissile costs,$ mills/kWhr
Fissile inventory€ : 0.215 0.358 0.338 0.350
Fissile burnup or production 0.169 —0.111 -0.0722 -0.112
Total fissile cost 0.384 0.247 0.266 0.238

4Refer to MSBR salt compositions, Table 2.4.
bRefer to feed compositions, Table 2.3.
¢Excluding material in processing.

dRecycled.

€0On the same basis as in Table 2.5, notes ¢ and d.

TCalculated as (recycle + purchases) — (production + recovery).
ZExcept for the equilibrium case, on the same basis as in Table 2.5, note e. The equilibrium cost is
based on the same material values and inventory charge rate, but assumes steady-state rates of

expenditure.
plutonium fuel and indicated some hardening of the
spectrum. We therefore reduced the thorium concen-
tration to 8 mole % for the second trial and obtained an
initial loading of 530 kg of fissile plutonium. The
results of this calculation are given as case P3A.1 in
Table 2.8. The second stage of the HISTRY calculation
was started with uranjum available at the end of the
first stage. Equilibrium uranium was supplied as needed
for criticality and as feed until the reactor became
self-sustaining. Continuous processing by the reductive-
extraction—metal-transfer process on a ten-day cycle
was simulated. The results of this calculation are given
as case P3B in Table 2.8.

The principal nuclide inventories from the two-stage
calculation are plotted as a function of time in Fig. 2.4.
The plot for the first 39 months closely resembles an
initial MSCR batch cycle, except that the duration is
shorter. We see that switching to uranium feed for the

24

last 36 months of this initial stage allows more than
enough time for the burnout of plutonium. In the
second stage (continuous processing), the nuclide inven-
tories start practically at their equilibrium values and do
not change much over the rest of the reactor lifetime.
The equilibrium inventories calculated in case SCC219
(indicated by points at the end of the plot) are in good
agreement with the end-of-lifetime values from the
HISTRY calculation. One advantage of this start-up
with plutonium and bred uranium is that the equilib-
rium concentration of *3°U is never exceeded. Con-
siderably higher >3®U concentrations would be ex-
pected in an enriched uranium start-up.

This example of a plutonium start-up requires 1040
kg of equilibrium uranium mixture, over a period of
4"/, calendar years, compared with only 1360 kg that
would be required if no plutonium were used. Several
ways could be considered to reduce the amount of

ORNL-DWG 72-14147

1600 | Ww “\fiv
~50,000 kg Th ~70,000 kg Th
{ 8 mole % ThF,) (12 mole % ThF,)
1400
233p, 4 233 .
/
1200 T
¥
1000 e
=
- 2335, 233 / g
x =
> >
% a
@ D —
S 800 :
& 3
2 o
600 / —
239p, ]
234U
vl °
400
240py y
200 . P T 242p, |
Pu 235
235 o $ ]
336 238y
o | JV\ A\A
0 1 2 3 4 5 6 7 L 23

OPERATING TIME {equivalent full-power years)

Fig. 2.4. Principal nuclide lifetime inventories for MSBR started with first-cycle plutonium and equilibrium uranium feed; cases
P3A.1 (0-3.25 efpy), P3B (3.25-23.25 efpy), SCC219 (equilibrium).
uranium and increase the amount of plutonium used in
the start-up. The plutonium feed time could be ex-
tended from 3 to 12 months with little penalty for the
discard of plutonium. The thorium concentration might
be increased to 10% in the preliminary cycle, but this
would significantly increase the inventory cost. Two
preliminary cycles could be used with, say, 8 and 12
mole % thorium. The uranium fed in the first cycle
could be paid back at the start of the second cycle, like
an MSCR case with borrowed recycled uranium (refer
to Table 2.5, case A52.3). |

2.3.3 ROD Code Modifications

" The HISTRY-2 option of the ROD code, in which the
reaction rate coefficients are recalculated at intervals, is

25

now operational. In addition to the extensive changes
reported last period, we have programmed HISTRY-2
to converge on the start-of-cycle critical concentrations.
The program iterates between MODRIC and HISTRY
until both converge on a single set of critical concen-
trations. This eliminates any bias from the estimate of
starting concentrations and gives more accurate initial
concentrations for cycles after the first.

Another new capability in HISTRY-2 is an option to
approximate neutron absorptions in transplutonium
nuclides by increasing the neutron absorptions in 2% 2Pu
by a fixed fraction, 0.4. The 2*?Pu burnout calculation
is not affected.
26

3. Systems and Components Development

Dunlap Scott

3.1 GASEQUS FISSION PRODUCT REMOVAL

3.1.1 Bubble Separator and Bubble
Generator :

The detailed designs of the bubble separator and
bubble generator for the gas system technology facility
(GSTF) were completed, and the drawings were issued
for fabrication. These units will be tested in the water
loop before installation in the GSTF.

Bubble Separator (C. H. Gabbard). The final design
for the bubble separator is shown in Fig. 3.1. This
design consists of a set of swirl vanes and a set of
recovery vanes separated by a 44-in.-long tapered
casing. The separated gas collects at a small-diameter
void on the center line and is removed through both the
swirl and recovery vane hubs along with 9 to 11 gpm of
entrained liquid. The separation efficiency of this
design was reported previously.'

Tests are currently in progress to determine the radial
and axial velocity and pressure distributions of the
vortex. Figure 3.2 shows these velocity and pressure
distributions for the design flow conditions at the two
locations indicated on the figure. The tangential veloc-
ity distribution at radii greater than the hub radius
(0.86 in.) was found to be that of a free vortex, that is,
the tangential velocity at a given radius was equal to a
constant divided by the radius. The velocity measuring
probe was found to. alter the tangential velocity
distribution in the vortex, particularly when inserted to
radii less than 0.8 in. This change in the tangential

1. MSR Program Semiannu. Progr. Rep. Feb, 29, 1972,
ORNL-4782.

GAS -LIQUID TAKE-OFF LINE

% in. OD x0042 in. WALL TUBING
- 48%, in

—1.72 in. diam

velocity distribution was indicated by the increase in
gas void static pressure difference from —91 to —71 ft,
shown in Fig. 3.2. The largest portion of this change
occurred after the probe was inserted to radii less than
0.8 in. Therefore, the unperturbed velocity distribu-
tions would be somewhat different from the measured
distributions, especially at radii below 0.8 in. Until a
better method of measuring this velocity is found, we
plan no further work in this area.

Bubble Generator (C. H. Gabbard). The final bubble
generator design for the GSTF is shown in Fig. 3.3. The
venturi dimensions were selected to produce the desired
bubble size and to achieve the largest pressure depres-
sion at the throat to facilitate the injection of gas while
still keeping the throat pressure above the point where
cavitation would be a problem. Previous test results
showed that the pressure required to inject gas into the
flowing salt increased with increasing gas flow rate.
Several attempts to reduce this pressure increase by
design revisions to the bubble generator were only
partially successful, and the increase was found to be
the summation of the following four mechanisms:

1. Increased mixing losses in the throat and diffuser
with the increased gas fraction. '

. The work of gas compression between the throat
and discharge pressures.

3. The pressure drop in the gas supply passages.

4. A pressure difference which was found to exist
between the gas plume and the bulk fluid at the
throat. This pressure difference is believed to pro-
vide the force to divert the liquid flow around the
gas plume,

ORNL-OWG 72-9608

5-in.
SCHED 40

PIPE rL.72 in, diam

ZFTTITFITILN

.

? T s =
— FLOW ’/7

- 3% in diom

S5-in SCHED 40 PIPE RECOVERY

VANES

-— FLOW

- _

SWIRL VANES —4 in. diam

GAS-LIQUID TAKE-OFF LINE
% in. ODx 0042 in, WALL TUBING

Fig. 3.1. Bubble separator design for gas system technology facility.
27

ORNL—-DWG 72— 13743

PRESSURE e

, fe— 4434 in, ——=]
16 | § GAS VOID PRESSURE f—— 36 iN. —] -1 -80
| OIFFERENCE PROBE — m—2Y2 in,
| FULLY WITHDRAWN |
I | + t
32 REFERENCE STATION { STATLON 2 4 —-80

GAS VOID PRESSURE DIFFERENCE

|
| C\ PROBE INSERTED TO MINIMUM RADIUS

(BN |
[y e  TANGENTIAL VELOCITY o
y —— —

4

24 |
! |
\ | |
% 20 |4-4 %
= g\
z \ a
: | N
[09]
Q 16 \ J2.| —40
|

\\. -30

AXIAL VELOCITY

: STATIC PRESSURE DIFFERENCE —] —20

GAS VvOID

AN
i
/

QN‘“- —10

STATIC PRESSURE DIFFERENCE FROM REFERENCE PRESSURE TO INDICATED STATION (fr of liquid)

O 0.4 0.8 1.2 {.6 2.0 24
RACIUS (in.)

Fig. 3.2. Velocity and pressure distribution of GSTF bubble separator at 500 gpm. Static pressures measured relative to the
separator inlet upstream of swirl vane. )

18 'g-in-diam

ORNL-DWG 72-9609
GAS INJECTION LINE
¥,-in. OD x 0072-in.
WALL TUBING

GAS INJECTION HOLES \

5047 in. — FLOW

diam
5563 in.
diam

CHANNEL

oz’ ) b 5047 in.
7’, 30 d|?m : diam \ diam
1 3
22°-3Q'
/
ANNULAR GAS DlSTF;lBUTlON
% in.x ¥ in. ,
= 41%g in—
19%5 in.

Fig. 3.3. Bubble generator design for gas system technology facifitj/.

A calculational procedure was developed to calculate
the required gas supply pressure as a function of liquid
and gas flow rates. The first three mechanisms listed
above are calculated directly from the fundamental
relations involved, but the pressure difference in item 4
is obtained from an empirical correlation derived from
the water test loop data using the liquid velocity and
the void fraction at the generator throat.

Figure 3.4 shows a comparison of the calculated to
measured pressure distribution of the prototype GSTF
bubble generator operating in the water test loop. The
inlet and gas injection pressures are referenced to the
discharge pressure of the bubble generator. A report is
in preparation to describe the water tests and the
development of the GSTF bubble generator.

Bubble Generator Analysis (T. S. Kress). The present
bubble generator consists of a converging-diverging
nozzle in which gas is injected into the throat. The fluid
turbulence breaks the plume into small bubbles in the
throat and diverging sections of the separator.

The following analysis was made to relate the size of
bubbles generated in a turbulent field to flow rate,
dimensions, and fluid properties.

At the turbulent Reynolds numbers of interest in the
bubble generator, regions of local viscous flow are small
compared with the size of the bubbles. Consequently,
one would not expect the bubble breakup to be the
result of viscous shearing. More likely, the bubble size is
controlled by a balance between turbulent forces and
surface tension forces.

An expression was developed earlier® for a mean
value of turbulent inertial forces F; in terms of the
power dissipation per unit volume ¢, .

2 ¢ 2/3
Ec

where
p = density,
d = bubble diameter,

g, = proportionality constant.
The surface tension forces F' . can be written as
F.~oad,

where ¢ = surface tension. Therefore, the ratio of
inertial to surface tension forces, F;/F, defines a
“turbulence” Weber number We in terms of the power
dissipation,

€ dg” 2/3
We = F/F, ~‘;—d (—";——c) /o
c

This dimensionless parameter is expected to control the
equilibrium bubble size in the absence of coalescence,
so that

£.0\3/5 2/5
d~|[= Ly (1)
p €,8p

This relation was originally proposed by Hinze3 for
droplets immersed in an immiscible fluid. To use it, the

2. T. S. Kress, Mass Transfer Between Small Bubbles and

. Liquids in Cocurrent Turbulent Pipeline Flow, ORNL-TM-3718,

April 1972,

3. 1. O. Hinze, “Fundamentals of the Hydrodynamic Mech-
anism of Splitting in Dispersion Processes,” AfChE J 1(3),
289-95 (September 1953).

ORNL-DWG 72-13714

PRESSURE DIFFERENCE (ft of liquid head)

12
[ ]
..o ® e
8 .. L_J ET
."".';T saLt -
/
4
5 SALT DISCHARGE PRESSURE (REFERENCE)
16 v\\’w\e
_ o
39 e
\“ .’
P
< ()
-20 Q’O D’/
‘5\)@ v
N2 o*
[ N ]
—24 / et
[/
/' CALCULATED
[ ® (OBSERVED
-28 %
[ ]
—-32
0 1 2 3 4 5

GAS FLOW RATE (ctm at throat conditions)

Fig. 3.4. Calculated and observed pressure distribution of
GSTF prototype bubble generator at 500 gpm.

power dissipation in the region of bubble generation
must be known. An expression has been recommended?
for the power dissipation in a region of constant-area
well-developed turbulent flow in a conduit. If the angle
of divergence is not large, then the local rate of energy
dissipation may be proportional to that for conduit
flow using local parameters. This would give

C 2Rell/4
eMuRe

, (2)
g.oD*

where u = viscosity, D = conduit diameter, and the
values of Re and D at the region of bubble generation
are to be used.
Substituting (2) into (1) results in
Expression (3) predicts that the bubble size produced
in a turbulent field will be proportional to the Reynolds

29

number to about —1.1 power. Data from tests of a
bubble generator described in ref. 2 agreed very well
with the —1.1 power. However, data taken with the
generator presently proposed for use with the GSTF are
better correlated by about —0.8 power. The two
generators may differ significantly in the mechanism of
bubble generation. In the former, bubbles are generated
in the central portion of the cross section, whereas in
the latter they are generated near the wall in a region of
high shear. It may be that in the present design bubble
size is determined more by local shear rates than by the
average power dissipation. Further analyses will be
attempted to clarify the difference.

3.1.2 Bubble Formation
and Coalescence Test

C. H. Gabbard

The bubble formation and coalescence tests on
2LiF-BeF, and 72LiF-16BeF,-12ThF; MSR fuel salt
were temporarily delayed due to a mechanical failure of
the shaker drive and clouding of the quartz furnace
liner. New capsules have been loaded with fuel salt
containing 0, 0.1, and 0.2 mole % UF,4. The full 0.3
mole % of UF, for MSR fuel was not loaded because
the darkness of the melt would probably prevent
photography of the bubbles. A test melt of the
72-16-12 salt was clear and did not contain any
indication of the impurities that had been observed in
the previous fuel salt capsules. The shaker drive and the
furnace liner have been replaced and testing will be
resumed.

3.1.3 Mass Transfer to
Circulating Bubbles

T. S. Kress

Mass transfer. The experimentally determined correla-
tion for mass transfer between cocirculating bubbles
and liquids was reported® to be

Sh/Scl/2 = 0,34 Re0-94 (dvs/D)l'o ’ (4)

where

Sh = Sherwood number,
Sc = Schmidt number,
Re = Reynolds number,

dvs = bubble Sauter-mean diameter.

4, MSR Program Semiannu. Progr. Rep. Aug 31, 1971,
ORNL-4728, pp. 41-44,
Although pipe diameter D was included as a logical
nondimensionalizing factor, no actual variations in pipe
size were made in arriving at the correlation. Conse-
quently, experiments with variations in pipe diameter
were planned to measure the actual dependence, and
the experimental facility was converted for use with a
conduit diameter of 1'% in. compared with the original
diameter of 2 in. This change necessitated recalibration
of instruments, rewriting of data reduction programs,
and making preliminary experiments to verify correla-
tions of bubble size distribution and volume fraction
that had been established for the original conduit size.
These changes and preliminary tests were completed,
and a new set of mass transfer data was taken fora 25%
mixture of glycerine in water. Other than for the above
validation of supporting parameters, these data have not
yet been reduced to mass transfer coefficients.

It was necessary to shut down the mass transfer loop
for routine maintenance of the bubble separator during
attempts to obtain a second set of data with a 37.5%
mixture of glycerine-water. Over extended periods of
operation, the special separator used in the mass
transfer measurements tends to accumulate particulate
debris that must be removed or the separator efficiency
decreases to unacceptable levels.

A theoretical analysis?> indicated that Sherwood
number Sh should depend on (d,/D)!/3. It was earlier
speculated that the observed higher dependence [Eq.
(1)] may have been due to a transition from rigid-
interface (small bubbles) to mobile-interface (large
bubbles) behavior. Additional examination of the data,
however, indicates that the interface was mobile
throughout the measured range. No new hypotheses
have been made to satisfactorily explain the linear
dependence.

3.2 GAS SYSTEMS TECHNOLOGY FACILITY
| R. H. Guymon

Work on the GSTF was suspended during three
months of the period because of a midyear cut in the
budget. Although the work was resumed in July,
rescheduling will not be done until it can be determined
what effect the delay will have on the availability of
craftsmen and on the construction plan. The following
describes the status at the end of the period.

The mechanical and electrical design of the facility
approached completion in August, and procurement of
components and construction of the loop was started.
The instrument application, control circuit, and panel
layout drawings were approved and issued. Most of the

30

instruments needed were removed from the MSRE and
have been checked out and repaired. Fabrication of the
instrument panels was started.

The Mark Il pump was cleaned and is being refur-
bished and modified to provide the pressures and flows
needed for the GSTF.

A preliminary system design description (PSDD)*, as
well as the system design description (SDD), was issued
in March. The master copy of the SDD, which is serving
as the primary design control document, has been
continuously updated since then, and revisions are sent
periodically to recipients of the SDD.

3.3 MOLTEN-SALT STEAM
GENERATOR INDUSTRIAL PROGRAM

J. L. Crowley

Foster Wheeler Corporation, the firm selected to
make a four-task conceptual design study of molten-salt
steam generators,® is proceeding with Task I — Concep-
tual Design of a Steam Generator for the ORNL
1000-MW(e) MSBR Reference Steam Cycle. Foster
Wheeler is being assisted by Gulf General Atomic
through a subcontract for a portion of this design
study. Of the 24 subtasks in Foster Wheeler’s Task |
work plan, 10 are now complete. The remaining
subtasks, including the Task I final report, are expected
to be finished by the middle of November, after which
they will proceed with Task II — Feasibility Study and
Conceptual Design of Steam Generators Using Lower
Feedwater Temperature.

A vertical “L” configuration of unit size (i.e., one
steam generator per secondary coolant loop) was
chosen by Foster Wheeler as best meeting the design
criteria established jointly with ORNL.

The preliminary sizing of the steam generator resulted
in a quite tall unit of about 115 ft. Foster Wheeler
found that when cross-flow baffles were placed as
required to meet the vibration criteria, excessive shell
side pressure drop resulted. As a result, Foster Wheeler
is now showing long flow (i.e., no cross-flow baffles) on
the shell side.

Foster Wheeler requested more definitive information
on the number of thermal transients to be expected in
the life of the MSBR for use in their analysis of thermal
stress in nozzles, heads, and tube sheets of the steam

5. R. H. Guymon, Preliminary Design Description of the Gas
System Technology Facility (GSTF), ORNL internal memoran-
dum, March 30, 1972,

6. MSR Program Semiannu. Progr. Rep. Aug 31, 1971,
ORNL-4728, p. 29.
generator. Although no extensive study of transients
has been made for the MSBR, a limited number of
hybrd computer simulations have been run of some of
the most severe transients expected in an MSBR
system.” Based on these simulations, the list shown
below was supplied to Foster Wheeler for use in their
design study.® Because there were limitations in both
the number of simulations and 'in the manner in which
the simulations were modeled, the list is somewhat
arbitrary and conjectural. However, the events selected
and the number of occurrences are believed to be
sufficiently representative for use in the conceptual
design stage.

Number of events
in steam generator

30-year life
1. Rapid change in salt flow rate, 120
100% to 75% in 5 sec (see Fig. 8
of ref. 9).
2. Rapid change in load demand, 50
100% to 40% in 3 sec (see Fig. 7
of ref. 9).
3. Reactor shutdown (2 rods inserted). 150
Reactor power from 100% to 12.5%
in 2‘/2 sec (see Fig. 10 of ref. 9).
4. Normal load change at rate of 4%/min.
a. Full to 20%. 300
b. 20% to full. 300
¢. Changes of 20% or less. 10,000
§. Complete heat-cool-heat cycles.
a. Room temperature to operating 50
heat.
b. Operating heat to room 50,
temperature.

Analysis of the sudden load-reduction transient (item
2 above, which results in a sudden increase in the outlet
steam temperature) indicated severe thermal stresses.
Foster Wheeler’s present design incorporates thermal
liners in both the steam-side outlet head and in the tube
sheet to reduce these thermal stresses to a tolerable
level.

7. MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL-4782.

8 J. L. Créwley, Proposed Transients for Use in the

31

Conceptual Design of the MSBR Steam (Generator, MSR-72-72 -

(Aug. 8, 1972),
9. O. W. Burke, Hybrid Computer Simulation of the MSBR,
ORNL-TM-3767 (May 5, 1972).

3.4 COOLANT-SALT TECHNOLOGY
FACILITY (CSTF)

A. I. Krakoviak

Construction and installation of the CSTF were
completed in August, and checkout of the control
circuitry and equipment was in progress at the end of
the period. Figure 3.5 shows the CSTF enclosure, pump
motor, cable trays, control and gas supply cabinets,
salt-monitoring vessel, and cold trap. Salt circulation is
scheduled for September.

The helium purification system and the salt-pump
lubricating oil system which were used during MSRE
operations were moved to 9201-3 and are now in
service for use in the CSTF. The oxygen getter
(titanium sponge) in the helium purification system and
the oil in the lube oil system were replaced prior to the
CSTF service. An on-line oxygen analyzer (Lockwood-
McLorie Model 0-1000) indicated that the purification
system reduced the oxygen content of the helium
supply from about I.1 to about 0.7 ppm.

The spare pump-rotary unit which was in readiness
for use in the MSRE coolant salt pump (but not used)
was refurbished and installed in the CSTF. With the
exception of calibration of the float-type level element
on the pump bowl, preoperational check-out of the salt
pump is complete. Installation of the alternative power
source for the salt pump (variable-frequency motoi-
generator set) is scheduled for completion in the latter
part of September. This unit is planned for use
primarily by the gas system technology facility but will
be used for the CSTF during startup and when required
by the experiments.

Except for the drain tank, which is charged with
sodium fluoroborate, the salt-containing portion of the
CSTF was heated to 500°F and purged with treated
helium at the rate of 3 liters/min (STP). From an initial
moisture content in the off-gas of 55 ppm (MEECO
electrolytic water analyzer), the moisture content in-
creased to a value in excess of 1000 ppm as shown in
the solid curve in Fig. 3.6. During the next temperature
increase (from 500 to 950°F), additional moisture was
carried off in the off-gas stream, as shown by the
dashed line. This additional moisture was given off by
the components of the salt-monitoring vessel and cold
trap that normally will operate at about 200 to 300°F,
while the salt-containing portion of the loop operates at
900 to 1150°F. The purge will continue until an
equilibrium is reached.
32

PHOTO 4438-72

PUMP MOTOR

T —

CABINETS

SALT MONITORING
VESSEL AND COLD

BF3 SUPPLY
CABINET

Fig. 3.5. Coolant salt technology facility.

3.5 SALT PUMPS

A.G.Grindell ~ W. R. Huntley
L. V. Wilson

3.5.1 Salt Pumps for MSRP
Technology Facilities

Pumps for the coolantsalt technology facility. The
reconditioning of the rotary assembly for the spare
MSRE coolant-salt pump'® was completed, and after
cold shakedown tests it was installed in the pump tank
in the CSTF. The oil-lubrication system was subjected
to preoperational testing, and the oval-ring gasket in the
flange joining the rotary assembly to the pump tank
was satisfactorily leak-tested. The electric drive motor
for the pump was cleaned, refurbished with new ball
bearings, and subjected to an electrical check-out. After
installation in the CSTF, the motor was operated to
establish the proper direction of rotation and to verify
that the motor and pump operate smoothly at the
normal rotation speed, approximately 1750 rpm.

In light of this experience, the assembly procedure for
the MSRE coolant pump was revised to incorporate
improvements in the installation of shaft bearings and
the assembly and installation of shaft seals. Several
pump drawings were revised to reflect the changes and
to bring the drawings up to date. New record drawings
were issued for the special equipment used in the cold
shakedown test and for the equipment used to deter-
mine the effective diameter of the bellows seal assem-
blies.

Pump for the gas system technology facility. The
MSRE Mark 11 fuel salt pump'" is being reconditioned
prior to installation in the GSTF. The rotary assembly
will be subjected to vibration tests to determine the
critical frequency of the shaft with an attached pump
impeller of the appropriate diameter.

10. A. G. Grindell, W. R. Huntley, L. V. Wilson, and H. C.
Young, MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL-4782, pp. 34-37.

11 Ibid., p. 35.

33

" ORNL-DWG 72-13715

WATER CONTENT IN OFF-GAS (ppm)

1200
T
| |
1000 & ‘
’ |
|
800 |
AN
P
s |1
600 + 4
! \
w a ‘|
!
8 i} ‘\
w0 LN} 5
! Ot [o N
400 —— 8t & \
of \
o \
o 04 O\\\
A’ Q \\
200 [ i NG
J£ T
Qv a
il
0
0 4 8 12 16

PURGE TIME (hr}

Fig. 3.6. Moisture content of loop gas during initial heatup.

Available vibration test equipment is being repaired
and checked out. After the vibration testing is com-
pleted, the rotary element will be disassembled, and
new bearings and mechanical seals will be installed in
accordance with the detailed assembly procedure that
was improved during assembly of the CSTF salt pump.
The pump impeller is being machined from an available
Hastelloy N casting, which was purchased in 1966.

The spare drive motor for the MSRE coolant-salt
pump will be used to drive the GSTF pump. It will be
supplied from a variable-frequency generator and will
provide variable pump speed operation up to 1800 rpm.
The. drive motor will be disassembled to install new
bearings. The detailed procedures for refurbishing
MSRE salt pump drive motors, recently revised to
include improvements made during refurbishing the
CSTF salt pump drive motor, will be used.

3.5.2 ALPHA Pump

Operation. The ALPHA pump'? was operated only
intermittently during the past six months due to an oil
leak at the pump shaft and operational problems related

to other test equipment. However, the pump was placed
in routine operation again on August 10, 1972, and has
now accumulated about 5800 hr of operation at
MSR-FCL-2 design conditions of 4800 rpm, pumping 4
gpm of sodium fluoroborate at 850°F.

A wattmeter was installed to measure the power input
to the variable-speed pump drive, a 5-hp Adjustospede
system. The input power was measured at 3.7 kW (4.96
hp) with the pump operating at MSR-FCL-2 design
conditions. The data were used to determine pump
efficiency, which at the present relatively low flow rate
is 8%. At a pump design flow rate of 30 gpm an
efficiency of 40% is anticipated.

Viton O-rings. The lower shaft seal for the ALPHA
pump was equipped with Viton O-rings'?-'* in August
1971, after the original Buna N O-rings failed by attack
from the effluent, a dilute mixture of BF; (<0.1%) in
helium. To obtain performance information concurrent
with pump operation, sample rings of Viton (Parker
compound 77-545) and Buna N were installed in the
effluent from the lower seal region of the pump. The
O-rings have been examined after 2000 and 4200 hr
exposure, and no apparent degradation of the Viton has

occurred. The Buna N was hardened and cracked at
2000 hr exposure. Hardness data for both exposed and
unexposed samples are:

Hardness
(Shore durometer A)

Buna N, unexposed 65—67
Buna N, exposed to dilute BF3; (2000 hr) 85-87
Buna N, exposed to dilute BF; (4200 hr) 85-89
Viton, unexposed 68—72
Viton, exposed to dilute BF3 (2000 hr) 65-70

68-74

“Viton, exposed to dilute BF5 (4200 hr)

The exposed sample O-rings were reinstalled for
further test operation in the dilute BF3 effluent.

Oil leak. An oil leak from the upper end of the
ALPHA pump shaft caused a shutdown of corrosion
loop MSR-FCL-2 on April 28, 1972, after nearly 5300
hr of satisfactory pump and loop operation. There was
no fire or serious damage to the test facility from the
leak. About 1 gal of oil was lost from the lubrication
system before the leak was stopped. However, oil flow
was maintained to the pump during the incident to

12, Ibid., p. 35.

13, 1bid,, p. 35.

14. A. G. Grindell, W. R. Huntley, H. C. Savage, L. V. Wilson,
and H. C. Young, MSR Program Semiannu. Progr. Rep. Aug. 31,
1971, ORNL-4782, p. 36.
provide cooling to bearings and seals until the salt was
drained and the pump was cooled to room temperature.

The oil leak occurred at the upper end of the pump in
a soft-soldered joint between the bore of the Hastelloy
N pump shaft and a 304 stainless steel plug on the oil
guide tube. The oil system was being operated at
normal design conditions with a flow rate of 0.2 psig.
Visual inspection of the joint showed that the soft

34

solder had cracked or failed around the entire circum-
ference of the joint.

Repair of the oil leak was accomplished without
removing the pump from the MSR-FCL-2 facility. The
joint surfaces were cleaned and resoldered. In addition,
an oil-resistant, rubber-gasketed seal was installed to
back up the soft-solder seal. In future pumps the joint
will be seal-welded rather than soft-soldered.
35

4. Heat Transfer and Physical Properties

H. W. Hoffman

4.1 HEAT TRANSFER
J. W. Cooke

Final analysis has been completed of the heat transfer

data for a proposed MSBR fuel salt (LiF-BeF,-

ThF,-UF,; 67.5-20.0-12.0-0.5 mole %) flowing by
forced convection through a 0.18-in.-ID X 24.5-in.-long
circular horizontal tube. The results are shown plotted
in Fig. 4.1 in the form of the Sieder-Tate heat transfer
function

NS*T = [NNuNPr_lla(#/#s)—O'l4]

vs the Reynolds modulus Vg .. The data were obtained
under the following conditions:

Reynolds modulus, Vg . 400 — 28,000
Prandtl modulus, Np, 4.9 —16.3
Average fluid temperature, 1065 — 1550

T,, CF)

Heat flux, /4 (Btu hr ™' ft72) 22,000 — 560,000

J. J. Keyes, Jr.

The physical properties of the salt were taken from
accepted published data.!

Within the above ranges, the heat transfer coefficient
was found to vary from 320 to 6900 Btu hr™' ft™?
°F7! (Nusselt modulus of 7.0 to 149.8). Statisticai
analysis of the data resulted in the correlations

Ng_ =189 [Np(D/D)] '3, (1)
with an average deviation of 6.6% for N, < 1000 (D/L
is the test section inside diameter to heated length
ratio);

Ng_5=0.107 (Ng2/? - 135), (2)

with an average deviation of 4.1% for 3500 <Ny, <
12,000;and

Ng_p=0.0234 N %8, (3)

1. S. Cantor, Physical Properties of Molten-Salt Reactor Fuel,
Coolant, and Flush Salts, ORNL-TM-2316 (August 1968).

ORNL-DWG 72—-13716

500
8
- P
S 200
: /]
. L/
i N, =0.027 N, 08w 3 0.1
o 100 Nu_o' e - (F-/F-s
w =
z (SIEDER- TATE) —
: 7 J
: T
= 50 =
Z 'nfl‘
w
I e
= J
S 20 1= 2 iy 044
<. NNu=o‘”6 (NRE /3 —125} NPr 3 (F'/F'S) .4?‘
’ N ®|
A
T 0 (HAUSEN) —_ /@
_}m II [ ]
& /
= 5 y
\ .‘.‘<\/ 1/3 i
2 0.14
i Ez L] 0{ Wny=1.86 [NRe Npr (D/L)] (F-/F-s) +
2 B (SIEDER-TATE) L.
1
102 2 5 103 2 5  10% 2 5 105

Nge» REYNOLDS MODULUS

Fig. 4.1. Final correlation of heat-transfer data for LiF-BeF,-ThF4-UF,; 67.5-20.0-12.0-0.5 mole %.
with an average deviation of 6.2% for Ng . > 12,000. In
the range 5000 < Mg, < 30,000 the data average 13%
below the correlations of Hausen? and Sieder-Tate?
depicted in Fig. 4.1. For Ng, < 1000, the data average
1.6% above the correlations of Sieder-Tate for laminar
flow.?

4.2 THERMAL CONDUCTIVITY
J. W. Cooke

A detailed analysis was made of the uncertainties
affecting the measurement of thermal conductivity
using the absolute, variable-gap apparatus.® The result
of this analysis is shown in Fig. 4.2. The estimated
maximum and standard error limits in the conductivity
measurements are plotted as a function of the specimen
conductivity. The effects of radiation and, hence, of the
specimen temperature level (40 to 950°C) and specimen
transparency (absorption coefficient O to =) do not

2. H. Hausen, “Darstellung des Warmeuberganges in Rohren
durch verallgemeinerte Potenzbeziehangen,” Z. VDI, Beihefte
Verfahrenstechnik, No. 4, 1943,

3. W. H. McAdams, Hear Transmission, 2d ed., p. 168,
McGraw-Hill, New York, 1942.

4. Ibid., p. 190.

5. J. W. Cooke, Development of the Variable-Gap Technique
for Measuring the Thermal Conductivity of Fluoride Salt
Mixtures, ORNL-4830 (in publication).

36

contribute greatly to the error and are shown by the
areas included within the narrow bands in Fig. 4.2. It is
apparent from the figure that the error is sensitive to
the magnitude of the specimen conductivity except at
larger values of the conductivity. Much of the increase
in error at low values of conductivity results from an
increased uncertainty in the measurement of the change
in the specimen thickness. The larger temperature
gradients caused by the lower specimen conductivities
increased the uncertainty in the thermal expansion
corrections to the specimen thickness.

Figure 4.2 also shows the deviation of the calibration
measurements [H,O, Hg, Ar, He, and HTS (KNO;-
NaNQ,-NaNQ;; 44-49-7 mole %)] from published
values. Several points are outside the estimated error
limits. Data for H,O and Hg were obtained using
Chromel-Alumel thermocouples. The error limits for
these data should be considerably larger than shown in
Fig. 4.2, which was based on the use of Pt vs Pt + 10%
Rh thermocouples. Also, we believe the published value
for the conductivity of HTS to be unreliable. Excluding
these points, 93% of the data lies within the maximum
error limits, and 68% lies within the standard error
limits.

Final analysis of the molten fluoride salt conductivity
data (880 data points) is being made. From the above
error analysis, the maximum error in the fluoride salt
conductivity measurements is expected to be less than
+5%.

ORNL-DWG 72-10542

[ I 1T 11 TTH | I T T TTTH

20 N ~BAND AREAS COVER 300-900°C TEMPERATURE .|
- \Q\ RANGE, WITH OR WITHOUT RADIATIVE HEAT
G NN TRANSFER
- B NN
o b 1N .
S 40 2 N
2 z |5 s
g c g |° SN
ro | & R Rari eSS
— QJ s XAl
E 0 —§<g . >Bgf’zzz7-.7 /@(//< LR
3 ~12]. fsf@s&fim
2 O S
[e) _10 L E [4p] 7 A o] HTS o
u- L ‘fi/ A e Ar
g Q@
o §§§§> & He
: '20 2 % A H20
= b Hg
v

® HTS (soutl))
-30 L i
o 2 5 0> 2 5 0% 2 5 0!

k, SPECIMEN CONDUCTIVITY (W em tech)

Fig. 4.2. Estimated probable and maximum error limits in the conductivity measurement as a function of specimen
conductivities over a 300 to 400°C temperature range with and without radiative heat transfer. Also shown as plotted points are the

deviations of the experimental results from published values.
Part 2. Chemistry

W. R. Grimes

The research and development activities described
below continue to deal with chemical problems posed
in the design and ultimate operation of molten-salt
reactor systems.

Experimental study of specimens and materials from
the MSRE has been completed. The final report of
complex fission product behavior in that system has
been drafted and is undergoing final editing.

Study of  intergranular attack by tellurium upon
metallic components of the MSRE (conducted in close
cooperation with MSRP metallurgists and reported in
more detail in Part 3 of this report) continues to engage
a considerable fraction of the chemical effort. Many

exposures of pertinent metals to tellurium (and to other -

potentially harmful fission products and contaminants)
have been completed, and others are in progress. New
techniques for a more readily controlled addition of
tellurium to the specimens have been devised and will
be put into service 'during the next reporting period. A
systematic study, primarily by mass spectrometry,’ of
the volatility, relative stability, and chemical reactions
of pertinent metal tellurides is producing useful data.
Spectrophotometric methods for the study of tellurium
species in molten fluorides have been developed for
application to this problem area.

The behavior of hydrogen isotopes in molten salts,
metals, and graphites is a continuing field for major
attention. Study of H, solubility is virtually concluded
with a few confirmatory studies yet to be done. Study
of permeation of metals by hydrogen isotopes seems to

-have demonstrated conclusively that the rate of permea-

37

tion for *“‘clean” metals depends on the square root of
the hydrogen partial pressure. Careful measurements of
permeation through metals with oxide films seem to be
yielding encouraging information; such studies are now
directed at Incoloy 800 and other metals of likely value
in steam generator systems. Significant uptake of
tritium by graphite at elevated temperatures has been
confirmed, and this phenomenon is being studied to
assess its value in tritium control and management
processes. Fluoroborate chemistry — aimed largely at
improved understanding of these coolant materials and
their corrosion mechanisms — is still under study at a
modest funding level.

Very little attention has been paid to protactinium
chemistry during this reporting period, and, although a
few confirmatory studies may prove necessary, none is
planned for the near future.

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. ThesQ
methods include electrochemical and spectrophoto-
metric means for determination of the concentration of
U3 and other ionic species in fuels and coolants and
adaptation of small on-line computers to electroanalyt-
ical 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.
38

5. Behavior of Hydrogen and Its Isotopes

5.1 THE SOLUBILITY OF HYDROGEN
IN MOLTEN SALT

D. M. Richardson  A. P. Malinauskas

Measurements of the solubilities of hydrogen and
deuterium in Li, BeF,; are being performed in order to
provide data to assess the importance of solubility on
tritium transport in a molten-salt reactor. These experi-
ments are being conducted using a two-chamber appa-
ratus which has been described previously.' *2

In an earlier report® the validity of Henry’s law was
demonstrated for hydrogen and helium in Li,BeF, at
600°C over the pressure range | to 2 atm; that is, the
solubility values were reportedly observed to increase
linearly with increasing saturation pressure, Since then,
measurements have been made at 600°C with deute-
rium in order to ascertain whether the solubility
phenomenon displays an isotope effect. Additionally,
the experiments have been extended to include studies
at 500 and 700°C.

The results obtained to date are presented in Table
5.1, wherein the Ostwald coefficients K, are listed as
functions of temperature. Because of the applicability
of Henry’s law, K., which is defined as the ratio of the
gas concentration in solution to its concentration in the
gas phase, is independent of saturation pressure. Note
also that the values are preliminary and are subject to
adjustment; however, there is presently no reason to
believe that the final values will lie outside the standard
deviations listed.

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

2. A. P, Malinauskas, D. M. Richardson, J. E. Savolainen, and
J. H. Shaffer, “Apparatus for the Determination of the
Solubility of Hydrogen in Molten Salts,” fnd. Eng. Chem.
Fundam. (accepted for publication).

3. D. M. Richardson and A, P. Malinauskas, MSR Program
Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782, p. 53.

The helium solubility determinations are being made
to provide a basis for comparison with the work of
Watson et al.* As is evident from the values presented
in Table 5.1, the inexplicable discrepancy noted earlier®
at 600°C persists at the other temperatures; our results
are consistently lower than the results obtained earlier.

The theoretical values which are listed for hydrogen
were derived from a modified form of the Uhlig
expression for gas solubility which ignores a possible
explicit dependence upon polarizability.® Comparison
of the theoretical results with experiment yields reason-
ably good agreement. Also, the data obtained thus far
give no substantive indication of a possible isotope
effect for solubility; within the limits of mutual
uncertainty, the solubilities of hydrogen and deuterium
appear to be identical.

It is important, however, to point out that significant
amounts of hydrogen (as H, and HD) have been
encountered in performing the deuterium solubility
determinations, and the K, values reported for deute-
rium have been evaluated under the assumption that all
hydrogen isotopes are indistinguishable in solution. In
other words, we assume that it is not possible to
separate the isotopes on the basis of solubility. The
results for K., of course, are not inconsistent with this
assumption.

The unexpected behavior of hydrogen which has been
observed does suggest the possibility of inhibiting
tritium transport through heat-exchanger materials by
pretreating these members with hydrogen; data which
indicate this as a possible method are presented in Table
5.2, which is an abbreviated chronology of the compo-
sitions of gas samples taken from the stripper chamber

4. G. M. Watson, R. B, Evans III, W. R, Grimes, and N. V.
Smith, J. Chem. Eng. Data 7, 285 (1962).

5. 1. E. Savolainen, J. H. Shaffer, and A. P. Malinauskas, MSR
Program Monthly Report September 1970, MSR-70-79, pp.
16-19.

Table 5.1. The solubilities of hydrogen, deuterium, and helium in Li,BeF,

10° K,
T :
°K) Hydrogen Deuterium Helium
Experimental Theoretical experimental Experimental Watson et al.
773 1.13 £ 0.08 1.46 3.86 £ 0.25 4.75
873 3.17 £ 0.09 271 2.74 £ 0.16 6.01 = 0.11 8.27
973 3.87 +0.37 451 4.17 £0.42 9.26 +0.19 11.92

of the apparatus during the course of the various
solubility experiments, The data are presented both in
terms of volume percent of sample and as volume of
component collected (reduced to standard conditions).
The remaining components, which are not listed (i.e.,
the deviations from 100 vol %), are of known origin
and are not believed to affect the observed behavior.
Note also that HD is equally apportioned between H,
and D, in calculating the volumes of these two species.

Run 41 gives an indication of the residual helium
which persists through the hydrogen solubility determi-
nations; the presence of this gas arises from its use as a
cover gas in the saturator chamber between hydrogen
experiments. Run 42, on the other hand, is a helium
solubility experiment;in this case the residual hydrogen
most probably arises from gas dissolved in the metal
apparatus. All of the runs between 42 and 48 were
conducted with helium, and a comparison of the
volumes of hydrogen in these two cases is typical of the
gradual reduction in hydrogen content which was
observed to occur. Runs 51 to 58 again employed
hydrogen.

Deuterium was first introduced into the system in run
59, during which time a rather large volume of
“residual” hydrogen, as H, and HD, was noted.
Contrary to previous behavior during the helium runs,

Table 5.2. Chronological listing of sample analyses

Run 41 42 48 51 58 59

Percent gas

H,  80.20¢ 394 272 43.85%7 69.767 10.20
HD ' 15.76
D, 48.18¢
He 0.18 B84.669 - 88.007 344 0.25

Standard cubic centimeters of gas
H, 1.353  0.093  0.057 0.235 0.565 0.152
D, : 0471
He 0.003  2.007 1.835 0.018 0.002
Run 68 69 71 72 76 77

Percent gas

H, 5.33 2.99 6.38 78719 82.097 13.19
HD " 13.8s 0.34 14.58 1.17 0.49 30.45
D, 60.67% 0.61 59.184 0.03 0.01 41.664

He 0.53 82.617 1.06 - 045 0.20 0.24
Standard cubic centimeters of gas

H, 0.136 0057 0.152 0.837 1.011  0.343

D, 0.750 0014 0.740 0.006 0.006 ~ 0.712

He 0.006 1503 0.012 0005 0.002 0.003

IMajor component.

‘the high values for hydrogen persisted throughout the

deuterium experiments, the last of which is designated
'as run 68. Surprisingly, the hydrogen (and deuterium)
content dropped markedly in run 69, a helium solu-
bility determination, but returned to its previous value
in run 71, which utilized deuterium.

Throughout runs 59—71, the apparatus was exposed
only to helium and deuterium, and so we decided to
perform some experiments with hydrogen to determine
whether an enhancement in residual deuterium would
be observed. As indicated by runs 72 and 76, the
deuterium concentration was not significant, but, as run
77 demonstrates, the reverse effect remains. The deute-
rium employed in these investigations contains about 1
vol % hydrogen, but this is a minor contribution to the
amounts of hydrogen encountered.

A plausible argument for this unexpected behavior is
that the metal saturator chamber dissolved significant
amounts of hydrogen during purification of the salt by
sparging with hydrogen and during the course of the
subsequent hydrogen solubility determinations. Some
of this dissolved gas is released during the helium
experiments, and this accounts for the higher quantities
of residual hydrogen in the helium runs as opposed to
the amounts of helium found in the hydrogen measure-
ments. Deuterium, on the other hand, enhances the
release of dissolved hydrogen, but in the process is itself
prevented from dissolving in the metal. Consequently,
effectively no residual deuterium is observed in the
hydrogen experiments.

We wish to emphasize that this description is con-
jectural; nonetheless, we believe the observations merit
further investigation of the phenomenon.

5.2 HYDROGEN PERMEATION
THROUGH METALS

H.C.Savage R.A.Strehlow

Our experimental program® continues to examine
two important questions concerning permeation of
metals by tritium. These questions are (1) whether the
theoretical (', power) dependence of permeation on
pressure is followed to the very low tritium pressures of
reactor interest and (2) the extent to which the oxide
film anticipated on the steam side of the steam
generator tubing will serve as an effective barrier to
tritium permeation. The work reported here concerning
the first question shows that permeation of nickel varies

6. R. A. Strehlow and H. C. Savage, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 54-55.
as the square root of the pressure over the pressure
range from 750 to 8 X 107 torr. Work on the second
question suggests that oxide films on some metals can
reduce permeation by hydrogen isotopes by substantial,
and practically useful, factors.

The square-root relation between pressure and per-
meation flow rate is expected because hydrogen gas and
its isotopes dissolve in metals with dissociation of the
molecule into two atoms. As a consequence, solubility
of hydrogen in the metal is proportional to the square
root of pressure (Sieverts’s law). Since diffusive flow of
hydrogen through the metal is proportional to the
concentration of hydrogen in the metal (Fick’s law),
the permeation of hydrogen gas is dependent on the
square root of the driving pressure. These relations are
summarized in the expression

cm?(STP)
_-_._.—T ’ (1)
sec-cm

where JH2 is the flux of hydrogen permeating a metal
of thickness AX cm, the diffusivity of hydrogen is D
cm? sec”', and the Sieverts’s law constant for hydrogen
is K, {cm3(STP)]/cm®-atm!/2. [The driving pressure is
shown as A(p'/?) atm!/? to allow for the “down-
stream” pressure. |

This square-root dependence has a sound theoretical
basis, but the literature contains many reports that at
pressures below about 30 to 100 torr the dependence
changes from the square-root relation to the first-power
(direct proportionality) relation. The extrapolation to
be used over the five or six orders of magnitude in
pressure to the range for the molten-salt reactor was
clearly of great concern. An apparatus was constructed
to measure this pressure dependence to very low
pressures using a mass spectrometer method and deu-
terium gas.® The results for nickel are shown in Fig.
5.1. The square-root dependence is quite good to
pressures well under 1073 torr.

Nickel was used because the oxidation potential
expressed as PHz O/PH2 is about 100 at temperatures of
interest, and oxidation by impurity water or air leaks
would not be expected to form any impedance to flow.
Stainless steel and indeed chromium alloys generally
have an oxidation potential near 10 to form Cr,0;
(chromium sesquioxide) from chromium metal. This
should permit the development of an oxide in which
the hydrogen solubility and diffusivity are low. Such an
oxide, if free of cracks and pores, is expected to serve as
a barrier to hydrogen permeation.

It is well known that oxide coatings yield a marked
reduction in the permeability of many metals to

40

hydrogen. No oxide films are possible on the metal
surfaces exposed in an MSBR to molten fluorides at
high temperatures; however, an oxide film will be
present, and perhaps one with low permeability can be
produced and maintained, on the steam side of the
MSBR steam generator. Studies have accordingly been
carried out to study hydrogen permeability of such
films.”

One model leading to a decrease of permeation rate
upon oxidation of a metal would be to view the oxide
as offering a resistance to flow in series with that of the
metal. If this were so, the pressure dependence of
permeation rate would be expected to vary from a
one-half power dependence at high pressures to a
first-power dependence at low pressures. An 18-8
stainless steel was studied as a metal which was
expected to form a protective oxide coating. Hydrogen
permeation data for the clean metal showed a one-half
power pressure dependence. After a light coating of
oxide was permitted to form, the permeation rate
decreased by a factor of 8. The pressure dependence,
however, was not substantially changed from the
half-power relationship. This indicated that the oxide
coating was not continuous and that defects in the
oxide permitted ready access of the hydrogen to the
metal.

A model was derived for the observed behavior based
upon the notion that the oxide defects (holes or cracks)
permit permeation flow to occur merely with a con-
striction resistance in addition to the bulk resistance of
the metal. The relations obtained in terms of defects in
the oxide film are based on the derivations from
potential theory as applied to the analogous phenom-
enon of electrical contact resistance. This is indicated in
Fig. 5.2, which shows the resistance due to the -
“necking down” of the permeation flow lines. An
example of a relation obtained from this model for
holes at average separation 2/ and average radius a is as
follows:

1
Iy =DiAK, (AX—-+ 12/a> * APL?)

where /. is the permeation flow rate through a
specimen, D.. is the diffusivity of tritium in the metal,
A is the area, K is the solubility constant (Sieverts’s
constant) of tritium in the metal, AX is the specimen
thickness, / is the average half separation of holes of

cm? (SCTP) ©

7. R. A. Strehlow, Reactor Chem. Div. Annu. Progr. Rep.
May 31,1971, ORNL4717, p. 54.
41

ORNL-DWG 72-7963

(P2 - (P)"2 (Torr'2)

0O 02 0.4 0.6 0.8 1.0 1.2 1.4
90 3.6
I /
80 Y 3.2
®
[o]
70 2.8
Py ®
60 / 2.4
o)
o

2.0

50 /‘

¥ /.

DEUTERIUM PERMEATION [cm3 (STP)/hr]

DEUTERIUM PERMEATION [cm3(STP)/hr]

- 1.6
/ b 0.02-0.90 Torr"2
(o] [ pe——
o
: O, —p— O
30 | 1.4-27.4 Torr"2 .2
/ L
20 0.8
7 — 1
25 AREA = 144 cm2, THICKNESS = 0.89 mm,
10 Aé‘z" T=626°C | — 0.4
&
0 0
0 5 10 15 20 25 30 35

(A2 - (P2 (Torr"e)

Fig. 5.1. D, permeation through nickel vs square root of pressure.

ORNL-0OWG T2—-5844

FLOW LINES —
1- TYPICAL
\ %UIPOTENTIALS
L |

J |
———I I—— AVERAGE DISTANCE =2 ¢ A—l

AVERAGE RADIUS =@

Fig. 5.2. Schematic representation of flow of hydrogen into a
metal through constrictions at the surface.

average radius g, and Pp_ is the pressure of tritium.
Although the model has not yet been tested, it appears
to fit the salient features of the phenomena associated
with oxide structure and with the permeation process.
This relation and a similar one based on the presence of
cracks indicate that the continuity of the oxide and its
freedom from cracks and holes is of crucial importance
in decreasing permeation through the metal. In Eq. (2)
it can be seen that the separation of holes / must be
quite large relative to their radii to achieve values of
I?/a as large as 100 to 1000 times the metal thickness
AX.

Continued study of 18-8 stainless steels and Hastelloy
N after oxidation in the permeation apparatus yielded
generally disappointing results. Measured decreases in
permeation for several materials approached factors of
only 10 to 12, and the observed pressure dependence
for the permeation was not greatly different from the
one-half power relationship. Part of the problem is
undoubtedly due to the ease with which the iron and
the manganese-chromium spinels can be reduced and
reformed as the redox potential of the gas mixture is
changed during measurement of permeation at various
partial pressures of deuterium. In addition, it is likely
that the permeability of deuterium varies with the
character of the particular metal oxide in the film.

To study the effect of one “best chance” oxide, a
nickel tube was electroplated with a 1-mil coating of
chromium, and the surface of this film was oxidized
within the permeation apparatus to Cr,O;. Oxidation
at 650°C showed only a relatively small decrease in
permeability from that expected for nickel. Since
Flint* had shown that more effective films could
develop on steel under oxidation at higher tempera-
tures, the chromium plate was oxidized at about 900°C.
The subsequent measurements of permeability were
encouraging; a marked reduction in permeation (to
about 1% of the preoxidation value at 1 torr and
650°C) was observed, and the pressure dependence was
found to be near first power. Much, but not all, of the
decrease remained after a thermal cycle to room
temperature and back to 625°C. Implementation of the
technique used in this experiment might not be possible
for a reactor system. However, we now believe that
substantial improvement in tritium management is
possible with the aid of proper oxide films.

Indeed, as this report goes to press, we have prelim-
inary evidence that the film that formed readily in the
permeability apparatus by oxidation of Incoloy 800
shows a markedly reduced permeability and a pressure
dependence near first power. Studies of the problem at
present are concentrating on this interesting heat-
exchanger material.

5.3 THE CHEMISORPTION OF
TRITIUM ON GRAPHITE

R. A.Strehlow H. E. Robertson
The presence of significant amounts of tritium on the

graphite from the MSRE® has prompted a study of the
chemisorption of tritium on graphites. The MSRE

42

graphite data suggested that the bonding of the tritium
was tenacious even at the high temperatures of the
reactor and that the kinetics of sorption were fast in
comparison with the fuel salt circulation time.

An apparatus has been constructed and operated to
permit exposure of graphite at elevated temperatures to
a gaseous mixture of tritium in helium. The tritium
concentration used is controlled to be in the range of 1
to 10 ppm by volume (partial pressures around 107>
torr). Preliminary data have been obtained on three
types of graphite: CGB (MSRE-type), Poco, and an
ORNL-made lampblack graphite. After the usual ex-
posure conditions of 4 hr at 750°C the graphite pieces
are sampled by removing successive 0.006-in.-thick
samples which were subjected to analysis for tritium.
The tritium contents of the samples were substantially
the same as the findings for the MSRE graphite
examination with values of several times 10 dis min ™
g~! for the outermost 0.006-in. cut, decreasing to
values below 10® for interior samples. Values of 6 to 8
X 10'% dis min™' g”' have been found for some
samples of high surface area.

The amount of tritium sorbed on the ORNL graphite
after a pretreatment to increase its surface area to 2
m?/g corresponded to 5 X 10'? tritium atoms per
square centimeter. Interior samples from these speci-
mens also contained tritium levels near this magnitude.
Several percent of the surface carbon atoms appeared to
have bonded the tritium atoms. That the tritium is
chemically bonded rather than being merely adsorbed
trace amounts of tritium oxide from an unknown
source is indicated by the fact that water and alcohol
rinsing of the specimens removes less than 1% of the
tritium.

Several questions remain as to whether this phe-
nomenon of chemisorption, if real, can be used to
effect a separation of tritium from a molten-salt
reactor. Notable among these questions is the appar-
ently sluggish kinetics of the reaction. There does not
yet appear to be a fundamental bar to the development
of a process based on the observed phenomena.

8. P. S. Flint, Diffusion of Hydrogen through Construction
Materigls, KAPL-659 (Dec. 14, 1951). :

9. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971,
ORNL-4728, p. 51. )
43

6. Fluoroborate Chemistry

6.1 SOLUBILITY OF BF; IN
MOLTEN LiF-BeF,-ThF,

S. Cantor

As stated previously' the purpose of this study is to
determine changes in fluoride ion activities in melts
composed of LiF, BeF,, and ThF,. The working
hypothesis in this investigation is that available or
“free” fluoride ions combine with dissolved BF; to
form tetrafluoroborate ions:

F7(in the melt) + BF;(dissolved)

- BF,;7(in the melt).

The results thus far reported '?. suggest that the above
equilibrium is shifted almost completely to the right.
Further, in LiF-BeF,; melts, the free fluoride concentra-
tion can be identified with the thermodynamic activity
of LiF.

Solubility measurements have now been carried out in
ternary solvents. The four compositions studied and the
BF; solubilities expressed in terms of Henry’s law are
listed in Table 6.1; also tabulated are the results of
measurements in a binary (LiF-BeF,) solvent.

The results are qualitatively consistent with observa-
tions in the LiF-BeF, system, namely, the greater the
LiF concentration (and activity) the greater the Henry’s
law constant. To calculate activities of LiF in the
ternary solvents containing 85, 80, and 76 mole % LiF,
liquidus temperatures were determined by the conven-
tional thermal-halt method. In these three solvents, the
activities of LiF are approximately proportional to the
values of K, Henry’s law constant. It should be noted,

however, that a similar proportionality between the
systems LiF-BeF, and LiF-BeF,-ThF,; does not seem
to be the case. For instance, the activity of LiF in
79.7-20.3 mole % LiF-BeF, is approximately 1.2 times
that in 80-10-10 mole % LiF-BeF,-ThF,; the ratio of
Ky, for these two melts, which would be expected to be
1.2, is 2.3. The significance of this discrepancy is
presently being explored. '

6.2 SOLUBILITY OF BF; IN MSBR PRIMARY
SALT CONTAINING 8 MOLE % NaF:
SOME SAFETY CONSIDERATIONS

S. Cantor

The object of these measurements is to aid in
assessing the consequences of a large inleakage of
fluoroborate coolant into the primary system of an
MSBR. In the preceding semiannual report,” equations
were presented with which to calculate BF; partial
pressures following a leak of coolant into the fuel salt.
For a “small” leak, in which 1 ft® of coolant mixes
with 1720 ft* [primary salt volume in a 1000-MW(e)
MSBR] of fuel salt at 1250°F, assuming only a 1% void
volume, we obtain an estimated BF; pressure of 0.17
atm. The BF; pressure would actually be less than this
because of the large expansion volume of the off-gas
system associated with the primary salt system.

1. S. Cantor and R. M. Waller, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 78.

2. S. Cantor and R. M. Waller, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 63—65.

Table 6.1. Solubility of BF? in molten fluoride solvents -

Solvent Temperature Liquidus Henry’s law constant vs temperature equation;? 103 K
{mole % range measured temperature Ky in mole fraction BF 3/atm; H.
LiF-BeF,-ThF ) (8] (0O temperature in °K 700°C 800°C

85-7.5-1.5 744-990 740 Ky = exp (—13.297) exp (7065/T) 2.39b 1.21
80-10-10 700-967 660 = exp (—14.457) exp (7470/T) 1.13 0.555
79.7-20.3 718-957 700¢ = exp (—13.937) exp (7800/7) 2.68 1.27

(LiF-BeF,) : '
76-12-12 647-907 601 = exp (~15.083) exp (7870/T) 0.915 0.431
72-16-12 530-795 = exp (—14.950) exp (7667/T) 0.849 0.407

4] east-squares fit of the data.
.bExtrap'olated.

‘K. A. Romberger, J. Braunstein, and R. E. Thoma, J. Phys. Chem. 76, 1154 (1972).
Provided that the leak is not truly massive,> when
coolant leaks into MSBR fuel salt the NaBF, fraction
of the coolant decomposes into NaF and BF;. This
NaF, along with the 8 mole % NaF of the coolant, will
dissolve in the fuel salt, while the BF; will be
distributed between the salt phase and ‘the available
vapor space. Thus, by measuring the BF; pressure in
equilibrium with MSBR fuel salt containing NaF, one
can estimate the partial pressure of BF; that would be
obtained following accidental mixing of coolant and
fuel salts. In a melt consisting of 8 mole % NaF and 92
mole % fuel solvent (72-16-12 mole % LiF-BeF,-ThE,),
solubilities were measured between 615 and 755°C by
the volumetric method previously described;® BF;
saturating pressures did not exceed 2 atm and Henry’s
law was obeyed. The data are represented by the
equation

9203 .
T(°K)

Ky =exp (—15.5673) exp , (1)

where Ky, is the Henry’s law constant in the units of
mole fraction BF3 dissolved per atmosphere of BF;
pressure. At 1250°F (677°C) the solubility of BF3 in 8
mole % NaF—92 mole % fuel solvent is three times that
in the fuel solvent alone.

The concentration of NaF in these measurements is
equivalent to mixing 416 ft* of fluoroborate coolant
with the 1720 ft> of fuel salt. The likelihood of such
massive mixing of coolant and fuel must be considered
extremely remote. There are now design provisions for
draining the shell of a primary heat exchanger,® so that
even in the event of a doubleended tubing rupture,
only a small fraction of the contents of a shell (whose
volume is about 500 ft*) need leak into the primary salt
circuit.

Nonetheless, an exploratory experiment has been

performed by Richardson and Shaffer® in which BF,
pressures were measured over a mixture of fuel salt and
fluoroborate in the approximate volume ratio of 3:1.
Their experiment was roughly equivalent to mixing 1.2
times the volume of coolant in the shell of one heat
exchanger with the entire salt volume in the primary

3. At a coolant-to-fuelsalt volume ratio greater than unity, it
is known (C. E. Bamberger et al.,, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1968, ORNL4254, pp. 171-73) that
there will be substantial liquid-liquid immiscibility, which
would limit the BF3 pressure upon mixing coolant and fuel.

4. E. S. Bettis, personal communication.

5. D. M. Richardson and J. H. Shaffer, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4528, pp.
136-38.

44

system of a 1000-MW(e) MSBR. Assuming that the
Henry’s law constant for the melt (containing about 10
mole % NaF) in the Richardson-Shaffer experiment is
25% greater than that given in Eq. (1) and applying this
to the appropriate equations,® we calculate a BF;
pressure at 647°C of 201 psia; at this temperature,
Richardson observed that the pressure was no less than
148.5 psia. Considering the approximations and as-
sumptions (one being that Henry’s law holds at a
pressure of about 13 atm and with a BF; concentration
of about 6 mole %), agreement between calculated and
observed pressures is reasonably good.

6.3 CORROSION OF HASTELLOY N
BY FLUOROBORATE MELTS

C.E. Bamberger C. F. Baes, Jr.

Previous measurements of the free energies of forma-
tion of NaNiF; and NaFeF;® had allowed us to
estimate the effect of redox potential on the oxidation
of Ni and Fe in Hastelloy N by fluoroborate melts.
Those estimates were based on measurements of the
equilibrium
NaMF;(c) = NaF(d) + MF,(c), K| =ay,F, (1)
where M is either Ni(Il) or Fe(II), in molten fluoro-
borate. The activities of NaF were estimated by
measuring the BF; pressures generated by the equilib-
rium

NaBF,4(d) = NaF(d) + BF3(g), (2)

a
NaF
Ky =Pgp,—2f
2 T BE3 aNaBF,

which has been measured by Cantor:”’
log K, = [5.772 — 6.513 (10°/T)] * 0.04.

After termination of the reported experiments, 2 to §
mm Hg of a mixture of mainly CO, and SiF4 remained
at room temperature, which raised questions regarding
the purity of BF, at the temperature range of the
measurements (420—600°C). Consequently, measure-

6. C. E. Bamberger, B. F. Hitch, and C. F. Baes, Jr., MSR
Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782,
pp. 65—68.

7. S. Cantor (ed.), Physical Properties of Molten Salt Reactor
Fuel, Coolant, and Flush Salts, ORNL-TM-2316, p. 34 (August
1968).
ments of equilibrium (1) were repeated® using a
modified system that permitted sampling of the gas
phase at temperature for analyses by mass spec-
trometry.” The concentration of BF; was used to
correct the measured pressures; only data where BF; =
92% were used. When the BF; concentration was below
that limit, the system was purified by repeated evacua-
tions at 25 to 200°C. The values of Pgy, measured in
millimeters of mercury in both equilibria (NiF, -NaNiF;
and FeF,-NaFeF3) were corrected to millimeters of
mercury at 0°C and standard acceleration of gravity,
and were fitted simultaneously by least squares to an
expression

1 1
Pop. =Ky (= 3
BF; 2 (Kl 7N3F> ( )

by assuming that log K, varies linearly with tempera-
ture, that is, log K; = a + b(10*/T), and that the
entropies of formation of NaNiF; and NaFeF;
[reaction (1)] are the same. Again, assuming that
NaF-NaBF, mixtures have an ideal entropy of mixing,
from Cantor’s results' © we estimate

log YnaF = (1 1;'9> 1 0.04.

The results obtained are

log K, (Ni) = [0.31 — 1.91(10%/T)] £ 0.04, (4a)
log K, (Fe) = [0.31 — 1.68(10%/T)]  0.04 (4b)
and
AG [reaction (1), Ni]

= [8.72 — 1.42(T/10%)] £ 0.2, (5a)
AG [reaction (1), Fe]

= [7.68 — 1.42(T/10%)] £0.2. (5b)

The agreement between measured and least-squares
calculated values of Pgp, is shown in Fig. 6.1. By

8. We acknowledge with pleasure the assistance of Eugene
McKissack, Pre-Coop. student from Tennessee State University.

9. We are thankful to O. Howard and his group at ORGDP
for performing the analyses,

10. S. Cantor, R. E. Roberts, and H. F. McDuffie, Reactor
Chem. Div. Annu. Progr. Rep. Dec. 31, 1967, ORNL4229, p.
55.

45

ORNL-DWG 73-1253
TEMPERATURE (°C)

3 600 550 500 450
LS N [ | |
8 \ -‘\
N AN
N o

—_ 2. 3
: NN
o Yf; \
Qm
102 i \ \g
"~ N
8 N
6 AN %,

1.1 1.2 1.3 1.4
1000/7 (=)

Fig. 6.1. BF;3 pressure generated by NaF-NaBF, melts satu-
rated with NiF,-NaNiF; and FeF,-NaFeF3. The lines were
calculated by a least-squares fit of the data according to Eq. (3).

combining (54) and (5b) with free energies of formation
of NiF,(c) and FeF;(c), respectively, known from
Blood’s measurements' ' and with AG/ of liquid NaF ' ?
we obtain

AGS [NaNiF;(¢)]

= [-295.44 + 58.49(T/10%)] ¢ 1.4, (6a)
AG! [NaFeF;(c)]
= [~306.69 + 53.82(T/10%)] £ 1.4. (6b)

Combining (42) and (4b), respectively, with the values
of equilibrium constants for the reduction of NiF,(c)
and FeF,(c) by hydrogen, based on measurements by
Blood,'' we obtain the values of the equilibrium

11. C. M. Blood, Solubility and Stability of Structural Metal
Difluorides in Molten Fluoride Mixtures, ORNL-CF-61-54
(September 1961).

12. JANAF Thermochemical Tables, 2d ed., US. Dept. of
Commerce, NSRDS-NBS-37 (June 1971).
ORNL-DWG 72-589GR
TEMPERATURE (°C)
550 500 450
10 T T T T

] i ! !

™~
N

Pur (otn;i) :
S
/

115 1.20 1.25 1.30 1.35 140 1.45
1900/ o

Fig. 6.2. Calculated HF particle pressures for the equilibrium
(7) (in NaBF 4-NaF eutectic).

constants of the reaction

2HF(g) + M (Hastelloy N)
+ NaF(d) = NaMF;(c) + H,(g), (7)

log K, (Ni) = [~8.98 + 7.58 (10*/T)] * 0.06, (7a)

log K;(Fe) = [-7.96 + 10.04 (10°/T)] + 0.05. (7b)
Assuming that the activities of nickel and iron in
Hastelloy N are ~0.70 and ~0.05, respectively, that the
activity of NaF in the coolant salt is ~0.08 (for the
eutectic composition), and that Py, = 1 atm, from (7a)
and (7b) we may calculate the partial pressures of HF at
which nickel and iron will be oxidized to NaNiF; and
NaFeF 3, respectively, in NaBF4-NaF eutectic:

(8)

(8b)

log Py p(Ni) = 5.12 — 3.79 (10%/T),
log Py (Fe) = 5.18 — 5.02 (10%/T).

A graphical representation of (82) and (8b) is shown in
Fig. 6.2.

Although the above new results are not significantly
different from those reported previously, they are more
accurate. :
47

7. Behavior of Simulated Fission Products

7.1 EFFECTS OF SELECTED FISSION
PRODUCTS ON METALS

J. H. Shaffer W. P. Teichert W. R. Grimes

A joint program of the Reactor Chemistry and the
Metals and Ceramics Divisions has been directed toward
identification of those fission products that are capable
of producing superficial grain boundary cracks in
Hastelloy N and investigation of other alloys of interest
to the Molten-Salt Reactor Program that might be more
resistive to this corrosive process. As previously re-
ported,! corrosive effects on Hastelloy N, not unlike
those found in the MSRE fuel system, have been
produced on metal tensile specimens exposed to rela-
tively low concentrations of tellurium vapors at 650°C
for 1000 hr. Similar tests with selenium, sulfur (in-
cluded for comparative purposes), iodine, cadmium,
arsenic, and antimony failed to show any detectable
attack on the alloy. These findings are based on
metallographic examinations of the exposed tensile
specimens following their stress to failure at room
temperature. Although tests with all these solute
elements have been continued to investigate long-term
effects, the interim program has been directed toward a
more complete evaluation of the implied contribution
by tellurium to the corrosion process.

The experimental method has consisted of a vapor
deposition technique by which the metallurgical tensile
specimens were exposed to tared quantities of tellurium
metal in evacuated quartz ampuls. Prior to the addition
of tellurium, the quartz tubes, with specimens inserted,
were evacuated of gases while heating to about 600°C
for removal of absorbed water. The deposition of
tellurium on the specimens was accomplished by vapor
transport while heating the ampuls to selected experi-
mental temperatures of 550 to 700°C at controlled
rates. Tellurium additions to the quartz ampuls were
based on its assumed uniform dispersion on the metal
surfaces and diffusion to a depth of 5 mils into the
surface layer. Tellurium concentrations of 100, 300,
1000, and 10,000 ppm by weight in the surface layer,
which corresponded to initial surface populations of
about 5.3 X 10'®, 1.6 X 10'7, and 5.3 X 10'® atoms
Te per square .centimeter, respectively, have been used

1. MSR Program Semiagnnu. Progr. Rep. Feb, 28, 1972,
ORNL-4782.

for the various tests in the program. The lower
concentration of 100 ppm approximates concentrations
of tellurium found on metal surfaces of the MSRE fuel
system, and the highest concentration has been cal-
culated to represent the expected tellurium exposure of
an MSBR fuel system during its proposed 30-year life.
Although these tests have been conducted at applicable
temperatures and tellurium concentrations, time at
temperature and tellurium deposition rates have not
corresponded to reactor operating conditions. Con-
sequently, the program has been directed toward
comparative analyses of the effects by tellurium be-
tween Hastelloy N and other alloys that might be of
interest to the MSRP. The selectivity of this potential
corrosion process was established by tests that showed
severe cracking of Hastelloy N under experimental
conditions that failed to produce recognizable attack on
type 304L stainless steel and that produced only very
mild attack on pure nickel.

Experimental conditions that would yield analytically
significant results for this comparative evaluation study
were based on a study of the time at temperature
dependence of the corrosion process on Hastelloy N.
Specimens exposed to 10,000 ppm of tellurium at
700°C were examined at several time intervals from 20
to 1000 hr. The results showed that recognizable grain
boundary cracking occurred within the first 20-hr
period but that the severity of the attack increased at
less than a linear rate dependence on time at tempera-
ture. Specimens withdrawn after 1000 hr under these
test conditions showed effects very similar to those
found on Hastelloy N samples taken from the MSRE
during postoperational examinations of components
from the fuel circuit.

Comparative evaluations of metals exposed to 10,000
ppm of tellurium for 100 hr at 700°C have been
obtained from eight experiments. Some 71 specimens
which represent 14 different alloys or metals and about
24 modifications of Hastelloy N were tested. In
addition to the analytical procedure described, repre-
sentative samples were also provided for examinations
by Auger spectroscopic and x-ray-diffraction tech-
niques. Significant metallurgical findings from analyses
of this specimen group are reported by H. E. McCoy in
Chap. 10 of this report. The results showed generally
that the severity of cracking was greatly reduced in
alloys that were rich in chromium but that minor
composition changes of Hastelloy N yielded no signifi-
cant improvement in resistance to cracking.
Because of the severity of these analytical tests and
the qualitative nature of the results, additional tests are
being conducted with a reduced tellurium concentra-
tion of 1000 ppm for 100-hr exposures at 700°C.
Preliminary results obtained thus far from these tests
indicate that the rate of tellurium deposition may be an
important factor in the corrosive process. Additional
long-term experiments will provide some control over
this experimental variable by using thermal transpira-
tion techniques.

7.2 STABILITY AND VOLATILITY
OF METAL TELLURIDES

C. F. Weaver J.D. Redman

Examination?*? of Hastelloy N specimens from com-

ponents of the Molten-Salt Reactor Experiment has
shown grain boundary cracks on metal surfaces that
were exposed to the fuel salt mixture during operation.
Recent studies*’® have suggested that tellurium may
play an important part in the grain boundary corrosion
of Hastelloy N and, perhaps, of other alloys.® Con-
sequently, the reactions, stability, and volatility of
tellurium fluorides”*® and of structural metal tel-
lurides® are being investigated by mass spectroscopy,
x-ray analysis,® and synthesis.

Two of the three known tellurium fluorides!® were
produced by the fluorination of Te, NiTe,, and Ni;Te,
with CoF;. The results of the reaction of CoF 5 with Te
as well as the mass spectrometric cracking patterns of
TeFgs, TeF,, CoF;, and tellurium were described
carlier.”*® The analogous reactions with NiTe, and
Ni;Te, have been observed to proceed in a similar

2. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1971, ORNL-4676, pp. 147—-66.

3. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 89-106.

4. H. E. McCoy et al.,, MSR Program Semiannu, Progr. Rep.
Feb. 29, 1972, ORNL-4782, p. 109.

S. J. H. Shaffer et al., Reactor Chem. Div. Annu. Progr. Rep.
May 31, 1972, ORNL-4801, pp. 31-33.

6. P. Murray, Reactor Technol 15(1), 23 (1972),

7. C. F. Weaver and J. D. Redman, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4872, p. 91,

8. C. F. Weaver and J, D. Redman, Reacror Chem. Div. Annu,
Progr. Rep. May 31, 1972, ORNL-4801, pp. 33—34.

9. The x-ray analysis was provided by R, M, Steele in the
Metals and Ceramics Division of ORNL.

10, D, R, Vissers and M, ], Steindler, Laboratory Investiga-
tions in Support of Fluid-Bed Fluoride Volatility Processes, Part
X. A Literature Survey on the Properties of Tellurium, [ts
Oxygen and Fluorine Compounds, ANL-7142, p. 22 (February
1966).

48

fashion but with lower pressures as expected. CoF; and
NiTe, produced both TeF, and TeF4 at temperatures
greater than 160°. The excess NiTe, was converted to
Te vapor and solid Ni3 Te, as the temperature increased
to a final value of 780°C. The CoF; and Ni;Te,
produced no vapors below 450°C. From 450 to 550°
both TeF, and CoF,. were evolved. At 600° TeF,,
TeFs, and CoF; were observed. At higher temperatures
complex mixtures formed containing unidentified
species, TeF,, Te,, CoF;, CoF,, and NiF,. The
cracking pattern for the CoF, was observed to be:

Ion Relative intensity
CoF," 35
CoF™ 100
Co®* 25

It was noted that the fluorinating agent CoF; tends to
adsorb rather large amounts of O, and smaller quanti-
ties of H,O and CO, during exposure for several
months in a dry box containing He dried with liquid N,
and with molecular sieve but containing no oxygen-
removing reagent. Exposure to laboratory air produced
essentially the same results in about 2 hr, causing a
1.2% weight gain. Baking this material for several hours
in vacuum at 175°C apparently removed these impuri-
ties completely. The use of this degassed material to
800°C produced no oxides or oxyfluorides in either the
vapor or the residues.

Tellurides of Ni, Fe, Cr, Mo, Bi, and Li are presently
being studied. Three tellurides of nickel, NiTe,, NiTe,
and Ni;Te,, have been observed. Tellurium vapor
reacted with nickel metal to form NiTe, at 412°, NiTe
in the range 550 to 600°, and Ni;Te, in the range 600
to 800°C. The Ni;Te,, which is apparently the most
stable telluride of nickel, was found to slowly decom-
pose at 800° to form nickel metal and tellurium vapor.
The pressures of monomer and dimer were 7 and 4 X
107* torr respectively. These pressures yield AF°g 500 c
of +30.8 and +59.3 kcal for the reactions:

800°C

NijTe,(c) — 3Ni(c) + Te,(g) |

and

Niy Tep(c) 25> 3Ni(c) + 2Te(g)

respectively. The reaction

o

1
3NiTe 2225 Ni, Te, +=Teyt
was used to synthesize small (gram) quantities of the
Ni3Te, from the commercially available NiTe. While
mass spectroscopy detected only tellurium vapor evolv-
ing from NiTe at 705°C, the off-gas trap in the
synthesis experiments was found to contain NiTe, as
well as tellurium. This suggests that NiTe decomposes at
705°C primarily by the above reaction, but also by the
reaction

4NiTe(c) = Ni;Te,(c) + NiTe, ,

producing NiTe, vapor with a pressure of <107¢ torr.
This is our first evidence that a structural metal
telluride may exist in the gas phase. We have no
evidence that Ni;Te, is thermally unstable at 705°C
over a period of several days at pressures in the range of
1075 torr. Under the same conditions at 800°C the
decomposition to nickel and tellurium is complete, with
no evidence of compounds richer in Ni than NijTe,.
CrTe was formed by the reaction of tellurium vapor
at 412°C and 2.4 X 107% torr with chromium in a
quartz container. This material was heated to 750°C in
a nickel container under hard vacuum, with no evidence

49

of volatility, decomposition, or reaction with the nickel
container.

MoTe, showed no measurable vaporization, but
evolved tellurium vapor at temperatures above 750°C.
At 800° the tellurium pressure was near 107 torr,
yielding AF g 4 of about +29 kcal for the reaction

MoTe, (c) > Mo(c) + Te,(g) .

FeTe was found to react with Ni at 725° to form
Ni;Te, and Fe.

Bi Te; produced no volatile species below 450°C.
Between 450 and 800°C, however, the material was
both volatile and unstable; vapor species of BiTe, Bi,
Bi,, Bi;, Te, and Te, were observed. The gas-phase
reaction Bi, + Te, = 2BiTe has been discussed in the
literature.!!

Li;Te was found to react rapidly with air at room
temperature, going completely to Te and LiOH.

11. R. F. Porter and C. W. Spencer, J. Chem. Phys. 32,
943-44 (1960).
50

8. Development and Evaluation of Analytical Methods
A. S, Meyer

8.1 IN-LINE CHEMICAL ANALYSIS
OF MOLTEN FLUORIDE SALT STREAMS

J.M, Dale D. L. Manning
A. S. Meyer

We have concentrated our in-line research and devel-
opment efforts on preparations for the analysis of the
coolant salt in the Coolant Salt Technology Facility
(CSTF). Our research studies have indicated the possi-
bility of measuring corrosion products and an active
protonic species in the NaBF,-NaF coolant stream.

The Salt Monitoring Vessel (SMV) that has been
installed at the CSTF is equipped with six risers, each
fitted with 1%-in. ball valves. Two of these risers are
fitted with Pyrex windows and are angled so that their
axes intersect the axis of one of the vertical risers at the
melt surface and thus provide for illumination and
visual observation of the melt. One of the vertical ports
has been reserved for surface tension experiments, and
the remainder are assigned to analytical measurements.
We have prepared a shielded assembly equipped with
three voltammetric electrodes for insertion in one of
these vertical risers. The shield is a 1-in. Inconel tube
whose open end penetrates the melt to below the level
of the electrodes. It can be purged with He-BF; to
provide fresh melt surfaces and an essentially quiescent

melt around the electrodes. In a second riser we will -

install four voltammetric electrodes where they can be
observed through the viewing ports. We plan to use the
third analytical port for future potentiometric measure-
ments with a proposed LaF; membrane reference
electrode that uses the potential of the Fe-FeF, couple
as a reference. Provision has also been made to vary the
flow through the SMV from 0.15 to 2 gpm so that it
will be possible to study the effects of flow on the
voltammetric measurements.

A large portion of our recent research has been
devoted to the studies of the proton wave described in
the next section. In earlier work we had investigated the
fundamental electroanalytical chemistry of iron in
NaBF,! and had observed the Cr(III) - Cr(0) reduction
wave in a region (~—1.0 V) near the cathodic limit of
the NaBF, melts.? Recently we performed vol-
tammetric measurements on a melt prepared from the
material used to charge the CSTF in order to determine
whether it presented any unusual features and to
measure the increase in the diffusion coefficients of the
corrosion product ions with temperature.

At 500°C the melt showed the anticipated iron wave
and a rather large proton wave (despite a low OH
analysis and careful prepumping) and no chromium
wave. On heating to 600° the cathodic limit of the melt
decreased several hundred millivolts at a platinum
working electrode so that it was not possible to observe
the reduction wave from 200 ppm of added Cr(III). An
even greater decrease in the limit was noted at the
palladium electrode. We attribute this change in the
limit to the alloying of the boron reduction product
with the electrode material.

We have screened a number of possible alternate
electrode materials and found that iridium, pyrolytic
graphite, and copper yield resolved chromium waves at
600° that are comparable to those obtained at
platinum at 500°C. On a gold electrode the chromium
wave appeared about 100 mV earlier, giving better
resolution from the limit apparently as a result of
alloying of the deposited chromium. Silver offered no
advantages, and rhenium and tantalum were too active
for use in NaBF, melts.

Our present research cells cannot be operated at the
maximum operating temperature of the CSTF
(1250°F). We have chosen a selection of three iridium
electrodes and one each of pyrolytic graphite, copper,
gold, and evacuated palladium in order to offer the best
opportunity to perform chromium analyses at 1250°F.,

8.2 VOLTAMMETRIC STUDIES OF PROTONATED
SPECIES IN MOLTEN NaBF,-NaF (92-8 mole %)

D. L. Manning  A. S. Meyer

We have continued studies on the voltammetric
behavior of protonated species in molten fluoroborate
salts at an evacuated palladium tube electrode® The
initial studies were conducted with the melt contained
in a graphite cell, which was then enclosed in a silica
jacket. However, to eliminate any effect of silica and to
achieve higher melt temperatures, a nickel cell enclosure
was fabricated. Both graphite and copper cells have
been utilized. Voltammograms recorded at palladium or

1. Fred R, Clayton, Ir.,“Electrochemical Studies in Molten
Fluorides and Fluoroborates,” doctoral dissertation, University
of Tennessee, December 1971, pp. 45—-47.

2. D. L. Manning, MSR Program Semiannu. Progr. Rep. Feb.
29, 1972, ORNL-4782, p. 82.

3. D. L. Manning and A. S. Meyer, ibid., p. 83.
palladium-silver (75:25) alloy electrodes revealed a
well-defined wave at about —0.1 V vs a platinum
quasi-reference electrode. The wave conforms to a
reversible one-electron reduction by voltammetric and
chronopotentiometric criteria. By electrolyzing at a
potential beyond the peak potential of the wave, an
increase in pressure was observed inside the electrode.
This is evidence that the wave is due to the reduction of
protons with part of the resultant hydrogen diffusing
through the palladium. It was first believed that it
would be possible to relate the pressure measurements
to the protonic content of the melt under controlled
electrolysis conditions. Experience with different elec-
trodes, however, has revealed that the “porosity” to
hydrogen among the electrodes is not necessarily
constant; also, pressure measurements at a given elec-
trode will change (usually decrease) with time in the
melt. Nevertheless, it is still possible to collect gas
samples by this technique for measurement of isotopic
ratios of hydrogen. '

Our first assumption was that the voltammogram
resulted from the reduction of —OH as NaBF;OH at
the palladium electrode. However, the voltammetric
results were consistently low when compared with
infrared data. Also, we observed that the wave height of
the voltammogram can be significantly reduced by
- cooling the top of the nickel cell enclosure and can be
returned to the original value by heating. We now
believe that the protonic species yielding the vol-
tammetric wave is in equilibrium with a vapor species
that can be condensed at cooler temperatures. More-
over, nomarked variation in the BF;OH ™ concentration
as measured by the infrared pellet technique appeared
to be associated with the marked changes in the
reduction wave. We have also noted that when the
vapor species is removed by sweeping with helium, the
wave decreases, but recovers slowly after the sweep is
discontinued. This suggests that the more electroactive
species may be replenished via slow reactions of a more
stable protonated component.

To explain these results we have postulated reactions
such as

2NaBF;0H + NaBF, == Na,(BF;),0
+ NaF + HBF ; OH = vapor,
NaBFgoH t+ 2N3BF4 = Nag(BF3)20

+ NaF + HBF, = vapor,

in which the melt equilibrium strongly favors the
NaBF;0H species. To be consistent with our observa-

51

tions, the diffusion coefficient of the more active
species must be higher than the typical value of 107%
cm?/sec in fluoride melts so that the active proton
concentration is lower than we suspected.

To test these hypotheses, we added crushed chro-
mium metal to a fluoroborate melt and followed the
voltammetric wave and the BF3;OH ™. Pertinent observa-
tions at S00°C are recorded below.

Proton wave H as

Time

height (uA) NaBF;0H (ppm)
~1 hr 3300 52
4 hr 500 42
1 day Not detected 32
2 days Not detected 29
Not detected 24

8 days

Of interest here is that after eight days, the BF;OH™
level has decreased approximately one-half, whereas the
voltammogram, for practical purposes, was lost within
the first day. This appears to be additional evidence
that the voltammetric and infrared techniques are not
observing the same protonated species.

After eight days at 500°C the temperature was
increased to 600°C, and a sample was withdrawn after
the melt was maintained at 600°C for three days. It was
also discovered that the cell had developed a leak while
at 600°C. The sample indicated 47 ppm H as NaBF;OH
(an increase from 24 ppm at 500°C) and 0.22% Cr.
Thus the NaBF;OH production from contamination,
most probably moisture, proceeded faster than its
reaction with the Cr metal. At no time, however, was a
voltammetric proton wave seen, and the equilibrium
potential remained cathodic (melt reducing). Acidic
protons (HG, HBF,, etc.) were apparently consumed
rapidly by the Cr metal as they were not observed
voltammetrically during the buildup of the NaBF;OH.

The 0.22% Cr in the filtered sample seems high. The
cathodic equilibrium would seem to indicate predomi-
nantly Cr?* in the melt, which may be more soluble
than Cr**. However, this is yet to be established
experimentally. _

After the leak was detected, the salt was frozen and
the cell dismantled for inspection. A white coating
analyzed to be predominantly NaBF, was deposited on
the flanged top of the cell. The route by which the salt
collected on the flanged top, about 12 in. from the
surface of the melt, is not clear. The gold thermowell
("4-in-OD tube, closed end, ~%;,-in. wall) was coated
extra thickly with green-colored salt and underneath
was a bright metallic plate deposited on the gold. A
sample of the metallic film was found to be chromium.
This would seem to support the existence of Cr(ll),
which could disproportionate at the coolest surface (in
this case the thermowell) into Cr metal and Cr®* as the
salt was gradually frozen. It now seems apparent that
the protonic species present and associated equilibria
are more complex than our original concepts. We are
continuing these studies, however, and attempts will be
made to evaluate the diffusion coefficient of the
electroactive protonated species so that the concentra-
tion of this substance can be determined.

8.3 INFRARED SPECTRAL STUDiES
OF NaBF,-NaF

J.P. Young A.S. Meyer

There is still some confusion as to the chemical
reactions and stable species of H, B, F, and O that exist

52

in NaBF,-NaF (92-8 mole %). Progress in our under--

standing of these melts has been made, however. From
infrared studies of NaBF,-NaF melts described in the
previous reporting period, we had established the
existence of BF;0D  ion in the melt but had not found
the route of entry of this species, whether by chemical
reaction or exchange. The usefulness of the NaBF,
coolant salt in the trapping of hydrogen isotopes
(tritium) is not necessarily based on an exchange
reaction with BF;OH™, even though thisis a possible
mechanism. The studies we have carried out® seem to
demonstrate a chemical reaction of diffused hydrogen
isotopes with the melt rather than an isotopic exchange.
This apparent lack of exchange in melts, coupled with a
continuing unsatisfactory correlation between the re-
sults of the deuterium exchange technique described in
an earlier report® and the infrared pellet technique® for
the determination of combined hydrogen in samples of
NaBF,4, has led to a discontinuance of the isotopic
exchange method. Based on present knowledge, the
apparent exchange we observed in the earlier studies
was probably caused by a side chemical reaction or
exchange with a species other than BF; OH™. Additional
confirmation of the original calibration® of the infrared
method has been obtained by mixing known quantities
of solid NaBF3;OH with solid NaBF ;-NaF and melting
these mixtures in vacuum-tight crimped copper con-
tainers maintained -under isothermal conditions. In
these tests, NaBF;OH additions corresponding to 25

4. Section 8.4, this report,

5. J. P. Young and A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, p. 83.

6. J. P. Young, J. B. Bates, M. M. Murray, and A. S. Meyer,
MSR Program Semiannu. Progr. Rep.. Aug. 31, 1971, ORNL-
4728, p. 73.

ppm of hydrogen or less were recovered quantitatively.
Less than quantitative recoveries of larger additions
(e.g., 40 ppm) are probably caused by the separation of
a volatile proton species which does not completely
reequilibrate with the salt when it solidifies. Indeed,
under nonisothermal conditions, a tan-colored viscous
liquid can be condensed on a water-cooled surface (see
also Sect. 8.2, this report). This liquid fumes in air and
was shown by NMR analysis to contain essentially equal
molar quantities of H and F. From the NMR chemical
shift, it is indicated that the hydrogen is not ionic.
Thus the liquid cannot be HBF 4 or HBF;OH but might
be BF;-H,0, BF3°2H20, ~or a compound of
BF,OH, ;. Although a very careful technique is
followed, it is not possible to melt NaBF, or NaBF,-
NaF mixtures of low BF;OH™ content (7 ppm H) in the
crimped isothermal containers without causing the
BF;OH™ concentration to increase by about 15 ppm
hydrogen based on the infrared pellet analysis. It
appears that this increase is caused by some other form
of proton, most likely H,O, that is converted to
BF;0H™ on melting. If the salt containing 20 ppm
hydrogen is ground in air and again melted, a further
increase of 15 to 20 ppm is observed. If the product of
the first melting is ground in an inert atmosphere box
and remelted, no furtherincrease in BF;OH ™ concentra-
tion is observed. We had reported® that sometimes we
could melt samples and keep the hydrogen concentra-
tion at approximately 6 to 8 ppm. These salts were
melted in a nonisothermal gas phase system of very
large gas-to-melt volume ratio, where it seems likely
that the BF;OH ™ was prevented from forming or the
excess was decomposed by an as yet undefined mecha- -
nism. From a consideration of spectral and electro-
chemical results, there are at least three kinds of
protonic species that can find existence in molten or
solid NaBF,. This fact undoubtedly is a partial cause of
our inability to obtain a precise analytical determina-
tion of hydrogen in these salts by the several methods
available. We plan to undertake further investigations in
the preparation or synthesis of NaBF, melts that
contain -essentially no BF;OH™ so that the various
proposed solute species and their interactions can be
better studied.

8.4 STUDIES OF PROTONS IN MOLTEN NaBF,

J.P. Young J. B. Bates
G. E. Boyd

Several studies of the reaction or exchange of D, or
H, with solute species in molten NaBF,-NaF have been
carried out. Our most definitive results, thus far, have
ORNL-DWG 73-1254

0.20
e—9-—0=0—9
/‘r.’fi—
.
.
- e
Ig 0.15 ./|
rd

o ./.
& /
o o
L /

0.10
W /
Q .
Z /
< *
@ 7
x .)
S ¢~ T=0 START D, DIFFUSION
@ 0.05 P=12 psig

$i0, CELL
0
0 5 10 15 20 25
TIME {min)

Fig. 8.1. Growth of BF30D  as D, is diffused into molten
NaBF4-NaF.

been obtained from experiments in which either D, or
H, was independently diffused through 0.004-in.-wall
nickel thimbles into molten NaBF,-NaF contained in a
Si0, cell at a temperature of approximately 425°C.
The presence or absence of the BF3;O0D ion and its
change in concentration was followed spectrophoto-
metrically by changes in the intensity of its character-
istic infrared absorption peak at 2690 cm™ . In a SiO,
cell one cannot observe similar changes in the BF;OH™
ion since that absorption peak is masked by absorption
peaks of the container. Under our experimental condi-
tions, if D, is diffused into a fresh melt, some BF;0D
ion is formed initially, but within 20 min the intensity
of the 2690-cm™ peak reaches a maximum and levels
off, as shown in Fig. 8.1; no further changes in the
intensity of this peak are noted if the diffusion of D, is
continued. If H, is diffused into the melt first (for 2 hr)
and then D, is allowed to diffuse into the melt, no
BF;0D absorption is seen. These results in the SiO,
apparatus suggest that D, does not exchange with
BF;0OH ion in the melt, but rather that D, (or H,) is
capable of reducing some species in the melt. One of
the ultimate products of this reduction is BF;0D . If
D, is first diffused, the BF;0D ion is observed; if H, is
first diffused, the reduction is completed with H;, and
no BF;0D is seen when D, is subsequently intro-
duced. We have carried out similar studies with melts
contained in the LaF; cell,” but the results have been
inconclusive because NaBF, creeps out of these small
cells when gas is passed into the melt.

7. 1. B. Bates, J. P. Young, and G. E. Bovd, MSR Program
Semiannu. Progr. Rep. Feb. 29, 1972, ORNL-4782, p. 59.

53

8.5 ANALYSIS OF COOLANT COVER GAS
R.F. Apple  A.S. Meyer

In order to calibrate transducers for the measurement
of hydrolysis products in coolant cover gas, it will be
necessary to measure the ‘‘water” that is present in
actual or simulated cover gas streams, probably as BF,
hydrates. Our earlier tests have shown that the BF; in
such gas streams interferes with the determination of
water by the Karl Fischer (K.F.) titration. It appears
likely that the BF; interference can be eliminated by
absorbing the water (and BF;) in cold pyridine and
then separating the water by azeotropic distillation.
Application of this technique will require a K.F.
titration of improved sensitivity. D. J. Fisher and T. R.
Mueller of the Analytical Chemistry Division’s Instru-
mentation group have designed an improved automatic
coulometric titrator for this purpose. This instrument
decreases the generating current as the end point is
approached and provides an automatic adjustment of
the compensating current that is required to maintain
the titrating solution at the end point.

Although the instrument functioned according to its
design specifications, we were unable to exploit its
improvements because of random variation in the
potential of the indicator electrode system. In the
coulometric titration the sample is added to a solution
of essentially depleted K.F. reagent, that is, a solution
of pyridine, pyridinium iodide, excess SO, , and a trace
of iodine in methanol. Iodine is then generated electro-
lytically until the original activity of free iodine is
reached. Thus each mole of water added consumes one
mole of coulometrically generated iodine.

The precision of the technique is primarily limited by
the precision with which the free iodine activity at the
end point can be reproduced. The free iodine can be
detected by measuring the potential that is developed
between platinum electrodes immersed in the titrating
solution and polarized with microampere currents.
Where excess water is present, iodine is generated at the
indicator anode, while hydrogen is generated at the
cathode. Thus the potential approaches the H,-I,
couple (typically 600 mV). Conversely, when a large
excess of iodine is present, the cathode is depolarized,
and the potential approaches zero. In practice the most
sensitive region of the titration curve is found at iodine
concentrations corresponding to about 200 mV poten-
tial between the electrodes, and the end-point potential
is selected at about this value. The iodine concentration
at the cathode, demonstrated to be the potential-
determining electrode, is determined by both the rate
of reduction (density of polarizing current) and the rate
the free iodine reaches the electrode. Because of
turbulent flow in stirred solutions the concentration of
iodine varies randomly, and variation in the potential
between the electrodes becomes excessive.

We have postulated that if a rapidly rotating cathode
were used the rate of iodine transport to this electrode
would be determined primarily by the constant velocity
of -the electrode through the solution rather than by the
random movements in the stirred solution. An electrode
was constructed of 0.040-in. platinum wire approxi-
mately % in. long attached at right angles to a Y-in.
rotating shaft. The electrode and shaft were insulated

54

with a combination of Teflon spaghetti and shrinkable

viny! tubing. Only the tip of the platinum was exposed,
so that a relative large velocity was achieved on
rotation. By a systematic variation of the parameters
associated with this electrode, we have found that good
results can be obtained with 7 mm of 0.040-in.
platinum exposed and rotated at 1800 rpm with a polar-
izing current of 35 pA. Under these conditions the re-
sponse is about 15 mV per microgram of water in approx-
imately 100 ml of titrating solution, and the output
voltage appears essentially free of noise on the damped
meter of the instrument. In the titration of test
* portions of methanol containing about 500 g of water,
values of 500 * 3 ug were obtained, whereas reproduci-
bility to better than 1 pg was obtained with smaller
injections (~30 ug). It is likely that this reproducibility
is now limited by the precision of our injection
technique. More optimum conditions can probably be
found using a mechanically improved rotating system
now under construction, but these results are adequate
for our present usage in determining water in coolant
cover gas.

These investigations were made with a prototype
system in which the electrode shaft passes through a
Teflon bushing. Alignment is very critical, and after a
few hours of operation some Teflon “‘chaff” enters the
cell and affects the electrode potential by covering part
of its surface. An improved bearing and shaft seal are
now being fabricated.

We are testing a technique for determining the
hydrogen in coolant cover gas by measuring the
pressure of hydrogen as it diffuses through a heated
palladium diaphragm. At equilibrium the same partial
pressures of hydrogen are present on both sides of the
membrane, and by using prereductive techniques it may
be possible to obtain total hydrogen as well as
uncombined hydrogen in an untreated gas stream.
Initial tests using a hydrogen purifier equipped with a
heated diffusion tube of Ag-Pd alloy showed that, while
the pressure outside the tube did approach the partial

pressure of hydrogen inside the flow-through system,
equilibrium was approached quite slowly, particularly
when going to lower pressure. This has been attributed
to a reservoir of hydrogen that dissolves at high pressure
in the cool ends of the diffusion tube. We are rebuilding
the system to provide a more uniform temperature for
the diffusion tube.

8.6 SPECTRAL STUDIES OF MOLTEN SALTS
~J.P. Young

As in the past, spectral studies of solute species in
molten salts of interest to the MSRP have been carried
out to investigate possible analytical applications and to
gain insight into the chemistry of these melts. Studies
are carried out in BeF,-based melts, of direct interest to
the MSR, fuel salt, and alkali fluoride melts which
provide change in solvent character and are also of
interest in developing general knowledge of molten
fluoride solution chemistry. Two further papers have
been written, one published® and one accepted for
publication,” in which our views are given on the
formation and stability of high oxidation states of
several solute species in the presence of O*” and/or F~
ligands in fluoride melts.

In cooperation with F. F. Blankenship, S. Kirslis
(Reactor Chemistry Division), and S. Katz (Chemical
Technology Division) we are studying the possibility of
using the reduction of Ce(IV) to Ce(Ill) with the
subsequent fluorimetry of Ce(IIl) as an analytical
method for determining the reducing power of amen-
able melts. Katz and Pitt'® have described this tech-
nique as a very sensitive means of determining organics
in aqueous solution, and we have considered using this
in an inorganic melt.

As a general analytical scheme the possible determina-
tion of reducing power by Ce(Ill) formation from
Ce(1V) seems of value; either spectral absorption or
fluorimetric measurements could be made. An interest-
ing reagent for use in such a scheme would be CeQ,.

8. F. L. Whiting, G. Mamantov, and J. P. Young, *‘Spectral
Studies of 0°~, NO, , and CrO4% in Molten LiF-NaF-KF and of
0% in Liquid Ammonia,” J. Inorg. Nucl Chem. 34, 2975
(1972).

9. F. L. Whiting, G. Mamantov, and J. P. Young, “Electro-
chemical Generation and Spectrophotometric Study of Solute
Species in Molten Fluoride Media,” J. Inorg. Nucl Chem.,
accepted for publication (1973).

10. S. Katz and W. W. Pitt, Jr., “A New Versatile and
Sensitive Monitoring System for Liquid Chromatography:
Cerate Oxidation and Fluorescence Measurement,” Anal. Lett.
5,177 (1972).
Based on analogous chemical behavior, one might guess
that CeO, would be very insoluble in fluoride melts,
while Ce(1ll) would be soluble even in the presence of
O?*. Several preliminary studies have been carried out
in the solvents LiF-NaF-KF and LiF-BeF,-ZrF,. In
either melt it was found that Ce(Ill) was sufficiently
soluble for use in this scheme. The absorption spectra,
in either melt at 500°C, in the spectral region of 600 to
200 nm, consist of three peaks located at 292, 244, and
224 nm. The spectrum corresponds to the absorption
spectrum seen for Ce(lll) in CaF, crystals.!’ No
fluorescence of the melt has been observed, but the
solidified salt did fluoresce. The molar absorptivity of
the most intense peak at 292 nm is estimated to be 700
liters mole ™ cm™!. Addition of CeQ, to a melt of
LiF-NaF-KF generated some Ce(IIl) which was seen
spectrally; at equilibrium some undissolved CeQO, still
remained. Based on the absorbance of the 292-nm peak,
3 ueq of some material in the salt was oxidized. Under
our experimental conditions, down to 0.1 umole of
Ce(Ill) per gram of sample could be determined by
absorption spectral measurement. The method was
originally proposed to be considered as an alternate
method for determining very. low concentrations of
U(III). We envision that its use as a general reducing
power indicator might be of more use, either by melt
absorption measurements or melt or solid fluorescent
measurements. The evidence to date would indicate
that CeQ, is not as strong an oxidant in these melts as,
for example, MnF;, but CeO, should be strong enough
to oxidize Ni°, Fe®, etc., that might be present in such
melts and that in the past have created problems in
analysis when present with Ni** and Fe?".

Tellurium is a fission product of the fuel burnup in
molten-salt reactors and has been implicated in some
stress cracking that has been observed in the metal alloy
which has been used in the MSRE. Because of this
interest in Te, spectral studies of the various oxidation
states will be carried out. Initially only Te*” or Te® has
been considered. 1t has been found, spectrally, that
both ZnTe and Na, Te are soluble in LiF-NaF-KF and
that Na, Te is soluble in molten LiF-BeF; (ZnTe was

55

not investigated). The spectra of the resultant rose-
colored solution probably arise from the Te? ion and
consist of a moderately broad peak at 520 nm and a
very intense absorption in the ultraviolet below 350
nm. The spectra are similar but not identical to that
reported by Gruen et al.'? for Li,Te and Cs,Te in
melts that were essentially chloride media. We do not
know the sensitivity of the absorbance peak at 520 nm,
but we estimate the molar absorptivity to be between
25 and 50 liters mole ™! ¢m ™. The solubility of either
telluride, therefore, is believed to exceed 1000 ppm in
LiF-NaF-KF at 525°C. The dissolved Te*  ion is
reasonably stable in an inert atmosphere at least for 24
hr, but has been found to be easily oxidized by
exposure of the melt to air, possibly an oxidation via
O, or by the addition of Fe?*.

In the last report we described the spectral studies of
LisBi in molten LiCl that we had undertaken in
cooperation with L. Ferris and his group (Chemical
Technology Division). We have now carried out further
spectral studies of Li3Bi in LiBr and of Li-Pb alloys in
contact with LiCl. To summarize our results, Li;Bi
seems to be the ionized solute species in either molten
LiCl or LiBr near the LiX melting point. The reddish-
yellow solution has an essentially featureless ultraviolet
absorption peak which appears similar to that obtained
for LizBi in molten LiCI-LiF (80-20 mole %).!® No
absorption spectrum, or color, was observed when Li-Pb
alloy was in contact with molten LiCl, and it seems
probable that no unusual “plumbide” species were .
present in the salt melt. These results are in line with
chemical analyses of samples of these melts.

11. P. P. Feofilov, “Absorption and Luminescence Spectra of
Ce*** lons,” Opt. Spectrom. (USSR) 6, 150 (1959).

12. D, M. Gruen, R, L. McBeth, M. S. Foster, and C, E,
Crouthamel, “Absorption Spectra of Alkali Metal Tellurides and
of Elemental Tellurium in Molten Alkali Halides,” J. Phys.
Chem. 70, 472 (1966).

13. M. S. Foster, C, E, Crouthamel, D. M. Gruen, and R. L.
McBeth, *‘First Observation of a Solution of LisBi, an Inter-
metallic in Molten LiCl and LiCI-LiF,” J. Phys. Chem. 68, 980
(1964). :
56

9. Other Molten-Salt Research

9.1 THE DISPROPORTIONATION EQUILIBRIUM
OF UF; SOLUTIONS IN GRAPHITE

L. O. Gilpatrick L. M. Toth

In the previous report! we showed that a UC, phase
was the stable carbide formed in the UF; dispropor-
tionation equilibrium:

4UF3(d) + 2C(graphite) = 3UF,4(d) + UC, (D)
when dilute (<1 mole % UF; plus UF;) molten
fluoride solutions were equilibrated in a graphite
spectrophotometric cell. In order to determine equilib-
rium quotients:

Q'= (UF,)* /(UF3)* (2)
for Eq. (1), the reaction has been measured by
approaching the equilibrium from both directions,
either by disproportionating UF; solutions in a graphite
cell or by reacting excess UC, with dilute UF,
solutions. We have found that the best method of
measuring equilibrium quotients is by the back reac-
tion, and for this reason the @ value is written as the

reciprocal of Q' [Eq. (2)] when applied to the back -

reaction of Eq. (1).

Because the equilibration reactions were conducted in
the spectrophotometric cell, UF; and UF,; concentra-
tions could be monitored continuously by absorption
spectroscopy. Details of this procedure have already
been given! and will be reviewed in a final report.2

Extensive quantitative data have been obtained for Q,
for the back reaction of Eq. (1), as a function of
temperature and melt composition. These data repre-
sent refinements of earlier results! as a consequence of
improvements in the data analyses. These improvements
are primarily due to better base-line determinations and
calibration data for the absorption spectra of UF; and
UF,;. We have found that the equilibrium is more
sensitive to temperature and solvent changes than was
originally anticipated by previous investigators.3

1. L. M. Toth and L. O. Gilpatrick, “The Disproportionation
Equilibrium of UF3 Solutions,” MSR Semiannu. Progr. Rep.
Feb. 29, 1972, ORNL-4782, p. 98.

2. L. M. Toth and L. O. Gilpatrick, The Stability of Dilute
UF5-UF4 Solutions Contained in Graphite (in preparation,
1972).

3. G. Long and F. F. Blankenship, The Stability of Uranium
Trifluoride, Part II, ORNL-TM-2065 (November 1969).

In Fig. 9.1 the logarithm of Q is plotted as a function
of Tx! (Tx is the Kelvin temperature) for the back
reaction of Eq. (1) in the solvent LiF-BeF, (66-34 mole
%). Also shown in the figure at the right ordinate is the
ratio UF;3/(UF,; + UF,), equal to R, and at the top of
the figure is the centigrade scale. Two lines drawn
through the data points represent the experimental
uncertainty, which arises mainly from the base line in
the absorption spectra.

Equilibria at various temperatures were approached
from both the high-temperature (open circles) and
low-temperature (closed circles) directions. To ap-
proach equilibrium from the high-temperature direc-
tion, the system was initially held ca. 50°C above the
temperature sought until the UF; concentration ceased
to grow [UF, reacting with UC, in the back reaction of
Eq. (1)]. Then, the temperature was dropped 50° and
the UF; concentration allowed to decrease by the
forward reaction of Eq. (1) until no further change

ORNL-DWG 72-10719R
TEMPERATURE (°C)

500 550 573 600 622 650 700
Il l | [ I /I
‘ / / 0.6
L
/.
-3
/ /¢ — 05
/ 5/ —{ 0.4
-4 : s
L 4
/ t )/
. —
- 0.3
i >/ / o
B . +
< . ™
o ‘ 5
-6 / u
Il
/ )
-7 §/ —1 0.4
- //
8 i/
-9 — 0.03
130 1.25 1.20 145 140  1.05  1.00
1000 /7 (o)

Fig. 9.1. Equilibrium quotients vs temperature for the reac-
tion UC; + 3UF4 = 4UF3 + 2C in LiF-BeF, (66-34 mole %).
occurred. Equilibrium could be shifted repeatedly in
this manner by varying the temperature of the system.
It is shown in Fig 9.1 that the equilibrium is very
sensitive to temperature changes, shifting by AQ = 107
in the range from 500 to 700°C. This indicates that the
stability of UF; is increased from ca. 5% at 500°C to
ca. 60% at 700°C.

Solvent effects on the equilibrium also have been
found to be very large. Changing the composition from

LiF-BeF, (66-34 mole %) to (48-52 mole %)* at 500°C

increases the Q value by 10°, as found by comparing
Fig. 9.2 with Fig. 9.1. The plot of log Q vs 1/Ty for the
equilibrium of the reverse reaction for Eq. (2) in
LiF-BeF, (48-52 mole %) is similar to that of the
previous figure. However, comparable concentrations of
UF; can be held at temperatures which are 150 to
200°C lower than in the LiF-BeF, (66-34 mole %) case.
The two sets of @ values differ from each other
simply by the ratio of activity coefficients, v of UF;
and UF,; in the two solvents raised to the appropriate
powers. Baes> has assigned a value y = 1 to UF; and
UF, in LiF-BeF, (66-34 mole %) and has shown on this
basis that a-y = 0.7 and 10 was determined for UF; and
UF,, respectively, in LiF-BeF, (48-52 mole %). This
yields a ratio: ' ‘

QOrLp/0L,p =4000.

Our data for Q; g when extrapolated linearly to 600°C
yield a Qp g/Q1,p = 10,000, which agrees with Baes’s
predictions. It should be noted that the data of Fig. 9.2
suggest a slight curvature which if extrapolated on a
nonlinear basis would yield even better agreement.

The data of Figs. 9.1 and 9.2 in the “laboratory
system’ of LiF-BeF, mixtures has been used as a basis
for study by which the stability of UF; in reactor-type
solvents may be understood. Our earlier predictions®
were that for LiF-BeF,-ZtF, (64-29-5 mole %) the log
@ vs 1/Tk plot of Fig. 9.1 would be shifted upward by
ca. @ = 10 for the entire temperature range. This shift is
due to the ZrF;, which decreases the fluoride ion
concentration and causes an increase in the UF;
stability. For LiF-BeF,-ThF, (72-16-12 mole %) we
also predicted that UF3 should be less stable than in
LiF-BeF, (66-34 mole %). However, equilibration
measurements have recently been performed as de-
scribed for the LiF-BeF, solvent systems, and they

4. LiF-BeF, (66-34 mole %) designated as “L,B”’; LiF-BeF,
(48-52 mole %) designated as “LB.”

5. C. F. Baes, Jr., MSR Semiannu. Progr. Rep. Feb. 28, 1970,
ORNL-4548, p. 153.

6. L. M. Toth and L. O. Gilpatrick, MSR Semiannu. Frogr.
Rep. Aug. 31, 1971, ORNL-4728, pp. 77-78.

ORNL-DWG 72-10718

TEMPERATURE (°C)

370 400 450 500
l | I |

0.6

0.5

_4 ha

/ rd
/ f/ 0.3 [
-] . <
_5 / 5
/ i
> g
Q o — 0.2 _
g ' e
-6 A o

=

74
W

1.55 1.50 1.45 1.40 1.35 1.30
1000/r(.,K)

Fig. 9.2. Equilibrium quotients vs temperature for the reac-
tion UC, + 3UF,4 = 4UF3 + 2C in LiF-BeF, (48-52 mole %).

show definitely that UF; is at least as stable as in the
LiF-BeF, (66-34 mole %) plot of Fig. 9.1. A detailed
log @ vs 1/Tk plot for LiF-BeF,-ThF, (72-16-12 mole
%) will be presented later,?

9.2 EMF STUDIES OF OXIDE EQUILIBRIA
IN MOLTEN FLUORIDES

R.G.Ross C.E. Bamberger C.F.Baes, Jr.

Oxide equilibrium studies have continued because of
their relevance to MSBR fuel reprocessing development
and to its oxide tolerance. Previously the solubility
product for ThO, has been estimated from Pa, Q5 and
PaO, solubility measurements,” and the solubility
product for NiO has been estimated from measurements
in LiF-BeF, solutions.® However, these estimates have
not been corroborated by direct measurement.

7. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR
Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782.

8. C. F. Baes, J1., in Nuclear Metallurgy (Symposium on the
Reprocessing of Nuclear Fuels, ed. P. Chiotti), vol. 15,
USAEC-CONF 690801 (1969). ' ’
58

D. D. Sood’s? attempts to determine the ThO,
solubility product by directly measuring the oxide in
solution were not satisfactory because the oxide anal-
yses had considerable scatter. It seemed desirable,
therefore, to utilize a method not solely dependent on
oxide analyses. Thus, we have initiated emf studies of
the equilibrium

ThO,(c) + NiF,(d) = ThF4(d) + NiO(c) (1)

in LiF-BeF,-ThF, (initially 72-16-12 mole %). Through
a combination of the emf measurements and material
balance we expect to obtain precise values for the
solubility products of ThO, and NiO in these melts. In
addition, by varying the composition of the melt, we
will obtain a measure of the activity coefficients of Ni**
and Th*', which can be compared with previous
estimates.!©

The NiF, concentration in these experiments is
followed potentiometrically with a reference electrode
designed by Bronstein and Manning.!! In this elec-
trode, Li, BeF,; saturated with NiF, is contained in a
single-crystal LaF; cup. Efforts to date have been
primarily concerned with the design and assembly of
equipment, development of analytical and handling
techniques, the preparation of “oxide free” starting
materials, and elimination of design inadequacies asso-
ciated with the LaF; reference electrode.

We have been able to prepare salt which, based on HF
evolved during stripping after hydrofluorination and the
equilibrium

OH™ + F (d)= 0*(d) + HF(g) , (2)

contained 20 ppm oxide. A direct analysis (using a
refined KBrF, fusion method) gave an oxide concentra-
tion of 31 + 3 ppm. We feel that this is quite good
agreement, particularly since the equilibrium quotient
for the above reaction was estimated from a value for
Li,BeF4. In addition, it reflects our ability to avoid
contamination of the samples in handling prior to
analysis. The LaF; reference electrode has been modi-
fied a number of times to eliminate problems such as
crystal fracture, loss of electrode contact, creeping of
salt, and shorting of insulators.

9. Guest scientist from Bhabha Atomic Research Centre,
Bombay, India, private communication.

10. C. F. Baes, Ir., MSR Program Semiannu. Progr. Rep. Feb.
28, 1970, ORNL-4548, pp. 149-52.

11. H. R. Bronstein and D. L. Manning, J, Electrochem. Soc.
119, 125 (1972).

In the several runs to date, the initial potential has
been stable and approximately the value we expected.
As reliability of the electrode is demonstrated, measure-
ment of equilibrium (1) will be completed, and emf
measurements will be extended to other oxide-fluoride
equilibria of the general type

MF,(d) +—§Ni0(c) = MO, /2(c)

+12‘N1F2(d)+y ;xNio(c) :

wherein the dissolved fluoride MF, may be UF,, ThF,,
ZrF,, TiF,, or CrF, and the corresponding oxide is,
respectively, UO,, ThO,, ZrQ,, TiO,, or Cr;, 0,.

9.3 THE LITHIUM FLUORIDE—
TETRAFLUOROBORATE PHASE DIAGRAM

A.S.Dworkin M. A. Bredig

Determination of this phase diagram rounded out the
studies of the alkali metal fluoride—tetrafluoroborate
binary systems at ORNL. A comparison of activity
coefficients calculated from Fig. 9.3 with those for the
sodium, potassium, and rubidium systems calculated by
Moulton and Braunstein!2 shows an increasing positive
deviation from ideality with decreasing cation size. At
1000°K and mole fraction Xgr,” of 0.2, 0.4, and 0.6,
activity coefficients vy ;F were 1.16, 1.32, and 1.40,
while values of yN,r were 1.10, 1.20, and 1.23. The

12. D. M. Mouiton and J. Braunstein, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 100.

ORNL~DWG 72-6104A

900 I \J>\
800 P
~
~ S~ T
2 ~ ~Nd
& N
T 500 IDEAL
% ~ \
& 500 AN Y
a
s \\
w
400
300 —-0> "S-@
200
LiF 20 40 60 80 LiBF,
LiBF; (moale %)

Fig. 9.3. Phase diagram of the system LiF-LiBF,.
potassium system shows very slight positive deviation,
while the rubidium system is essentially ideal. This
trend most likely reflects the smaller distance and
consequently greater multipole interaction between
BF,  ions in the presence of smaller cations.

9.4 CHRONOPOTENTIOMETRY BASED
ON DIFFUSION OF MOBILE NONELECTROACTIVE
SPECIES

J. Braunstein H. R. Bronstein J. Truitt

It has previously been demonstrated that the elec-
trical transference number of Be(Il) relative to fluo-
ride ion as the reference constituent is zero in molten
LiF-BeF, mixtures at compositions between 0.3 and
0.7 mole fraction BeF,,!3 and probably beyond. We
have pointed out!# that if alkali ions remain the sole
current carriers (relative to fluoride) in these systems as
the BeF, content is increased (and the BeF, network
structure develops), a novel and potentially useful
response to electrochemical scanning methods is ex-
pected. We also demonstrated the existence of a
well-defined chronopotentiometric transition time for a
beryllium metal anode in molten NaF-BeF, (75 mole %
BeF,) and interpreted it qualitatively as due to the
formation of a nonconductive film of BeF, at the
electrode surface.!4

We here present the results of a quantitative interpre-
tation of the chronopotentiograms based on the diffu-
sion of BeF, formed at the beryllium electrode surface
into the melt (interdiffusion of BeF, and BeF, —alkali
fluoride mixtures).

The potential of a beryllium anode measured relative
to a beryllium reference electrode in melt of uniform
composition during a constant-current electrolysis rises
sharply as it sees increasingly high BeF, concentrations,
the time required for the rise being governed by the
current density and the interdiffusion of BeF, and
alkali fluoride, The reversible emf of the resulting
concentration cell, I:

BeF,-MF BeF,-MF
Be

Yot Ay =y
polarized electrode

Be I
Yo
reference

13. K. A. Romberger and J. Braunstein, fnorg: Chem. 9, 1273
(1970).

14. H. R. Bronstein, J, Truitt, and J. Braunstein, MSR
Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782,
pp- 96—98.

59

jsl3
g 7o 1,20y 12 HBeFa 1
-5 75 5y ()
Yo
If tg, is zero, the cell emf becomes
2FE y Yo 2FEE N
RT —ln(1 Rt _ln(l—y0)2+ RT - (2a)

In the above equation, y and ug.p, are the mole
fraction and chemical potential of BeF,, y, is the
initially uniform melt composition, and tg, is the
transference number of Be(Il). The first two terms on
the right-hand side of (22) give the “ideal” emf, and £E
is the “excess” cell emf, which can be calculated if the
activity ccefficients are known:

+y dinyper,
4 dy

2FEE _
RT

Y 1
(1 —IBez+) I
Yo

dy . (2b)

[f only the alkali ions are mobile and carry electrical
current by migration from the anode under the in-
fluence of the electric field, the Faradaic current can be
sustained only by alkali ions that diffuse back to the
anode. The situation is analogous to chronopotentiom-
etry of a dilute electroactive constituent whose elec-
trolysis product is soluble in the solution; however, the
emf is given not by the Nernst equation, but by the
expression for the binary molten salt concentration cell,
Eq. (1) or (2). The solution of Fick’s second law, '

3Gy
0x?

is subject to the usual initial condition
{t=0}> {Cy* =% , 0 <X <00} (i)

(i.e., uniform concentration initially) and to the bound-
ary conditions
{X =00} > {Cp* = CO\*, 0< ¢ K o0} (ii)

[uniform (initial) concentration at infinite distance
from the electrode| and

oCy*
0X

(iii)

X=0
(electric current limited by back diffusion of mobile
species). The third condition is unusual in that it is a
consequence of depletion of a nonelectroactive con-
stituent, alkali ion, by migration rather than the usual
depletion of electroactive species by electrode reaction.
The solution of the diffusion equation with the above
initial and boundary conditions is, however, formally
identical with the solution for ordinary chronopoten-
tiometry, and the concentration at the electrode sur-
facelS is
Cur =CoMp —at!/?, (3)
where C° and C are the initial and the instantaneous
concentrations at the electrode surface and

The beryllium concentration at the electrode is, by
stoichiometry,

VMF

B('.‘.FQ

— 0
CBer, =C Ber, T2 atl/?

(4)

The concentrations are in equivalents per cubic centi-
meter, and the ¥ are partial molal volumes.

The transition time resulting from settingy = 1 in Eq.
(2) is identical with that of the Sand equation:15

_ 1=Yo
Vmr t (Veer, — VMFDo

arl/? :COMF:

(5)

The chronopotentiogram may be calculated, with the
probably poor assumption for these melts of a concen-
tration-independent diffusion coefficient, by combining
Egs. (2), (4), and (5), making use of the relations
among the concentrations:

C P — 2y —- "_"—21
BeF2  Vmp+t (Veer, — VMEY Vi’
_1-y.
CMF = 7 (62)

yFEideal . (11— VMFCMF)[I ‘*‘(f/—Ber_ VMF)CMF]
RT (VBer,Cmp)?

lrl(1 —VMrC'MPI[1 + (Vg — VurF)COMF]

— (6b)
[VEer,CoMF]?

60

ORNL-DWG 72-T601A

600

500 A

-~ POTENTIAL AT WHICH POLARITY
WAS REVERSED

400
S ANODIC
Ea HALF CYCLE
2 |
a
o 200 T = {70 sec
o " (p=3x107T)
o
[N

100 | CATHODIC HALFI CYCLE

| EXPERIMENTAL
(¢ t
OPEN-CIRCUIT
POTENTIAL ~—
-100 L |
0 100 200 300 400
TIME (sec)

Fig. 9.4. Chronopotentiogram in molten NaF-BeF, (about
25-75 mole %) at 520°C, current density about 40 mA/cm?>.
Potential of beryllium working electrode, relative to a beryllium
reference electrode, vs time. Dashed lines are calculated from
the transition time r = 170 sec.

The excess emf for the system NaF-BeF, was
estimated from activity coefficient data in LiF-BeF,
mixtures!® since extensive data for NaF-BeF, are not
available, leading to the estimate from Eq. (25) and ref.
16:

2FEE

~ _ s
RT = 3.5<y y0+lny0).

(7)

Figure 9.4 shows a typical chronopotentiogram,
measured in 0.75BeF,-0.25NaF, with “ideal” and
“real” ones calculated from Egs. (1), (2), and (3—7) for
a transition time of 170 sec, leading to an average
diffusion coefficient of 3 X 1077,

Values of it1/2 were found essentially independent of
current density in the range 20 to 100 mA/cm?. This
estimate of the diffusion coefficient is two orders of
magnitude smaller than normal liquid diffusion coeffi-
cients and an order of magnitude smaller than values
reported in a less viscous LiF-BeF,-ZrF, mixture.l”?
The present results demonstrate that a transition time is
predicted for nonelectroactive species at high concen-
tration in molten salt mixtures and is observed experi-
mentally. A corollary of these results is that special care

15. P, Delahay, New Instrumental Methods in Electrochem-
istry, Interscience Publisher, New York, 1954, pp. 179-80.

16. B. F. Hitch and C. F. Baes, Inorg. Chem. 8, 201 (1969).

17. G. Mamantov and D. L. Manning, Anal. Chem. 38, 1495
(1966).
may be needed in interpreting electrochemical scanning
data in mixed molten salt solvents approaching uni-
cationic conduction.

Initial measurements with LiF-BeF, mixtures (10-90
mole %) show the anticipated increase in transition time
corresponding to the higher mobility of lithium ion.
The measurements were made with an ORNL 2943
cyclic voltammeter, designed by T. R. Mueller,!8 the
salts being contained in a nickel vessel in a controlled-
atmosphere envelope that was heated in a controlled-
temperature furnace.

Analogous equations may be developed for other
molten salt- systems, such as AICl;-MCl melts. The

61

analog of (1) and (2) for the emf of the concentration
cell with aluminum electrodes is

y 1+ 2y 9du
2wFE= | 7 (1-tay) 21C g4y,
Yo U iy
2 FEideal _ ¥ Yo
RT gy oo

18. T. R. Mueller and H. C. Jones, Chem. Instrum. 2, 65
(1969). .
Part 3. Materials Development

H. E. McCoy

Our materials work is currently involved with several
important areas including (1) gaining a better under-
standing of the intergranular cracking of Hastelloy N
that occurred in the MSRE fuel system and a method of
controlling this cracking in future reactors, (2) develop-
ing a graphite with improved dimensional stability
under radiation, (3) development of surface sealing
methods for reducing the permeability of graphite to
135 Xe, (4) modification of the composition of Hastel-
loy N to obtain an alloy that is more resistant to
neutron irradiation, (5) evaluation of Hastelloy N for
use in steam generators, (6) construction of a molyb-
denum system for use with bismuth and fluoride salts,
and (7) the evaluation of other materials for use as
structural materials in the chemical processing plant.

The first area has continued to occupy a high priority,
and a rather extensive program of laboratory tests has
been pursued. The laboratory tests are directed at
determining the cause of the cracking, the rate of
progression with time and temperature, and a reason-
able solution to the cracking problem. The emphasis is
on the fission product tellurium, since considerable
evidence exists that this element is the cause of the
cracking.

The lifetime of a breeder core will be determined by
the dimensijonal stability of the graphite. Dimensional
changes can be accommodated to some extent, but
volume expansion is usually accompanied by increasing
permeability to xenon, which can become intolerable.
Thus the problem of dimensional stability cannot be
separated from the requirement of low permeability to
135Xe. We are evaluating graphites made by com-
mercial vendors and locally in an effort to understand
what types of graphite have the best dimensional

62

stability. This information is being fed back to com-
mercial vendors and into our own experimental fabrica-
tion program. Pyrolytic carbon coatings derived from
propene are currently being studied as a means of
reducing the permeability to ' 5 Xe. Improved methods
of characterizing the coatings are being developed.

The program to develop a modified Hastelloy N with
improved resistance to radiation indicated that alloys
with additions of 1.5 to 2.0% titanium were most
promising, and recent efforts have concentrated on
these. Material has been obtained from three vendors
that has been made by two basic melting practices. The
evaluation includes weldability tests, unirradiated me-
chanical property tests, and postirradiation mechanical
property tests. :

A corrosion facility in TVA’s Bull Run Steam Plant is
being used to evaluate the corrosion of Hastelloy N in
steam. Both unstressed and stressed samples are in-
cluded in the tests.

Most of the work on chemical processing materials is
going into the construction of a reasonably complex
test facility constructed of molybdenum. Many new
techniques have been developed which overcome or
circumvent the basic difficulties of molybdenum fabri-
cation. (Welds in molybdenum are inherently brittle,
and fabrication into large sizes is hampered by the need
for high temperatures and large forces to fabricate large
parts.) The mutual requirement of compatibility with
bismuth and salt narrows the choice of materials for the
processing plant, but our screening tests offer encour-
agement that besides molybdenum, both tantalum and
graphite will be compatible. Experiments are being run
to determine whether these materials can be used and
under what operating conditions.
10.

63

Intergranular Cracking of

Structural Materials Exposed to Fuel Salt

H. E. McCoy

Examination of Hastelloy N from the MSRE revealed
that all surfaces exposed to fuel salt had intergranular
cracks to a depth of 1 to 13 mils. Some of the cracks
were visible when the parts were removed from service,
whereas others did not appear until after the parts
were deformed. The existing evidence indicates that the
cracking is likely due to the intergranular penetration of
the fission product Te. Our experimental program has
concentrated on the effects of Te, but some work is
continuing on the effects of other fission products.

Our work involves (1) studies of alloys exposed to
small amounts of various fission products, (2) studies of
Hastelloy N containing small additions of fission
products, (3) tube-burst and strain-cycle tests in salt of
several alloys plated with Te, (4) creep tests of several
materials in an Ar-Te environment, and (5) studies of
the tendency of Te to produce cracks in several
structural materials other than Hastelloy N. Our ob-
jectives are to determine the cause of the cracking; the
dependence of the rate on temperature, time, and
concentration; and a reasonable solution to the prob-
lem.

10.1 CRACKING OF SAMPLES ELECTROPLATED |

WITH TELLURIUM ‘
H.E.McCoy  B.McNabb

We have compared the susceptibilities of a wide
variety of metals to cracking due to penetration by
tellurium by electroplating tensile samples with tellu-
rium, then annealing, straining, and examining them.
Two different combinations of tellurium concentration
and annealing conditions were tried. We used the
method described in the last semiannual report! to
plate samples to nominal tellurium concentrations of
0.03 mg/cm? (1.4 X 10'7 atoms/cm?) and 0.1 .mg/cm?
(4.7 X 10'7 atoms/cm?). The materials included (1)
standard and modified Hastelloy N; (2) Hastelloys B, C,
X, S, and W; (3) Inconels 600, 601, and 718; (4)
Incoloy 800; (5) stainless steel types 201, 304, 309,
310, 316, 321, 347, 406, 410, 446, 502, and 17-7 PH;
(6) nickel; (7) copper; (8) Monel; and (9) cobalt-base
Haynes alloys 188 and 25. Groups of samples were

1. B. McNabb and H., E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, p. 128.

encapsulated with an argon atmosphere in stainless
steel; the first group, with 0.03 mg/cm? of tellurium,
was annealed for 1000 hr at 650°C, and the second
group, with 0.1 mg/cm? of tellurium, was annealed for
200 hr at 700°C. The samples were subsequently
deformed at 25°C and sectioned for metallographic
examination to determine whether intergranular crack-
ing had occurred.

Standard Hastelloy N cracked intergranularly under
both test conditions, as shown by the typical photo-
micrographs in Fig. 10.1. Pure Ni, Hastelloy B (1% Cr),
and most of the modified compositions of Hastelloy N
also formed intergranular cracks. Incoloy 800 (Fe—21%
Cr—34% Ni) formed very shallow intergranular cracks.
Generally, all of the nickel- and cobalt-base alloys with
15% or more Cr did not crack. Exceptions were
Hastelloy W, with only 5% Cr, which did not crack, and
Hastelloy S, with 15% Cr, which did crack. Armco iron
and the numerous stainless steels in the experiment
were free of cracks (Fig. 10.2). Copper and Monel
(Ni—35% Cu) also did not crack (Fig. 10.3).

The heats of modified Hastelloy N that did not crack
in this experiment offer encouragement that this alloy
can be modified to resist intergranular cracking due to
tellurium. Heats 70-727 (2.1% Ti), 72-503 (1.9% Ti),
70-786 (0.6% Nb, 0.8% Ti), 70-835 (2.6% Nb, 0.7% Ti),
and 21543 (0.7% Nb) did not crack and offer strong
evidence that Ti and Nb are beneficial in preventing
cracking. These same elements improve the resistance of
Hastelloy N to neutron irradiation.

10.2 INTERGRANULAR CRACKING
OF SEVERAL ALLOYS WHEN EXPOSED
TO TELLURIUM VAPOR

H. E. McCoy

As discussed previously, we observed that intergran-
ular cracks are formed in Hastelloy N that has been
exposed to Te vapor and strained.? During this report
period we have shown that the extent of the cracking

~ depends on the concentration of Te, the temperature

and duration of exposure, and the chemical composi-
tion of the alloy being exposed. In these experiments,

2. H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb.
29, 1972, ORNL-4782, pp. 134-37.
64

Y-114326

15.010 in.

0.035 INCHES
100X

toos0m

Y- 116306

0.035 INCHES
™ 100X

T

(6) 1N

Fig. 10.1. Hastelloy N (heat 5065) electroplated with tellurium and strained at 25 C (@) l-lecuoplal:d with 1.4 x 10'7
atoms/cm? of tellurium and annealed 1000 hr at 650°C. (b) Electroplated with 4.7 X 10'7 atoms/em? of tellurium and annealed
200 hr at 700°C. As polished.

65

=

0.035 INCHES
T 100X

™

Fig. 10.2. Type 304L stainless steel electroplated with 4.7 x 10'7 atoms/cm? of tellurium, annealed 200 hr at 700°C, and

strained at 25°C. As polished.

tensile specimens were sealed in quartz ampuls, together
with the amount of tellurium required to produce a
desired surface concentration of tellurium on the
specimens. When the ampuls were heated for annealing,
the tellurium vaporized and transferred to the metal
surfaces; the discussion which follows it is the
resulting surface concentration that is specified.

The effect of Te concentration on the severity of
cracking in Hastelloy N at 650°C is shown in Fig. 10.4.
These samples were exposed for 1000 hr at 650°C to Te
concentrations ranging from 0.5 to 54 X 10'7 atoms/
cm?
tration of tellurium was increased.

Samples of Hastelloy N were exposed to a Te
concentration of 5.4 X 10'® atoms/cm? for 1000 hr at
550, 650, and 700°C. The microstructures in Fig. 10.5
show that cracking became more severe as the exposure
temperature was increased.

The progression of the cracking with time was
followed by exposing samples at 700°C to 5.4 X 10'®
atoms/cm?® of Te for 20, 48, 100, 500, and 1000 hr.
The photomicrographs in Fig. 10.6 show that signifi-
cant cracking occurred after only 20 hr of exposure.

in

The cracking became more severe as the concen-

Increased exposure time did increase the cracking
severity.

In the foregoing experiments with Hastelloy N, we
observed that the depth of cracking after 100 hr at
700°C was slightly greater than that after 1000 hr at
650°C. We therefore chose the shorter time period and
higher temperature for screening tests of different
alloys.

Alloy composition had a large effect on the resistance
to cracking, and much of our testing was involved with
investigating this variable. Type 304L stainless steel,
Monel (Ni—35% Cu), Cr, Inconel 606 (Ni—20% Cr—3%
Mn-2.5% Nb—-0.6% Ti), and modified Hastelloy N
(Ni—12% Mo—7% Cr—2.6% Nb—0.7% Ti) resisted crack-
ing under all test conditions. Several alloys resisted
cracking in the presence of 5 X 10" 7 atoms/cm? of Te,
but did crack some when exposed to 5 X 10'%
atoms/cm?® of Te. These included Hastelloy N modified
with 0.75% Ce, Hastelloy S (Ni—15% Mo—15% Cr—
0.01% La), Inconel 600 (Ni-15% Cr—7% Fe), and
Inconel 601 (Ni-23% Cr—14% Fe—1.4% Al). Incoloy
811E (Fe-32% Ni-21% Cr—2% Al) did not crack at

66

(6)

=

0.035 INCHES
T 100X

™ 100X

Fig. 10.3. Materials electroplated with 4.7 X 10'7 atoms/cm? of tellurium, annealed 200 hr at 700°C, and strained at 25°C. (a)

Copper. (b) Monel. As polished
5.4

Y- 13099

54

Fig. 10.4. Hastelloy N exposed to various concentrations of tellurium (expressed in atoms/cm? x 10'7) for 1000 hr at 650°C

and strained at 25°C.

the lower Te concentration, but has not been tested at
the higher concentration

These observations illustrate clearly the effects of the
chemical composition of the alloy and the concentra-
tion of Te on the severity of cracking. The effect of Te
concentration raises some questions about how well the
type of test being run simulates reactor operation. An
MSBR would produce a Te concentration of 2 X 10'?
atoms/cm? over a 30-year period. Although the maxi-
mum concentration of Te used in our tests was 5.4 X
10'® atoms/cm?, the sample was exposed to this entire
amount as the test was brought to temperature over a
few hours. If we accept the model that Te diffuses
selectively along the grain boundaries, the tendency to
embrittle the boundary may be offset by the reaction

of the Te with some reactive element such as Cr or Ce
in the alloy. The availability of these elements at the
grain boundary depends upon diffusion, and a slower
rate of introducing the Te would be less severe. Thus,
our test conditions are likely more severe than those
that would be encountered in an MSBR. Methods of
adding the Te at a more reasonable rate are being
investigated.

Chromium seems effective in reducing the cracking.
On the other hand, since the dominant corrosion
mechanism in fluoride salts containing U is the selective
removal of Cr, the Cr concentration must be limited (to
some yet undefined value) to limit the amount of
corrosion. Tests are in progress to better define the
minimum amount of Cr to prevent cracking and the

550°C

700°C

650°C

Y-143100

|=0.03in~

Fig. 10.5. Hastelloy N exposed to 5.4 X 10'® atoms/cm? of tellurium for 1000 hr at various temperatures and strained at 25°C.

Table 10.1. Stress-rupture properties of several alloys at 650°C

Siess Rupture Rupture . )
Heat number (ps) life strain Basic alloy Addition
(hr) %)
351 47,000 59.1 112 Hastelloy N None
40,000 198.5 9.5
30,000 19362 204
373 40,000 65.0 39 Hastelloy N 600 ppm Te
30,000 2153 3.0
375 47,000 76.0 15.5 Hastelloy N 60 ppm St
40,000 1418 13.6
374 20,000 155 9.5 Type 304 stainless steel 60 ppm Te
Standard stainless steel? 20,000 366.6 175 Type 304 stainless steel None

4E, E. Bloom, In-Reactor and Postirradiation Creep-Rupture Properties of Type 304 Stainless Steel at 650°C, ORNL-TM-2130

(March 1968).
69

20 hr 48 hr

|

100 hr

Y-115902

Fig. 10.6. Hastelloy N exposed to 5.4 x 10'* atoms/cm® of tellurium for various times at 700°C and strained at 25°C.

maximum amount that will result in an acceptable
corrosion rate. The tests of various alloys indicate that
other elements such as Al, Nb, and Ce may also be
effective in preventing intergranular cracking.

10.3 MECHANICAL PROPERTIES
OF STRUCTURAL MATERIALS
CONTAINING STRONTIUM AND TELLURIUM

H. E. McCoy

Small heats of Hastelloy N were prepared that
contained small additions of several fission products
and sulfur. Creep tests at 650°C revealed that S, Se, and
Te reduced the time to rupture at a given stress and the
rupture strain.?> Four additional alloys were made and
tested (1) to reproduce the detrimental effects of Te on
Hastelloy N, (2) to determine whether Sr has detri-
mental effects on Hastelloy N, and (3) to determine
whether Te has detrimental effects on stainless steel.

The results of limited mechanical property tests on
these new alloys are given in Table 10.1. Alloy 351 was
a laboratory melt of Hastelloy N without additions, and

the properties are included for comparison. Alloy 373
had an analyzed addition of 600 ppm Te, and the
detrimental effects on the rupture life and the fracture
strain are quite large. Alloy 375 contained 60 ppm Sr,
and the properties are about equivalent to those of
alloy 351 without any addition. Thus, Sr at this
concentration does not appear to be detrimental. Heat
374 is a laboratory melt of type 304 stainless steel with
an addition of 60 ppm Te. The heat without a Te
addition was not completed, so a value from the
literature* for type 304 stainless steel is included in
Table 10.1 for comparison. Based on this comparison,
Te appears to reduce the rupture life and strain of type
304 stainless steel. This is not a firm conclusion,
however, since the literature value is for commercial
material and the validity of the comparison is question-
able.

3. H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb
29, 1972, ORNL4782, pp. 136-41

4. E. E. Bloom, In-Reactor and Postirradiation Creep-Rup-
ture Properties of Type 304 Stainless Steel at 650°C, ORNL-
TM-2130 (March 1968)

10.4 THE ISOTHERMAL DIFFUSION
OF '27Te TRACER IN HASTELLOY N, Ni-200,
AND TYPE 304L STAINLESS STEEL SPECIMENS

P.T.Carlson L. C. Manley
H.E.McCoy R. A.Padgett
B. McNabb

The evidence cited in the preceding sections strongly
indicates that tellurium is the cause of intergranular

70

cracking in Hastelloy N and that the depth of embrittle- -

ment is dependent upon the diffusion rate of tellurium
into the metal, particularly along the grain boundaries.
Similar cracking has been produced in Ni-200 by
exposure to tellurium, but type 304L stainless steel has
resisted embrittlement. One possible explanation of the
differences among alloys could be the Te diffusion rate.
To test this hypothesis, we have measured the diffusion
rates of tellurium in Hastelloy N, Ni-200, and type
304L stainless steel at 650 and 760°C. Generally, the
experiment consisted in electroplating '?7Te on the
flat surfaces of samples of each material, annealing the
samples for various lengths of time to obtain penetra-
tion, removing successive layers from each sample by
abrasion, measuring the ' 27 Te activity of the removed
material, plotting the penetration profiles, and analyti-
cally determining the volume and grain boundary
diffusion coefficients. In some cases the annealing
conditions did not give sufficient penetration to deter-
mine these coefficients very accurately.

The specimens were 1-in.-long right circular cylinders.
One-half of each specimen was '4 in. in diameter, and
the remainder was ¥, in. in diameter. The circular face
of the ¥, -in.-diam end of the cylinder was plated with
the radioactive tracer and was the surface from which
diffusion occurred. The samples were sealed in quartz
under vacuum and heated slowly to the diffusion anneal
temperature. One set was annealed for 1000 hr, another
for 3000 hr, and a third set of samples is still being
annealed, with a target time of 10,000 hr. After being
annealed, samples were broken out of the quartz and
decontaminated sufficiently to be sectioned. Prior to
sectioning, the diameter of the ¥,-in. end was reduced
to ', in. to eliminate the effects of surface diffusion.

Sectioning was accomplished by hand grinding the
specimens on silicon carbide metallographic papers.
Six-hundred-grit papers were used to obtain thin sec-
tions and 320-grit papers to obtain coarse sections.
Straight-line strokes were used on a marked area of
the silicon carbide paper, with the specimen rotated
90° after each five strokes to provide visual indication
that the sectioning was proceeding parallel to the
original surface. After each section was taken, the

specimen was cleaned with tissue dampened with
acetone. The cleansing tissue together with the marked
area of the metallographic paper were then enclosed in
a packet for gamma counting. Section thicknesses were
calculated from the measured weight difference of the
specimen before and after the grinding of each section
and the known cross-sectional area and density of the
material. The activity of '?7Te in each section was
determined by gamma counting using a scintillation
spectrometer with a multichannel pulse-height analyzer
and a 3 X 3 in. Nal(Tl) scintillation crystal enclosed in
a lead shield.

The volume diffusion coefficients D, were obtained
from the first linear portion of each plot of In 4 vsx?,
using the standard thin-film solution® to Fick’s second
law for diffusion. (A4 is the activity and is proportional
to the Te concentration, and x is the distance from the
surface.) The product of the grain boundary width &
and the grain boundrary diffusion coefficient Dyp, was
determined from the second linear portion of each plot
of In ¢ vs x®/% using a form of Whipple’s®>7 solution for
grain boundary diffusion. The appropriate relation is

o 1lnc\ 3/3 /4D \!/?
Bng= ax6/5 \t_

0 lnc 5/3
X (1)
a(nfi—l/2)6/5
where
_ [ Dgb . o
B‘(ov “) N @
and
n=x/\D, . (3)

The volume and grain boundary diffusion coefficients
are presented in Table 10.2 together with the appro-
priate diffusion anneal information. In addition, the last
two columns give the depth to which the volume
diffusion region was seen to extend (Harrison’s8 type
A region) and the maximum detectable penetration of
the tracer.

5. Paul G. Shewmon, Diffusion in Solids, McGraw-Hill, New
York, 1963,

6. R. T. P. Whipple, Phil. Mag. 45, 1225 (1954).

7. A.D. LeClaire, Brit. J. Appl. Phys. 14,351 (1963).

8. L. G. Harrison, Trans. Faraday Soc. 57,1191 (1961).
71

Table 10.2. Diffusion of tellurium into several alloys

Anneal Anneal Extent of .
:3:5:} Alloy time temperature (cmlz’)/‘,sec) (crig/gst;c) volume region Pene(tr;itlon
(hr) CO) (1) K
324 Ni-200 3000 650 452x107'% 992x 1076 20 210
34 Ni-200 3000 760 1.11x 1072 sgox 107! 48 220
370 Type 304L stainless steel 1000 650 8.67x 107'*  970x%x 1073 2 32
40 Type 304L stainless steel 3000 650 1.23x 107" 10x107'8 5 18
42 Type 304L stainless steel 3000 760 2.68x 107" 8o03x 107! 6 36
2b Hastelloy N 1000 650 1.12x 107"%  881x 107" 7 37
3 Hastelloy N 1000 760 1.06 x 107'*  461x107'® 8 68
9 Hastelloy N 3000 650 2.11x 107'%  ga2x 107" 3 28
10 Hastelloy N 3000 760 -1.01x107**  1.18x107'® 6 44

INo confidence in Dy value because of irregularity in specimen surface.

bTime and temperature of anneal insufficient to yield accurate volume diffusion coefficients with sectioning method used.

Harrison’s types A, B, and C regions were observed in
nearly all the samples examined, and the volume
diffusion coefficients D, and the grain boundary values
6Dgy, were determined from regions A and C respec-
tively. It should be noted that the annealing times at
the chosen temperatures were too brief to allow
accurate determination of volume diffusion coefficients
by the serial grinding sectioning method used. As can be
seen in Table 10.2, the volume diffusion region in all
cases except those for Ni-200 specimens extended to
less than 10 um. The newly developed rf sputtering
technique, which is designed for removal of precise,
ultra-thin layers, could not be employed on these
specimens due to their size and geometry. With speci-
mens of proper size sectioned by sputtering, it should
be possible to obtain more reliable volume diffusion
coefficients within the anneal times and temperatures
used in this study.

Some specimens were found to have an oxide layer
and scratches on the surface. It is not known whether
these conditions existed prior to the diffusion anneal or
whether they occurred as a result of handling after the
anneal. If they were present prior to the anneal, the
volume diffusion process and the calculated D, values
could be seriously affected. One specimen (No. 32) had
a concave surface, which resulted in the first six
sections being incomplete. Naturally, a minimum degree
of confidence should be placed in the D, values
obtained from this sample. Furthermore, the cSng
values reported in Table 10.2 share the same degree of
accuracy, since they have been calculated from the D,
values [see Eqs. (1) to (3)].

The data in Table 10.2 show clearly that tellurium
diffuses into Ni-200 most rapidly, into Hastelloy N less
rapidly, and into type 304L stainless steel least rapidly.
The shallow penetration of tellurium into stainless steel
may account for this material being resistant to
cracking.

10.5 TUBE-BURST EXPERIMENTS

H. E.McCoy J.W.Chumley E.J. Lawrence

The influence of stress on the rate of crack propaga-
tion in structural materials in the presence of tellurium
is a matter of concemn in reactor design. One type of
test being used to study this effect is tube-burst tests in
which specimens electroplated with 5 X 10'7 atoms/
cm? of tellurium are compared with unplated speci-
mens. Some are immersed in salt while being stressed;
others are stressed in helium. The tests were all run at
650°C, and the salt composition was LiF-BeF,-ZF,-
UF, (65.4-29.1-5.0-0.5 mole %).

Results for Hastelloy N described previously? showed
that tubes plated with tellurium failed more quickly
than those without tellurium. Metallographic examina-
tion showed that numerous intergranular cracks formed
that penetrated about halfway through the 0.020-in.
tube wall. Recent posttest dimensional measurements
have shown that the uniform diametral strains were
reduced by the presence of tellurium (Fig. 10.7). Thus,

9. H. E. McCoy and J. W. Koger, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 126-28.
MAXIMUM PRINCIPAL STRESS {psi)

MAXIMUM PRINCIPAL STRESS (psi)

40,000

20,000

10,000

8000

40,000

20,000

10,000

UNIFORM DIAMETRAL STRAIN (%)

72

ORNL-DWG 73-831

’ ’ [RREIE
o He
12 a SALT
h o SALT +FeF,
0 N OPEN PCINTS~CLEAN TUBES
“0 CLOSED PCINTS-Te COATED
8
A
] ia O
6 o
4 e,
™ a
[ ]
2
° A'n
A
o e
O W
100 2 5 oL 5 102 2 5 103 2 5 104

RUPTURE TIME (hr)
Fig. 10.7. Uniform strains of Hastelloy N tubes stressed at 650°C.

ORNL-DWG 73-832

-u-.-__..-.
“‘h-
\N
oA
® He—Te PLATED Rua
A SALT-Te PLATED T~
® SALT - NOT PLATED e WY
\...\ -
‘\
ol
N
1ot 2 s 109 2 5 10 2 s 102 2 s 103 2 5 10%
RUPTURE TIME (hr)
Fig. 10.8. Stress-rupture properties of type 304L stainless stee!l tubes stressed at 650°C.
ORNL-DWG 73-833
® He —-Te PLATED
A SALT—Te PLATED
s SALT—NOQOT PLATED LT
LT
-1!'/
1]
/,.%
—‘/
-”""
. |1
»—"/
At T
. /I/JQ r
—'/.
o4 1073 1072 107! 109 10!

8000
1

MINIMUM CREEP RATE (%/hr)
Fig. 10.9. Creep properties of type 304L stainless steel stressed at 650°C.
stress does accelerate the rate of crack propagation in
Hastelloy N with resultant reductions in rupture life
and strain.

Several tube-burst tests have been run on type 304L
stainless steel to determine whether tellurium has a
deleterious effect on crack propagation in stressed
specimens of this material. Three groups of four tubes
each were run under the conditions of (1) tellurium-
plated and stressed in a helium environment, (2)
tellurium-plated and tested in a salt environment, and
(3) not plated and tested in a salt environment.

The stress-rupture properties of the type 304L stain-
less steel samples are shown in Fig. 10.8. There is no
detectable effect of test environment on this property.
The uniform strains were divided by the rupture time to
obtain an apparent minimum creep rate. As shown in

73

Fig. 10.9, this parameter also did not appear to be
influenced by the test environment. It should be noted,
however, that the uniform strain determinations on
these specimens involve considerable uncertainty be-
cause of the large deformations that occurred. Often
tube samples will uniformly deform to some diameter
before some region becomes unstable and fractures. In
this case most of the tube will have a uniform diameter,
with only a small region near the fracture having a
larger diameter. In other cases the strain may occur
nonuniformly with the tube having several enlarged
regions so that the uniform strain is difficult to define.
Type 304L stainless steel has the latter type of
behavior; hence the creep rate values in Fig. 10.9 and
the uniform strain values in Fig. 10.10 are not very
accurate.

CRNL-DWG 73-834

12
=10 *
53
— [ ] Y
Z
Py
A
x g
@ |
2 ® He-Te PLATED o
T s A SALT-Te PLATED
s " SALT-NOT PLATED
= .
a A
= 4 ) L
1l
(@]
1
=2
o 2 4
Y
0
10° 2 5 100 2 5  10% 2 5 100 2 5 10?

RUPTURE TIME (hr}

Fig. 10.10. Variation of uniform diametral strain with rupture life for type 304L stainless steel tubes tested at 650°C,

ORNL-DWG 73-B35

50
e
B ah
Z 40 N
<L \ N
x N M
o e He-Te PLATED \\
2 30 A SALT-Te PLATED AN .
& 8 SALT-NOT PLATED \\
= A
< 20 I =
[an}
=
s } s ] ®
=10
>
<
=
0
10® 2 5 100 2 5 102 2 5 10 2 5 104

RUPTURE TIME (hr)

Fig. 10.11. Variation of maximum diametral strain with rupture life for type 304L stainless steel tubes tested at 650°C.
The maximum diametral strains for the type 304L
stainless steel specimens are shown in Fig. 10.11. The
tubes tested in helium had significantly lower maximum
fracture strains than the tubes tested in salt. It appears
that there may also be a significant difference between
the tubes plated with tellurium and those not plated,
but the data are too limited to support this as a firm
conclusion.

Sections of each tube were examined metallographi-
cally. A photomicrograph of a tellurium-plated sample
tested in helium is shown in Fig. 10.12. Although
numerous intergranular cracks formed in this sample,
there were not as many as in the sample shown in Fig.
10.13 that was tested in salt without tellurium present.
This implies that some factor other than Te may have
been responsible for the cracks in the sample tested in
helium. Some oxidation obviously occurred on the
surfaces of this sample due to impurities in the helium
(Fig. 10.12). All of the samples tested in salt ex-
perienced considerable corrosion, resulting in the for-
mation of intergranular voids that linked together to form

74

cracks. The amount of cracking was equivalent in
samples not plated with tellurium (Fig. 10.13) and
those that were plated (Fig. 10.14).

The foregoing observations lead us to conclude that
tellurium does not accelerate intergranular cracking in
type 304L stainless steel. There was considerable
corrosion in the salt that may have been worsened by
the presence of impurities.

Tests are also in progress on tellurium-plated speci-
mens of Inconel 600 (Ni—15% Cr—7% Fe) and Incoloy
811E (Fe—30% Ni—21% Cr—2% Al). The stress-rupture
properties of these materials along with those of type
304L stainless steel and Hastelloy N are shown in Fig.
10.15. There does not seem to be any progressive
deterioration of the properties of Inconel 600 and
Incoloy 811E with increasing rupture time (as seems to
be the case for Hastelloy N), but the effects of
tellurium on these materials cannot be fully assessed

until the tests are completed and the samples examined
metallographically.

=

0.035 INCHES
™ 100X

Fig. 10.12. Photomicrograph of a section of type 304L stainless steel tube electroplated with 0.1 mg/cm” of tellurium and
stressed at 12,000 psi and 650°C in helium. Discontinued after 1570.8 hr with 6.8% uniform strain. The outside diameter of the tube

is on top of the photomicrograph. As polished.
~
a

Fig. 10.13. Photomicrograph of a section of type 304L stainless steel tube stressed at 12,000 psi and 650°C in salt. Failed after
1663.9 hr with 2.1% uniform strain. The outside diameter of the tube is on top of the photomicrograph. As polished

100X

s

Fig. 10.14. Photomicrograph of a section of type 304L stainless steel tube electroplated with 0.1 mg/cm® of tellurium and
stressed at 12,000 psi and 650°C in salt. Failed after 1415.6 hr with 8.2% uniform strain. The outside diameter of the tube is on top
of the photmicrograph. As polished

76

ORNL-—- DWG 73-836

- I »
= 40,000 i e LA
& \-.. \'
o "'-.____“‘ \
n e T Py
E:J -....‘.-..... \__‘-‘-. .-\\
= b iy
w N
g \-\\ Ne
g ] ~ \
@ 20,000 = o
I Ty
g \ ...-""h
o [ Y s
= ® HASTELLOY N ‘-~\ TR
= a TYPE 304 STAINLESS STEEL ~] el
x = INCONEL 600 ~Nlll e =
= 10,000 — 4 ncoLoY 8HE =
8000 Lol prl
0! 1o° 10! 108 10° T

RUPTURE TIME (br)

Fig. 10.15. Stress-rupture properties of several materials electroplated with 0.1 mg/cm2 of tellurium and stressed at 650°C in

salt.

10.6 STRAIN CYCLE EXPERIMENTS
H.E.McCoy K.W.Boling J.C. Feltner

Heat exchangers normally involve two fluids at
different temperatures, and transients in operation can
lead to large thermal stresses and strains. These severe
operating conditions and the relatively thin-walled
tubes involved make the primary heat exchanger of a
molten-salt reactor particularly susceptible to fajlure
from the intergranular penetration of fission products.
The design criterion for the reference MSBR heat
exchanger is a strain range of 0.3% for 10* cycles.!® We
have initiated a program to determine the properties of
materials under cyclic strain conditions in a salt
environment with tellurium present.

Sufficient work was done previously to show that the
ability of a material to withstand thermal strains can be
approached experimentally in terms of either mechani-
cal or thermal strain cycling. Both types of test data can
be expressed in terms of a total strain range consisting
of elastic and plastic components and the number of
cycles required to cause failure under these test
conditions. Because of relative experimental simplicity,
we have chosen the mechanical strain cycling approach.

The mechanical strain cycling equipment that we have
designed and fabricated is shown in Fig. 10.16. The test
sample is tubular with a diameter of about | in. and a
gage section 1 in. long with a 0.030-in.-thick wall. The
top of the sample is welded to a tubular support, and
the bottom is welded to a rod that passes up through

10. Private communication, C. W, Collins, ORNL.

the test sample and the tubular support to the piston.
The force for cycling is generated by alternately
pressurizing the two sides of the piston, and the
magnitude of the force is controlled by the air pressure
and measured by the load cell. The strain is measured
by extension rods that are attached to the sample just
outside the gage section. There are four extension rods
on the extensometer; a pair from the top and bottom of
the sample is used to measure strain in the compressive
direction, and the other pair is used for measurements
in the tensile direction. Small microswitches and adjust-
able micrometers are attached to these rods to limit the
strain in a given direction. A control circuit is used to
obtain the desired cyclic action. The top of the piston is
pressurized, and the sample strains in tension until the
limit switch is closed. The pressure is then bled off the
top of the piston and the bottom pressurized. The
pressure is sustained until the sample strains the desired
amount in compression. The pressure is then bled off
the bottom side of the piston and the top pressurized to
begin a new cycle.

The sample can be plated with tellurium after it is
welded in place but before it is assembled into the test
apparatus. The lower test chamber is sealed so that it
can be filled with salt and a helium cover gas main-
tained. The inside of the sample is pressurized with
about 10 psig of helium to detect propagation of the
first crack through the sample wall.

Although this test concept is not new, we have made
several changes and improvements. The initial tests were
accompanied by numerous experimental difficulties,
which included a poorly functioning pressure valve,
solenoid valves that leaked, movement of the exten-
someter on the specimen, self-welding of the exten-
someter attachment set screws, impact loading of the
sample, shorting out of the level indicator, stress
concentration due to the specimen design, and changes
in ambient temperature. All of these problems have
been dealt with adequately except the changes in
ambient temperature and the specimen design. The
total movement being controlled is only 0.0032 in., and
very small shifts in temperature can cause the dimen-
sions to change by this amount. The specimen design

77

was changed to reduce the buckling tendencies, but the
stress concentration at the end of the gage section still
caused all failures to occur at this location.

The test can be run in several ways. The variables that
can be controlled are time (or cycle frequency), stress,
and strain. Thus, the stress and cycle frequency could
be fixed and the resultant strain and the number of
cycles to failure measured. If the sample strain hard-
ened, the strain would become progressively less. If the
strength were different in tension and compression, the

“sample would become progressively longer or shorter.

ORNL-DWG 72-10495

STRAIN LIMIT =
BRIDGE
SOLENOID ‘
! : |
]
' CYLINDER ‘ ! E PISTON
AIR — | ' |
\
0
I 1 i 1 I
b} '..;‘“_‘ “. :
@ l .|: ® ! |
f Ik |}. } ! ' '
1 ! FAILURE -
STRAIN ot [F® | DETECTOR ; |
INDICATOR I | NN
—/ y | :
1 I | .
| i |
STRAIN ; ! . |
ADJUSTMENT— ; ' i 1
) ' |
H 1 £ |
] - i
LOAD CELL: ;
L
FURNACE
7.
WL LEVEL
PROBE
I DUMP VESSEL
FLOOR | ; EXTENSOMETER
\ ! /
1
SPECIMEN
§ _—SALT
Ll i) —
INCHES

SECTION A-A

Fig. 10.16. Schematic drawing of strain cycle equipment.
78

Another test would be to fix the strain limits and the and let the frequency vary. This requires that the stress
frequency. This would require that the pressure (or level and the strain limits be chosen so that a reasonable
stress) be adjusted progressively to account for changes cycle frequency (a few minutes to a few hours) results.
in the material properties. A third mode, and the one The tests run to date are summarized in Table 10.3.
that we plan to use, is to fix the stress and strain limits The test results have become more reliable as we have

Table 10.3. Summary of strain cycle tests at 700°C on several alloys

Teat Cycles
=2t Material Condition Test conditions to
nimber .
failure
103 Hastelloy N No tellurium Fixed frequency of 870
1 hr/cycle, 35,100 psi
101 Hastelloy N No tellurium Fixed frequency of 178
1 hr/cycle, 38,700 psi
104 Hastelloy N No tellurium Strain range of 0.32% 2154
102 Hastelloy N Tellurium (1 mg/cm?) Strain range of 0.32% 472
202 Type 304L stainless steel No tellurium Strain range of 0.32% 7574
204 Type 304L stainless steel Tellurium (1 mg/cm’) Strain range of 0.32% 974
301 Inconel 600 Tellurium (1 mg/cm?) Fixed frequency of 68.5
1 hr/cycle, 26,400 psi
ATest in progress.
Y-145270
!
H
= 23
- HE
B

Fig. 10.17. Strain cycle specimen 101. Hastelloy N without tellurium plating that was exposed to 178 cycles at a stress level of
38,700 psi and a frequency of 1 cycle per hour. Outside surface (at top of photomicrograph) was exposed to salt. As polished.

corrected the problems mentioned previously. The
metallographic observations are probably the most
significant results to date.

Three Hastelloy N samples have been examined. The
two samples tested without tellurium plating are shown
in Figs. 10.17 and 10.18. Only a few intergranular
cracks were present near the fracture, and the intergran-
ular cracking did not extend down the gage section of
the specimen. One Hastelloy N specimen (Fig. 10.19)
that had been plated with 1 mg/em® (5 X 10'®
atoms/cm?) of tellurium was examined. Intergranular
cracks were formed along the entire gage length of the
sample.

One stainless steel sample that was plated with
tellurium has been examined. The
shown in Fig. 10.20 reveals some general surface attack
with shallow intergranular penetration. It is not possible
to conclude whether the intergranular cracking is

microstructure

associated with salt corrosion or the presence of the
tellurium. One Inconel 600 sample that was plated with
tellurium was examined (Fig. 10.21). The surface of
this specimen had intergranular cracks along the entire
gage length

Y-116317
i

79

10.7 IDENTIFICATION
OF REACTION PRODUCTS
FROM TELLURIUM-STRUCTURAL-MATERIAL
INTERACTIONS

R.E. Gehlbach  Helen Henson

Our investigations of the problem of fission-product-
induced edge cracking of Hastelloy N are directed at
understanding the nature of interactions between Te
and Hastelloy N at elevated temperatures. Since many
nickel-base alloys do not suffer the edge cracking
reproduced in Hastelloy N after exposure to Te (Sect.
10.1), it is possible that in these alloys Te may be tied
up in a stable, relatively innocuous compound. We are
currently identifying the reaction products which are
formed when various structural materials are exposed to
tellurium at elevated temperatures. The emphasis is
being placed on Hastelloy N of the MSRE type,
although we are also examining various other types of
Hastelloy N, Inconel 600, and Ni-200. Unirradiated
specimens have been exposed to Te by electroplating,
vapor deposition, creep testing in an argon environment
with a partial pressure of Te, and Knudsen-effusion-

=

0.035 INCHES
™ 100X

Fig. 10.18. Strain cycle specimen 103. Hastelloy N without tellurium plating that was exposed to 870 cycles at a stress of 35,100

psi and a frequency of 1 cycle per hour. Qutside surface (at top of photomicrograph) wa

exposed to salt. As polished.

=

0.035 INCHES
T 100X

™

Fig. 10.19. Strain cycle specimen 102. Hastelloy N plated with 1 mg/cm” of tellurium and exposed to 472 cycles at a strain range
of 0.32%. Outside surface (top of photomicrograph) was exposed to salt. As polished.

G s Bt e S

0.035 INCHES
T~ 100X

Fig. 10.20. Strain cycle specimen 204. Type 304L stainless steel that was plated with 1 mg/cm? of tellurium and exposed to 974
cycles at a strain range of 0.32%. Outside surface (top of photomicrograph) was exposed to salt. As polished

81

=

™ 100X

Fig. 10.21. Strain cycle specimen 301. Inconel 600 plated with 1 mg/cm? of tellurium and exposed to 68.5 cycles at 26,400 psi
and a frequency of | cycle per hour. Outside surface (top of photomicrograph) was exposed to salt. As polished

cell—mass-spectrometer experiments. Reaction tempera-
tures are in the range of 400 to 750°C (primarily 650
and 700°C).

The above techniques for introducing Te do provide
for the diffusion of this element into the material along
grain boundaries as evidenced indirectly by observed
edge cracking and directly by Auger spectroscopy (Sect.
10.8). The fraction of Te which actually diffuses into
the specimen is probably quite small, with the re-
mainder reacting to form various telluride intermetallic
compounds on the specimen surfaces. Figure 10.22
shows a Hastelloy N creep specimen fractured in an
environment of Ar plus Te at 650°C before removal
from the specimen grip. The “fuzzy” network of
particles and fibers on the specimen surface is tellurides
formed during the test. The reaction products have
been identified primarily by x-ray diffraction and
electron probe microanalysis and are those products
which form on exposed surfacés of the metal or alloy
rather than on grain boundaries. The x-ray-diffraction
technique used for identification of the tellurides has
been described previously.!'! We are able to get
excellent diffraction from 500 A of Te electroplated on
Hastelloy N. A layer approximately 100 A thick does

not provide a strong pattern, but the Te is detectable
and the pattern is adequate for identification. Whenever
possible, the tellurides are scraped from sample surfaces
for x-ray analysis to eliminate matrix lines and to
minimize preferred orientation and background effects.
Tellurides removed as sheets or large pieces are ground
to powder before analysis.

10.7.1 Knudsen Cell Reactions

Four foils of standard Hastelloy N (0.25 X 0.004 in.)
have been reacted with Te by J. D. Redman of the
Reactor Chemistry Division. Three of the foils were
electropolished prior to reaction. The fourth was
electropolished and then oxidized about 30 sec at650°C
in air, resulting in a light discoloration. Each foil, along
with a small quantity of Te metal, was placed in a
Hastelloy N Knudsen effusion cell coupled to a time-of-
flight mass spectrometer. After degassing, the foils were
reacted at 412°C until Te vapor became undetectable.
The first unoxidized foil was examined after the 412°C

11. R. E. Gehlbach and S. W. Cook, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL4782, pp. 173-74.

82

Y-112489

Fig. 10.22. Typical appearance of nickel telluride on a fractured Hastelloy N specimen exposed to tellurium during a 650°C

creep test. The sample is still in the specimen grip.

reaction. The second and third foils were reacted at
412°C and heated to 650 and 750°C for 96 and 725 hr,
respectively, the times required for effusion to cease.
The oxidized specimen was heated to 650°C for about
7hr.

Each foil was removed from the cell after its run with
its reaction products intact. With the exception of the
foil held at 750°C, thick black reaction layers formed
on both the Hastelloy N foil and cell surfaces. The
outer reaction layers were removed from the foils with
tweezers or by light scraping and usually flaked off as
sheets. The Knudsen cell reaction products were
scraped from the surface after each run and kept
separate from the foil products. The foil exposed at
750°C did not have a visible reaction product on the
surface. Instead, the foil appeared “pimply” and sagged
under its own weight in the effusion cell. In addition,
fairly tight bonding occurred between the corners of
the foil and the cell. Reaction products were removed
from the cell after the 750°C exposure. Pieces of each
foil were mounted for metallographic examination and
electron probe microanalysis. Although the outer reac-
tion layers had previously been removed, the surfaces
remained dark, indicating that an adherent product still
remained.

10.7.2 Possible Tellurides in Hastelloy N

Based on the composition of Hastelloy N we might
expect tellurides of nickel, chromium, molybdenum,
iron, and possibly combinations of these. The binary
tellurides of these systems are listed in the ASTM
Powder Diffraction File. An extensive study of the
nickel-tellurium system was made by Barstad et al.,'2
who identified beta, gamma, and delta nickel tellurides.
The beta field extends from NizTe, (NiTeg 7) to
NiTeg.7¢. Within this field, three crystallographic struc-
tures are observed due to slight variations in cell
parameters with composition. At NizTe,, a monoclinic
structure is observed with the angle = 91.22°. With
increasing Te concentration, § becomes 90° to give an
orthorhombic structure at NiTe, 9. At the composi-
tion NiTeg 5o (very slowly cooled) the continuing
changes in the @ and b parameters result in a tetragonal
structure. The gamma phase is located at NiTey ;4 and
is orthorhombic. The delta phase is hexagonal with an
extensive homogeneity range between NiTe, and ap-
proximately NiTe. The ¢ and a parameters and c/a

12. ). Barstad et al ‘On the Tellurides of Nickel,” Acta
Chem. Scand. 20(10), 286579 (1966).

ratios vary nonlinearly with composition across the
homogeneity range.

Two chromium tellurides are reported in the ASTM
Powder Diffraction File. a-CrTe is hexagonal with an
approximate homogeneity range of 50 to 54% Te. A
monoclinic modification at slightly higher Te concen-
trations is designated $-CrTe. Only one molybdenum
telluride, hexagonal MoTe, , has been identified. Several
iron tellurides are reported in the ASTM data, none of
which have been observed in any of our studies. We
have not found any reported temary tellurides.

Positive identification of many of the reported.

tellurides is difficult from x-ray data alone because
many have virtually identical interplanar spacings.
Monoclinic distortions can generally be observed from
x-ray-diffraction peak splitting. Many of our diffraction
patterns have lines of various intensities which cannot
be accounted for by reported tellurides, possibly
resulting from ternary phases. Thus, information on the
chemical composition of each phase is a useful supple-
ment to the x-ray data in phase identification.

10.7.3 Tellurides from Knudsen Cell Reactions

We identified NiTe, as the major phase formed by the
reaction of Hastelloy N with tellurium at 412°C in the
Knudsen cell. The foil reacted at 412°C and subse-
quently held at 650°C for 96 hr contained NiTeq ¢
and an additional strong phase which is probably a
high-parameter monoclinic $-CiTe intermetallic. The
interplanar spacings appear to be closer to the hexag-
onal a-CrTe, but several diffraction peaks were split,
suggesting the monoclinic distortion. No NiTe, was
detected at 650°C. As previously mentioned, the foil
exposed at 750°C for 725 hr did not have a removable
reaction product. An x-ray-diffraction pattern from the
surface of the slightly bent foil was poor, but weak lines
were indexable as Ni3Te, and CrTe in addition to
strong Hastelloy N diffraction peaks. However, the
reaction product removed from the Hastelloy N effu-
sion cell after the 750°C exposure did provide a good
x-ray-diffraction pattern. §-CrTe and Hastelloy N were
present in this reaction product, but nickel tellurides
were not detectable. The Hastelloy N—type lines prob-
ably resulted from decomposition of the nickel and
(possibly) mixed tellurides originally formed at 412°C
rather than from the original alloy.

Metallographic examination of these foils expectedly
revealed reaction layers along the surfaces of the foils
reacted at 412 and 650°C. Electron microprobe analysis
showed the reaction layer on the 650°C foil to consist
of Ni, Mo, and Te with less than 1% Cr. There were no

83

detectable concentration gradients of Ni, Mo, Cr, or Fe
at the edge of the foil itself.

The foil exposed at 750°C after the 412°C reaction
was markedly different from those exposed at the lower
temperatures. Figure 10.23 shows the microstructure of
this extensively reacted foil with striated particles along
the edges. The outer surface was mostly coated with
material which looked metallographically similar to the
Hastelloy N matrix. Electron microprobe analysis
showed the particles to be nickel and chromium
tellurides. Figure 10.24 shows the elemental distribu-
tion of Te, Fe, Cr, Mo, and Ni in the particles and
surrounding matrix of Hastelloy N. The striated por-
tions of the particles are basically chromium telluride
within nickel telluride. (The striated structure is not
visible in the backscattered electron image of Fig.
10.24; see instead Fig. 10.23.) Figure 10.25 is a
lower-magnification elemental distribution of a cross
section of the foil and tellurides. Most of the iron and
molybdenum were concentrated in the “foil area.”
However, large concentration gradients were observed
in the Hastelloy N—type material surrounding many
particles (Fig. 10.26). The center of the foil (area A) is
essentially Hastelloy N. The material at the edge of the
remaining foil (area B) is enriched in Mo and depleted
in Cr. The material around the telluride particles (area
G) is enriched in Fe and depleted in Cr. The tellurides
range from binary (Ni, area £') through ternary (Ni-Cr,
areas D, F, H, and J) to essentially binary (Cr, area C)
tellurides with up to 2% Mo and/or Fe. (The J telluride
is not shown in Fig. 10.26.) Beam overlap from the
striated tellurides may indicate more alloying than
actually occurred. The Mg C carbides in the foil (area /)
remained seemingly unaffected by the reaction with
Te. The reactions of this foil resulted under extremely
severe conditions, but they indicate the complexity of
the reactions which may occur between Hastelloy N
and tellurium,

10.7.4 Tellurides from Other Reaction Methods

Hastelloy N. We removed reaction products from
several specimens of Hastelloy N exposed to tellurium
vapor during creep tests at elevated temperatures.
Particles scraped from a sample exposed for 1500 hr at
593°C gave a very complex x-ray diffraction pattern.
The major phase was NiTeq 9, but NiTeqy 7, and/or
CiTe and possibly NiTe were present. Specimens
exposed at 650°C were found to form either NiTeq 49
or Ni;Te,, depending on the particular sample. Addi-
tional diffraction lines were obtained from some
samples that have not been identified.
84

Y-116133

=

T

0,005 INCHES
750X

T

™

Fig. 10.23. Optical micrograph of Hastelloy N foil exposed to tellurium at 412°C and heated to 750°C in Knudsen effusion cell.
Note striated tellurides along the foil, which has undergone severe reaction.

One of the creep specimens tested at 650°C was
examined by optical metallography and electron probe
microanalysis. The specimen had a reaction layer along
the edges which appeared similar to oxidation, with
preferential grain boundary attack. Some oxidation
may have occurred due to the impurity of the test
environment. This layer contained 30% Mo, 17% Te,
12% Cr, 4.8% Fe, and only 2.4% Ni (wt %). Presumably
this reaction layer is an oxide, but we have not
identified it structurally. Metallic particles adhering to
the specimen contained approximately 40% Ni, 62% Te,
and 0.8% Fe. No Mo or Cr was detected.

A 0.010-in.-thick sheet of standard Hastelloy N,
exposed to tellurium vapor for 100 hr at 700°C by J. H.
Shaffer of the Reactor Chemistry Division for Auger
spectroscopy (Sect. 10.8), formed NijTe, and CrTe
plus several unidentified lines. Employing the same Te
exposure technique we observed NizTe, on a low-sili-
con vacuum-melted heat of Hastelloy N (heat 5911)
and on a 2% Cr modification of Hastelloy N (heat 114).
In addition, CrTe was present on the specimen of heat
5911.

We have also examined the surface of a strain cycle
specimen that had been electroplated with 1 mg/cm? of
Te and tested in salt at 650°C. The only telluride
identified was Ni3Te,. We will be removing this thin
layer electrolytically to extend our investigation.

Nickel 200. One specimen of Ni 200 exposed to
tellurium vapor at 650°C during a creep test exhibited
only slight discoloration of the surface. X-ray-diffrac-
tion patterns from the darkened surface itself as well as
from the reaction film electrolytically stripped from the
surface were indexed as Ni3Te,. Transmission electron
microscopy of this film revealed large particles of nickel
telluride on a thin film covered with numerous small
thin platelets. A second specimen also exposed during a
creep test exhibited a large amount of reaction product.
The x-ray pattern from this material is best indexed as a
mixture of NiTeq 69, NiTe, and NiTe, with several
additional unidentified lines.

A thin sheet of Ni 200 exposed by vapor deposition
at 700°C for Auger spectroscopy was observed to
contain Niy Te, (plus several unidentified lines).
CrKa

Fig. 10.24. Electron beam scanning images showing elemental distribution i

85

MolLa

Y-115860

NiKa

striated tellurides. The nickel- and chromium-rich

tellurides are surrounded by an iron-rich “Hastelloy N matrix. Same specimen as Fig. 10.23.

Inconel 600. Particles scraved from specimens of
Inconel 600 creep tested at 650°C were identified as
NizTe, for one specimen and NiTeq.q9, CrTe, and
possibly NiTe for a second. These particles made up
what appeared visually to be an outer layer. Increased
emphasis will be placed on evaluation and analysis of
telluride formation in this system.

Molybdenum. Our only observation of the MoTe,
intermetallic made to date is from the residue in an Mo
capsule used to add Te to a salt vessel. The exposure
was for 24 hr at 650°C.

10.7.5 Summary

Our observations of tellurides formed on Hastelloy N,
Ni 200, and Inconel 600 in the temperature range 600
to 700°C indicate that the stable and most prevalent
phase is the f-Ni;Te,. The f-NiTeg 4o is characteristic-
ally observed when an excess of Te is present during the
reaction. The richer NiTe, which forms at 412°C is

unstable and transforms to the beta tellurides when the
temperature is raised. Chromium telluride, CrTe, is
observed to form on both Hastelloy N and Inconel 600.
Electron microprobe analysis of tellurides formed at
750°C suggests that alloying of the compounds may
oceur.

Although Ni;Te, appears to be the most stable nickel
telluride in the 600 to 700°C temperature range, the
presence of CrTe and “Hastelloy N as the Knudsen cell
reaction products after the 750°C exposure indicates
that CrTe is much more stable than NizTe,. The
Hastelloy N lines observed probably result from de-
composition of tellurides which formed during the
original reaction at 412°C. Strong evidence for this
arises from (1) the relative absence of nickel tellurides
(the nickel tellurides observed in the 750°C foil are
often surrounded by the iron-rich “Hastelloy N"), (2)
the extensive chromium depletion and large composi-
tional variations observed by microprobe analysis, and

1+0.0025in.+ %
BACKSCATTERED ELECTRONS

CrKa

MoLa

86

Y-115858

NiKa

Fig. 10.25. Electron beam scanning images showing elemental distribution across Hastelloy N foil reacted at 750°C. Same

specimen as Fig. 10.23 with a lower magnification.

(3) the fact that the foil was essentially sintered to the
Knudsen cell after the 750°C exposure.

The effect of molybdenum is not clear at this time.
We have not observed MoTe, in reaction products from
Hastelloy N, even though Mo is present in larger
quantities than Cr. The high Mo observed with the
electron microprobe in the reaction layers on the
Knudsen cell foils is unexplained, unless it is present as
a telluride which has not been previously identified.

The emphasis of our work will be shifted toward
microanalysis to investigate more thoroughly the beha-
vior of chromium tellurides, alloying of tellurides, and
the role of molybdenum. Reaction products which we
have examined by x-ray diffraction will be studied by
energy-dispersive x-ray analysis using the electron
microprobe and scanning electron microscope. We will
also study alloys which do not exhibit edge cracking
(both stainless steel and other nickel-base alloys), those
which are borderline (Inconel 600), and compare these

results with those from Hastelloy N and other alloys
which are embrittled. A major effort will also be
expended on a new series of small modified Hastelloy N
alloys with higher chromium concentrations and small
lanthanum and cerium additions in an attempt to tie up
tellurium or otherwise prevent its embrittlement of
Hastelloy N.

10.8 AUGER ELECTRON SPECTROSCOPY
OF INTERGRANULAR FRACTURE SURFACES
OF NICKEL AND HASTELLOY N EXPOSED
TO TELLURIUM VAPOR AT 700°C

R.E.Clausing  D.S.Easton R.E. Gehlbach

Auger electron spectrometry combined with sputter
thinning and in situ sample preparation provides a
powerful technique for studying the role of tellurium or
other fission products in producing intergranular frac-
ture or cracking. We have exposed several samples of

0.0015"

Y-115868

4.7 Ni D. 39.8 Ni
1.9 Mo 1.0 Mo
19.9 Cr 5.8Cr
74.8 Te 59.7 Te

0.8 Fe

E. 37.1 Ni
3 0.7 Cr
B. 53.1 Ni A
28.9 Mo

3.1Cr F. 10.1 Ni
5.5Fe 0.9 Mo
13.2 Cr
A. 71.7.Ni 75.6 Te
L G. 74.7 Ni
o 13.1 Mo
40T 2.8Cr
12.0 Fe

J.13.5 Ni 1. 29.6 Ni H. 13.5 Ni

1.5 Mo 58.1 Mo 1.5 Mo

19.6 Cr 4.7 Cr 19.6 Cr

55.5 Te 0.9 Fe 55.5 Te

1.0 Fe

Fig. 10.26. Quantitative electron microprobe analysis of various areas in Hastelloy N foil reacted at 750°C.

type 304 stainless steel, nickel, and Hastelloy N to
tellurium at high temperature to cause them to be
embrittled. These were then fractured in vacuum, and
the intergranular surface thus exposed was examined
immediately by Auger electron spectroscopy. The
elemental composition of the first few atomic layers
was determined and the sample sputtered to remove a
few atomic layers and expose a new surface for analysis.
Thus the composition of the original fracture surface
could be determined and information obtained regard-
ing the concentration gradients near the exposed grain
boundary. Information was obtained with depth resolu-
tion of one or a few atomic layers depending on the
depth below the original surface.

10.8.1 Method

We previously described a form of Auger electron
spectroscopy used for analysis of fission product
deposition on graphite in the MSRE core region.!? The
present techniques are much improved; although the
basic physical processes remain the same, the detection

and analysis of Auger electrons from the sample is now
by means of a high-resolution double-pass cylindrical
mirror analyzer.'* Auxiliary equipment permits the
fracture of samples in ultrahigh vacuum and has
provision for sputtering away a few atomic layers at a
time using 1000V argon ions.

The samples of type 304 stainless steel, nickel, and
Hastelloy N were in the form of miniature sheet metal
tensile specimens 0.010 in. thick. They were cleaned
and sealed in evacuated quartz capsules with a small
amount of tellurium. The capsules were then heated to
700°C for 100 hr to expose the samples to tellurium
vapor. To avoid unnecessary contamination the samples
were kept in the quartz capsules until immediately
before testing. Samples for analysis were obtained by
breaking them from the original tensile sample by
repetitive bending. These pieces of 0.010-in.-thick foil

13. R. E. Clausing, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL4676, pp. 143-47.

14. Physical Electronics Industries, Inc., model 15-25G, High
Resolution Cylindrical Auger Electron Optics.

were examined by optical and by scanning electron
microscopy to determine that the fracture indeed did
produce intergranular surfaces. They were then
mounted so they could be fractured again under
vacuum. One of the Hastelloy N pieces was abraded to
remove surface tellurium compounds and then coated
with titanium prior to being placed in the vacuum
system for fracture and analysis. All samples were
fractured under vacuum and spectra obtained from the
fractured surface within a few minutes.

Spectra were obtained using the smallest-diameter
electron beam available, which varied from about 0.015
in. FWHM (full width at half the maximum intensity)
for some of the initial data to less than 0.003 in. FWHM
for some of the latest data. Electron currents of 1 or 2
A at 2.5 or 5 keV were used. To enhance the
detectability of the Auger electrons, the pass energy of
the spectrometer was modulated by a 2-eV peak-to-
peak ac potential at 1.97 kHz, and a lock-in amplifier
was used to detect the ac component in the electron
current through the spectrometer. This arrangement
provides an output signal proportional to dN/dE (ref.
13), where N is the number of electrons in a given
energy interval and £ is the electron energy.

The data are obtained as an x-y plot of AN/dE vs E. 1t
has been shown!S that the amplitude of dN/dE is
directly related to the intensity of the Auger signal. The
signal is thus also proportional to the concentration of
the element producing the characteristic Auger elec-
trons. Standards must be used to convert signal intensi-
ties to concentration if accurate quantitative results are
required, but approximate compositions can be deter-
mined by comparison with published data.

We have examined one Ni 200 specimen and two
Hastelloy N specimens. Although the examination of
the second Hastelloy N sample is not yet complete, the
major results are clear. '

10.8.2 Nickel

The fracture surface of the nickel specimen was about
50% intergranular in a zone about 0.0025 in. thick just
below the original surface of the 0.010-in.-thick speci-
men, while the center 0.005-in. zone remained rela-
tively ductile and failed transgranularly. Unfortunately
the electron beam diameter was too large to permit
analysis of the 0.0025-in. zone without including
material from the ductile region. It is estimated by
comparison with computer-generated spectra that the

15. P. W. Palmberg and T. N. Rhodin, J. Appl. Phys. 39,
2425 (1968).

88

concentration of Te on the fracture surface may be 20
to 30 at. %. Sputtering away the original surface to an
estimated depth of 100 atomic layers removed tellu-
rium to below the limit of detectability (estimated to
be about 1 at. % under these conditions). Figure 10.27
shows the tellurium signal strength as a function of
sputtering time. The sputter etching removed about two
atomic layers per minute. This sputtering rate is based
on a plane perpendicular to the sputter beam. Since
rough surfaces will sputter unevenly and some sputtered
material may be redeposited on other parts of the
fracture region, sputtering of rough samples may
proceed at a slightly slower rate, and the depths
calculated should be considered to be slightly too great.

The results on the nickel sample show a very high
tellurium concentration at the surface of the brittle
fracture (along the grain boundaries) that drops off
rapidly as the original fracture surface is sputtered
away. The tellurium is thus highly concentrated in the
original grain boundary. This could result either because
of an affinity of tellurium for the grain boundary or
because the diffusion of tellurium along grain bound-
aries was much faster than through the bulk.

10.8.3 Hastelloy N

The fracture surface of a Hastelloy N sample is shown
in the scanning electron micrograph of Fig. 10.28. Note
the intergranular fracture in the zone near the original
surface and the cracks in the surface of the foil near the
fracture. Two Hastelloy N samples were examined
under similar conditions, but the results from the
transverse scan of the fracture surface of the second
sample were superior and will be presented. The second
sample differed from the first in that the tellurium

ORNL-DWG 72-10581A

25

oy J
SI? Ni-A3
4

2 e

> |

- 3

[7p]

=

- \

S 2

-

3 \

& '

@ R .

W 0

§ 0 10 20 30 40 50 60 70 80

SPUTTERING TIME ( min)

Fig. 10.27. Intensity of 482-eV tellurium Auger signal as a
function of sputtering time. The initial point was from the
-as-fractured surface of the nickel. The sputtering rate is about
two monolayers per minute,
" PHOTO 0085-73

Fig. 10.28. Fracture surface of Hastelloy N sample. Note the
brittle intergranular fracture in the zones near the original
surfaces of the foil and the cracks and grain boundary
separation in the areas adjacent to the fracture.

reaction layer on the foil surface was removed by
abrasion and titanium was vapor deposited on the foil
surface to hide any tellurium-rich surface which might
remain. This would avoid possible errors due to Auger
emission from the original foil surface.

Figure 10.29 shows a beam traverse of the broken
surface along a path perpendicular to the plane of the
foil. The diameter of the electron beam was still too
large to permit simple interpretation of the data.
However, the data show that the brittle zones have
tellurium present, while tellurium is much lower or
undetectable in the center ductile zone. Figure 10.30
shows the tellurium signal strength as a function of
sputtering time for the beam centered on one edge of
the sample. The sputtering rate is no more than 2
atomic layers per minute. Note that the tellurium
concentration decreased nearly a factor of 10 in 3-min
sputtering, or 6 atom layers.

Figure 1031 is a sputter profile from the first
Hastelloy N sample, showing variations in Auger signals
for various elements as a function of sputtering time.
This sample was sputtered much longer than the second
one, and the initial points were from a surface that had
been sputtered for 2 min. Note the very slow decline in

ORNL-DWG 73-837

175

150

100

T T
NICKEL
CHROMIUM
MOLYBDENUM
IRON —
TITANIUM
TELLURIUM

|

INTENSITIES

75

AUGER SIGNAL

50

-

0 2 4 6

8
DISTANCE (~0.001 in. /division)

10 12 14 16 18

Fig. 10.29. Auger intensities from the fractured edge of a Hastelloy N foil as the sample was scanned in the thickness dimension.
The Hastelloy N foil was 0.009 in. thick. Note that tellurium was found only at the edges and not in the center of the sample.

ORNL- DWG 72-11302

40

30

20

AUGER SIGNAL INTENSITY

Te

-

T ———

0
FRACTURE 1 2 3
SURFACE
SPUTTERING TIME (min)

Fig. 10.30. Auger signal intensity as a function of sputtering
time for the second Hastelloy N A41 foil sample. The initial
point was from the as-fractured surface. The sputtering rate was
approximately two monolayers per minute.

tellurium with increasing depth. Tellurium is still easily
detected after over 400 min of sputtering. This would
indicate that the tellurium-rich zone at the fracture
surface is thicker in Hastelloy than in nickel.

These data provide strong evidence connecting teilu-
rium with brittle intergranular fracture in Hastelloy N:
tellurium is strongly concentrated in the affected grain
boundary region. QOther observations include (1) a
strong depletion of iron from the grain boundary and
(2) an indication that chromium is higher in the brittle
region than in the ductile zone.

The analysis of the second Hastelloy N sample is not
yet complete. A composition profile as a function of
depth below the fracture surface is being made.

A new electron gun has been obtained that is
expected to be capable of producing an electron beam
with a diameter of only 0.001 in, (compared with 0.003
to 0.015 in. in the experiments reported here). This
would greatly improve the resolution of the composi-
tion gradients across the thickness of the foil.

90

10.9 DESIGN OF AN IN-REACTOR EXPERIMENT
TO STUDY FISSION PRODUCT EFFECTS

ON METALS
R. L. Senn H. E. McCoy
P. N. Haubenreich

J. H. Shaffer

During this report period we began design of an
experiment in which three prospective MSBR container
materials will be exposed to molten fluoride salt in
which fission products are being produced. Fission
product concentrations, temperatures, and times will be
the same for each material and will be in the ranges that
were found to produce surface cracking of the Hastel-
loy N material that was used in the MSRE. The relative
resistance to cracking of the different materials will be
determined by straining specimens of each material
after exposure.

10.9.1 Design Criteria

In order to accomplish the purposes of the experi-
ment in an economical, timely, and safe manner, the
following design criteria were adopted.

1. Three fuel pins shall be irradiated simultaneously
for one cycle in the ORR.

2. Each pin shall be made of ' -in. tubing.
MSRE-type fuel salt shall contact 3 in. of the
cylindrical portion of each pin.

4. At least 0.5 in. of the cylindrical portion of each pin

shall be above the salt surface.

5. The inner surfaces of the walls in contact with the
salt shall operate at 700°C, as nearly as is practical.

Fuel pin temperatures shall be monitored by one or
more thermocouples attached to the outside of
each pin; interior temperatures shall be estimated
from the observed wall temperatures.

Heating provisions shall be adequate to ensure that
salt temperatures do not go below 100°C for more
than 2 hr at a time from the beginning of
irradiation until the experiment is removed from
the ORR shortly after irradiation (to avoid F,
evolution).

Insertion and withdrawal of the capsule in the ORR
poolside irradiation facility shall be used to regu-
late the nuclear heating in the experiment.

The 233U content of each pin shall be sufficient to
produce at the end of one cycle in the ORR a
tellurium inventory of at least 5 X 10'® atoms per
square centimeter of area exposed to salt.
ORNL-DWG 73-838

12
'l
LA Mo
) / //
> & ///A
@ 8 -
= e
= L
2 d
S 6 d
n [ // b c
QU [ —— r o
A -‘fi-.7<tv-"‘—-.~_ [ ___—--"'—"—_.—_-‘
2
. // — | —_____U,—o-
g 4 // — —
2 | 0 2.0
.—-—‘—"—-_ r
- I
‘-—._l-—.____‘_-_-
, --l.._..._____ Fe
--->== — D |
o aaran & Q I -
_____._-—"" Te
JL——-—'—'—'————P——— |
.0
100 2 5 10! 2 5 10° 2 5 10°

SPUTTERING TIME (min)

Fig. 10.31. Auger signal intensity as a function of sputtering time for Hastelloy N A41 sample. The initial point was on the
as-fractured surface. The sputtering rate is about two monolayers per minute.

10. Containment shall be provided by two barriers
outside the pins.

10.9.2 Configuration

The configuration conforms to the ORR poolside
facility and is similar to previous poolside capsule
designs.!© This is particularly true in the upper portions
of the capsule, which include the instrument lead tube
and connector, gas tubing and connectors, and heaters
and their connectors. The lower portion of the capsule
will provide the necessary containment and a controlled
high temperature in the fuel pins while exposed to
neutrons from the ORR core.

The lower portion of the capsule is shown schematic-
ally in Fig. 10.32. (TeGen-1 stands for Tellurium
Generator No. 1.) Each of the three sealed salt
containers (fuel pins) is made of a 4-in. section of
1,-in.-OD, 0.035-in.-wall tubing. The three materials
‘chosen for this experiment are: standard Hastelloy N,
which is highly resistant to corrosion but which cracked

16. D. B. Trauger, Some Major Fuel-Irradiation Test Facilities
of the Oak Ridge National Laboratory, ORNL-3574 (April
1964).

-in the MSRE; type 304 stainless steel, which resists

intergranular attack by tellurium but is corroded more
by fluoride salts, and Inconel 601, a nickel-base
high-chromium alloy that may combine adequate resist-
ance to both tellurium attack and fluoride corrosion.
Each pin contains 7.14 cm?® of salt, which fills 3 in. of
the length and contacts 27.1 cm?® of metal surface. The
salt composition is ’LiF-BeF,-ZrF,-UF, (63.1-29.3-
5.1-2.5 mole %); except for slightly higher UF, this is
practically the same as the fuel in the MSRE where the
cracking was first observed. The %, -in. space above the
salt in each pin is filled with helium.

10.9.3 Thermal Considerations

Fission heating in the salt is used to produce the
desired high temperature in the pins during ORR
operation. Heaters are also provided around each pin
for use in minimizing temperature differences among
the pins and in keeping the salt warm enough to prevent
F, gas evolution during shutdown periods. Pin surface
temperatures are measured by attached thermocouples,
four on each pin. Fuel pins and heaters are immersed in
NaK inside the primary containment vessel. Thermo-
couples in the bottom of the NaK pool would sense
ORNL-DWG 72- 9654 A

STAINLESS STEEL

ARGON ‘
HELIUM —_ B

4
71— 304
=

”
MoK T PRIMARY CONTAINMENT
R Ty = 304 STAINLESS STEEL
T="—1[If ~ SECONDARY CONTAINMENT
TE-101 ' § ] — FUEL_PIN NO.1: 304
R 1l __ (el STAINLESS STEEL
==
TE 102 el J [l — MSR FUEL SALT
B U —U 7"‘/
TE-103 el =1 ) ]
|__[ §jfi — HEATER coiL No.
TE-104 R o Ry 1)
1 10 s 1
11
N U
TE-105 ] g / FUEL PIN NO. 2
e ;E’!‘_fl/"”/ HASTELLOY- N
TE 106 —p e~ ol f MSR FUEL SALT
B
TE-107 kil HEATER COIL NO.2
z ?/___fi _‘1/_ )
TE-108 Rd HHH
] 1
T
1
10 7 | FUEL PIN NO.3:
TE-109 1R ,8/‘/ INCONEL 604
11 : )
TE - {10 ;/~ f H— MSR FUEL SALT
1] r“ —_
TE-111 is ~
i YU —— HEATER COIL NO.3
TE-112 el
Te-113~_{I{ !
TE-114 7

TE—145

Fig. 10.32. Schematic drawing of TeGen-1 capsule.

abnormal heating if salt should leak from a pin and
react with NaK to precipitate uranium. An argon-filled
annulus between the primary and secondary contain-
ment vessels is sized to provide the thermal resistance
that is required between the hot NaK and the water
surrounding the capsule.

The fissile material in the salt is 233U, chosen because
its fission product tellurium yield is greater than that of
2351 (0.040 atom stable Te per fission compared with
0.024) so that more tellurium can be produced with a
given amount of heat generation. The goal of 5§ X 10!
atoms of stable Te per square centimeter of surface
contacted by salt was chosen as being low enough to be
attained without inconveniently high heat production
and high enough to produce cracking. (Specimens
removed from the MSRE in July 1966 showed cracks,
although the Te concentration had reached only 1.0 X
10' ¢ atoms/cm?.) Producing 5 X 10'® atoms of stable

92

Te per square centimeter in 1100 hr from 222U in a
0.43-in.-diam cylinder requires a fission power density
of 34 W per cubic centimeter of fuel (1.19 kW per foot
of fueled length).

When the ORR first starts up with the experiment
installed, the distance of the capsule from the core will
be adjusted to bring the temperatures on the middle
fuel pin into the desired range. The installed heaters on
the upper and lower pins (which will have about 14%
lower fission rate because of the axial flux variation)
will be used to bring the temperatures on these pins
into the same range. If the heat-transfer calculations
used to size the argon annulus are accurate, the fission
rate (and tellurium production) will then be as ex-
pected.

Figure 10.33 shows radial temperature profiles calcu-
lated for a long cylindrical model with heat transfer
only by conduction under two conditions. The lower
profile is for the design condition of 1.19 kW/ft of
fission power. The upper profile shows what would
happen in the same system, at the same location, if the
reactor power should increase to 120% of nominal full
power. The increase in pin wall temperature would not
be hazardous. Actual temperature differences from the
wall to the center line will tend to be less than shown
because of thermal convection effects which were not
considered in this simple calculation. Even the center-
line temperatures shown in Fig. 10.33 are not excessive,
however, as there are no changes of any consequence in
the properties of the salt over the ranges shown.

In the actual 4-in.-long pins, heat flow out the ends
would tend to cause axial variation of the temperature.
The spiral heater elements were therefore concentrated
near the ends of the pins to counteract this effect to a
large extent. Figure 10.34 indicates the location of the
heater coils and gives temperature distributions calcu-
lated!? with the three-dimensional heat conduction
code HEATING-3. (The 1.2-kW/ft fission power and
1.1 kW/ft heater power refer to the fueled sections and
the heated sections only, not to the averages over the
length of the capsule.) The calculated distribution for
normal operation indicates that the spread among
temperatures at different points on a fuel pin will be
acceptably small. Thermal convection effects, neglected
in these calculations, will tend to lower center-line
temperatures but possibly increase the spread of temp-
eratures along the pin wall. The calculated temperatures
for the case of zero fission heat also show that the
design power of the heaters is ample to keep the salt

17. H.C. Roland, ORNL, private communication.
¢

CAPSULE

93

ORNL—DWG 72-9655A

TEMPERATURE (°C)

1073
262

DESIGN POINT —m—

RADII (in.)
0.215

!

0.2155

0.250

0.388

0.428

0.453

0.498

UPPER LINE REPRESENTS TEMPERATURE PROFILE WITH
CAPSULE OPERATING AT 120 % NOMINAL OPERATING POWER.
ASSUMES 1.43 xW/ft FISSION POWER AND 0.2 W/g GAMMA HEAT.

LOWER LINE REPRESENTS TEMPERATURE PROFILE WITH CAP-

SULE OPERATING AT NOMINAL DESIGN POWER.

ASSUMES

1.19 kw/f1 FISSION POWER AND 0.2 W/q GAMMA HEAT.

Fig. 10.33. TeGen-1 schematic temperature profile from GENGTC calculations. Temperatures are in °C.

temperature well above the threshold of less than
100°C for fluorine evolution from irradiated salt.

10.9.4 Containment

Four gas lines are used in the experiment: two for
filling and maintaining a 20-psig helium pressure over
the NaK (primary containment internal pressure) and
two for filling and maintaining a 40-psig argon pressure
in the gas gap (secondary containment pressure).
Pressures over the NaK and in the gas-filled annulus are
monitored for any indication of leakage in either the
primary or secondary containment.

10.9.5 Fuel Salt Chemistry

The fissile uranium concentration in the fuel salt was
chosen as follows. With 1.0 mole % 233 UF, in the fuel

salt, calculations showed that a thermal neutron flux of
about 0.9 X 10'? neutrons cm ™ sec™ would produce
the required fission power density. This flux is readily
achieved in the ORR poolside positions with the
capsule retracted 10 to 12 in. from the face of the
reactor. Furthermore, at this rather large distance from
the reactor, the vertical flux profile is relatively flat, so
all three fuel pins will be irradiated in approximately
the same flux (*~7%). Thus a different fuel loading for
each pin will not be required if the fissile loading is high
enough to operate this far away from the core.
Therefore, the concentration of the 233U isotope was
set at 1.0 mole %.

The purposes of the experiment require that the fuel
salt chemistry be carefully controlled. Specifically, it is
desirable that the U3 /U* ratio be in the range
experienced in the MSRE and expected in future
94

ORNL-DWG 73-839

<1
Z 5 C NOMINAL DESIGN OPERATION WITH
2 S = POINT OPERATION HEATER ONLY
p— o puny
S £ ¥8 1.2 kW/ft FISSION HEAT O kW/ft FISSION HEAT
- 3 1.1 kW/ft HEATER POWER 1.1 kW/ft HEATER POWER
o
FUEL FUEL PIN FUEL FUEL PIN
CENTER LINE INNER WALL CENTER LINE INNER WALL
) TEMPERATURE (°C) | TEMPERATURE (°C) | TEMPERATURE (°C) { TEMPERATURE (°C)
[ POSITION A POSITION B POSITION A POSITION B
906 (e99) 386 396
FUEL PIN
966 704 328 327
970 707 286 286
32 in. FUEL SALT
972 708 267 266
; 971 707 269 268
L. 968 706 291 290
’—*— ‘ 713 700 340 336
Yo in 259 14 1o 130
TYPICAL ‘
[] maximMum TEMPERATURE
MAXIMUM
A/‘-\X:AL) () MINtMUM TEMPERATURE
T(oC

Fig. 10.34. TeGen-1 capsule two-dimensional temperature profiles at various operating conditions.

molten-salt reactors (on the order of 0.01). Two

mechanisms tend to decrease the ratio. They are.

impurity reactions, such as
2NiO + 4UF; - UO, + 3UF, + 2Ni,

and fissioning of some of the uranium to produce
lower-valence products. Great care will be used in
cleaning and loading the pins to minimize impurity
reactions, but the amount of U3* going to U*" as a
result of fissioning cannot be changed. The change in
the U**/U*"* ratio can be reduced, however, by increas-
ing the total amount of uranium in the pin. If the total
uranium fluoride concentration were 1.0 mole %, the
U3*/U*" ratio would decrease by about 0.008 as a result
of fissioning the amount of 223U required to produce
the desired amount of tellurium. This change in ratio is
large compared with the desired range, so some increase
in uranium is desirable. A total concentration of 2.5
mole % uranium fluoride was selected as being reason-
able. With this much uranium, the anticipated change in
the U**/U*" ratio due to fissioning is reduced to 0.003.
The 233U concentration will be kept at 1.0 mole %

while the total uranium is raised to 2.5 mole % by
mixing depleted uranium with the 233U procured for
the fuel charge.

Plans for salt production and pin filling were designed
to ensure the desired U>*/U*" ratio in the pins at the
beginning of operation. The fuel salt mixture will be
prepared from 23**UQ, and other constituents as
fluorides by treatment with anhydrous hydrogen fluo-
ride admixed with hydrogen at 600°C in nickel equip-
ment. The U3*/U*" ratio will be determined in situ by a
voltammetric probe and will be increased if necessary
by the addition of beryllium metal. The carefully
cleaned fuel pins will be connected between a salt
reservoir and overflow vessels and will be filled, drained,
and refilled sequentially by control of the gas pressure
differential imposed on the system. Following these
operations, the salt fill lines will be cut from each fuel
pin and the final closure weld performed in a controiled
atmosphere. Because of radiotoxicity considerations of
233U, operations involving salt preparation, fuel pin
filling, and welding will be performed in glove boxes
within an established alpha facility.
11.

95

Graphite Studies

W. P. Eatherly

The objectives of the Graphite Program continue to
be the development of improved radiation-resistant
graphites and the development of techniques to seal the
graphite against ' ** Xe. The major emphasis remains on
the sealant problem.

The period covered by this progress report has seen
two significant advances, described in detail below. The
first ‘of these is the successful irradiation of our
experimental graphites to 3.5 X 10?? neutrons/cm? (£
> 50 keV) with a high degree of dimensional stability
demonstrated. We interpret this as at least partial
demonstration of our concepts on the nature of
radiation damage in polycrystalline graphite.

The second advance is in methods of fabricating
pyrolytic coatings to seal out xenon from the bulk core
graphite. Flaw-free coatings with the desired density
and isotropy are now laid down routinely on the small
cylindrical test pieces. Samples are now under irradia-
tion testing, and preliminary results will be available in
the next reporting period.

Development work has been initiated to obtain
anisotropy measurements on pyrolytic coatings by their
optical anisotropy under polarized light. The optical
equipment is performing well, and standards for both
optical and x-ray anisotropy measurements are being
fabricated.

11.1 THE IRRADIATION BEHAVIOR OF GRAPRITE
C. R. Kennedy

The purpose of this program is to interrelate the
lifetime expectancy and dimensional stability of graph-
ite under neutron irradiation to fabrication variables.
During this report period, two graphite irradiation
experiments have been performed in the HFIR; both
were geometrically the same as previous experiments.
The first, run at 715°C, was allowed to accumulate 2 X
10%2 neutrons/cm? (E > 50 keV) before removal. The
second was run with neon replacing the helium cover
gas to increase the specimen temperature to a calculated
950°C. (The temperature monitors have not been run as
yet to confirm the operating temperature.) The 950°C
experiment was only allowed to accumulate 1 X 10%2
neutrons/cm2 before removal from the reactor. The
grades irradiated in one or both experiments are listed
in Table 11.1.

All of the samples in the 715°C experiment were
ORNL grades, some made with and some without a
coke modification process. Previously irradiated first-

generation samples made with unmodified coke reached
an accumulated fluence of 3.5 X 10?? neutrons/cm?.
Second-generation modified materials, which had no
previous irradiation, accumulated a fluence of 2 X 10?2
neutrons/cm?. The length changes as a function of
fluence, shown in Figs. 11.1—11.4, demonstrate several
significant factors relating to the dimensional stability
of the ORNL graphites. The first is that they are very
dimensionally stable, generally changing less than 1% in
length up to a fluence of 2 X 10%?* neutrons/cm? (E >
50 keV), then expanding linearly with fluence at a rate
of 2.5% per 10?? neutrons/cm?®. No accelerating
breakaway in expansion has been observed to 3.5 X
10*% neutrons/cm® for the ORNL materials. (This
behavior is in contrast to the Poco grades, which remain
stable slightly longer but then expand at an accelerating
rate.) The lifetime expectancy of these first-generation
(unmodified coke) grades would be greater than the
reference grade GLCC H-337, and could approach the
Poco AXF lifetime, depending upon the expansion
allowed. For example, these ORNL experimental grades
tolerate a higher fluence before reaching volume expan-
sions of 5% or more than the Poco AXF grades do.
Although the second-generation (modified coke) graph-
ites have only been exposed to a fluence of 2 X 1022
neutrons/cm?, up to this fluence, at least, they show a
slightly better stability than the earlier ORNL grades.
The lifetime of the newer grades will be determined by
further irradiation.

Several subtle effects of fabrication and raw materials
were observed in these irradiations. The first is that
although these grades are very isotropic, they always
have a slightly greater initial shrinkage in the direction
normal to the molding axis. After this initial difference,
the dimensional changes are very isotropic. The effect
of original bulk density is very small and much less than
that observed for conventional grades. The addition of
up to 40% carbon black also appears to have only a
minor influence. The grades with carbon black all had a
higher original bulk density. Thus, the slightly earlier
initiation of volume expansion may be related more to
the higher bulk density than to the addition of the
black. Among the binders used — coal-tar pitch,
petroleum pitch, furan-pitch mixtures, and nitroben-
zene-pitch mixtures — no differences in behavior were
observed.

The 950°C experiment was designed to extend our
knowledge of the effect of irradiation temperature.
Representative commercial and experimental samples of
Table 11.1. Graphites irradiated in the HFIR

96

Grade Mf(:)t:;]?:‘gf desfiil:trion Binder Impregnation 27:151;?) Source”
Conventional grades
AGOT Extrusion Coarse acicular Pitch Probable 1.70 UCC
NL-19 Extrusion Coarse acicular Pitch Probable 1.72 UCC
NL-31 Extrusion Coarse acicular Pitch Probable 1.74 UcCcC
H-327 Extrusion Coarse acicular Pitch Probable 1.75 GLCC
H-429 Extrusion Coarse isotropic Pitch Probable 1.75 GLCC
ATI-S Molded Medium acicular Pitch Multiple 1.82 UCcC
RY-12-29 Extrusion Fine acicular and Thermax Varcum None 1.90 Y-12
Carbon black grades
A-680 Molded Carbon black Pitch None 1.49 SP
A-681 Molded Carbon black Pitch None 1.66 Sp
SA-45 Molded Carbon black Pitch Possible 1.55 ucc
T-30 Molded Thermax Furan-pitch None 1.51 ORNL
Apparently binderless grades
AXF Molded Fine semiacicular ? None 1.83 Poco
AXM Molded Fine semiacicular ? None 1.71 Poco
H-337 Molded Medium semiacicular ? Multiple 2.00 GLCC
ORNL grades

116 Molded 60% Robinson, 40% Thermax 15V pitch No 1.83 ORNL
129 Molded Robinson 15V pitch No 1.65 ORNL
157 Molded 60% Robinson, 40% Thermax 350 pitch No 1.87 ORNL
S-31 Molded Robinson 15V pitch No 1.76 ORNL
518 Molded 65% Robinson, 35% Thermax 15V pitch No 1.89 ORNL
NB-16 Molded 65% Robinson, 35% Thermax 15V pitch, nitrobenzene No 1.83 ORNL
511 Molded Santa Maria 15V pitch No 1.66 ORNL
N-2 Molded Santa Maria 15V pitch, nitrobenzene No 1.74 ORNL
S-14 Molded 65% Santa Maria, 35% Thermax 15V pitch No 1.77 ORNL
T-61 Molded Modified Robinson 240 pitch No 1.86 ORNL
T-58 Molded Modified Robinson Nitrobenzene No 1.90 ORNL
T-12 Molded Modified Robinson Furan-pitch No 1.79 ORNL
10-25 Molded Modified Santa Maria Furan-pitch " No 1.83 ORNL
11-1 Moided Modified Santa Maria Furan-pitch No 1.85 ORNL
11-23 Molded Modified Robinson Furan-pitch No 1.84 ORNL
11-43 Molded Modified semiacicular Furan-pitch No 1.78 ORNL
22-28K Molded Modified acicular Furan-pitch No 1.73 ORNL

4UCC ~ Union Carbide Corp., Carbon Products Division.
GLCC — Great Lakes Carbon Corp.
SP — Stackpole Carbon Co.
Y-12 — Y-12 Plant,
Poco — Poco Graphite Co.
ORNL - Oak Ridge National Laboratory.
97

ORNL— DWG 73—849

8
| | |
o MOLDING DIRECTION
5 6| —*® TRANSVERSE DIRECTION
" | | | o
Z GRADES 129 AND S-3f ~
T 44—
Q / ®
- ///
— (o]
D
E 2 ° // hd
- o} 0 " © /
(a7 Oo.—'—'a-___c [ /
~J ,/. b et @
~ o) ] [
~J
Q
-2
0 5 10 15 20 25 30 (x102Y)

FLUENCE (neutrons /¢cm2, £>50 keV)

Fig. 11.1. Length changes in Robinson coke graphites irradiated in the HFIR at 715°C.

ORNL-DWG 73-850

8
—
(s}
Se >
3 GRADES 116,157, 5-18, NB-16 //
+
- O MOLDING DIRECTION >/
< ® TRANSVERSE DIRECTION
4 o— -
W
[ ]
g > /{ /
s o :)(1___00_.—-0'/ ™
0] et
L ¢ o* o
& o o ‘E/J ———
| Y [ ]
2 21
0 5 10 15 20 25 30 (x10°")

FLUENCE (neutrens/cm2, £> 50 keV)

Fig. 11.2. Length changes in Robinson and Thermax mixture graphites irradiated in the HFIR at 715°C.

~

three broad types of graphites' — conventional, carbon
blacks, and apparently binderless — were included in
this experiment. The behavior of each type is compared
with the results obtained at 715°C in Figs. 11.5 and
11.6.

Figure 11.5 gives the results of several conventional
grades which present a consistent behavior pattern.
First, the samples irradiated at 715 and 950°C densify

1. C. R. Kennedy, MSR Program Semiannu. Progr. Rep. Aug.
31,1969, ORNL-4449, pp. 175-78.

at about the same rate. The so-called “breakaway” or
transition to volume expansion occurs much earlier at
950°C and reduces the lifetime expectancy by about a
factor of 2. Grade RY-12-29, bindered with a furfuryl
alcohol binder (Varcum), behaves similarly at- 715 and
950°C except that the changes at 950°C are intensified.
This is the binder that produces a carbon phase of ~1.5
g/cm® which densified during irradiation to 2.15 g/cm?®.
This process creates voids between filler particles, and
the structure degenerates. The heavily impregnated
ATJ-S also appears to begin expanding more rapidly
than the conventional electrode stock graphites. Ex-
cluding the RY-12-29 and the ATJ-S grades, all of these
graphites would have a lifetime expectance at 950°C,
based on volume expansion, of around 10 to 11 X 10*'
neutrons/cm?.

The results of the carbon-black grades are shown in
Fig. 11.6. Again, the effect of temperature is consistent
with the conventional graphite behavior. The densifica-
tion is almost identical to the behavior at 715°C, since
these materials all densify very rapidly. The expansion
rate does appear to be larger than that observed at
715°C; however, the irradiation exposure should be
extended before quantitative values are obtained. As
observed at 715°C, the major difference between

ORNL-DWG 73-851

8 .
a

I
MOLDING TRANSVERSE o
— DIRECTION  DIRECTION GRADE

o . S-11 AND H-2
a - 5-14

Yl

[}

n
N

o

o
-3

]
o
[=]

& b //
" a

LENGTH CHANGE [100 In (1+A4/Lg)]
o ro

0 5 10 i5 20 25 30 (x102)
FLUENCE (neutrons /cm?, £ >50 keV)

Fig. 11.3. Length changes in graphites made with all Santa
Maria and with a mixture of 65% Santa Maria and 35% Thermax

98

1000°C heat-treated grade A-680 and 2800°C heat-
treated grade A-681 is the initial densification. The
lifetime expectancy of these grades, as observed at
715°C, does appear to be greater than the conventional
grades.

The apparently binderless grades and ORNL experi-
mental grades are compared in Fig. 11.7. All of these
grades are very isotropic except grades 11-43 and
22-28K, which were made with anisotropic cokes (grade
22-28K being more anisotropic than 11-43). The
irradiation is actually too small to give a valid indication
of the expected lifetime or general behavior. The trends
indicate that there is little difference in AXM and AXF,
which are expanding linearly. Grade H-337 seems to
have densified slightly and then begun to expand at a
rate greater than the Poco grades. The ORNL grades
have almost too much scatter to define a trend;
however, they do appear to have a slight initial
densification, then some expand and some do not. They
seem certainly to be as stable as H-337, if not more so.
One point demonstrated by the results for grades 11-43
and 22-28K is that isotropy is not required for
volumetric stability. Grade 22-28K has experienced
linear dimensional changes of over 4% growth in the
molding direction and 2% shrinkage in the transverse
direction with virtually no volume change.

These results yield the following general conclusions:

1. First-generation ORNL grades have a lifetime ex-
pectancy greater than the reference grade H-337 at
715°C.

2. There are no growth rate or densification rate
increases from irradiation temperatures of 715 to

irradiated in the HFIR at 715°C. 950°C.
6 ORNL-DWG 73-852
° MOLDING ~ TRANSVERSE GRADE
} a DIRECTION DIRECTION
b ) ° T-61
N A A X-58
- g . T-12
Cg) R 10-25 pd
o c@fl /
| S—
N ¥
§ C — >4 .v. ] "%’L
£ ¢ ~—1r |
Q
=
-2
=
L
|
-4
20 25 20 35 (x10%))

o 5 10 15

FLUENCE (neutrons/cm2, £>50 keV)

Fig. 11.4. Length change of ORNL graphites made from modified coke irradiated in the HFIR at 715°C.
99

ORNL-DWG 73-853

» >

6
11.2
4 7
7
Ve
— é ‘/
g 2 ' 7
= o
< -
b N
Z 0 <
= ]
£ o° [} -~
o X , - \
2, S j@ - 715°C
e -
g -2 \ un:-._______/‘__ ¥ A
O \Q/ 8 //
% 7 - \Q L S //
" -4 e sl 4
% \\\;/v/ \\ _ L' //
> _gl— 9 NL-1S -~ 6
& NL-31
e H-327
= H-429
_g | & RY-12-29
v ATJ-S
| I
0 2 4 6 8 10 12 14 16 18 (x102h
FLUENCE (neutrons/cme £ > 50 keV)
Fig. 11.5. The volume change of conventional graphites at 950°C.
ORNL-DWG 73-854
4
3
~
~J Q 7?15 °C — et
p n’ AT
<0 / _E._—-@L‘/
< fl’l—-‘ SA-—45 _.._—-"——_-_-—-
P n_____—"‘. —_________—
E H—#— _— ___—/
w _2
[F1} "
- 80 -—
o g
a—"
a—
5 —4 d o /"’r -__‘—— ———
X . __——-—"'__-.
- —————
N LENGTH diom GRADE
W _g o e A-680
a . A-681
A A SA-45
-8 | ]
0 2 4 6 8 10 12 14 16 18 (x10%")

FLUENGE (neutrons/cm?, £>50 keV)

Fig. 11.6. Length changes of carbon black grades irradiated in the HFIR at 950°C.
ORNL - DWG 73-8535

8 i
— e AXM
© A AXF
X gl W H337
< o -4
= o 1-23
£ ¢ 1—43
8 4 - a 22-28K
g /u)
Z o 847 /
;: ._//
O a
w
= n/l ¢
5 0O I
8 0 m=t= a
> o o
-2
0 2 q (<) 8 10 12

FLUENCE (neutrons /cm?, £>50 keV) -

Fig. 11.7. Volume change of apparently binderless and
ORNL grades of graphite irradiated in the HFIR at 950°C.

3. The transition to volume expansion occurs more
rapidly, by approximately a factor of 2, at 950°C
compared with 715°C.

4. Isotropy is not required for volumetric stability.

5. The bindering systerfi and particle boundary mor-
phology are more important than growth rates in
determining lifetime.

The general behavior at 950°C for virtually all of the
graphites is indeed identical to the 715°C behavior with
one major exception: the earlier transition to volume
expansion. The implication is that this is not caused by
growth rate increases but by a lessening of the
graphites’ ability to absorb the shearing deformation.

11.2 PROCUREMENT AND CHARACTERIZATION
OF VARIOUS GRADES OF GRAPHITE

W.H.Cook J.L.Griffith

Graphite procurement and evaluations in the MSRP
are currently directed toward two applications: (1) core
moderators and reflectors for MSBR’s and (2) equip-
ment for processing the reactor’s fluoride salt fuels.

For the studies -on the nuclear application, our
objectives are to identify grades that have superior
resistance to dimensional changes induced by fast
neutrons and to attain a better understanding of the
mechanisms of neutron damage in graphite. Our sources
of graphite for these studies are experimental grades
fabricated by us (Sects. 11.1 and 11.4) plus experi-
mental and production grades fabricated by commercial

100

manufacturers. Currently, our work is almost com-
pletely limited to experimental grades, since typical
examples of all existing production grades known to
have potential for nuclear application have been ex-
amined. As we acquire various grades, we make initial
limited determinations of their physical properties to
establish reference data. These data are used to deter-
mine if they warrant more extensive evaluation through
additional physical, chemical, and fast-neutron irradia-
tion studies. We have recently received samples of three
new production grades of graphite: grades H395 and
H395X from the Great Lakes Carbon Corporation and
grade A800 from the Stackpole Carbon Company.
None of the three has been optimized to the specific
requirements of molten-salt reactors other than possible
neutron damage resistance. Specifically, the pore en-
trance diameters are much larger than the I-um
maximum required for MSBR core moderator graphite.
They were supplied to us for general evaluation and are
of interest in our studies on the relationship of
microstructure and dimensional stability under irradia-
tion.

The Great Lakes grades H395 and H395X are
advanced grades beyond grade H337, which has good
stability under fast-neutron irradiation.! Grade H337
has a bulk density of 2.00 g/m?®, whereas grades H395
and H395X have densities of 1.797 and 1.875 g/cm®
respectively. The higher bulk density of the H337 is due
to a greater degree of impregnation with liquid hydro-
carbons that have been carbonized and then graph-
itized. This is apparently the reason grade H395X also
has a higher bulk density than grade H395. The
microstructure of grade H395 is fairly uniform, whereas
that for grade H395X varies because the impregnation
fills some zones of the porosity more effectively than
others.

The basic difference in grade H337 and grades H395
and H395X is their optical domain sizes (as revealed in
the examination of their microstructures using a rotat-
able sensitive tint plate in a light microscope). Grade
H337 has a fine matrix in which the optical domains are
small, randomly oriented, and have an appearance
similar to the most dimensionally stable graphite, grade
AXF-5Q,! except that there are larger highly aniso-
tropic filler particles scattered throughout the matrix
which range approximately 5 to 25 times larger than
the general matrix optical domains. Grades H395 and
H395X do not have these larger anisotropic filler
particles. Their optical domains appear to be 1% to 2
times that of grade AXF-5Q, which typically ranges
from 4 to 10 um. Rarely in grade AXF-5Q and only
occasionally in grades H395 and H395X do we see filler
particles with optical domains that are aligned preferen-
tially to a degree sufficient to give the whole filler
particle a preferred orientation.

The cross-sectional views of grade AXF-5Q show the
pore structure to be rough but approximately equiaxed,
while that for grades H337, H395, and H395X is linear
and branching,

Based on their microstructures, we anticipate that
H395 and H395X would have similar dimensional
stability under fast-neutron irradiation and would react
in a way equal to or slightly more dimensionally stable
than grade H337.

Stackpole grade A800 was supplied in connection
with our core moderator studies directed toward
understanding the relationship between dimensional
stability and microstructure under fast-neutron irradia-
tion, hence our interest in this material despite its
density of only 1.642 g/cm?® and its large open pores, It
represents a different method of fabrication and again
has not been optimized toward MSBR requirements. Its
microstructure reveals that it is composed of well-
graphitized filler particles and carbon-black particles on
an estimated volume ratio of 1.5:1. Additional char-
acterization examinations are necessary to determine if
this material warrants inclusion in our neutron-
irradiation studies.

The graphite for constructing equipment for process-
ing the reactor fuel salt has more latitude in physical
properties because the material will not be subjected to
fast-neutron radiation, The primary requirements of the
graphite are that it be (1) chemically compatible with
the lithium-bismuth alloys containing from O up to 3 wt
% (48 at. %) lithium at 650°C, (2) impermeable to the
salts and lithium-bismuth alloys used in the chemical
processing of the salts and their fission products, and
(3) easily fabricated into relatively complex structures.

Two approaches are being considered for making the
graphite impermeable to the salts and the lithium-
bismuth alloys. One is to coat or to seal a polycrystal-
line graphite structure with pyrocarbon or pyrograph-
ite. The other is to adequately reduce the porosity and
pore entrance diameters of the base stock by repeated
impregnations with liquid hydrocarbons that are car-
bonized after each impregnation. These carbonized
impregnations probably would be graphitized to im-
prove chemical inertness in our applications.

The impregnation of the graphite is the approach
favored by the Chemical Technology Division personnel
who are developing the technology for the chemical
processing of the fuel salts. Through their efforts we
have acquired 6-in. cubes of graphite grades PGX and
PGXX from the Carbon Products Division of the Union

101

Carbide Corporation. Grade PGX will be the reference
material that will be upgraded by impregnations. [t is a
base stock material that can be made into large shapes
- for example, a 67-in.-diam, 80-in.-long piece is listed
as a catalog item. Large-sized stock is necessary for one
configuration of the chemical processing facilities under
consideration by the Chemical Technology Division.
Our samples of the base stock, grade PGX, and the
impregnated base stock, grade PGXX, have bulk den-
sities of 1.791 and 1.851 g/ecm® respectively. These
suggest that grade PGX may be too porous to be used
without modification. Additional tests are necessary
and are in progress to determine if the pore entrance
diameters of grade PGXX are small enough to be
impermeable to the media cited above.

We have acquired from the Stackpole Carbon Com-
pany additional quantities of grade 2020, which has an
average bulk density of 1.745 g/cm®. This is good
density control for a production-grade graphite of this
type. This grade of graphite typically has almost all of
its accessible porosity with pore entrance diameters
within 1 to 3.5 um and none more than 3.5 um. This
restrictéd range of pore entrance diameters makes this
grade of graphite of direct potential for low-pressure
regions (<25 psia) in the systems for the chemical
processing of the fluoride fuels, and it could be
upgraded relatively easily by impregnation or with
pyrocarbon coating for use in regions at higher pres-
sures. At this time we plan to use it without modifica-
tion in compatibility studies with the lithium-bismuth
alloys at 700°C.

11.3 CHARACTERIZATION OF ORNL GRAPHITES
O.B.Cavin C. R. Kennedy

Our program to develop graphite with improved
irradiation resistance is directed toward obtaining a
product having a virtual monolithic structure. That is,
we would like for all of the carbon in the body, both
filler and binder, to have the same structure. This would
minimize the interparticle shearing which must be
accommodated during the irradiation-induced ani-
sotropic dimensional changes. There are numerous ways
of evaluating these bodies to determine the extent to
which we have accomplished this, each sensitive to one
or more structural parameters.

Several samples were fabricated using modified Ro-
binson coke with and without natural flake additions
and bindered with furan--petroleum-pitch mixtures.
These graphites, listed in Table 11.2, were all molded at
90°C and air cured at 150°C for 16 hr. The physical
Table 11.2. Physical and mechanical properties of ORNL graphites

| Binder Bulk Electrical ‘Modulus (psi) Sonic E;Stm:‘:s Crystallite
Filler Content density resistivity , Sonic Sonic Poisson’s layer‘;l:nes size
(pph) Type (g/cm®)  (uohm-cma)  Rupture  Young’s longitudinal shear ratio (A) (A)
X 10° X 10° X 10°

Modified Robinson 10 QX229 1.55 2520 2,860 0.80 0.90 0.40 0.06 6.810 244

Modified Robinson 15 QX229 1.60 2360 4,810 1.13 1.14 0.49 0.13 6.800 257

Modified Robinson 20 QX229 1.70 1860 7,590 1.62 1.39 0.60 0.17 6.806 257

Modified Robinson 25 QX229 1.77 1790 10,370 1.84 1.71 0.74 0.15 6.809 233

Modified Robinson 30 QX239 1.80 13,840 2.03 6.820 223

Modified Robinson 25 QX234 1.73 1770 6.806 213
plus 10% natural flake

Modified Robinson 25 QX234 1.78 1630 6.790 258
plus 15% natural flake

Modified Robinson 25 QX234 1.78 1680 6.798 169

plus 20% natural flake

¢o1
and mechanical properties of these bodies are also given
for comparison. It is immediately obvious that in-
creasing the binder content results in a body of higher
bulk density, strength, and elastic constants, but de-
creased resistivity. The carbon yield of the binder is
about 50%, so the maximum carbon contribution from
the binder to the bodies would be approximately 15%.
There is a 16% increase in density in going from 10 to
30 parts per hundred of binder. This increase in density
is an expected result of binder optimization, a complex
procedure of creating more efficient bindering to
produce a more uniform shrinkage. The significant
increases in mechanical properties reflect the reduction
in the critical defect size by the optimization of binder
to reduce porosity.

The electrical conductivity increases significantly with
increased density. This is in contrast to the x-ray
parameters, which show a decrease in the average
crystallographic perfection with increased binder con-
tent. The same conflict is observed by comparing the
graphites with natural flake additions. The electrical
conductivity of the samples is virtually identical to the
all-Robinson coke sample with the same density and
independent of the average crystallographic perfection
as indicated by the x-ray parameters. This emphasizes
the sensitivity of electrical conductivity, not only to
crystallographic perfection, but to the effective cross-
sectional areas and structural morphology. 1t is shown
in Table 11.2 that the conductivity is a very strong
function of the density or effective cross-sectional area.
However, the structural morphology is also significant
in that the location of the binder around each filler
particle is a limiting factor in the increase of the
conductivity. The cured binder is certainly less graphitic
than the filler materials, particularly the natural flake
filler, as shown by the x-ray parameter., Thus, the
binder will have a much higher resistivity and act as an
insulator in the structure. A further increase of the
conductivity can therefore only come from an improve-
‘ment in the crystallinity of the binder.

11.4 GRAPHITE FABRICATION?
C. R. Kennedy

Graphites for improved resistance to irradiation must
have a binder system that produces an interparticle
boundary with a very high mechanical integrity. We
have concentrated our efforts in this program to

2. This work was supported in part by the Naval Ordnance
Laboratory.

103

understanding the binder characteristics in order that
representative graphites of improved structures can be
made and evaluated. The binders studied include
coal-tar pitches, petroleum pitches, and furan—
petroleum-pitch mixtures, both with and without nitro-
benzene as a diluent and plasticizer.

The initial graphite fabrication of the green cokes
produced reasonably dense structures; however, these
hard dense cokes have excessive springback after mold-
ing, causing high localized shrinkage during carboniza-
tion. These effects produced the structure shown in Fig.
11.8, which is fairly representative of our first-
generation graphites. This graphite has been irradiated,
and the results are given in Figs. 11.1—-11.3. The
reasonable stability of these graphites is very promising
and suggests that the bindering of the individual filler
particles is fairly good.

To eliminate the linear defects shown in Fig. 11.8, we
have, through a coke modification process, produced
graphites with the pore morphology as shown in Fig.
11.9. The mechanical integrity of this structure is very
good; however, irradiation results are as yet very
preliminary, as shown in Fig. 11.4. We have produced
this type of structure using all the binders listed above.

The strength and modulus relationships of the various
graphites are compared in Fig. 11.10, demonstrating
again the similarity in pore morphology of our experi-
mental materials and the Poco grade AXF-5Q, both of
which are superior to the specialty grade ATIJ-S
graphite. This is a reasonably conclusive demonstration
that the pore morphology and structural integrity can
be maintained for a wide variety of raw materials
through this modification process. The electrical and
thermal conductivities, however, are found to be very
sensitive to the methods of fabrication as well as raw
materials. Controlling the conductivity of these graph-
ites is, in fact, one of the major problems in fabrication.

We are concerned about the thermal conductivity, not
only because of the reduction of thermal gradients in
the reactor core, but because it is also a sensitive
measure of the binder carbon characteristics. We desire
a monolithic structure in which all of the carbon has
the same structure and behaves uniformly under irradia-
tion. This would in turn reduce the strain concentra-
tions in the binder carbon between the filler particles.
Since the binder coats all the filler particles, the
conductivity of this layer can dominate the overall
conductivity of the graphite body. This is very evident
in the sharp effect of air curing the binder upon the
electrical resistivity, as shown in Fig. 11.11. Differences
in the filler materials are completely masked by the
curing, as is shown by the similarity of the semiacicular
Fig. 11.8. Representative pore texture of firstgeneration ORNL experimental graphites.

-,,.“, “.Y-15647

5

Fig. 11.9. Representative pore texture of secondgeneration ORNL experimental graphites.

]

0,035 INCHES
To 100X

0.035 INCHES
T toox

Tooi0 .

[CLECT

=

filler coke with and without the addition of Robinson
coke fines. Molding at too high a temperature also
appears to have a deleterious effect on resistivity. We
have now observed that the petroleum pitch 240 is even
sensitive to the air retained in the packed bed during
the bake cycle. Removing the air from the baking

ORNL-DWG 72-7553

(X103) T T
MAIN o o
FILLER BINDER
e SA PITCH Loy
a A Qx o o
A A PITCH H e
10— oRoOB Qx .
@ ® ROB PITCH LY
3 ® : ‘.:DD
E 8 0 s Ao
o o 04
& oy ®
o u]
[ 2%
w 6 [s] [AY
O o | a%0
z b5,
M 0‘ [¢]n] 0\ }
4 © e ATJS
0
L
le]}
2 (e}
(]
° 6
o] 0.5 {.C 1.5 2.0 {(x10

MODULUS OF ELASTICITY (psi)

Fig. 11.10. Strength as a function of Young’s modulus for
various green-coke-based experimental graphites.

ORNL-DWG 72-7555

(10" | ! 1 | |
® ROBINSON FILLER ADDED
o RoBINS } CURED 250°C (AIR)
4 MOLDED 150°C NO CURE
= 5| D ROBINSON FILLER ADDED-NO CURE |
§ & MOLDED UNDER 100°C - NO CURE
E
o
§ c\c
€ 4
T N
- \
>
E .
w
;] 3 \\
i
L )
2 \
o
E 2 \\ »
A
O A e 0O
*\‘afl\a;‘
{
éj .
= i
I
0

1.6 1.7 1.8
DENSITY (g/cm>) .

2.0

Fig. 11.11. Effect of air cure on semiacicular coke graphites.

105

)

fumace and backfilling with argon is now a standard
practice to eliminate this effect. Therefore, it becomes
important to mold the graphites at temperatures as low
as possible and to subsequently protect the molded
bodies with a nonoxidizing environment.

The relation between density and electrical resistivity
is also obvious in Fig. 11.11, and density is indeed a
major consideration in the strength of a graphite. We
have found that by eliminating the oxidation of the
very aromatic petroleum pitch 240 the coke yield
dropped from 50% down to 30 to 35%. In terms of
texture, reasonably good graphites have been fabricated
with the 240 pitch, but it is exceedingly difficult to
obtain a high-density graphite with this low yield. The
carbon yield can be increased by the addition of furans
to the pitch or by the addition of Thermax or fines to
the filler. Both of these additions, however, do increase
the electrical resistance.

It has long been a practice to add fines to fillers to act
as binder “extenders.” This practice increases the
density of the product by increasing the binder yield
and improves the packing of the filler particles. Since
the fines are more intimately associated with the
binder, the influence on the properties is again to those
that are more sensitive to binder effects. Thermax or
other carbon blacks have long been used for this
purpose, but even small additions seriously decrease the
conductivity of the graphite. We have been substituting
fines of Robinson, acicular, and semiacicular cokes
made by grinding in a fluid mill. We have added the
superfines of these cokes still in the green state or
prefired to 600°C. The addition of the 600°C calcined
fines was very deleterious. Not only was the conduc-
tivity seriously decreased, but the mechanical properties
of these graphites were reduced to more like conven-
tional grades. The green Robinson and acicular coke
fines appear to be more effective in obtaining high
densities than the semiacicular coke. However, all the
graphites with the fines additions have a lower conduc-
tivity, particularly with Robinson fines. The use of the
Robinson fines also has the ability to make the graphite
more isotropic. All semiacicular coke graphites fabri-
cated to obtain a Bacon anisotropy factor of 1.1 can be
made isotropic (BAF of 1.0) by a 30 to 40% Robinson
fines addition.

The addition of furans to the petroleum pitch 240
does make the binder more controllable, but some
slight increase in electrical resistivity does occur. The
purity and controllability of the 240 petroleum pitch
and furan mixtures make these binders otherwise look
attractive. We continually find for these mixtures a
wider latitude in the binder content required to
106

optimize a particular filler mixture. This may result
from a more gradual weight loss during the carboniza-
tion cycle than the more abrupt losses from petroleum
pitches and some of the coal-tar pitches. Also, these
binders are available with very low viscosities at room
temperature, allowing mixing and fabrication tempera-
tures to be reduced. The use of diluents such as
nitrobenzene has also been shown to be very control-
lable and useful in processing the pitch-furan mixtures.
In general, these binders may offer considerably more
control and uniformity than the variable coal-tar
pitches.

One additional method of increasing yield and quality
is the use of pressure baking. We have fabricated a
group of samples using both the 240 pitch and 240
furan mixture as binders. These have been pressure
baked to 650°C, using the facilities at the Y-12 Plant,
and then conventionally baked to 1000°C. The evalua-

tion is not complete, but the apparent yield of the
binders was increased by some 15 to 20%, and the
overall quality of the blocks looked very good.

In summary, the graphite development program is
continuing toward an optimization of the monolithic
structure. Most of our attention has been devoted to
studies to ensure the coherency of the binder carbon
and to match its structure to that of the filler particles.

11.5 THERMAL PROPERTY TESTING

J.P.Moore T.G.Kollie
D. L. McElroy

Thermal conductivity (M) and electrical resistivity (p)
measurements have been completed on the AXF-5Q
graphite which was irradiated to 4.2 X 10*' neutrons/
em? (E > 50 keV) at approximately 650°C. As shown
in Fig. 11.12 the thermal conductivity of this specimen

ORNL-DWG 72-753A

.2 <

AN

AN

0.8

UNIRRADIATED
AXF -5Q

0.6
=====4.2X10

THERMAL CONDUCTIVITY (W/cm-deg)

IRRADIATED AXF-5Q

L — — 2.6 X102 neutrons/cm2 AT 550°C
" neutrons/cm2 AT 650°C

—.— 7x102! neutrons./cm2 AT 650°C

~C

BAKER

0.4

o

.’. . IR N ¥

— e [T e

-~ ——F i —e |

_— ¢ esees——— e

0.2 -

300 500

700 900

TEMPERATURE (K)

Fig. 11.12, Effects of neutron irradiation on the thermal conductivity of graphites.
107

Table 11.3. Thermal conductivity of irradiated graphites from ORNL and Baker®

A(Wem !deg™h)

AXF-5Q
Temperature After After Baker,
CK) Unisradi 2.6 % 102! 4.2 x 10%! 7 X 102!
nirradiated 2 2 2
neutrons/cm neutrons/cm neutrons/cm
at §50°C at 650°C at 650°C?
300 1.31 0.32 0.30 0.19
400 1.17 0.34 0.31 0.24
500 1.07 0.345 0.32 0.25
600 0.97 0.345 0.32 0.25
700 0.88 0.34 0.32 0.25
800 0.80 0.33 0.32
900 0.73 0.31

2From D. E. Baker, J. Nucl. Mater. 12(1), 120 (1964). The thermal conductivity of the graphite

used by Baker was about 1.1 Wcm ™! deg™! at 400°K before irradiation.

was a few percent below that of the specimen irradiated
at 550°C to a fluence of 2.6 X 10*' neutrons/cm?.
Also shown in Fig. 11.12 are representative data for
unirradiated AXF-5Q from our previous results and
data from Baker® on a needle coke graphite which had
been irradiated at 650°C to a fluence of 7 X 10%!
neutrons/cm?. Baker’s data are in good agreement with
ours, and we have listed the various results in Table
11.3.

The specimen of H-337 graphite that was irradiated at
550°C to a fluence of 2.6 X 10?! neutrons/cm?® has
been assembled, and measurements of o and A are under

holder tube assembly were refabricated to tighter
tolerances. This was done to reduce movements in the
assembly due to vibrations and differential thermal
expansions. Repeat measurements with the NBS quartz
sample yielded reproducibilities of *8 X 107® cm,
which is within the #25 X 107® ¢m accuracy of the
electronic micrometer. After calibrating with the NBS
copper gage, measurements are planned on AXF-5Q and
H-337 graphites in the pre- and postirradiated con-
ditions.

11.6 REDUCTION OF HELIUM PERMEABILITY

way. During disassembly of the irradiation capsule, the OF GRAPHITE BY PYROLYTIC
H-337 graphite that had been irradiated at 750°C was CARBON SEALING
inadvertently broken. This prevents the planned A C B.Pollock W.H.Cook

measurements on this sample at high temperatures, but
we have fabricated holders which will enable us to
obtain some A data from the sample near room
temperature in another apparatus.

Thermal expansion is related to the structure of
graphite, and measurements of this parameter may be
useful in understanding irradiation damage in graphite.
Thus we are setting up the equipment required to make
these measurements accurately, and we plan to make
thermal expansion measurements on the same samples
being used in the thermal conductivity studies. At-
tempts to calibrate the quartz dilatometer between 25
and 725°C using NBS-certified quartz and copper sam-
ples yielded a large scatter (¥5 X 10™% cm) in the
length-change measurements. Since repeatability is es-
sential to the accuracy of thermal expansion measure-
ments using this technique, the quartz pushrod and

3. D. E. Baker, J. Nucl. Mater. 12(1), 120 (1964).

The breeding performance of molten-salt reactors is
significantly affected by the extent to which '3*3Xe
accumulates in the core graphite. A procedure for
stripping the gaseous fission products with a bubble
purge system has been developed, but its effectiveness is
dependent upon the surface condition of the core
graphite. Therefore, it is desirable to seal the surface of
the graphite to reduce '>5Xe permeation, and the
preferred material is pyrolytic carbon. The first tech-
nique used for this purpose was a carbon impregnation
process in which surface pores were plugged with

pyrolytic carbon. It was found that the helium per-

meability of this material increased rather rapidly with
irradiation, and alternate schemes for sealing were
investigated.

Previous investigations by Hewette, among others,
had shown that pyrolytic carbon coatings derived from
propene (CH3;CH:CH,) at 1300°C were extremely
stable under the influence of fast neutrons at 715°C
(ref. 4). The dimensional changes of dense low-
temperature coatings derived from propene are sum-
marized in Fig. 11.13. The extremely small change in
the parallel direction (approximately 2%) is significant
in that it matches very closely the observed changes in
Poco graphite under the same conditions of fluence and
temperature. The coatings in this study had densities
greater than 1.99 g/cm® and Bacon anisotropy factors
of less than 1.1. Furthermore, it was determined that
these classes of carbonous materials were able to relieve
internal stresses by creep. An examination of the
known properties of the candidate graphites and the
low-temperature coatings led to the belief that the two
materials would be compatible under molten-salt reac-
tor conditions of fluence and temperature.

A furnace designed for coating small graphite samples
with pyrolytic carbon was built, and coating techniques
were developed.® The excellent behavior of low-
temperature pyrocarbon can be attributed to several
factors that include a Bacon anisotropy factor less than
1.1 in the structure and a density of 2.0 g/fcm® or
greater. Therefore, our coating process parameters were
directed toward achieving these properties. Previous
studies indicated that deposition temperatures between
1100 and 1400°C were necessary to achieve the high
- density and required isotropy with a propene source.®
We found that coatings deposited in the temperature
range 1100 to 1300°C with a carbon deposition rate of
1.3 to 14.0 um/min had a density of 2.0 gfem? or
greater, But, in order to achieve the isotropic coating, it
was necessary to simulate the action of a fluidized-bed
furnace. This was accomplished by levitating a small
quantity of carbon beads around the sample, which was
fixed in place by a carbon rod. All coatings were heat
treated to at least 1800°C before being removed from
the furnace.

Coated samples were prepared by coating at 1100,
1150, 1200, 1250, and 1300°C. The coatings were
deposited on a substrate of grade AXF-5Q graphite
cylinders, nominally 0.40 in. OD, 0.14 in: ID, and 0.50
in. long.

108,

Part of our evaluation of the coatings is to examine a -

selected quantity of the samples in detail nondestruc-

4. D. M. Hewette and C. R. Kennedy, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp.
215-18.

5. C. B. Pollock, MSR Program Semiannu. Progr. Rep. Feb.
29 1972, ORNL-4782, pp. 151-55.

- 6. R. L. Beatty, J. L. Scott, and D. V. Kiplinger, Minimizing
Thermal Effects in Fluidized-Bed Deposition of Dense, Iso-
tropic Pyrolytic Carbon, ORNL-4531 (April 1970).

ORNL-DWG 70—4918R
FLUENCE {(neutrons/cm?> 50 keV)

o} 5 10 15 x10%h

* |' 1 1 1
~ STRUCTURE PARALLEL PERPENDICULAR
B 30 NO. DIRECTION DIRECTION
4 15 o .
D‘I,_ 16 ° .
|U_) 17 o .

20
(L)
=
b
Q
O 10
;*T . |' L =
q‘l _.——-—-—"::._—_0—- =
" — .
e = ]
= =1 : g
= e —— b
< ’ _____1_______.
o —1{0
o

-20

0 2 4 6 8 10 12 14 (x10%)

FLUENCE (neutrons/cm2 > O.18 MeV)

Fig. 11.13. Effect of fast-neutron exposure at 715°C on the
dimensional changes of propene-derived pyrolytic carbons.

tively with the scanning electron microscope (SEM) and
destructively by conventional bright-field metallo-
graphic techniques. These are complementary types of
examinations which we use to characterize the as-
deposited coatings from the bases of the effects of the
deposition parameters plus the nature and quality of
the coatings.

A summary of the SEM observations on this series of
coated samples is as follows:

1. The rounding of the sharp edges of the substrate
samples appears to have eliminated the cracking and
spalling of the pyrocarbon coatings which sometimes

occurred at these sharp edges.

The quality of the pyrolytic coating appears to be
improved by decreasing the roughness in the
machined surfaces of the substrate graphite.

The pyrolytic carbons deposited at the lower tem-
peratures appear to be more continuous and have
less bridging of pyrocarbon particles than those
deposited at higher temperatures.

. The basic pyrolytic deposition particles start as
oblate spheroids at 1100°C, become almost flat
platelets at 1200°C, and finally become more
spheroid-like at 1300°C.

. At 1100 and 1150°C the longer coating period
appears to improve the overall physical appearance
of the coatings. (Lack of time and samples prevented
such evaluation for the other deposition tempera-
tures.)
PHOTO 0087 -73

10 um

1200°C
DEPOSITION TEMPERATURE

1100°C

Fig. 11.14. SEM photomicrographs showing typical microstructural topography of as-deposited and heat-treated pyrolytic carbon coatings derived from propene and

deposited on grade AXF-5Q graphite at the temperatures indicated. S000X.

601

6. The physical nature of a coating at a particular
deposition temperature is sensitive to the geometry
and orientation of the sample in the coating furnace.
For example, the coating on the cylindrical surface
of a sample is not always the same as that on the
rounded edges of the sample.

Figure 11.14 illustrates the typical range of micro-
structures observed in this series. The pyrocarbon
deposited at 1100°C (Fig. 11.14a) appeared to have a
fine-grained component that had the appearance of a
“solidified silt” cementing around the larger pyro-
carbon particles and the grains of the substrate graphite.
The quantity of the silt portion varied with coating
time at 1100°C. Figure 11.14b illustrates the change to
smaller and more spherical particles at a deposition
temperature of 1200°C. Note that there is more
porosity due to particle bridging at the higher deposi-
tion temperatures.

Our bright-field metallographic examinations of trans-
verse sections through the coatings indicate that they
are dense, bonded to the substrate graphite, and the
degree of anisotropy of the coatings is estimated to be
less than 1.3 in the extreme case by examination under
polarized light. To acquire more quantitative data on
the isotropy, samples have been prepared for measure-
ment by the OPTAF technique.”

During this report period we have prepared an
extensive array of coated samples for the first irradi-
ation experiments that cover the full range of coating
conditions, all of which were sealed to a helium
permeability of less than 107® cm?/sec. Table 11.4
describes the samples being irradiated in HFIR capsule
HTM-20, which was inserted in August 1972, Plans are
to remove this capsule in November 1972, after it has
accumulated a total peak fluence of 1 X 10%? neu-
trons/cm? at 700°C. The capsule consists of a stack of
samples threaded on a graphite spine assembly which is
enclosed within an aluminum can, Figure 11.15 shows
some of the samples selected for the first irradiation
experiment.

We have used the SEM to selectively examine about
25% of the pyrolytic-carbon-coated samples before
installation in the HFIR irradiation capsule HTM-20.
We have made SEM photomicrographs as permanent
records on the outside and on one of three rounded
corners of each of the selected coated samples. Un-
doubtedly, fast-neutron damage will not manifest itself

7. H. B. Gruebmeier and G. P. Schneidler, An Optical Method
for the Determination of the Local Anisotropy of Pyrolytic
Carbon Layers and Graphite, ORNL-Tr-2127.

110

exactly in the small regions photographed, but the fact
that the sample being irradiated is its own control is
superior to using a similar sample as a control. We plan
to monitor these selected samples with the SEM as they
acquire increments of fluence. We hope to be able to
see how fast-neutron damage starts and progresses with
increasing fluence and how the damage varies with
structures. We shall attempt to correlate these observa-
tions with other data such as dimensional changes and
gas permeability. The goal of this program is to
determine the deposition parameters that will produce
the most reliable types of coating under fast-neutron
irradiation.

Table 11.4. Coated graphite samples irradiated
in the HFIR capsule HTM-20

Coatin Effective
Sample temoaera tire helium Coating thickness
number l()"C) admittance (in.)
(cm?/sec)

MS-82 1100 1.4x 10710 0.0008
A

MS-84 1100 1.2x 10710 0.0015
B 0.0017

MS-85 1100 1.1x107'° 0.0013
B

MS-86 1100 1.1 x 1071¢ 0.0010
A 0.0012

MS-87 1100 2.2x 10710 0.0014
A

MS-88 1100 1.1x 1071 0.0010
A 0.0012

MS-89 1150 9.0x 107! 0.0022
B

MS-91 1150 6.4 x 107! 0.0028
B 0.0032

MS-92 1150 6.8x 107! 0.0012
A 0.0016

MS-93 1150 5.9x 107! 0.0026
B

MS-94 1150 4.1 x 107! 0.0020
B

MS-95 1150 3.8x 107! 0.0020
A

MS-97 1150 38x 10" 0.0021
A 0.0024

MS-98 1200 59x 107! 0.0023
B ' 0.0026

MS-99 1200 72x 1071 0.0020
A

MS-99 1200 52x 10711 0.0024
B .

MS-101 1300 1.6 X 107° 0.0053
A 0.0057

MS-103 1300 1.1x 1078 0.0029
A 0.0031

111

PHOTO 2634-72A

900000

MS-119-A MS-118-A MS-106-A MS-103-A MS-101-A MS-99-8 MS-99-A
MS-129-B MS-129-A MS-128-A MS-127-A MS-126-A MS-125-B  MS-123-A
o 0.5 1.0 1.5 2.0
| CEE ] =t a7
INCHES

Fig. 11.15. Graphite samples coated with pyrolytic carbon for HTM-20.

11.7 IRRADIATION OF PYROCARBONS

D. M. Hewette I C. R. Kennedy

The ability of pyrocarbon coatings to retain their low
permeability depends upon how their dimensional
changes match those of the substrate graphite. We
recognized early from our results® and those of Bokros®
that the growth rates of the individual pyrocarbon
crystallites are exceedingly high; for the coatings to be
dimensionally stable, the pyrocarbon must be almost
perfectly isotropic. The results of Beatty' ® showed that
it would be impossible to obtain an isotropic high-
density carbon from methane (CHg4). It is possible,
however, to obtain the necessary high-density isotropic
pyrocarbon from propene (CH3CH:CH;) deposited at
relatively low temperatures (around 1100 to 1300°C).
To demonstrate the dimensional stability of these
pyrocarbons, several propene- and methane-derived
pyrocarbons were irradiated in the HFIR to fluences up
10.3.3 X 10*? neutrons/cm? (£> 50 keV).

The results of this irradiation experiment are shown
in Figs. 11.16 and 11.17 for both methane- and
propene-derived pyrocarbons. The materials irradiated
are listed in Table 11.5. The methane-derived materials

8. D. M. Hewette Il and C. R. Kennedy, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, pp.
215-18.

9. 1. C. Bokros and R. J. Price, Carbon 5(3), 301 (1967).

10. R. L. Beatty, F. L. Carlsen, Jr., and J. L. Cook, Nucl.
Appl. 1,560 (1965).

were very unstable, with the most isotropic coatings
shrinking over 10%. The propene coatings were fairly
isotropic and dimensionally stable. The heat-treated
YZ-131 had the greatest stability, with little effect of
heat treatment at temperatures from 1900 to 2500°C.
The slightly lower density YZ-196 had the expected
high initial densification of the as-deposited material.
However, there was one YZ-196 sample heat treated to
1900°C that expanded when irradiated, which suggests
that the structure was not fully stabilized. It appears
that the preferred coating would have a density greater
than 2.03 g/em? and would be heat treated to 2000°C.
It is doubtful that a strain differential between the
coating and substrate of much greater than 2 to 3%
could be sustained without cracking or spalling.

11.8 EXAMINATION OF UNIRRADIATED AND
IRRADIATED PYROCARBON STRIPS WITH THE
SCANNING ELECTRON MICROSCOPE (SEM)

W. H. Cook

The results (previously described' ') of the examina-
tion of pyrocarbon coatings on graphite prompted us to
make similar examinations on pyrocarbon strips being
studied by Kennedy and Hewette.® These materials
were derived from several hydrocarbon sources under
different deposition and heat-treating conditions. We

11. W. H. Cook, MSR Program Semiannu. Progr. Rep. Feb
29,1972, ORNL-4782, pp. 155-59.
LENGTH CHANGE [100 In (1 +AL/L )]

LENGTH CHANGE [1001n (1+AL/Ly)]

112

ORNL-0WG 73-892

6 T T T T
YZ-196 YZ-134
s AS DEPOQSITED (t250°C) o 13900°C
4 |- @ 1900°C & 2200°C .
A 2200°C o 2500°C
2 //
®
A
O A / /
2 }:_::
—— [ ]

N %_——"———-—" # \%}O\\ R
. o
= -.___.--.

-4 — n

—--'l-—___
\ =

-6 ' 21

0 5 10 15 20 25 30 (X10¢)
FLUENCE (neutrons/cm?, £> 50 keV)
Fig. 11.16. In-plane dimensional changes in propene-derived pyrocarbons.
ORNL-DWG 73-893
O _g
-10 \"‘—\‘B—-—-____—G
o)
-20 Dfl N \.
TS
B \A
-30 \\ £
-40
g
o YZ-166 E
-50 ——— A YZ-1914 n| N,
o YZ-192
0 YZ-193
-60 \
¢
-80
0 5 10 15 20 25 30 (x102")

FLUENCE (neutrons/cm2) (£>50 keV)

Fig. 11.17. In-plane dimensional changes in methane-derived pyrocarbons.
113

Table 11.5. Pyrocarbons irradiated in the HFIR

" Liquid
Deposition . .
Structure . immersion
Material Source temperature Heat treatment .
number ©0) density
' (g/cm?)
18 YZ-166 Methane 1600 As deposited 1.62
12 YZ-191 Methane 2000 As deposited 1.93
13 YZ-192 Methane 2000 As deposited 2.05
14 YZ-193 Methane 2000 As deposited 2.08
6 YZ-131 Propene 1250 1900°C 2.08
7 YZ-131 Propene 1250 2200°C 2.09
8 YZ-131 Propene 1250 2500°C 2.08
15 YZ-196 Propene 1250 As deposited 2.01
16 YZ-196 Propene 1250 1900°C 2.03
17 YZ-196 Propene 1250 2200°C 2.03

have examined several controls and irradiated specimens
derived from methane and propene. Some of the
irradiated specimens examined had accumulated
fluences as high as 3.2 X 10*? neutrons/cm?® (F > 50
keV) at 715°C.-

For the unirradiated controls, those deposited from
methane at 2000°C had virtually no porosity, and those
deposited from propene at 1250°C had a signifcant
amount of porosity. Under the deposition conditions
cited, the pyrocarbon derived from methane was
anisotropic, and that derived from propene was iso-
tropic. The rougher starting surfaces of the control for
the methane-derived pyrocarbon (Fig. 11.18¢) vs that
derived from propene (Fig. 11.19¢) can be explained by
the fact that the methane-derived material is more
anisotropic and more graphitic. This caused it to cleave
and shear rather than polish, as the pyrocarbon derived
from propene did.

The results of the SEM examinations of these
materials after irradiation show graphically the changes
created by the fast-neutron damage and the differences
in the resistance to the damage by the different
materials. In the unirradiated condition, the polished
specimen shown in Fig. 11.18a is only 0.0032 in. thick
in the direction parallel with the ¢ axis of the graphite
crystallites of the pyrocarbon. Under a fluence of 3.2 X
10?? neutrons/cm?® (E > 50 keV) accumulated at
715°C, this thickness increased by a maximum of more
than 500%, while the width and length decreased by 50
and 53% respectively.!? This observation clearly illus-
trates the ability of graphite to make these large
dimensional changes plastically without cracking.

12. Personal communication from C. R. Kennedy and H.

Keating.

In contrast, the dimensions of the isotropic pyro-
carbon derived from propene decreased approximately
5% under the same irradiation conditions. A com-
parison of Figs. 11.19d and 11.19¢ shows that the
original surface roughened during irradiation.

11.9 TEXTURE DETERMINATIONS
E.S.Bomar 0. B. Cavin

We described previously an optical technique used for
determining the anisotropy of graphites of interest to
the Gas-Cooled Reactor Program.!?® This technique is
being adapted to evaluate the anisotropy of surface
sealants placed on graphites for MSR applications. The
optical technique is suitable for highly localized anisot-
ropy determinations, while x-ray diffraction can be
used to measure the average bulk properties. In order to
correlate the values obtained from the optical and x-ray
techniques, samples are being prepared which are
suitable for either technique.

We continued to use the sphere technique described
previously to arrive at the average anisotropy of bulk
graphites.'* The spherical sample, however, is not
amenable to texture determinations of a surface coat-
ing. Therefore, we are using thin flat samples of the
coating extracted from the surface of Poco AXM
substrate. In the past it has been necessary to have
samples 10 to 20 mils thick, most of which was
substrate, mounted onto a glass backing plate for
holding during the spectrometer run. The glass con-

13. 0. B. Cavin and D, M, Hewette II, MSR Program
Semiannu. Progr. Rep. Feb. 29, 1972, ORNL-4782, pp.
149-50.

14. O. B. Cavin, MSR Program Semiannu. Progr. Rep. Aug.
31, 1969, ORNL-4449,p. 172.
UNIRRADIATED CONTROL

114

f-20c

al
8
z
5
8
g
2
]

700

f-a0

SNONOIN 091
T
X008

120

f-1a0

PHOTO 0088-73

3.2 x 1022 [n /cm? (£ >50 keV)] AT 715°C

Fig. 11.18. Pyrocarbon strip derived from methane at 2000°C. (2, ¢) Unirradiated control and (b, d) after an accumulation of 3.2

X 1022 neutrons/cm? (E > 50 keV) at 715°C.

115

PHOTO 0089-73

f-200

SNOHDIW 008

{600

700

|
3
H
3
2
2
H
%

UNIRRADIATED CONTROL 3.2 x 1022 [n /em? (£>50 keV)] AT 745°C

Fig. 11.19. Pyrocarbon strip derived from propene at 1250°C. (a, ¢) Unirradiated control and (b, ) after an accumulation of 3.2
X 1022 neutrons/cm? (E > 50 keV) at 715°C

tributes to the background scatter, particularly at the
low scattering angles near the basal plane peak. We are
now using silicon single-crystal wafers for mounting
plates. The crystal is cut so that no crystallographic
planes which diffract the copper Ka (A = 1.5418 &) x
rays are parallel with the surface on which the sample is
mounted. This gives an extremely low background
count which makes the weak peaks easier to detect. As
a result of this low background, we can reduce the
required thickness of the sample. Now, instead of the
usual 10 to 20 mils, much of which was substrate,
samples 2 to 3 mils thick are used from which
essentially all the substrate has been removed. Hence, it
is possible to determine the x-ray anisotropy values of
graphite coatings independent of the substrate by using
the spectrometer and counter technique. We prefer the
counter over the film technique used by others' 5! ¢
because an intermediate step of film developing and
recorded density measuring is eliminated. In the
counter technique, the intensity values are recorded
directly on a computer-compatible paper tape, which
greatly reduces the man-hours required for arriving at
an anisotropy value.

Samples from which the x-ray data have been
obtained will then be studied optically to arrive at an
optical anisotropy factor (OPTAF) value,

We have done additional work with the Leitz micro-
scope and photometer to improve the quality of the
OPTAF measurements. The information obtained led to

15. G. E. Bacon, J. Appl. Chem. 6,477 (1956).
16. R. J. Price and J. C. Bokros, J. Appl. Phys. 36, 1897
(1965).

116

several mechanical alterations to the equipment, de-
fined operational limits for light intensity and voltage
applied to the photomultiplier tube, and helped in the
selection of an operational procedure to give repro-
ducible results. : .

After examining the elements in the optical path of
the microscope and after a discussion with Leitz, we
removed a plain-glass reflector and a prismatic diffuser
plate to avoid an error in the apparent intensity of
elliptically polarized light reflected from pyrolytic
carbon samples.

We tested the performance of the photomultiplier
tube and the associated electronic components by
measuring the OPTAF of a graphite single crystal over a
range of incident light intensities and voltages applied
to the photomultiplier. The resulting OPTAF was not
constant, although the values formed an irregular
plateau for tube voltages ranging from 830 to 1120 V.
Scatter in the data became very large under conditions
requiring a correction for dark current: namely, very
low light levels or low voltage to the tube. We found a
marked improvement in the consistency of the OPTAF
for the single-crystal graphite sample when a 10-in.
recorder was substituted for the 3'%-in. meter on the
picoammeter to read the current from the photomulti-
plier tube. The improvement was due in part to the
increased precision of the larger scale, but, in addition,
a comparison with the picoammeter results showed a
bias in the latter instrument.

Pyrolytic carbon standards are being prepared by C.
B. Pollock to permit correlation between our particular
OPTAF equipment and the x-ray-determined Bacon
anisotropy factor.
12.

117

Hastelloy N

H. E. McCoy

Prior studies have shown that a modified alloy of
Hastelloy N containing about 2% Ti will likely have
adequate resistance to embrittlement by neutron irra-
diation. Several small commercial heats of this composi-
tion have been obtained for evaluation. Samples of
these alloys and standard Hastelloy N were irradiated to
a high thermal-neutron fluence in the HFIR to evaluate
the effects of high helium concentrations on the
mechanical properties. Corrosion experiments involving
Hastelloy N and two austenitic stainless steels exposed
to fuel and coolant salts are in progress. The compati-
bility of Hastelloy N with steam for potential use in
steam generators is being evaluated in a facility at
TVA’s Bull Run Steam Plant.

12.1 MECHANICAL PROPERTIES OF SEVERAL
COMMERCIAL HEATS OF MODIFIED
HASTELLOY N

H. E. McCoy  B. McNabb

Five small commercial melts have been tested during
this reporting period. All alloys were received in the
form of Y“-in-thick plates. The vendor’s chemical
analyses are given in Table 12.1. The first two alloys
were made by the same vendor, both with nominal
additions of 2% titanium but utilizing two different
melting procedures. A 120-1b melt was made by vacuum
induction melting (VIM), and two electrodes were cast.
One electrode was remelted by consumable electrode
remelting (CEVM) under vacuum, and the other elec-
trode was remelted under a protective slag (ESR). The
third alloy, also with a nominal addition of 2%
titanium, was prepared by another vendor using VIM

and CEVM processes. The last two alloys have hafnium
additions. Although the electrode that was remelted by
the ESR process was initially overcharged with haf-
nium, the slag extracted so much of the hafnium that
the final alloy contained only 0.3% hafnium. The alloy
that was melted by the VIM and CEVM processes had
0.7% hafnium.

These alloys were first subjected to weldability tests.
Filler wire was prepared from the same material as the
test plates by cutting '4-in.-square strips and swaging
them to %, in. in diameter (for use in the first three
weld passes) and " in. in diameter (for the remaining
passes). The weld test plates were prepared with a bevel
so that the included angle for the weld deposit was
100°. The welds were examined visually after every pass
and periodically with dye penetrant. Side-bend samples
Y% in. thick were cut from the welds and bent around a
Ys-in.-diam mandrel. All of the alloys passed the
weldability test except heat 72-115, which contained
0.7% hafnium. The root pass cracked in the first test
weld. A second weld was made in which the first three
passes were made with standard Hastelloy N fillér metal
(heat 5005). These three passes looked excellent, but
when the weld sequence was continued with three
passes of heat 72-115, numerous cracks formed. This
weld was ground out, and the entire weld was made
with standard Hastelloy N filler metal. The weld looked
excellent, and the side-bend specimens were free of
cracks (Fig. 12.1). -

Transverse weld samples that include base metal,
fusion zone, and weld metal were prepared from all
alloys. Samples from two of the heats that were
modified with about 2% titanium have received limited

Table 12.1 Vendor chemical compositions on several heats of modified Hastelloy N plate

Heat Concentration (wt %) Melting®
number Cr Mo Fe C Si Mn P S Al B Ti Hf Zr process
71-114 7.24 12.03 006 0.05 0.04 001 0.002 0005 0.12 0.002 196 VIM + CEVM
17-583 7.21 12,16 0.06 004 005 0.01 0002 0.004 0.12 0.001 1.79 VIM + ERS
72-503 696 1295 0.32 0.06 0.01 0.01 0.001 0.003 0.001 194 VIM + CEVM
72-604 6,87 11,90 0.07 009 009 007 0003 0003 012 <0.001 0.3 002 VIM+ESR
72-115 6.87 1190 0.07 0.09 0.09 007 0.003 0003 0.12 <0.001 0.7 0.02 VIM+CEVM

2VIM = vacuum induction melt.
CEVM = consumable elecirode vacuum melt.
ESR = electroslag remelt.
118

PHOTO 0086-73

0.5

1.0

INCHES

Fig. 12.1. Side-bend specimens of a weld made using heat 72-115 (0.7% Hf) base metal and heat 5005 (standard Hastelloy N)
filler metal. The samples were examined with dye penetrant, and no flaws were visible.

Table 12.2. Creep properties at 650°C and 55,000 psi of the base metal and transverse welds of two experimental alloys

: Minimum : Reduction

:x:;:r Postweld anneal R“"(':;)' e creep rate R“"":;)““‘“ in area
(%/hr) (%)
71114 Base metal, annealed 1 hr at 1180°C 119.0 014 443 458
As welded 68.0 0.034 611 316
8 hr at 870°C 39 0.25 76 332
1 hr at 980°C 56 026 8.1 25
71-583 Base metal, annealed 1 hr at 1180°C 63.1 011 314 26.0
Aswelded 68 0.055 34 9.9
8 hrat 870°C 24 0.25 49 21
1 hr at 980°C 23 0.38 6.6 29.8

testing, and the results are summarized in Table 12.2.
The following points seem important concerning the
creep behavior of these heats at 650°C.

1. Welds have lower rupture lives, creep rates, and
fracture strains than the base metal.

2. Postweld-heat treatments at 870 and 980°C decrease
the rupture life over that of the as-welded material,
increase the minimum creep rate, and have only
slight effects on the fracture strain.

Samples of the base metal of all five heats listed in
Table 12.1 have been creep tested at 650°C. As shown
in Fig. 12.2, the stress-rupture properties of all five
heats were superior to those of standard Hastelloy N.
Within the scatter of the data, the three titanium-
modified alloys and heat 72-602, containing 0.3%
hafnium, had about equivalent properties. Heat 72-115,
with 0.7% hafnium, had superior properties. The creep

rates of these same heats at 650°C are shown in Fig.
12.3. The three titanium-modified alloys had about
equivalent creep strength, and the data indicated that
they were generally stronger than standard Hastelloy N.
The two hafnium-modified alloys were even stronger
than the titanium-modified alloys.

Some testing of these five heats was also done at
760°C. The stress-rupture properties (Fig. 12.4) of heat
72-604, with 0.3% hafnium, were about equivalent to
those of standard Hastelloy N, and the properties of the
other alloys were superior. The creep rates of these
heats are shown in Fig. 12.5. Heat 72-115, with 0.7%
hafnium, was considerably stronger than standard Has-
telloy N, and the other alloys were generally slightly
stronger.

Samples of alloy 71-114 (1.96% Ti) and 71-583
(1.79% Ti) were irradiated to a thermal fluence of 3 X
10%° neutrons/cm? at 650 and 760°C and postirradia-

119

tion creep tested at 650°C. The stress-rupture prop- in carbide structure." The microstructures of the
erties are shown in Fig. 12.6. The rupture time at a current alloys have not been characterized, but it is
given stress level was reduced by irradiation, but is still
superior to that of standard Hastelloy N irradiated at 1. H. E. McCoy and R. E. Gehlbach, “Influence of Irradiation
650 and 760°C. This large effect of irradiation tempera- Temperature on the Creep-Rupture Properties of Hastelloy N,”
ture on standard Hastelloy N was attributed to changes  Nucl Technol. 11(1), 45 (1971).

ORNL-DWG 73-840

80 Q\
\\\
et
70 SN N0
= o
NG \
L ~
N \1 Y
SN ™
o0 e 72-115
~N ™ N
NN
4 (el [ r'y
™ N N
<\ ) \
= 50 - ¥ ™~ ~ N
3 ‘\\ \G \hl
3 ~ NN
bal NN DY
= ~ \\\"\
— 40 STANDARD HASTELLQY N —me= 2 3]
)
@ A \\
% \\ \\\7\1—114
[ N
» 30 AN T1-583 %y L 1
\\ S\\“
N
o 71-114 (1.96 Ti) AN
20 & 71-583 (1.79 Ti) ™
o 72-503 (1.24 Ti)
e 72-604 (0.3 Hf}
10 A 72-115 {0.7 Hf)
100 2 5 10t 2 5 102 2 5 10° 2 5 10
RUPTURE TIME (hr)
Fig. 12.2. Stress-rupture properties at 650°C of several heats of modified Hastelloy N.
ORNL- DWG 73-841
70 @,
A L [n] 20
60 —
: L~
® A [= 4}/
50 -
A ™ do A/a/
2 /
8 L~
8 40 ! * 5 O STANDARD HASTELLOY N —+
e p
— ’/
@ . 41
'fi':" 30 /,ac
‘_
[75]
.of’/
20 pg ' o T1-14 (1.96% Ti)
// a 71-583 (1.79 % Ti)
10 7 o 72-503 (1.94 % Ti)
e 72-604 (0.3% Hf)
A 72-15 (0.7 % Hf}
0 LI I
10-4 03 . 102 107! 100

MINIMUM CREEP RATE (%/hr)

- Fig. 12.3. Creep-rates at 650°C of several heats of modified Hastelloy N.
likely that the poorer properties observed after irradia-
tion at 760°C compared with 650°C are due to changes
in carbide structure. The minimum creep rates of these
samples are shown in Fig. 12.7. Before irradiation, these
alloys had minimum creep rates about equivalent to
those of standard unirradiated Hastelloy N (Fig. 12.3).
Irradiation caused an increase in the minimum creep
rate at high stress levels, but had little effect at low
stresses. Irradiation temperature did not have any
detectable effect on the creep rate.

The property change of most concern in Hastelloy N
during irradiation is the reduction in the fracture strain.
This parameter is shown in Fig. 12.8 for alloys 71-114
and 71-583. Both alloys had fracture strains many times
those for standard Hastelloy N. Alloy 71-583, with
1.79% titanium, had slightly lower fracture strains when
irradiated at 760°C than at 650°C, but alloy 71-114,
with 1.96% Ti, did not show any detectable effect of
irradiation temperature.

120

In summary, small commercial heats of an alloy
containing nominally 2% titanium have shown satisfac-
tory weldability, unirradiated mechanical properties,
and mechanical properties after irradiation. Further
development of this alloy will await a clearer under-
standing of the effects of composition on the suscepti-
bility to intergranular cracking, discussed in Chap. 10.

12.2 IRRADIATION OF HASTELLOY N
IN THE HFIR

H. E. McCoy

There is increasing evidence that Ni will transmute to
helium by a thermal-neutron reaction.? We had as-
sumed previously that ' °B would be the primary source
of helium in Hastelloy N irradiated in an MSBR, and,

2. A. A. Bauer and M. Kangilaski, “Helium Generation in
Stainless Steel and Nickel,” J. Nucl. Mater. 42, 91-95 (1972).

ORNL—DWG 73— 842

40,000
"N 20
-@O—Chir
= \o\\ h AD
2 N
- By
o 20,000 | ™ o | aa
a9 N
w \
@« O 71514 (1,96 % Ti) N :
» A 71-583 (1.79 % Ti) T °
D 72-503 (1.94 % Ti) /\
e 72-604 (0.3 % Hf) STANDARD HASTELLOY N N
10,000 ==~ & 72-1{5 (0.7 % Hf) N
l P "
s000 L T T 1
io~' 2 5 100 2 5 1o 2 5 {02 2 5 103 2 5 104
RUPTURE TIME (hr)
Fig. 12.4. Stress-rupture properties at 760°C of several heats of modified Hastelloy N.
ORNL-DWG 73 —-843
40,000 :
V{
N s
O 74— 114 ({.96 % Ti} -
= AT71-583 (.79 % Ti) A p ot e
& : L~
— 20000 D72—503 (4.94°’o TI) . | | /
o SO0 972604 (0.3 % Hf) “TII
u A72-115 (0.7 % Hf) A
e cp/;\.c/
/LSTANDARD HASTELLOY N
/fl
A
10,000
8000
T Juad 10-3 1072, 10~ 10° 10*

MINIMUM CREEP RATE (%/hr)

Fig. 12.5. Creep rates at 760°C of several heats of modified Hastelloy N.
121

ORNL-DWG 73-844

70 _
NIRRT
7=
\\>< UNIRRADIATED
60 n ‘\\
NN
o UNIRRADIATED —=t™N| NS
N \\<,/— 71-114
50 NS UNIRRADIATED
N T~
[ ™~ \\.
= ~ NN
a8 ~ NN
Q 40 SN AN \\\‘\
@] —~ r = N
9 ~ ~—. < \\ \
- [ -~ Ml 0Ԥ \
%) ™~ IRRADIATED | [/ ~ N N
& 30 S~ AT 650°C ] ~la N X
| ot ™ S N ™ ~
w ~d F ~al .
o - S ~— \
kS IRRADIATED >
20 [ {
AT 760°
o 71-H4 (1.96% Ti} T~ 60°C
& 71-583(1.79% Ti) ™I~ L
10 |~ — — STANDARD HASTELLOY N P
SOLID POINTS — IRRADIATED AT 760°C
OPEN POINTS— IRRADIATED AT 650°C
0 R L1 11l
ITo xd 5 To 5 102 2 5 o: 2 5 10%

RUPTURE TIME (hr)

Fig. 12.6. Stress-rupture properties at 650°C of two heats of titanium-modified Hastelloy N following irradiation to a thermal

fluence of 3 X 10%° neutrons/cmz.
ORNL~-DWG 73 -845
70 v
]
!
I T T )
e
STANDARD HASTELLOY N pd
60 UNIRRADIATED, IRRADIATED AT 650°C 7 t
T
!
L
50
[ ] A
o 40 e B—O— ik » =
O ’/ ___4"
Q s
= 54§ ”,’
2 20 _// 4 STANDARD HASTELLOY N
g A L-FT IRRADIATED AT 760°C
= l—”'
W L
-
20 =
o 71-114 (1.96 % Ti)
o a 71-583 (1.79 % Ti)
SOLID POINTS - IRRADIATED AT 760°C
OPEN POINTS - IRRADIATED AT 650°C
0 | [P R
0" 2 5 02 2 5 o' 2 5 1° 2 5 10

MINIMUM CREEP RATE (% /hr)

Fig. 12.7. Creep rates at 650°C of two heats of titanium-modified Hastelloy N following irradiation to a thermal fluence of 3 X
10%° neutrons/cm?.
122

ORNL-DWG 73-B46

14
IR
o T1-114 (1.96 % Ti)
12 ol a T1-583(1.79 % Ti)
SOLID POINTS-IRRADIATED AT 760°C
OPEN POINTS-IRRADIATED AT 650°C | [
10
;\6 D
; FaY . /éé
a A
z ° . » . ‘ y
" > Yy
€ g § %449
e Y
-
: 7 %%2/ fl
TE— Al /// 'é
1
STANDARD HASTELLOY N A 2/ 4
IRRADIATED Wi
> //, . \ IV /7 é;’ ;‘
£ éé%’w A
29%%, i
g
1007 2 5 1072 2 5 107! 2 5 10° 2 5 10

MINIMUM CREEP RATE (%/hr)

Fig. 12.8. Fracture strains at 650°C of two heats of titanium-modified Hastelloy N following irradiation to a thermal fluence of

3 x 10%° neuttons/cmz.

since this reaction would approach completeness at a
thermal-neutron fluence of about 5 X 102° neu-
trons/cm?, higher neutron fluences would have no
further effect on the amount of helium. The possibility
of low-cross-section thermal transmutation that would
produce helium from Ni presents the possibility that
the above assumptions are incorrect and that helium
could continue to build up with increasing neutron
fluence. There are many ways of limiting the thermal-
neutron fluence received by an MSBR vessel, but in
optimizing the reactor design we need a better under-
standing of how the properties of Hastelloy N vary with
helium content.

The current experiment was run for three cycles in
the HFIR at a design temperature of 650 + 25°C
to a peak thermal fluence of 1.6 X 10?2 neu-
trons/cm?. This fluence is considerably above the
maximum that we have considered for an MSBR.
Engel® has calculated on the basis of the information
currently available that the helium content was most
likely 1300 atom ppm in the Hastelloy N at the center
line of the experiment and 350 ppm at the ends of the
experiment. The minimum estimates for these positions
are 150 and 50 ppm.

The geometry of the test samples is shown in Fig.
12.9. The samples fitted in a %-in.-OD aluminum tube
that was positioned in a peripheral target position in the
HFIR. The samples were maintained at an elevated
temperature by a balance between gamma heating and

the He-filled gap between the specimen and the cooled
holder. The experiment included four heats of material:
(1) heat 5065, standard air melted, (2) heat 5911,
standard vacuum melted, (3) heat 70727, modified with
2.1% Ti, and (4) heat 72115, modified with 0.7%
hafnium.

The irradiated samples were postirradiated creep
tested at 650°C, and the results are shown in Fig.
12.10. Although previous tests of these heats after
lower fluences exhibited marked differences in postirra-
diation creep properties among the different heats, at
this extremely high fluence the results for all four heats
fall into a rather narrow scatter band. The rupture lives
and fracture strains were reduced drastically compared
with those for unirradiated Hastelloy N and are
significantly below those for another heat of standard
Hastelloy N irradiated to 1.5 X 102! neutrons/cm? in
the MSRE.* This MSRE exposure was the highest
fluence to which Hastelloy N had been irradiated prior
to this experiment. Although this was well beyond the
point of boron burnout, it is clear from the results
shown in Fig. 12.10 that the factor of 10 higher fluence
in the current experiment had a more deleterious effect.

3. J. R. Engel, ORNL, personal communication.

4. H. E, McCoy, An FEvaluation of the Molten-Salt Reactor
Experiment Hastelloy N Surveillance Specimens — Fourth
Group, ORNL-TM-3063 (1971).
123

ORNL-OWG 73-847

SQUARE CORNERS’

+0000 D, D (TYPICAL)
0.1133 Z-55= DIAM 2 roo0
(TYPICAL) ° D, 0.1605 12232 DiaM
| (TYPICAL) # £
0.047£0.002 ~#| |=a—  |fe——— 0.720 ——=|
(TYPICAL)
1.250 -
CENTERS NOT
PERMISSIBLE
ALL DIMENSIONS IN INCHES p, =0.0798 13582 piam
_ +0000
Dy =0.0808 12552 DIAM

A =0.125*0.005

Fig. 12.9. Hastelloy N creep sample used in the HFIR.

ORNL-DWG 73-848

80
mr T 11 T il
0 HEAT 5065, HFIR, 1.6 x 1022 neutrons/cm?2
70 A HEAT 59M, HFIR, 1.6x10%2 neutrons/em? | |1/l
™. 0 HEAT 70727 HFIR, 1.6x1022 neutrons/cm?
\ 0 HEAT 72115, HFIR, 4.6x4022 neutrons/cm?
60 ] @ HEAT 5085, MSRE, 1,5x10%" neutrons/ecm®  +H
™
| »
3 50
: ] 1]
o N ||| STANDARD, UNIRRADIATED
< 40 1.1 N HASTELLOY N
) ~N
[72) \
g 0.5
n » q
30 1.5 0.6 v
il ' LTI
0.6 N
: 1.3001.3 . o NG
20 ag33 | 0.7
1.3 N O.3CK)O.1 130 Iq
' 2.3 2.4 0.3
10 fa) ju] 0
0 :
107! 10° 10! 102 103 10*

RUPTURE TIME (hr)

Fig. 12.10. Stress-rupture properties of severa! heats of Hastelloy N irradiated at 650°C to the indicated thermal fluence and
tested at 650°C. The numbers by each point indicate the strain.
The data in Fig. 12.10 suggest again the possibility
that there may be some stress below which irradiation
has practically no effect on the stress-rupture prop-
erties,. The deterioration of properties in material
containing He is generally attributed to the stress-
induced growth of He bubbles to the extent that
cohesion between the grains is reduced. The conditions
required for a bubble to grow as rapidly as vacancies (or
more He) can diffuse to the bubble are described by:?

0. =0.76 y/ry ,

where o, is the critical stress, 7y is the surface tension,
and ry is the bubble radius. Assuming that the stress
.below which the presence of He has no effect is about
10,000 psi and that the surface energy is 1500
dynes/cm, the He bubbles larger than 165 A could
undergo unlimited growth. Such a diameter is not
unreasonable based on our electron microscope observa-
tions on other samples.

The equally poor properties of the diverse heats in
this experiment after a thermal fluence of 1.6 X 10?2
neutrons/cm?® indicate that the properties of any
Hastelloy N irradiated to such an extremely high
fluence will probably not be acceptable. Further experi-
ments will be required to determine the maximum
tolerable helium level in our modified alloys.

12.3 SALT CORROSION STUDIES
J.W. Koger

" The success of an MSBR is strongly dependent on the
compatibility of the container materials with the
molten salts used in the primary and secondary circuits
of the reactor. Nickel-base alloys, more specifically
Hastelloy N and its modifications, are considered the
most promising for use in molten salts and have
received the most attention. The major constituents of
Hastelloy N are nickel, molybdenum, chromium, and
iron. Chromium constitutes about 7% of the alloy and
forms the most stable fluoride. Thus corrosion is
normally manifested by the selective removal of chro-
mium. Stainless steels that have more chromium than
Hastelloy N are generally more susceptible to corrosion
by fluoride melts.

Salts of interest to us are LiF-BeF, based with
additions of UF, (fuel), ThF, (blanket), or ThF, and
UF, (fertile-fissile), and a NaBF ,-NaF mixture (coolant
salt). Lithium, beryllium, sodium, and thorium fluo-

5. D.R. Harries, J. Brit. Nucl. Energy Soc. 5, 74 (1966).

124

rides are stable under anticipated operating conditions.
However, several different oxidizing reactions may
occur, depending on the salt composition and impurity
content. Among the most important reactants that can
cause corrosion are UF,, FeF,, and HF. The reaction
of chromium (in the metal) with FeF, and HF proceeds
to completion at relevant temperatures. The reaction of
UF, with Cr (in the metal) to form CrF, and UF; soon
reaches equilibrium in an isothermal system. The
equilibrium constant has a small temperature depend-
ence, however, so a mechanism exists in nonisothermal
systems for continued chromium removal in hot regions
and deposition in cooler regions, while a steady-state
amount of corrosion-product chromium remains in the
salt. This phenomenon, called temperature-gradient
mass transfer, would be expected to occur in a nuclear
reactor, where the temperature in the primary circuit
would be a maximum at the reactor outlet and a
minimum in the heat exchanger. Because the products
of oxidation of metals by fluoride melts are quite
soluble in the corroding media, passivation is precluded,
and the corrosion rate depends on other factors,
including the thermodynamic driving force of the
corrosion reactions. At steady state the rate of corro-
sion is generally limited by the rate of diffusion of the
chromium through the alloy to the hot surface where it
is being removed. Corrosion by temperature-gradient
mass transfer involving selective removal of chromium
occurs in all of our nonisothermal systems. Under
abnormal oxidizing conditions (as in fluoroborate sys-
tems containing water), other constitutents of Hastelloy
N besides chromium can be removed. Therefore, design
of a practical system utilizing molten fluoride salts
demands the selection of salt constituents that are not
appreciably reduced by available structural metals and
alloys whose components can be in near thermo-
dynamic equilibrium with the salt medium.

The experiments discussed in this section are being
conducted primarily to determine quantitatively the
amounts of corrosion in various salt-alloy systems and
were designed and operated to show the effects of

variables such as alloy constituents, temperature, salt

impurities, salt velocities, and exposure times. These
experiments involve 11 thermal-convection loops which
provide nonisothermal dynamic conditions. Nine of the
loops are constructed of Hastelloy N and two of
stainless steel. Some have been in operation for up to
nine years, others were started during this report
period.

The status of the thermal-convection loops in opera-
tion at the end of this period is summarized in Table
12.3. Recently there has been increased interest in
125

Table 12.3. Status of MSR program thermal-convection loops through August 31, 1972

Salt Max, AT Op_erating
Loop No. Loop material Specimens Salt type composition temp. c0) time
(mole %) ("0 (hr)
1258 Type 304L Type 304L stainless Fuel LiF-BeF,-ZrF 4-UF 4-ThF 4 688 100 79,367
stainless steel steel® (70-23-5-1-1)
NCL-13A  Hastelloy N Hastelloy N; Ti- Coolant NaBF 4-NaF (92-8) plus 607 125 33,579
modified Hastelloy tritium additions
N contro]sb’c
NCL-14 Hastelloy N Ti—n;)odified Hastelloy Cooland NaBF4-NaF (92-8) 607 150 42,154
N L, C
NCL-15A  Hastelloy N Ti-modified Hastelloy Blanket LiF-BeF , -ThF 4(73-2-25) 677 55 35,416
N; Hastelloy N
controls™"€
NCL-16 Hastelloy N Ti-modified Hastelloy Fuel LiF-BeF4-UF,4 704 170 37,942d
N; Hastelloy N (65.5-34.0-0.5)
controls? ¢
NCL-16A  Hastelloy N Hastelloy N, one Te- Fuel LiF-BeF,-UF, 704 170 1,729
coated Hastelloy N%€ (65.5-34.0-0.5)
NCL-17 Hastelloy N Hastelloy N; Ti- Coolant NaBF 4-NaF (92-8) 607 100 27,817
modified Hastelloy plus steam additions
, N controls?*
NCL-18A  Hastelloy N Hastelloy No¢ Fertile-fissle = LiF-BeF,-ThF 4-UF4 704 170 672
(68-20-11.7-0.3)
NCL-19A  Hastelloy N Hastelloy N; Ti- Fertile-fissile =~ LiF-BeF,-ThF 4-UF,4 704 170 22,203
modified Hastelloy (68-20-11.7-0.3) plus
N controls™€ bismuth in molybdenum
hot finger
NCL-20 Hastelloy N Hastelloy N; Ti- Coolant NaBF4-NaF (92-8) 687 250 19,928°¢
modified Hastelloy
N controls™¢
NCL-20A  Hastelloy N Hastelloy N; Ti- Coolant NaBF 4-NaF (92-8) 687 250 1,682
modified Hastelloy
N controls ™€
NCL-21 Hastelloy N - Hastelloy NbBe MSRE fuel LiF-BeF,-Z1F4-UF 4 650 110 9,793
(65.4-29.1-5.0-0.5)
NCL-22 Type 316 Fertile-fissii  LiF-BeF,-ThF4-UF, 650 110 342

stainless steel

Type 316 stainless
steel?€ :

(68-20-11.7-0.3)

4Specimens in hot leg only.

bRemovable specimens,
“Specimens in hot and cold legs.

dSalt and specimens changed, restarted as NCL-16A.,
€Salt and specimens changed, restarted as NCL-20A.
stainless steels because of their apparent resistance to
cracking by fission products (see Chap. 10). Thus,
experiments to more carefully define the compatibility
of stainless steels and other candidate alloys with
various salts have begun. The results obtained from two
pumped loops containing NaBF,-NaF are discussed in
Sect. 12.4.

12.3.1 Fuel Salt

Loop 1258, constructed of type 304L stainless steel,
has now completed nine years of operation with
LiF-BeF,-ZrF,-ThF;-UF, (70-23-5-1-1 mole %). Dur-
ing the current report period the specimens were not
examined to determine weight changes. The chromium
content of the salt remained at about 580 ppm. The
loop continues to operate satisfactorily.

NCL 21 is a Hastelloy N thermal-convection loop,
with removable Hastelloy N specimens in each leg,
containing salt of the same composition as the MSRE
fuel. The loop is equipped with electrochemical probes
to measure the U¥/U* ratio of the salt. Such probes
had been used suceessfully in small static systems, and
this experiment is designed to evaluate their possible
use for on-stream analysis in a large system. Descrip-
tions of the loop and the probes have been given
previously.®>7 No corrosion specimens were in the loop
during this report period because various experiments
were being conducted by members of the Analytical
Chemistry Division (see Analytical Chemistry section of
this report). The maximum corrosion rate while speci-
mens were in the loop was 0.02 mil/year.

The experiment in which we added FeF, to the
Hastelloy N thermal-convection loop (NCL-16) to study
cracking under severe oxidizing conditions has ended.
The loop contained LiF-BeF,-UF, (65.5-34.0-0.5 mole
%) at a maximum temperature of 704°C and a AT of
150°C. After 29,509 hr of operation, two additions of
FeF, (500 ppm each) were made, and the loop
operated 6901 hr after the additions. Cracks to a depth
of 0.5 mil were seen 2887 hr after the first FeF,
addition, but only voids were seen after the last 1000
hr. Extensive examination of the specimens will now be
undertaken,

New salt, of the same basic composition but less
oxidizing, was added to the loop (now designated
NCL-16A), and operation was resumed at the tempera-
ture conditions of NCL-16. Standard Hastelloy N

6. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug. 31,
1971, ORNL-4728, pp. 143-45.

7. 1. M. Dale and A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 69-70.

126

specimens and one tellurium-coated Hastelloy N speci-
men were placed in the loop, with the main object
being to determine if the tellurium will transfer through
the salt to other specimens in the loop. Since then the
loop has operated for over 1700 hr. After 1300 hr, the
loss of material from the hottest Hastelloy N specimen
was only 0.06 mg/cm?, but the tellurium-coated speci-
men at the same position had lost 1.6 mg/cm?. The
amount of material lost was equal to the amount of
tellurium placed on the specimen initially. Analyses
showed the presence of 1 ug/cm?® of tellurium on a
specimen from the cold leg which had a total weight
gain of 0.8 mg/cm?.

12.3.2 Fertile-Fissile Salt

A fertile-fissile MSBR salt has circulated for over
22,000 hr in Hastelloy N loop NCL-19A, which has
removable specimens in each leg and includes bismuth
in a molybdenum vessel located in an appendage
beneath the hot leg of the loop. 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
weight changes of various corrosion specimens are
shown in Fig. 12.11. Assuming uniform loss, the
maximum weight loss of 0.7 mg/cm? is equivalent to a
corrosion rate of only 0.02 mil/year. A modified
Hastelloy N alloy (Ni-13.0% Mo—8.5% Cr—0.1%
Fe—0.8% Ti—1.6% Nb) has lost less weight than a
standard alloy at the same temperature. We attribute
this difference to the low iron in the modified alloy
compared with about 5% Fe in the standard alloy.
Principal corrosion reactions in this loop appear to be

2UF,4(s) + Cr(m) == 2UF3(s) + CtF;(s)
and
FeF,(s) + Cr{m) = Fe(m) + CrF,(s),

where m and s refer to the metallic and dissolved (in
salt) states respectively. The chromium content of the
salt increased 178 ppm in 15,000 hr, and the bismuth
concentration remained below the limit of detection
(10 ppm). The bismuth in contact with the salt seems to
have had no effect on our mass transfer results.

Loop NCL-18, constructed of standard Hastelloy N
and containing removable corrosion specimens in each
leg, operated for over two years with the fertile-fissile
salt mixture.® The maximum weight loss at the highest

8. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug. 31,
1970, ORNL-4622, p. 168.
127

ORNL—DWG 71— 6990A

o~
g
N . % 550°C
E .--—.—__-—____—-
E e ——t—
Iz é.
zZ 0 o1y MODIFIED HASTELLOY N
% .\...___=;_——____-__;:::__'/ * e 685°C
o s . ® 705°C
T ——* 705°C
o -
w
z

-2

o] 2000 4000 000 8000 10,000 12,000 14,000 46,000 18,000 20,000

OPERATING TIME (hr)

Fig. 12.11. Weight change vs time of Hastelloy N specimens exposed to LiF-BeF,-ThF 4-UF 4 (68-20-11.7-0.3 mole %) at various
temperatures in thermal-convection loop NCL-19A, in which liquid bismuth is in contact with the salt at the bottom of the hot leg.

temperature, 704°C, was 1.5 mg/cm?, which is equiva-
lent to a corrosion rate of 0.05 mil/year, assuming
uniform attack. Operation of NCL-18 ceased when an
electrical short in one of the main heaters burned a hole
in the loop piping. The loop was repaired by replacing
the hot leg, new specimens were installed, and the
loop was filled with new salt. The rebuilt loop,
designated NCL-18 A, has now operated for over 600 hr.

As discussed in Chap. 10, stainless steels appear to
resist cracking by several fission products more than
many of the nickel-base alloys, particularly Hastelloy N.
We have constructed and are operating a type 316
stainless steel thermal-convection loop (NCL-22) which
contains the single-region MSBR fertile-fissile salt. The
purpose of the experiment is to determine the compati-
bility of the stainless steel with the highly purified salt.
In this first test, our maximum temperature is 649°C
(1200°F) with a AT of 110°C. The loop has now
operated over 300 hr.

12.3.3 Blanket Salt

Loop NCL-15A, constructed of standard Hastelloy N
and containing removable specimens in each leg, has
operated over four years with the LiF-BeF,-ThF,
blanket salt proposed for a two-fluid MSBR. Mass
transfer, as measured by the change of chromium
concentration in the salt, has been very small. The
maximum corrosion rate indicated by specimen weight
changes is 0.06 mil/year (assuming uniform removal of
all alloy constituents).

12.3.4 Coolant Salt

Loops NCL-13A and NCL-14, constructed of stand-
ard Hastelloy N and containing removable specimens in

each leg, have operated for 3.8 and 4.8 years, respect-
ively, with the fluoroborate mixture NaBF,-NaF (92-8
mole %). The maximum corrosion rate (assuming
uniform removal of all constituents) at the highest
temperature, 605°C, has averaged 0.7 mil/year for both
loops. Examination of specimens has shown that
corrosion has generally been selective toward chro-
mium, but there have been short periods when there
was general attack of the Hastelloy N. These periods
resulted when leaks in the gas lines or in ball valve seals
admitted gaseous impurities into the salt. The valves are
exposed to a mixture of He and BF; gas and not to salt.
We attribute the leaks of the ball valves to corrosion
induced by air and moisture inleakage from lines into
the gas mixture. Although the overall corrosion rates in
these loops are not excessive, the rates observed in the
absence of leaks have been an order of magnitude lower
than the average rate.

Loop NCL-17, constructed of standard Hastelloy N
and containing removable specimens in each leg, is
being used to determine the effect of steam injection on
the mass transfer characteristics of the fluoroborate salt
mixture. After 1000 hr of normal operation, steam was
injected into the salt.” The loop has now operated over
26,000 hr following that injection. There was a large
increase in weight change during the first 239 hr after
steam injection, and another abrupt change in the rate
of weight change occurred at 10,178 hr because of a
leak in the cover gas system; however, the overall mass
transfer rate decreased steadily following these two
events.

9. J. W. Koger, MSR Program Semiannu, Progr. Rep. Aug. 31,
1970, ORNL-4622, p. 170.
A leak in the gas line between the two surge tanks of
NCL-17 was detected in the last six-month period.
Following this discovery the specimens were removed,
examined, and weighed. Over the 4841]-hr exposure
prior to the leak the maximum weight loss was 2.9
mg/cm?, which corresponds to a corrosion rate of 0.2
mil/year (assuming uniform material removal). Over the
entire life of the loop, including 1000 hr of normal
operation, steam addition, and 22,375 hr after steam,
the maximum corrosion rate is 1.6 mils/year. The leak
was reparied and the loop is continuing operation.

Loop NCL-20, constructed of standard Hastelloy N
and containing removable specimens in each leg, was
terminated after 19,928 hr of operation with the
fluoroborate coolant salt at the extreme temperature
conditions considered for the MSBR secondary circuit
(687°C max and 438°C min). Forced-air cooling of the
cold leg was required to obtain this A7. For the first
11,900 hr of operation the maximum corrosion rate
was 0.2 mil/year. Then a regulator failure allowed some
moisture to leak into the loop and increased the
maximum corrosion rate to 0.7 mil/year. During the
last 4797 hr, the average rate (assuming uniform loss)
was 0.4 mil/year and steadily decreasing.

During the last few hundred hours of NCL-20
operation, variations in flow were noted. The salt was
frozen in the loop, and the heaters were removed to
examine the loop and the heaters. We found that the
loop tubing had been severely damaged and pitted
because of direct contact with the heater wire. The
insulating bushing normally used to separate the heater
and the tubing had slipped out of place. We decided to
make repairs by overlaying Hastelloy N material on the
damaged portions of the tubing. Because of the

128

possibility of a penetration through the tubing while
overlaying, the heaters were temporarily wired back in
place and the loop was drained. This significant repair
was made without problems and without penetration of
the loop wall. New salt was then placed in the loop
(now designated as loop NCL-20A), and the loop began
operation at its previous conditions.

After the termination of NCL-20, the specimens from
the hot and cold legs were weighed and analyzed to
evaluate the mass transfer in the system. Figure 12.12
shows the weight change of several specimens for the
entire run. The specimens in the hottest position of
NCL-20, 685°C, showed the greatest loss, 14.4 mg/cm?,
for the modified Hastelloy N in 19,300 hr. The
maximum corrosion rate (assuming uniform loss) was
0.29 mil/year. The maximum weight gain, measured on
a specimen in the cold leg, was 4.0 mg/cm?.

The micrographs of the hottest and coldest standard
Hastelloy N specimens from NCL-20 are shown in Fig.
12.13. The hottest and coldest specimens were also
analyzed by scanning electron microscopy (SEM) and
x-ray fluorescence (x-ray fluorescence was induced
from bombardment by the electron beam of the
SEM).!® A view of the surface of the hot leg specimen
is shown in Fig. 12.14. The surface has a semipolished
appearance with many areas which have been under-
mined with labyrinths of holes and channels. The areas
pictured were composed primarily of nickel and molyb-
denum (71 and 29% respectively) with extremely low
iron (0.37%) and chromium (<0.1%). Figure 12.15

10. Analysis performed by L. D. Hulet of the Analytical
Chemistry Division, Quantitative analysis numbers are subject
to errors as high as 20%.

ORNL-DWG 71-8059A

10 T
NCL-20 ‘
. 460°C COLDEST
NE 5 SPECIMENS ——
. e ——
% L] e Le . b
E g1 o]
— 0 —
| B Py
O —— ,
g \.... -\
S i, o W
S -5 ~—~ 660°C
e e
(&)
r u_J ‘ \... -..-."""0
= -0 S
685°C HOTTEST "
SPECIMENS
-5 | | \'
0 2000 4000 6000 8000 10,000 12,000 14,000 16,000 {8,000 20,000

TIME OF OPERATION {hr])

Fig. 12.12. Weight changes of Hastelloy N specimens from NCL-20 exp0>sed to NaBF4-NaF (92-8 mole %) as a function of time

and temperature. Air inleakage occurred after 11,900 hr.
129

t6001

(a) . ; & tlo

6,603 in.

looo7 ™

(b) 1l

Fig. 12.13. As-polished optical micrographs of Hastelloy N specimens from NCL-20 exposed to NaBF-NaF (928 mole %) for
19,300 hr. (a) 685°C, weight loss 14.4 mg/cm?. (b) 460°C, weight gain 2.8 mg/cm?

130

Fig. 12.14. Scanning electron micrographs of the surface of a hot leg specimen from NCL-20. Exposed to NaBF4-NaF (92-8

mole %) for 19,300 hr at 685°C. (a) 1000X. (b) 5000X .

shows the surface of the specimen from the cold leg.
Many of the grains appear to have a semblance of cubic
symmetry with possible growth steps. One would not
expect the grains of the original alloy substrate to have
this much cubic symmetry. The concentrations of
elements from this area are 65% Ni, 23% Fe, 0.5% Cr,
and 11% Mo.

Loop NCL-20A has now operated for over 1600 hr.
After the corrosion specimens had been exposed to the
salt for 884 hr, they were removed and examined. At
the same time we added an amount of KBF;OH to the
loop equivalent to a concentration of 500 ppm in the
salt. (KBF;0H was used because it was easier to
prepare than NaBF;OH.) The BF3;0H ion is the
product of the reaction of H, O and NaBF,, and we are
interested in its effect on mass transfer. Our experience
has indicated that corrosion in fluoroborate systems is
due largely to impurities, particularly water. Reactions

that are believed to be involved include'' the follow-
ing:

H,0 + NaBF, == NaBF;0H + HF ,

NaBF;0H = NaBF,0 + HF ,

6HF + 6NaF + 2Cr == 2Na; CrFq + 3H, .
These may be combined to give

6NaBF,OH + 6NaF + 2Cr
= 6NaBF, 0 + 2Na;CrF, + 3H,

11. Although chromium is the most readily oxidized constit-
uent of Hastelloy N, under certain conditions HF will attack all
the alloy constituents.
131

Fig. 12.15. Scanning electron micrographs of the surface of a cold leg specimen from NCL-20. Exposed to NaBF4-NaF (92-8
mole %) for 19,300 hr at 460°C. 5000x.

and
3H,0 + 3NaBF, + 6NaF
= 3NaBF, 0 + 2Na, CrF4 + 3H, .

Thus both the BF;OH™ initially in the salt and any
H,0 that may be admitted later can play a part in
corrosion. This experiment is designed to quantitatively
measure the effect of BFE;OH ™ on the mass transfer.
During the first two months of operation of loop
NCL-20A, a heater has burned out and two gas leaks
have been found in the ball valve above the hot leg. One
of the gas leaks occurred after the KBF;OH addition.
With the gas leaks and the KBF;OH addition, weight
changes of the corrosion specimens have been fairly
large (Table 12.4). Although the data suggest that the
KBF;0H addition has increased the rate of mass
transfer, the experimental complications preclude any
firm conclusion, and the experiment will be repeated.
We have initiated isothermal capsule tests in which
type 304 stainless steel specimens in four type 304

stainless steel capsules are exposed to NaBF,-NaF (92-8
mole %) at 482, 538, 593, and 649°C. During the first
300 hr all specimens lost weight, with those in the
hottest capsule losing the most. In the next 475 hr the
specimens exposed at 593 and 649°C gained weight and
had brown splotches and pitlike areas on the surface,

Table 12.4. Weight changes of Hastelloy N
specimens from NCL-20A

Weight change (mg/cm?)

Time Max?mum
(hey  Hotleg specimen  Cold-leg specimen  corrosion rate
(687°C) (482°C) (mils/year)
3107 ~1.1 +0.5 1.37
479” -4 +15 379
433° ~68 +3.07 6.07

“Specimens removed because of heater failure.
b8 pecimens removed because of leaking ball valve.
“Specimens removed because of leaking ball valve. KBF ;0H
added during this period.
'Metallic crystals found on specimen.

while the lower-temperature specimens continued to
lose weight. During the last 835 hr very little change
was measured for any of the specimens. Table 12.5
gives all the weight change results to date. In compari-
son, Hastelloy N specimens exposed to the fluoroborate
mixture in similar isothermal capsules showed no
measurable weight changes after 4830 hr at tempera-
tures up to 649°C.'?

12.3.5 Corrosion of Type 304L Stainless Steel and
Hastelloy N by Mixtures of Boron Trifluoride,
Air, and Argon

In the years that we have operated corrosion and
engineering systems with the fluoroborate mixture,
questions have arisen concerning the compatibility of
BF; with various container materials. Results have
ranged from poor to excellent depending mainly on the
purity of the BF; and the gases mixed with it. Because
of the variation of results, we conducted an experiment
in which type 304L stainless steel and Hastelloy N were
exposed to argon, air, BF3, and mixtures of each for
100 hr at temperatures from 200 to 600°C. In addition
the metal specimens were placed with approximately 20
g each of the single-region fertile-fissile salt [LiF-BeF,-
ThF4-UF, (68-20-11.7-0.3 mole %)] and the fluorobo-
rate mixture [NaBF,-NaF (92-8 mole %)] while ex-
posed to the gases.

Table 12.6 gives the results for the metal-salt combi-
nations exposed to various gases at 600°C for 100 hr.
As expected, air in combination with the fluoride salts
produced highly corrosive conditions which destroyed
not only the specimens but the nickel boats. None of
the weight changes measured for the .Hastelloy N
immersed in salt and exposed to any of the gas mixtures
other than air were significant. For the stainless steel
specimens immersed in salt, BF; probably caused the
greater attack. The fluoroborate mixture was more
aggressive toward the stainless steel than the fuel salt.
The weight losses (as opposed to weight gains) occurred
because the corrosion products were removed from the
specimens by the salt and dissolved.

Table 12.7 gives the results for alloys exposed to the
various gas mixtures at temperatures from 200 to
600°C for 100 hr. In most cases, weight gains were
found, since the corrosion products remained on the
specimens and were not carried away.

At 600°C, the only significant changes measured for
Hastelloy N were in the mixtures containing air. These

12. J. W. Koger, MSR Program Semiannu. Progr. Rep. Feb.
28, 1970, ORNL-4548, p. 246.

Table 12.5. Weight change of type 304 stainless steel specimens
exposed to NaBF4-NaF (92-8 mole %) for 1300 hr

Temperature (°C) Weight change (mg/cm?)
482 -0.16
538 -0.32
593 0
649 +0.8

Table 12.6, Weight changes of Hastelloy N and type 304L
stainless steel exposed to various gases while immersed
in fluoride salts at 600°C for 100 hr

Gas flow 100 ¢c/min

Weight change (mg/cmz)

Material . Salt a b c
Argon BF, Air Ar1-BF,
Hastelloy N d -0.03 +0.03 e 0
f +0.06 0 e -0.06
Type 304L stainless d -0.3 -0.3 e ~1.6
steel [ -12 =52 e 2.4

224 ppm moisture.
50 ppm moisture.
€7.5 ppm moisture.
91 iF-BeF5-ThF 4-UF 4 (68-20-11.7-0.3 mole %) fuel salt.
®These specimens were completely destroyed.
INaBF 4-NaF (92-8 mole %) coolant salt.

changes can be attributed to the small amount of
chromium in Hastelloy N and its low resistance to air
oxidation. At 200 and 300°C, all changes were rather
small.

Figure 12.16 shows type 304L stainless steel exposed
to several gas mixtures. The stainless steel specimen in
Fig. 12.16a was exposed to air at 200°C. The air-BF;
mixture produced a large amount of chromium oxide
and iron oxide on the stainless steel at 600°C which was
easily removed (Fig. 12.16b). The stainless steel speci-
men on which the oxides were formed is shown in Fig.
12.16¢. Boron trifluoride, by itself, at 600°C also
produced a large amount of corrosion product on the
stainless steel and was more aggressive than argon.
Significant weight gains of the stainless steel specimens
were produced by all gas mix tures except air at 300 and
600°C. At 200°C, most changes were rather small.
Reaction of the stainless steel with argon, air, and an
argon-BF; mixture resulted in a red corrosion product,
while reaction with BF; by itself resulted in the
formation of white material on the surface. These red
and white corrosion products could not be identified by
x-ray analyses, but contained iron, chromium, oxygen,
133

Y- 142326

0o Oi4 0[2 0,3 0.4 0.5
i

INCHES

Fig. 12.16. Type 304L stainless steel exposed to several gas mixtures. (a) Exposed to air at 200°C for 100 hr. (b) Exposed to
Ar-BF; for 100 hr at 600°C. (¢) Reaction product formed on the sample in b,

134

Table 12.7. Weight changes of Hastelloy N and type 304L stainless steel exposed to various
gases at 200 to 600°C for 100 hr

Gas flow 100 cm/min

Weight change (mg/cm?)

Material Temperature
0 Argon®  Argon® BF;¢  Aird  ArBF3?  Ar-BF3®  AirBF3;  Ar-Air?
Hastelloy N 600 +0.06 +0.03 0 +0.2 —0.03 +0.03 +0.2 0.1
300°¢ 0 0 +0.2  +0.03 +0.03 +0.3 0 0
200 +0.03 -0.03  -0.03 +0.03 0 0 0 -0.03
Type 304L stainless steel 600 +0.1 +0.05 +5.2 +0.03 +O.22_ +0.05 +119.2" +0.8
300° +0.3¥ 4017 +337 003  +58§ +2.0 +25.4 +10.0/
200 +0.05 +0.03 0 0 0 0 -1.2 +40.2

%4 ppm moisture in the argon.

2 ppm moisture in the argon, passed through heated Ti sponge.

€<50 ppm moisture.
475 ppm moisture.
®Near gas entrance,

Large amount of oxide on surface. Identified by x-ray diffraction as 25 mole % Cr,04 and 75 mole % Fe, 0.

€ Adherent red layer on surface.

“Large amount of white material on surface. Nonadherent, 0.0290 g removed.
’Large amount of red material on surface. Nonadherent, 0.0745 g removed.

7L arge amount of material flaked off.

and fluorine where BF3; was involved. Thus, the
corrosion products were probably complex mixtures of
metal oxides and fluorides.

The use of drier argon (2 ppm moisture as opposed to
17 ppm moisture) lowered the weight gain of the
stainless steel specimens by about a factor of 2. The
change was more dramatic when the argon was used
with BF ;. Even though air had almost no effect on the
stainless steel, the air-BF; mixture had the worst effect
of any of the gases, even much worse than just BF;.
The air in combination with argon also produced more
corrosion products than either of the two gases by
themselves. In combination with salt, BF; had the
worst effect. These results underline the problems that
can result in a system which allows air or moisture to
come in contact with molten fluoride salts or with the
BF; vapor..

With the exception of the effect of air at 600°C,
Hastelloy N was much more resistant to corrosion than
the type 304L stainless steel.

12.3.6 Oxide Additions to Sodium Fluoroborate' >

The possibility of tritium retention in fluoroborate
salt by the introduction of a small concentration (10 to
50 ppm) of hydroxide is being investigated. Normally,

LY

.13, D. D. Sood, Bhabha Atomic Research Centre, Bombay,
India, assisted in this work while on assignment at ORNL.

OH™ present in the salt reacts with the constituents of
Hastelloy N to yield hydrogen, which diffuses through
the metal. One of the possible reactions for this is:

2NaBF;0H + Cr = CrF, + 2NaBF,0 + H, .

This reaction indicates the possibility of retaining a
higher concentration of OH™ in the salt if sufficient
NaBF, O can be dissolved in the coolant salt thhout
appreciable corrosion of Hastelloy N.

Investigations along these lines were conducted in a
static system for 1703 hr. A mixture of approximately
33% NaBF,0, 33% NaBF,4, and 33% NaF (NaBF,0
was prepared from NaBF;OH obtained from the
Reactor Chemistry Division) was added to 3.3 kg of an

- NaBF,;-NaF (92-8 mole %) mixture to give approxi-

mately 350 ppm oxide in the salt in which Hastelloy N
specimens were placed. The Hastelloy N corrosion
specimens had a weight loss of 1.5 mgfcm? after 200 hr
of exposure. An analysis of a salt sample taken at this
time showed that the nickel concentration of the salt
had increased from 20 to 100 ppm. There was a slight
increase in the concentration of iron (from 170 to 250
ppm), while the concentrations of chromium and
molybdenum remained essentially constant. The O*"
and OH"™ concentrations were found to be 650 and 17
ppm respectively. It is likely that impurities were
responsible for the corrosion. In all other time intervals
the specimens gained weight, and the concentration of
nickel generally decreased. The concentration of all the

135

other constituents remained constant with an increase

of chromium from 15 to 28 ppm. In the time period
between 474 and 804 hr the specimens gained an
average of 5 mg/cm?, contrasted to a gain of about 0.2
or 0.3 mg/cm® in other intervals. Metallic crystallites
were deposited on the cooler regions of the specimen
holder during this period, and the nickel concentration
decreased to its original level. The crystallites were
found to be essentially pure nickel, with 0.4% copper,
0.2% iron, and 1.5% molybdenum. Only small weight
gains were then found for the last 900 hr of the test,
resulting in an average specimen weight gain of 5.3
mg/cm? for the entire test.

We concluded from the test that the NaBF, O did not
cause corrosion of the Hastelloy N specimens, but
reduced impurity fluorides that deposited on the metal
surface.

12.4 FORCED-CONVECTION LOOP
CORROSION STUDIES

W. R. Huntley J. W. Koger

12.4.1 Operation of Forced-Convection
Loop MSR-FCL-1A

Corrosion loop MSR-FCL-1A continued to operate
during this report period. This test is evaluating the
compatibility of standard Hastelloy N with NaBF4-NaF
(92-8 mole %) coolant salt at temperatures similar to
those expected in the- MSBR secondary circuit. The
loop is fabricated of % -in.-OD, 0.042-in.-wall tubing,
and the nominal salt velocity is 5% fps. Hastelloy N
corrosion test specimens are exposed to circulating salt
at 620, 548, and 454°C.

A scheduled shutdown for examination of the cor-
rosion test specimens occurred on March 14, 1972,
after loop operation for 3994 hr at design conditions.
The corrosion specimens were removed by the normal
procedure of cutting sections out of the loop piping
after the loop had been drained and cooled to room
temperature. The specimens were examined and other
loop repairs made before resuming operation on April
13, 1972. The next scheduled shutdown was made on
July 12, 1972, after 2108 hr of additional exposure for
a total operating time of 6102 hr. The corrosion
specimens were again examined, and  operation was
resumed on August 16, 1972.

The allowable operating time for bearings and seals in
the LEFB salt pump was increased from 2000 to 4000 hr
due to the long bearing life expected at the present
pump speed of 3000 rpm. Oil leakage rates from the

mechanical seal were normal throughout the first
4000-hr operating period and ranged from 1 to 7
cc/day. The bearings and seals appeared in good
condition after 4000 hr of operation, but they were
replaced as a precautionary measure when the pump
was reassembled.

We encountered problems with the brushes and slip
rings of the Adjustospede motor that drives the LFB
salt pump. Frequent honing of the slip rings is required
to prevent arcing and loss of rotating speed. Several
types of graphite and carbon brushes were used in an
attempt to improve brush life and reduce arcing, but
this proved ineffective. Several of the brush failures
resulted in automatic shutdown of the loop and loss of
operating time. New brushless variable-speed motors
have been ordered to eliminate this recurring problem.

12.4.2 Corrosion Results from Forced-Convection
Loop MSR-FCL-1A

The Hastelloy N specimens were weighed and ex-
amined after 3994 hr of operation. The specimens had
been previously removed after 2000 hr.'* As expected,
the three specimens exposed to the salt at the highest
temperature, 620°C, showed the most weight loss. The
average weight loss for this second 2000-hr interval was
10.8 mg/cm?, as opposed to 19 mg/cm? for the first
2000 hr. One of the specimens exposed at 548°C was
badly damaged due to an undefined erosion-corrosion
process and was not included in the weight-change
evaluation. The average weight loss for the other two
specimens at this position was about 1.0 mg/cm?. One
of the specimens exposed at 454°C, the lowest tempera-
ture, lost weight, while the other gained weight,
resulting in an average weight change of zero at this
position.

The salt chemistry changes throughout the 2000 hr
were quite small and indicated small amounts of mass
transfer. However, the corrosion rate was quite high,
with the specimens exposed at 670°C losing weight at a
rate of about 2 mils/year (assuming uniform removal).

The damaged specimen mentioned above was the first
of the three exposed to the salt at 548°C. The damage
was primarily at the leading edge. The leading edge of
the Hastelloy N specimen holder immediately adjacent
to the damaged specimen was not harmed.

During the last test period of 2108 hr, the specimens
exposed at 620°C lost 4.6 mg/cm?, the specimens at
548°C lost 0.5 mg/cm?, and the specimens at 454°C

14. W. R. Huntley and J. W. Koger, MSR Program Semiannu.
Progr. Rep. Feb, 29, 1972, ORNL-4782, p. 179.
136

gained 0.28 mg/cm®. The maximum corrosion rate
during the 2108-hr period was 0.8 mil/year, assuming
uniform dissolution. The weight changes were much
smaller this time period than in the two previous
2000-hr runs, and the maximum corrosion rate was half
that of the previous 2000 hr. The overall maximum
corrosion rate for the entire 6102-hr run is about 2
mils/year.

12.4.3 Operation of Forced-Convection
Loop MSR-FCL-2

Forced-convection loop MSR-FCL-2 is being used
to evaluate the corrosion and mass transfer of stand-
ard Hastelloy N in sodium fluoroborate coolant
at nominal velocities of 10 and 20 fps. The test
contains three sets of removable corrosion specimens
exposed to salt at 620, 537, and 454°C. This test has an
improved design which permits easy removal and
insertion of corrosion specimens and allows individual
study of the effects of velocity, temperature, and time
on mass transfer,

Corrosion loop MSR-FCL-2 operated routinely
throughout March and April 1972 except for minor
problems, For example, the test was automatically
placed in “‘standby” condition due to an electrical
power dip in March, and was automatically shut down
twice in April due to faulty brushes on the drive motor
of the pump. Each of these outages was brief, and the
test was placed in operation again quickly. The experi-
ment was stopped on April 17 for a scheduled
examination of the corrosion specimens, and operation
was resumed on April 20,

Operation of the loop was halted on April 28, when
an oil leak occurred at a soft-soldered seal plug in the
upper end of the ALPHA pump shaft. The salt was
drained from the piping system and the loop cooled as
soon as possible after the incident. No fire or major
damage occurred as a result of the leak, but it was
necessary to remove insulation and shim stock below
the pump to clean out the oil leakage. There were no
indications that the corrosion loop or corrosion speci-
mens were damaged as a result of the oil leak. The leak
in the pump shaft was repaired without removing the
pump from the piping system, as described in detail in
Sect. 3.5.2. All loop and pump repairs were completed
and the test was readied for salt filling on May 25.

During refilling of the salt system, two of the three
drain lines leading from the dump tank to the system
piping did not pass salt due to some type of restrictive
plugging. This caused helium bubbles to be trapped in
portions of the loop piping as the salt level rose and

caused the salt level to exceed the desired levels in the
metallurgical sample stations. The loop was drained and
cooled to room temperature so the ball valves and gas
lines above the metallurgical sample stations could be
disassembled and freed of salt which had frozen in these
unheated regions. The ball sections of the three ball
valves were removed and replaced with flat copper
gaskets so the lines could be heated to allow the molten
salt to drain back into the loop piping. The ball valves
were then reassembled and the corrosion specimens
were withdrawn and examined to see if significant
corrosion damage had occurred during these operational
problems. The maximum weight loss of the specimens
was 0.3 mg/cm? (0.6 mil/year corrosion rate, assuming

-uniform dissolution), which indicated that slight con-

tamination had occurred since the preceding inspection.

The cause of the plugged drain lines is not known. In
retrospect it can be seen that they became plugged
during the 5300-hr period of operation preceding the
oil leak from the pump shaft. During that period, the
salt was never drained from the piping system, so static
salt remained in these small (*;-in. OD, 0.035-in. wall)
unheated lines. The plugged lines were cleared by
heating to temperatures several hundred degrees higher
than those used during two preceding filling operations
early in the loop’s history. Temperatures up to 650°C
were required to loosen the plugged lines. This suggests
that some long-term phenomenon such as deposition of
a high-melting compound may have occurred in the
sections of drain lines near the main loop piping. A
similar problem was encountered on thermal-convection
loop NCL-20, where prolonged heating of the drain
lines at 650°C was necessary to drain the sodium
fluoroborate mixture after 20,000 hr of operation. To
reduce the probability of future plugging in MSR-
FCL-2, we are now operating all static drain lines
adjacent to the circulating salt system at temperatures
greater than those of the circulating salt.

The vent line from the seal oil catch basin of the
ALPHA pump has been troubled with solid crystal
growths and acid formation due to moist air reaction
with BF; (<0.1%) present in the helium purge flow. A
small room-temperature trap filled with Ascarite (NaOH
granules in. an asbestos base) has been installed at the
end of the vent line to remove the BF; by formation of
sodium borate and sodium fluoroborate. The effective-
ness of the trap will be evaluated in forthcoming
operation.

Normal operation of MSR-FCL-2 was resumed on
August 10 and continued smoothly through the end of -
the report period, by which time a total of 5789 hr had
been accumulated at design conditions.
12.4.4 Corrosion Results from Forced-Convection
Loop MSR-FCL-2

Examination of the corrosion specimens from MSR-
FCL-2 after the April 17 shutdown disclosed continu-
ously decreasing corrosion rates through 5100 hr of
exposure (Table 12.8). Over the last 1250 hr, the
corrosion rate of the hottest.specimens (620°C) was
0.06 mil/year (assuming uniform dissolution). No veloc-
ity effect has been seen for the last 2300 hr. The
corrosion rate at 620°C for the entire 5100-hr run was
0.48 mil/year for specimens exposed to salt at 10.8 fps
and 0.67 mil/year for specimens exposed to salt at 20.9
fps. Half the entire weight loss occurred during the first
450 hr. Almost no weight changes were measured for
the specimens exposed at 454 and 537°C for the last
1250 hr. Our data indicate that the controlling corro-
sion mechanism over the last 2300 hr (no velocity
effect) was probably solid-state diffusion in the alloy.

12.5 CORROSION OF HASTELLOY N IN STEAM
B.McNabb  H. E. McCoy

There is more favorable experience with Hastelloy N
in fluoride salts and fluoroborate than with any other
structural metal. If the cracking due to tellurium
described in Chap. 10 can be overcome, it would be
desirable to use this same material of construction
throughout the MSBR salt systems, including the steam
generator. A small program is therefore being con-
ducted to determine whether Hastelloy N is compatible
with steam.

"Unstressed specimens of Hastelloy N appear to be
compatible with steam at 538 and 593°C. The average
weight gain after 13,000 hr exposure in TVA’s Bull Run
facility at 538°C for air- and vacuum-melted Hastelloy
N is 0.45 mg/cm?®. Assuming uniform removal of metal,
this would be equivalent to less than 0.25 mil/year of
metal being oxidized.

Some of the Hastelloy N specimens were removed for
detailed examination after 10,000 hr exposure at
538°C, and preliminary observations were reported
previously.'®> The compositions of these heats of
Hastelloy N are given in Table 12.9. Microprobe
examination of the metallographic samples has been
completed.!® Occasional nodules or blisters of oxide
having a maximum penetration into the metal of 0.4
mil were observed in the standard Hastelloy N speci-

15. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, p. 183.

16. H. Mateer, R. S. Crouse, and T. J. Henson, ORNL, private
communication.

137

Table 12.8. Corrosion rate of Hastelloy N specimens exposed
to NaBF 4-NaF (92-8 mole %) at 620°C
as a function of time and velocity

Corrosion rate? (mils/year)

Time
interval Velocity of Velocity of

(hr) 10.9 fps 20.8 fps
Entire 5100 0.48 0.67
Last 4650 0.27 0.36
Last 4200 0.17 0.25
Last 3320 0.09 0.13
Last 2330 0.08 0.08
Last 1250 0.06 0.06

¢ A ssuming uniform material removal.

mens. Figure 12.17 shows a microprobe display of
backscattered electrons (20-kV accelerating voltage)
from the surface of air-melted Hastelloy N. The oxide is
complex. The outer layer appears enriched in Fe and
Cr, and the interior of the nodule contains some of this
phase with a second phase rich in Mo and an Si-rich
network. The composition of the matrix appears
unaffected just a short distance below the nodules.

Heat 2477 is a typical vacuum-melted heat with low
manganese and silicon (0.05 wt % each). Figure 12.18
shows an optical photomicrograph of a cross section of
this material after exposure to steam for 10,000 hr at
538°C. The compositions of the various microconstit-
uents were determined by microprobe analysis, and the
results are shown in Fig. 12.18. The outer layer is
enriched in Fe and Cr, and the second network layer is
enriched in Cr and Mo. The third region has a metallic
appearance, and is depleted in Fe, Cr, and Mo. The
Si-rich network is not present, since this heat of
material contains only 0.05% Si. Figure 12.19 shows
x-ray-fluorescence images of a region of the same
sample shown in Fig. 12.18. The surface is enriched in
Fe, Cr, and Mo, and a region beneath the oxide is
depleted in these elements. Even though the surface
layer is slightly depleted in Ni, the oxide is primarily
NiO.

A sample of modified Hastelloy N, heat 21546 (Table
12.9), had grain boundary penetrations of oxide up to
about 10 mils after exposure to steam at 538°C for
10,000 hr. Figure 12.20 shows backscattered electron
and x-ray-fluorescence images of a typical area having
grain boundary penetrations. The grain boundary ap-
pears to be slightly enriched in Fe and Cr and depleted
in Ni. Another heat of modified Hastelloy N, heat
70-727 (see Table 12.9), appeared to resist grain
boundary penetration during 7330 hr exposure. Figure
Table 12.9. Compositions of several heats of standard and modified Hastelloy N

Alloy - : Concentration (%)
number Mo  Cr Fe Mn C Si Cu Co A W Al Ti B Nb Hf Zr Mg
5065 160 7.1 4.0 0.55 0.06 0.57 0.01 0.07 0.23 0.1 <0.03 <0.01 0.001 <0.05 <0.1 <0.1 0.02

2477 160 6.9 4.1 0.055 0.057 0.047 0.01 0.0 <0.01 0.03 0.03 0.02 0.002 <0.0005 <0.001 <0.001 <0.005

21546 123 7.3 0.046 0.16 0.05 0.009 0.01 <010 <0.10 <0.10 0.02 0.10 0.002 0.005
70727 13.0 7.4 0.05 0.37 0.044 <0.05 <0.01 <0.01 <0.01 <0.01 <0.03 2.1 0.00006 <0.01 <0.01 <0.001 0.015

8€1
A: 42 Ni
0.55Si
10 Cr
11 Mo
7 Fe

B: 51 Ni
0.6 Si
8Cr
18 Mo
2 Fe

C: 60 Ni
0.8Si
7 Cr
12 Mo
1 Fe

D: 72 Ni
0.5Si
7 Cr
16 Mo
4 Fe

Y-114387

Backscattered Electrons

SURFACE ANALYSIS
(5065, M-581, Met. 72877)

Fig. 12.17. Display of backscattered electrons from a cross section of standard air-melted Hastelloy N after exposure to steam at

538°C for 10,000 hr. 2000X .

12.21 shows backscattered electron and x-ray-fluores-
cence images of a cross section of a sample of heat
70-727. The specimen had two surface layers, with the
outer one enriched in Ni and Fe and the subsurface
layer enriched in Ti, Cr, and Mo and depleted in Ni.
Figure 12.22 shows the backscattered electron and Ti
x-ray-fluorescence images of a typical grain boundary.
Titanium is concentrated in the grain boundaries (likely
as carbides) and may stabilize them against grain
boundary attack of the type observed in heat 21546.
X-ray-diffraction patterns were run on the flat oxi-
dized surface of each of the above specimens by
Gehlbach.!” A typical set of data is shown in Table
12.10. The peaks in the x-ray-diffraction patterns were
indexed as NiO, MoO,, a spinel, Cr, O3, and “matrix”.
The matrix lines consisted of Hastelloy N (¢ = 3.53 A),

17. R. E. Gehlbach, ORNL, private communication.

essentially pure Ni (¢ = 3.53 A), and either a third set
(a = 3.56 A) or a nearly continuous range between Ni
and Hastelloy N. This indicates a depletion of Mo and
Cr from the matrix due to oxidation with a resultant
thin layer of Ni near the surface. Heat 70-727 had the
thickest nickel layer, and heats 21546, 2477, and 5065
had significantly less. Valid estimates of the relative
amounts of the oxides cannot be made at this time due
to the lack of absolute intensity data. However, the
x-ray spectra are by far the strongest for NiO and
weakest for Cr,05. The relative amounts of the oxides
among the specimens appear similar, although the total
amounts may vary considerably

Although the compatibility of Hastelloy N with steam
looks acceptable in the absence of stress, steam genera-
tors are stressed, and the effects of stress on the
compatibility must be evaluated. Tube-burst specimens
of two heats of Hastelloy N (N15095 and N25101)
have been fabricated and installed in the Bull Run

140

A. 17 Fe
11 Cr
2 Mo
50 Ni

10 Cr
19 Mo
65 Ni

6Cr
12 Mo
86 Ni

7Cr
16 Mo
72 Ni

Light Optics
SURFACE ANALYSIS
(2477, M-582, Met. 72878)

Fig. 12.18. Optical photomicrograph of a cross section of vacuum-melted Hastelloy N after exposure (o steam for 10,000 hr at
538°C. The results of microprobe analyses are indicated. 100X
vorra303

ELECTRON BEAM
SCANNING IMAGES
OF SURFACE
(2477, M-582, Met. 72878)

CrKa

NiKa FeKa

Fig. 12.19. Backscattered electron and x-ray-fluorescence images of the vacuum-melted Hastelloy N sample shown in Fig. 12.18.
2000x.

ELECTRON BEAM
SCANNING IMAGES
(21546, M-575, Met. 72879)

BACKSCATTERED ELECTRONS

CrKa

FeKa

141

¥-114391

MoLa

Fig. 12.20. Backscattered electron and x-ray-fluorescence images of modified Hastelloy N, heat 21546. 2000x

facility after slight modifications to the specimen
holder. There are presently ten instrumented tube-burst
specimens in test at stresses from 28,000 to 72,000 psi
(specimen wall thicknesses vary from 0.0108 to 0.0302
). The instrumented specimens have a capillary tube
connected to the annulus of the double-wall tube-burst
specimens to detect failure when steam enters the
annulus. Thermocouples on the capillary tubes are
connected to a multipoint recorder. When steam enters
the capillary tubing, it causes a temperature rise on the
recorder that indicates rupture of the specimen.

Figure 12.23 shows three specimens that failed at 4.0,
27.4, and 99.7 hr. The reduced gage sections of these
specimens were stressed at 66,000, 56,000, and 55,300
psi. The specimens were not removed until after 1000
hr exposure to steam. The specimens at the two highest
stresses failed with considerable deformation, and the
specimen at the lowest stress failed after less deforma-
tion. These differences in the amount of deformation
were reflected in the appearance of the fractures of

these specimens (Fig. 12.24). The specimen at the
higher stress had the most ductile fracture, being almost
entirely shear (Fig. 12.24¢). The sample at the inter-
mediate stress had a mixed shear and intergranular
fracture (Fig. 12.24b). The sample at the lowest stress
had an almost entirely intergranular fracture that is
characteristic of high temperatures.

There has been a considerable amount of scatter in
the stress-rupture data for tubes tested in argon and
steam environments. We suspected that flaws could be
important in causing the scatter. Several specimens were
ultrasonically inspected and compared with an elec-
trical-discharge-machined flaw 1 mil deep and 62 mils
long. Of 42 specimens inspected, 11 gave indications
equal to or greater than the standard. The specimen
wall thicknesses ranged from 0.011 to 0.020 in. With an
applied pressure of 3500 psi, a decrease of 1 mil in the
wall thickness would increase the stress on a
0.020-in.-wall specimen by about 2000 psi, and on an
0.011-in.-wall specimen by 6000 psi. A notch or flaw of

142

BACKSCATTERED ELECTRONS

v-114388

TiKa MoLa

NiKa

FeKa

CrKa

ELECTRON BEAM SCANNING IMAGES OF SURFACE
(70727, M-580, Met. 72929)

Fig. 12.21. Backscattered electron and x-ray-fluorescence images of modified Hastelloy N, heat 70-727. 2000 .

1-mil depth could have a greater effect than just the
1-mil reduction in wall thickness, due to stress concen-
tration at the tip of the notch or flaw. The length of the
flaw and the sensitivity of the material to cracks or
flaws would influence the stress concentration factor.
These stress concentrations due to the flaws present
likely account for some of the scatter of the stress
rupture data. Figure 12.25 shows a control specimen
tested in argon with an argon internal pressure of 3500
psig. It was tested at 538°C and a stress of 40,300 psi.
Rupture occurred at 565.2 hr, which was much less
than a similar specimen exposed to steam, still in test
after 7000 hr. There are numerous cracks on the inside

diameter which were probably present before test,
although this specimen was not ultrasonically inspected
before testing. The ultrasonically inspected specimens
without flaws should help in defining the stress rupture
properties in steam at 538°C. Some of the specimens
with flaws will be tested in argon to determine the
effect of flaws of this size on the stress-rupture
properties. Due to the scatter in the data, no definite
conclusions can be drawn at this time regarding the
effect of steam on the stress-rupture properties of
Hastelloy N tubes. In general the data points obtained
in steam fall on the weak side of the scatter band of
tests run in argon.

143

Y-114389
s B
- z ¥
’ .
3
.
v 2
.
P
/
. 2
. e
td
v ‘
BACKSCATTERED ELECTRONS Tika

ELECTRON BEAM SCANNING IMAGES
OF PRECIPITATES
(70727, M-580, Met. 72929)

Fig. 12.22. Backscattered electron and x-ray-fluorescence images of a typical grain boundary of modified Hastelloy N, heat
70-727. 2000%

Y-114994

HASTELLOY N TUBE BURST SPECIMEN
FAILED IN 4.0 hr IN STEAM AT 66,000 psi

N TUBE BURST SPECIMEN
N STEAM AT 55,300 psi

HASTELLOY N TUBE BURST SPECIMEN
FAILED IN 27.4 he IN STEAM AT 56,000 ps

Fig
assembly

.23. Photograph of Hastelloy N tube-burst specimens stressed in steam at 538°C. The inner annulus of the concentric tube

exposed to the steam.

144

Table 12.10. X-ray-diffraction pattern from surface of Hastelloy N, heat 70-727, exposed

to steam at 538°C for 10,000 hr

Relati Planar Planes for the indicated compound having the spacing shown at the left
clative .
intensity Sp?;l)ng Faci;‘:;';e“”d NiQ? Spinel® MoO, Cr,03
0.3 4.82 111
0.3 3.63 104
13 3.414 110,111
0.4 2.945 220
0.5 2.706
0.5 2.667 121
1 2.515 311
1 2471 110
80 2414 111 (Several) e
0.4 2.173 210,121 120
30 2.090 200 400
100 2.069 111
75 2.036 111
35 1.792 200
10 1.770 200
25 1.764 200
0.2 1.717 311,220
2 1.707 422 (Several)
0.3 1.679 321
0.1 1.601 003
0.5 1.528 131,313
60 1.478 220 440
0.3 1.429 221
0.3 1.402 131,202
30 1.266 220 533
0.5 1.260 220 311
5 1.247 220
4 1.207 222 411

“The diffractions indicate that the lattice cell has a dimension varying from 3.526 to 3.584 A. Hastelloy N has a cell size of 3.58

A and Ni has a cell size of 3.53 A.

by = 4.180 A.

€%4a=8.35A. .
145

Fig. 12.24. Photomicrographs of the fractures of Hastelloy N tube-burst samples exposed to steam at 538°C. (a) Stressed at
66,000 psi and failed in 4.0 hr, (b) stressed at 56,000 psi and failed in 27.4 hr, and (c) stressed at 55,300 psi and failed in 99.7 hr. As

polished

146

\
&b
S
H
3l
8
y-1soet |
A
g
g
H
3¢
8

Fig. 12.25. Photomicrographs of a Hastelloy N sample tested in argon at 538°C and 40,300 psi and failed in S65 hr. (a) Fracture.
(b) Region away from fracture showing frequent cracks on the inside diameter of the tube. As polished
- 147

13. Support for Chemical Processing
J. R. DiStefano

Processes involving the selective chemical reduction of
materials from the fuel salt into liquid bismuth are
being developed for high-performance molten-salt
breeder applications. Materials for this application must
withstand the corrosive effects of a number of environ-
ments at 500 to 650°C, such as HF-H, mixtures, F,,
molten-salt mixtures, and bismuth containing lithium
and thorium. Molybdenum, tantalum (or tantalum
alloys), and graphite appear promising for use with
bismuth, and we have concentrated our studies on these
materials.

One major undertaking has been the construction of a
molybdenum reductive-extraction test stand that will
allow us to obtain metallurgical data as well as chemical
processing data. Prior to construction, fabrication and
joining procedures for molybdenum were developed,
and during construction equipment has been developed
for specific applications.

We are also continuing our program to evaluate
molybdenum, tantalum, T-111 (Ta—8% W-2% HI),
Ta—10% W, brazing alloys, and graphite with bismuth-
lithium solutions under reprocessing conditions.

13.1 CONSTRUCTION OF A MOLYBDENUM
REDUCTIVE-EXTRACTION TEST STAND

J. R. DiStefano A.J. Moorhead

Work was continued on the construction of a molyb-
denum reductive-extraction test stand for chemical
processing studies. An unjoined mockup was con-
structed with actual test-stand components to check for
proper alignment and to reaffirm the step-by-step
fabrication sequence. A photograph of the mockup is
shown in Fig. 13.1. The principal components of the
test stand are a 1%-in.-OD X 5-ft-long packed column
through which bismuth and salt streams will counter-
currently circulate, two 37%-in.-OD enlarged end sec-
tions where the fluids will be separated, and two
37%-in.-OD X 8-in.-long feed pots containing removable
orifices for controlling the flow rate. Four different
tube sizes (!, in. OD X 0.020 in. wall, % in. OD X
0.025 in. wall, % in. OD X 0.030 in. wall, and 7% in.
OD X 0.080 in. wall) interconnect the various compo-
nents. Details of the design of the test stand have been
reported previously.!-2

Three steps in construction of the test stand have
been completed: (1) fabrication of back-extruded half
sections and column, (2) machining of subassembly

H. E. McCoy

components, and (3) construction of unjoined mockup
from test-stand components. We are now fabricating the
feed pot and column subassemblies. This involves
joining tube stubs to the pots by either electron-beam
welding or roll bonding and then, after assembling all
internal components, joining the pot half sections
together with an electron-beam girth weld. Intermediate
steps in this phase of construction are helium leak
checks or dye-penetrant inspections of the partially
completed assemblies to ensure the integrity of each
joint, and chemically vapor depositing tungsten on the
inside of each pot where a roll-bonded joint has been
made. The final step in subassembly fabrication will be
to back braze each welded or roll-bonded tube-header
joint and to braze a reinforcing band around the girth
weld. Although the filler metal 42M (Fe—15% Mo—5%
Ge—4% C—1% B) was previously selected for use in
back brazing, experience gained in fabricating a proto-
type head pot has shown the %-in.-OD molybdenum
tubes brazed with this alloy to be brittle in the braze
area.? Therefore, we are evaluating several commercial
filler metals to determine if they would be more
satisfactory for back brazing where mechanical support
of the joint is the primary requirement.

The final step in construction will consist of intercon-
necting the various subassemblies by field orbiting-arc
welding. Each tube-to-tube weld joint will then have a
sleeve brazed around it for reinforcement. Details of the
fabrication and joining studies related to construction
of the test stand are reported in the following sections.

13.2 WELDING OF MOLYBDENUM
A.J. Moorhead

During this report period all of the various lines of the
molybdenum test stand were bent to shape and
installed along with the extruded and machined compo-
nents on the field assembly jig. The tubing was then
stress-relieved for 60 min in vacuum at 900°C and cut
at selected locations using our small abrasive-wheel

1. E. L. Nicholson, Conceptual Design and Development
Program for the Molybdenum Reductive-Extraction Equipment
Test Stand, ORNL-CF-71-7-2 (July 1, 1971).

2. 1. R. DiStefano, MSR Program Semiannu, Progr. Rep. Aug.
31, 1971, ORNL-4728, pp. 163—69.

3. J. R. DiStefano and A. J. Moorhead, MSR Program
Semiannu. Progr. Rep. Feb. 29, 1972, ORNL-4782, p. 194.
148

Fig. 13.1. Unjoined mockup of molybdenum test stand.

cutoff device. We also completed development of our
welding procedures and began fabrication of the subas-
semblies of the test stand.

The final welding procedures developed were for the
- and 1%-in.-diam tube-tube welds and the Y-, %-,
-, and Tg-in.-diam tube-tee welds. All of these welds
were made in a vacuum-pumped, argon back-filled glove
box. By extensively modifying a commercial orbiting-
arc weld head (Rytik), we were able to successfully join
lengths of %-in.-OD X 0.050-in.-wall and 1%-in.-OD X
0.050-in.-wall molybdenum tubing to matching stubs
which had been electron-beam welded to back-extruded
half sections. The welding parameters for the 7-in.
tube-tube weld (using a “T-shaped Mo insert) were:
125 A, Y ¢-in. arc length, and a travel speed of 8.7
in./min. The 1%-in.-diam weld (required for the packed
column of the test stand) was made at 120 A, Y ¢-in.
arc length, and a travel speed of 11 in./min without

using a weld insert. Two welds of each size were made,
and all four were free of defects when inspected with
fluorescent penetrant.

Procedures were developed for making the tube-tee
welds (shown in Fig. 13.2) with another orbiting-arc
head (Astroarc) in the argon-filled glove box. Although
we have successfully demonstrated the ability to make
tube-tube welds of this size in the “field” with this
device,* these tube-tee welds will be made in a chamber
to avoid the problem of trying to locally shield these
complex parts from oxidation when brazing a reinforc-
ing sleeve around each weld. The welding parameters
were basically those which had been developed earlier
for tube-tube welds in the field. However, we found

4. A. 1. Moorhead and T. R. Housley, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, pp.
220-21.
149

Y -416310

INCHES

Fig. 13.2. Tubes (', and % in. diam) welded to molybden
used in joining the two tubes is indicated by the arrow.

that we had to increase our welding current about 10%,
apparently due to the close proximity of the tee body
to the weld joints.

We have begun fabrication of the test-stand subassem-
blies by making the eight required tube-to-header
electron-beam welds. Two of these welds attach stubs
of the 1%-in.-diam column tube to the disengaging
sections. The other six are %-in.-diam welds required to
attach the probe tube and other lines to the back-
extruded half sections. All of these welds were free of
defects when inspected with fluorescent penetrant
except for the %-in. weld in the upper half of the
bismuth feed pot. This weld contained pinholes in the
outer circumference of the weld bead and cracks in the
trepan. The weld was machined away and the joint
reprepared. However, the second weld was no better
than the first, indicating that this extrusion contains a
band of contamination which produced porosity and
cracking in the weldment. This part is presently being
replaced by a backup extruded half section.

13.3 DEVELOPMENT OF BRAZING TECHNIQUES
FOR FABRICATING THE MOLYBDENUM
TEST LOOP

N. C. Cole

Portions of the molybdenum test stand will be brazed
for two reasons: (1) mechanical support of welded or

um tee using an orbiting-arc weld head. The weld insert that will be

roll-bonded joints and (2) a backup seal should a leak
develop in a roll-bonded joint or a cracked weld. For
the latter reason, we developed an iron-base brazing
filler metal (Fe—15% Mo—5% Ge—4% C—1% B) with
adequate corrosion resistance to bismuth.S As with any
new filler metal, considerable development effort has
been necessary to learn to apply it properly.

Braze joints of several different configurations are
required, and a different method of brazing each is
being developed. Figure 13.3 shows a large molyb-
denum component joined to several molybdenum tubes
and is a mockup of one of the feed pot subassemblies.
All joints are identical to those on the test stand. On
the ends, the small tubes were welded or roll bonded
and then back brazed. A split sleeve was brazed over an
electron-beam weld around the girth. All of these brazes
were made at one time in a large resistance-heated
vacuum furnace, and the process was monitored by
thermocouples and by visual observance of the braze
flow.

Figure 13.4 shows another type of brazed joint in
which a head pot was joined to a short length of the
1%-in-diam molybdenum column and a split sleeve
brazed around the weld in a large vacuum chamber. A
helical induction coil was slipped over the sleeve, and

5. N. C. Cole, MSR Program Semiannu. Progr. Rep. Feb. 28,
1971, ORNL-4676, pp. 221-25.

Fig. 13.3. Mockup of molybdenum upper disengaging section. (a) Overall, () end view of tube-to-boss braze, and (c) side view
of sleeve brazed over girth weld.

PHOTO 3938-72

0ST
151

A S TS

Fig. 134. Mockup of column joined to upper disengaging
C-19 B over a tungsten arc weld in the column.

brazing was completed in 35 min by heating only the
immediate area of the pot. Because of the mass involved
and the characteristics of the induction machine, the
heating rate was closely controlled, and the brazing
temperature was monitored by a thermocouple placed
under the edge of the split sleeve. Alternatively, we
brazed this same 1'-in.-diam configuration by heating
locally with a portable resistance heater. We now have
the option of using either type of heating, and ease of
operation or accessibility will dictate which is used
when fabricating the test stand. The portable resistance
heater is also capable of brazing all of the smaller tube
joints, provided that there is enough space around the
tubing to place a 3-in-diam X S5-in.-long insulated
heater and the completed assembly will fit into a
vacuum chamber.

Figure 13.1 shows several nozzles protruding through
the stainless steel flange of the test stand. The nozzles
are made of nickel or stainless steel, and all are brazed
to at least one molybdenum tube. In the joint design,
we are using a combination of trepans and feeder holes
to obtain proper flow of brazing filler metal into the
joint. We have successfully brazed several mockups of
these dissimilar metal joints (Fig. 13.5).

A small amount of bismuth vapor may reach the
stainless steel nozzle in the area of the nozzle-to-molyb-
denum-tube braze. Therefore, it will be brazed with the
iron-base filler metal. None of the nickel nozzles should
be exposed to bismuth, so we will braze them with a
commercial brazing filler metal. Several filler metals
were evaluated, but Au—18% Ni was selected because it
flowed as well and had a lower brazing temperature
than the nickel-base brazing filler metals. Each of the
nozzle brazes will be made vertically in a vacuum
furnace by resistance heating in the area of the nozzle
only. To avoid several additional welded and brazed

section. A split sleeve was brazed with Fe-15% Mo-5% Ge-4%

Y- 111576

Fig. 13.5. Mockup of 2-in.-diam stainless steel nozzle brazed
to molybdenum tubes.

joints in the molybdenum tubing, an extension has been
added to the furnace to accommodate tubing up to 6 ft
long.

13.4 COMPATIBILITY OF MATERIALS
WITH BISMUTH AND BISMUTH-LITHIUM
SOLUTIONS

0.B.Cayin  J. L. Griffith L. R. Trotter

We have continued studying the compatibility of
potential structural materials with bismuth and bis-

muth-lithium solutions at 600 to 700°C in static
capsules and flowing thermal convection loop tests. At
present, tantalum alloys, molybdenum, and graphite
appear to have acceptable compatibility.

13.4.1 Tantalum Alloys

We tested T-111 (Ta—8% W—-2% Hf) in quartz
thermal convection loops containing Bi and Bi—0.01 wt
% (0.3 at. %) Li at 600 to 700°C for 3000 hr and
reported that it had excellent resistance to mass
transfer; however, its room-temperature ductility was
drastically reduced.®

The T-111 specimens picked up approximately 158
ppm O during test, and Liu et al.7 have shown that
T-111 is seriously embrittled by the addition of small
amounts of oxygen at temperatures below 1000°C.
Accordingly, we ascribed the ductility loss to oxygen
acquired by the T-111 when tested in a quartz
envelope.

More recently, a T-111 test was conducted in an
all-T-111 thermal convection loop circulating Bi—2.5 wt
% (44 at. %) Li for 3000 hr at a maximum temperature
of 696°C. The T-111 samples had a maximum weight
loss of 2.73 mg/cm? (0.19 mil/year), assuming uniform
dissolution, but remained ductile. Although the loop
specimens showed no ductility loss, the oxygen content
of the specimens was 240 ppm compared with 63 ppm
before test. Thus these specimens showed the same
order of oxygen increase as the quartz loop specimens.
However, the specimens in the T-111 loop were in the
form of Ҥ-in.-diam tensile bars, while the quartz loop
specimens were 0.020-in.-thick flat strips. Also, the
T-111 loop was vacuum annealed at 1400°C imme-
diately prior to operation, and some of the oxygen
pickup may have resulted from this anneal. Oxygen
acquired at the higher annealing temperature, if it is in
the 100-ppm range, does not contribute to a measurable
ductility loss.

Metallography of the tested tensile samples near the
fracture ‘indicated a ductile fracture; that is, no brittle
surface layer existed to suggest surface contamination.
Therefore, we conclude that insufficient oxygen was
added at the test temperature to cause room-tempera-
ture embrittlement.

Samples of Ta—10 wt % W tested for 3000 hr at
700°C in a quartz thermal convection loop containing

6. O. B. Cavin and L, R. Trotter, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4728, p. 173.

7. C. T. Liu, H. Inouye, and R. W. Carpenter, Mechanical
Properties and Structure of Oxygen-Doped Tantalum-Base
Alloy, ORNL report (in press).

152

bismuth had a maximum weight loss of 3.3 mg/cm?
(0.23 mil/year), about the same as T-111. A minimal
oxygen pickup of 50 ppm max was observed but did
not cause embrittlement. However, there was some
scatter in the tensile data, probably due to sample
fabrication. A second loop containing Ta—10 wt % W
samples in Bi—0.01 wt % Li is now in progress.

13.4.2 Molybdenum

We tested Mo in Bi and Bi—0.01 wt % Li at 600 to
700°C and showed that mass transfer is negligible after -
3000 hr in quartz loops.® An all-metal Mo loop
containing Mo tensile samples in Bi—2.5 wt % (44 at. %)
Li is still operating at 696°C with a AT of 116 + 5°C.
This loop, originally scheduled for 3000 hr, now has
accumulated about 5500 hr and will continue operation
to accumulate long-term test data. No problems have
been encountered with its operation.

13.4.3 Graphite

The extremely low solubility of carbon in bismuth
makes graphite attractive for parts of the chemical
processing plant. We previously reported from metallog-
raphic observations that at 700°C graphite is chemically
inert in contact with Bi containing low concentrations
of Li.6 However, we are continuing tests of graphite to
determine its static compatibility with and intrusion by
Bi, Li, or Bi containing up to 3.0 wt % (48 at. %) Li. We
have concentrated primarily on ATJ graphite, which is a
relatively low-density graphite and typical of a graphite
that could be fabricated into the large bodies needed
for a processing plant.

It has been reported that in the absence of some
metallic ions lithium does not form intercalation or
lamellar compounds with graphite at or below 500°C.?
However, Secrist and Wisnyi found that lithium and
graphite held at 700°C for two days in an iron capsule
formed Li,C,.!1%® To check these observations, we
tested AXF-5Q, pyrolytic, and hot-worked pyrolytic
graphites in a pyrolytic graphite crucible containing
high-purity lithium for 1000 hr at 700°C. All of the
graphite was vacuum degassed at 900 to 1000°C for
1000 hr prior to adding the lithium. At the completion
of the test, there was no metallic lithium in the

8. O. B. Cavin and L. R. Trotter, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 173.

9. M. L. Dzurus et al., Proceedings of the Fourth Conference
on Carbon, 1960, p. 165. '

10. D. R. Secrist and L. G, Wisnyi, Acta. Cryst. 15, 1042
(1962).
crucible, and the walls had moved inward. Each of the
graphite samples appeared to be unaffected except for
the slight delamination near one end of the pyrolytic
graphite. This limited delamination is not too surprising
because of the high degree of preferred orientation and
the inherently weak interplanar bonding in these
pyrolytics. However, each had a weight increase that
ranged up to approximately 7% for the AXF-5Q
sample. X-ray-diffraction results on two of the samples
showed the presence of Li,C, and an expanded
graphite lattice. From relative x-ray-diffraction intensi-
ties, the AXF-5Q contained more Li,C, than the
pyrolytic graphite. Since x-ray diffraction is more
sensitive than metallography for detecting interrelation
and compound formation, we are including x-ray
samples in subsequent crucible tests.

We reported no observable chemical incompatibility
of graphite with bismuth or bismuth-lithium solutions
at 700°C, but the liquid metal penetrates the open
porosity of the graphite. This problem is illustrated by
the photomicrographs in Fig. 13.6 which were made
after S00 hr at 700°C from two different regions of the
same ATJ crucible which contained Bi—3% Li. In an
apparent low-density region, the liquid metal pene-
trated the entire wall of 0.19 in., while in a higher-
density .region very little penetration has occurred. This
is not surprising because of the inherent density and
pore entrance diameter gradations present in this

153

graphite. The effect of time on the extent of penetra-
tion of Bi—3 wt % Li into ATJ is seen in Fig. 13.7.
Considerably greater depth of penetration occurred
after 3000 hr. In addition, there was some cracking near
the inner surface which was also roughened. This
roughening of the inner surfaces was caused by the
mechanical removal of the solidified metal from the
sliced crucible. Typical areas are illustrated, but the
great density variations in this material must be kept in
mind. Photomicrographs of wall cross sections of two
separate ATJ crucibles containing 0.01 and 0.17 wt %
Li in Bi and tested for 1000 hr at 650°C are shown in
Fig. 13.8. These crucibles had identical pretest treat-
ments of alcohol wash and oven dry at 190°C for 16 hr.
The one having the greater amount of metal penetration
contained a lower concentration of lithium in the
bismuth at the start of test.

From these results there does not appear to be a
chemical reaction between lithium and graphite in Bi-Li
mixtures with up to 3 wt % Li. A surface impregnation
or sealant would exclude the metal from the graphite,
and crucibles with a surface impregnation seal have
been ordered for evaluation. We are also looking at the
effects of pretreating the graphite by dedusting with an
air blast, washing with alcohol, oven drying at 190°C
for 16 hr, and vacuum degassing for 100 hr at
900—1000°C.
PHOTO 3939 - 72

= 1 15 = sox F 13 = 1

Fig. 13.6. Wall cross section of an ATJ graphite crucible that contained Bi—3 wt % (48. at. %) Li and was tested for 500 hr at 700°C. 50 X. (a) Region near top of
melt; (b) area near bottom of crucible. The graphite metal interface is at the right. The darkest regions are holes, some of which are filled by lighter-colored Bi.

pST

155

PHOTO 3940- 72

0035 INCHES
T 100X

Fig. 13.7. Photomicrographs of ATJ crucible walls near melt that contained 3 wt % (48 at. %) Li in Bi at 700°C for (@) 500 hr,
(5) 1000 hr, and () 3000 hr. 100x. The darkest regions are holes, and the lightest material is Bi that entered the holes.
156

PHOTO 3960-72

=

0.035 INCHES
TS 100X

Fig. 13.8. Photomicrographs of ATJ crucibles tested at 650°C for 1000 hr that contained (a) 0.01 wt % Li and (5) 0.17 wt % Li.
The darkest regions are holes, and the lightest material is Bi.

Part 4. Fuel Processing for Molten-Salt Reactors

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. Work was continued on a
computer program that is being used for calculating
steady-state concentrations and heat generation rates in
an MSBR processing plant. Recent changes in the
program allow the representation of new flowsheets
without programming changes. Several flowsheets that
differ appreciably from the reference flowsheet were
treated successfully with the program.

Studies of equilibria in fused salt-liquid alloy systems
were continued. The solubility of Li;Bi in molten LiCl
was found to be about the same as that reported for
LiCl-LiF (70-30 mole %); the molar ratio of “free”
lithium to bismuth in the salt phase is 3. Lithium and
bismuth were found to distribute between liquid Li-Bi
alloys and LiBr or LiF in a manner similar to that with
LiCl. Experiments were conducted to determine
whether lead could be substituted for bismuth in any
part of the metal transfer process. The concentration of
free lithium in LiCl in equilibrium with a 50 mole %
Li-Pb solution is about 0.001 mole fraction; this is
about the same as that observed with a 50 mole % Li-Bi
solution. Measurements of the solubility of thorium in
Li-Pb alloys indicate that Li-Pb alloys could be substi-
tuted for Li-Bi alloys in the contactor where divalent
rare earths are removed from the LiCl. However, the
solubility of thorium in lead is only about 10% of the
solubility in bismuth; for this reason, the substitution
of lead for bismuth in other parts of the system could
be made only if the metal phase flow rates in the
process were increased considerably or if the distri-
bution coefficients with lead are more favorable than
expected.

The reductive-extraction behavior of titanium, a
component of modified Hastelloy N, was determined. It

is believed that a large fraction of the titanium will be
volatilized during the fluorination step; however, the
remaining titanium would be easily extracted into
bismuth in the protactinium-removal column. This
material would then be accumulated in the salt in the
protactinium decay tank along with other corrosion
product fluorides, and additional quantities of titanium
would be volatilized from this salt by fluorination.
Thus, the presence of titanium in the fuel salt will not
adversely affect operation of the processing system.

Studies of the chemistry of fuel reconstitution were
continued. Spectrophotometric evidence showed con-
clusively that UFs is the product of the reaction of
gaseous UF¢ with UF, dissolved in MSBR fuel salt.
Results of the studies continue to indicate that UF¢ can
be readily combined with processed fuel carrier salt by
this method.

Further progress was made in studying the equilib-
rium precipitation of protactinium from LiF-BeF,-
ThF, (72-16-12 mole %) by sparging the salt with
HF-H, O-Ar gas mixtures.

The purification and charging of the salt and metal
phases to metal transfer experiment MTE-3 were
completed, and operation of the experiment was
initiated. Mass transfer coefficients were measured for
the transfer of 2®®Ra during the first two runs; the
resulting values are in good agreement with a literature
correlation that is based on mass transfer rates between
aqueous and organic phases in a stirred-interface-type
contactor similar to that being used in the present
experiment. The mass transfer coefficient values meas-
ured for europium and lanthanum during four addi-
tional runs were lower than the predicted values. The
values for europium were about 10% of those expected,
and those for lanthanum were considerably lower than
expected. It is believed that the low mass transfer -
coefficient values are due to the presence of an oxide
layer at one or more of the salt-bismuth interfaces in
the system. These data point out the importance of
preventing the ingress of impurities into metal transfer

157
systems. Discussions with a graphite manufacturer
concerning the design and fabrication of the three-stage
salt-metal contactor for use in the fourth metal transfer
experiment indicate that manufacture of the required
graphite components is within the capability of the
graphite industry. Additional data must be obtained on
the compatibility of graphite with bismuth containing
reductant and on mass transfer rates in mechanically
agitated contactors before further design work can be
carried out.

Additional experiments were carried out to measure
mass transfer rates in simulated mechanically agitated
salt-metal contactors. The important physical properties
of molten salt-bismuth systems are quite similar to
those for the water-mercury system, which allows study
of this type of contactor under conditions highly
desirable for experimentation. Measurements of the rate
of extraction of lead from an aqueous phase containing
lead nitrate into mercury containing zinc at low
concentrations indicate that the mass transfer coeffi-
cients in a water-mercury system are predicted reason-
ably well by an existing correlation that is based on
mass transfer in aqueous-organic systems. The results
also imply that the mass transfer rates in salt-bismuth
contactors of this type will be adequate for MSBR
. processing requirements. Further studies were per-
formed in the mild-steel reductive-extraction system in
which the rates of transfer of *?Zr and 2°7U from

158

molten salt to bismuth in a packed column are
measured. The resulting data indicate that the height of
an overall transfer coefficient based on the bismuth
phase is about 4.3 ft for extraction factors near unity,
and that the mass transfer performance of packed
columns is adequate for their use in processing systems.

The feasibility of flaring commercially available
molybdenum tubing was demonstrated, and several
types of flare fittings made of molybdenum were
designed and fabricated. Demonstration of the capa-
bility for making mechanical joints between sections of
molybdenum tubing and between machined molyb-
denum parts and graphite vessels will advance our
technology with regard to the construction of engi-
neering experiments using these materials.

A series of experiments on autoresistance heating of
molten salt was carried out in a simulated fluorinator
protected from corrosion by means of a frozen wall.
The results indicate that an electrically insulating frozen
salt layer can be formed reliably under conditions suit-
able for operation of a fluorinator experiment. Achieve-
ment of this objective is an important step in the devel-
opment of continuous fluorinators. Additional work
must be carried out for demonstrating a means for in-
troducing the high-voltage electrode into a fluorinator
before design of the first fluorination experiment can
be initiated.
159

14. Flowsheet Analysis

Work was continued on the development of a
computer program that can be used for calculating
‘steady-state concentrations and heat generation rates in
an MSBR processing plant. The behavior of 687
nuclides in about 70 regions allows an adequate
representation of present processing plant flowsheets.
Several recent modifications in the program facilitate its
use and allow an improved representation of equipment
items.

14.1 MULTIREGION CODE FOR MSBR
PROCESSING PLANT CALCULATIONS

C.W._Kee L. E. McNeese

We have reported previously' on the development of
a computer code that can be used for calculating
steady-state concentrations and heat generation rates in
an MSBR processing plant. The behavior of 687
nuclides in as many as 250 regions can be considered;
however, the use of only about 70 regions has been
adequate for representing processing plant flowsheets
considered thus far. Several modifications made in the
computer program facilitate its use and allow an
improved representation of equipment items. The more
important modifications are as follows:

1. The extent of transfer of materials between phases
present in a region can now be limited by rate

1. C. W. Kee, M. J. Bell, and L. E. McNeese, MSR Program
Semiannu, Progr. Rep. Feb, 29, 1972, ORNL-4782, pp.
205-206.

and/or equilibrium considerations rather than by
equilibrium considerations alone.

The volumetric flow rates of streams in a flowsheet
and the distribution ratios for materials that dis-
tribute between salt and bismuth phases are now
calculated by the program rather than being speci-
fied as input data, as was done initially.

An improved method was developed for specifying a
flowsheet. This method, which requires only a
description of the regions and specification of the
streams in terms of origin and destination, permits
us to represent a new flowsheet in approximately 30
min rather than 2 days, as was required previously.

. A means was found for decreasing the required
computer time to about 30% of the time needed
previously.

. The computer code ORIGEN was incorporated in
the program to permit the characterization of waste
streams produced by a flowsheet in terms of waste
disposal hazard, levels of radioactivity, and heat
generation after discharge of the streams from a
processing plant.

The modified program has been used successfully for
evaluating a number of flowsheets that, although based
on fluorination—reductive-extraction—metal transfer,
differ appreciably from the reference flowsheet.
Attempts are being made to reduce the rates at which
thorium, lithium, beryllium, and fluorine are discarded
from the processing plant; the results obtained thus far
are encouraging.
160

15. Processing Chemistry
L. M. Ferris

Certain aspects of the chemistry of the metal transfer
process'*? for the removal of rare earths and other
fission products from MSBR fuel salt were studied
further. Some of this work involved measurement of
the equilibrium distribution of lithium and bismuth
between liquid Li-Bi alloys and molten lithium halide
salts. Other experiments were conducted to determine
whether lead could be substituted for bismuth in any
portion of the system, since the availability of bismuth
is expected to be limited by the year 2000.®> These
experiments included measurement of the equilibrium
distribution of lithium and lead between Li-Pb alloys
and molten LiCl and determination of the solubility of
thorium in liquid Li-Pb alloys. In addition to the studies
relating to the metal transfer process, the reductive-
extraction chemistry of titanium and cesium was
determined, and the reaction of gaseous UF¢ with UF,
dissolved in LiF-BeF,-ThF, (72-16-12 mole %) received
further study, since this reaction provides the basis for
the fuel reconstitution step.-Studies of the precipitation
of protactinium oxide from MSBR fuel salt by sparging
with H; O-HF-Ar gas mixtures were also continued.

15.1 EQUILIBRIA IN FUSED SALT-LIQUID
ALLOY SYSTEMS

L. M. Ferris J. F. Land

Previously,® we studied the equilibrium distribution
of lithium and bismuth between liquid Li-Bi alloys and
molten LiCl, since LiCl is the acceptor salt used in the
metal transfer process!*2 for removing rare earths. The
data were interpreted in terms of the distribution of
salt-like Li3Bi between the two phases.* We concluded
work on this system by measuring the solubility of
Li3 Bi in molten LiCl. The LiCl was equilibrated with a
Li-Bi alloy having a lithium concentration of 70 at. %.
According to the Bi-Li phase diagram,® this alloy was a
two-phase equilibrium mixture of solid Li;Bi and a
liquid Li-Bi alloy in which the lithium concentration
varied from about 52 to 62 at. % as the temperature

1. L. E. McNeese, MSR Program Semiannu. Progr. Rep. Feb.
28, 1971, ORNL-4676, p. 234.

2. D. E. Ferguson et al., Chem. Technol. Div. Annu. Progr.
Rep. Mar. 31, 1971, ORNL4682, p. 2.

3. M. J. Bell, Availability of Natural Resources for Molten
Salt Breeder Reactors, ORNL-TM-3563 (Nov. 11, 1971).

4. L. M, Ferris and J. F. Land, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 207-209.

was increased from 650 to 800°C. It is important to
realize that the activity of the Li;Bi in each phase of
the mixture was the same. The temperature of the
salt-alloy system was randomly varied between 650 and
800°C. A period of at least 24 hr was allowed for
attainment of equilibrium at each temperature before
the salt phase was sampled with a molybdenum pipet.
Quenched samples of the salt were very deep red-brown
in color. The salt samples were analyzed both for “free”
lithium and for bismuth.

The values determined for the solubility of Li;Bi in
molten LiCl, expressed in terms of the respective mole
fractions Ny 4y and Ngjcq), are given in Table 15.1.
Within experimental uncertainty, the ratio of free
lithium to bismuth in the LiCl, NLi(d)/NBi(d)a was 3, as
expected. In Fig. 15.1, the data are compared with the
values obtained by Foster et al.® over the temperature
range 650 to 1000°C using LiCI-LiF (70-30 mole %) as
the salt phase. Within analytical error, the solubility of
Li3Bi is the same in each salt phase. The data can be
represented by log S (mole %) = [3.437 — 4107/T(°K)]
+ 0.06.

Preliminary data have also been obtained on the
equilibrium distribution of lithium and bismuth be-
tween liquid Li-Bi alloys and molten LiBr or LiF. These
data are presented in Table 15.2 as the Li;Bi concen-
tration in the salt phase at various alloy compositions,
and can be represented by the same model developed to
describe the distribution of Li;Bi between Li-Bi alloys
and molten LiCL.* At a given temperature, the distri-

5. M. Hansen and K. Anderko, Constitution of Binary Alloys,
p. 316, McGraw-Hill, New York, 1958.

6. M. S. Foster, C. E. Crouthamel, D. M. Gruen, and R. L.
McBeth, J. Phys. Chem. 68, 980 (1964).

Table 15.1. Solubility of Li;Bi in molten LiCl
in the temperature range 650 to 800°C

Temperature
0, 10°Ngiay  10°Miiey  MeiayVice)
650 1035+ 100 3720 £ 40 3.6+ 0.4
675 157035 5160+ 390 3.3£0.3
700 155010 4935 * 365 3.2£0.3
700 1410+15  4205:165 3.0:0.2
725 2420+ 55 7805 350 3.2£0.2
750 1760+ 10 5610 + 870 3.2:0.5
800 3815:450 108101000  2.8:0.7

161

bution can be expressed as follows: LiCl, T would be about 1.3 at 900°C when LiBr is used
_ as the salt phase. Our earlier results® yield I' = 0.63
log NLisBe(d) with LiCl as the salt phase at 900°C. The above results

indicate that, at a given temperature and alloy compo-

: N{, sition, the equilibrium Li3Bi concentration in the salt
1(m) . 1 . .

= log ; +logI', (1)  phase increases with increasing molecular weight of the

[3- NLi(m)] [3- 3NLi(m)] lithium halide salt. This effect undoubtedly reflects the

corresponding increase in the halide anion radius and,

in which /V is mole fraction and (d) and (m) denote salt  hence, an increased capacity for accommodating the
and alloy phases respectively. The results given in Table large Bi*~ ions from the Li;Bi in the salt lattice.

15.2 yield values of about 0.7 for I' with LiBr as the To determine whether it is possible to use lead instead

salt phase at 650°C, and about 0.2 for " with LiF as the of bismuth in parts of the metal transfer process, we

salt phase at 900°C. Assuming that the temperature made preliminary measurements of the distribution of

dependence of log I' is the same with both LiBr and  lithium and lead between liquid Li-Pb alloys and molten

LiCl at 650°C. The data are summarized in Table 15.3.

JORNL-DWG 72-68794 Both the lithium and lead concentrations in the salt

800 _ISE g' PERATURE (7'82) 650 phase increased regularly with increasing lithium con-

, T | centration in the alloy. The extremely low lead-to-

:\ . lithium ratio in the salt shows that the observed

\ (ANALYSES USED behavior cannot be explained in terms of the distri-

3.0 : i: _ bution of LiPb, LizPb, or Li;Pb between the two
phases. Since the equilibrium lead concentrations in the
+ salt are very low, the primary reaction is the distribu-
tion of lithium between the two phases. A possible
N equilibrium?” is:

2Li(m) = Li,(d) , (2)

3
10 NL|3B| (d)
n
(@)
[ N J

e & p

\ * in which (m) and (d) denote alloy and salt phases
respectively. An equilibrium constant can be written as

»

_NLiyayLiza)

, (3)

. Y 2
H 9.5 10.0 10.5 NLi(m)YLi(m)

19000/

Fig. 15.1. Solubility of Li3Bi in molten LiCl. The line
represents data obtained by Foster et al.® with molten LiCl-LiF

7. M. A. Bredig, personal communication, May 18, 1972.

Table 15.3. Equilibrium distribution of lithium and lead

(70-30 mole %). between liquid Li-Pb alloys and molten LiCl at 650°C
Table 15.2. Distribution of Li3Bi betwe.en liquid Li concentration Cor;::]elljitcr:aitlon .
Li-Bi alloys and molten LiBr or LiF Experiment in alloy (Wt ppm) 10° Ny icay
(at. %) T E——
Li concentration Li Pb

Salt Tem%erature in alloy 106 Ny, Bicay

phase ) at. %) Ehe 112 9.98 63 < 0.1 385:23
105 15.5 35.1 214 £+ 18

LiBr 650 27.3 854+ 15 103 17.8 14.4 0.6 88.3 £ 20
102b 35.2 139 1.4 851115

LiF 900 24.7 9.3:2 102a 50.1 153 78  935%55

46.3 492 + 90 101 53.3 144 12 879 + 30

- 162

in which A is mole fraction and 7 is activity coefficient.
Since

Npi, @y = 4NLic) » (4)
Eq. (3) can be written, with rearrangement, as
_ 2KN{i(m)YLi(m)
Niiy = e (5)
le (d)
or, in logarithmic form,
2K i(m)
log Nyicgy=2logNy ooy tlog | ————1 . 6
Li(d) Li(m) [ @) (6)

If the term containing the activity coefficients is
constant, a plot of log Ny ;4 vs log N i,y would be
linear with a slope of 2. As seen in Fig. 15.2, the
distribution data obtained at 650°C can be represented

ORNL-DWG 72-10400A

1000
/
o |/ é'
/
/
o
500 //
(]
= /LOPE =2
.t °
coo | J
100
//
/
/[
50 /
|
0A 0.2 0.5 q
NLi(m)

Fig. 15.2. Equilibrium distribution of lithium between liquid
Li-Pb alloys and molten LiCl at 650°C.

by Eq. (6), suggesting that the ratio ¥{ ;(m)/TLi,(q) 1
practically constant over the range of conditions investi-
gated. The line in Fig. 15.2 was positioned visually.
Extrapolation of this line to Nyi(m) = 1 yields a value
of about 0.0044 for NLi(d)- Interestingly, this value is
about the same as the value of 0.005 * 0.002 obtained
by Dworkin et al.® for the solubility of lithium in liquid
LiCl at 650°C. It should also be noted that, although
the equilibrium lead concentrations in the salt are low,
the lithium concentrations are about as high as those
obtained with Li-Bi alloys of comparable composition.
Consequently, the use of lead instead of bismuth in the
metal transfer process'*? would not significantly reduce
the rate of loss of lithium from the strip solutions.

15.2 SOLUBILITY OF THORIUM IN LIQUID
Li-Pb ALLOYS

F.J.Smith C.T. Thompson

In order for lead to be considered as a substitute for
bismuth in either the reductive extraction or the metal
transfer portions of the reference flowsheet! ? for the
chemical processing of MSBRs, the solubilities of
thorium in Li-Pb alloys would have to be comparable to
those in Li-Bi alloys. Reported data®''® on the solu-
bility of thorium in lead are in disagreement, and no
information regarding the solubility of thorium in Li-Pb
alloys could be found; consequently, we determined the
solubility of thorium in lead and selected Li-Pb alloys
over the temperature range 400 to 800°C. The appa-
ratus and experimental procedure were the same as
those used in our studies of bismuth systems.! ™' 4

Measured values of the solubility of thorium in lead
and in three Li-Pb alloys are shown in Fig. 15.3. The
solubilities in lead are in fair agreement with but
slightly lower than those reported by Shaffer et al.,’
and are markedly lower than those reported by
Bryner.'® As seen in Fig. 15.3, the thorium solubility
at a given temperature increases regularly with in-
creasing lithium concentration in the alloy. This be-
havior is similar to that observed previously for Li-Bi

8. A. S. Dworkin, H. R. Bronstein, and M. A. Bredig, J. Phys.
Chem. 66, 572 (1962).

9. J. H. Shaffer et al., Reactor Chem. Div. Annu. Progr. Rep.
Dec. 31, 1965, ORNL-3913, p. 43.

10. 1. S. Bryner, TID-7502, Part 1 (1960), p. 230.

11. L. M. Ferris, J. C, Mailen, J. J. Lawrance, F. J. Smith, and
E. D. Nogueira, J. Inorg. Nucl, Chem. 32,2019 (1970).

12. C. E. Schilling and L. M. Ferris, J. Less-Common Metals
20, 155 (1970).

13. F.J. Smith, J. Less-Common Metals 27, 195 (1972).

14. F.J. Smith, J. Less-Common Metqgls 29, 73 (1972).
ORNL-DWG 72-10401A
TEMPERATURE (°C)

900 800 700 600 500 400

10,000 Yy T T T T

™ 1 T T 1

A

N

A\\
[ oA N
N PR
—_ C\\K \
€ -\B\\
(=%
a A
% 1000 \8 % -
= AN
= N\ AN N
o NS S
[ g NI\
3 N\ AN
Q F Y
7z \) N\ \,u
z 0\.\
g °\NN M
© 400 ~ N
= [ W] \\
= O NCa&
= AAW; AN
z g
3 \
S A\
\
2 Li CONC.IN SOLUTION 9
= z
o 10 {ot %) I\
(8] A\
I (o] (0] \\
= ® 6.2 AN
A 27.0 AN
A 44.0 Y
L
1
9 {0 1" 12 13 14 15
1o.ooo/T °K)

Fig. 15.3. Solubilities of thorium in Li-Pb alloys in the
temperature range 400 to 800°C. -

alloys.!® Under all conditions investigated, the solu-
bilities of thorium in Li-Pb alloys were much lower than
those in Li-Bi alloys of comparable lithium concen-
tration. Consequently, Li-Pb alloys are not attractive as
substitutes for Li-Bi alloys in those parts of the
processing system where thorium is involved.

Since practically no thorium is present when the
divalent rare earths are stripped from LiCL'? a
scouting experiment was made to determine whether
Li-Pb alloys could be used as strip solutions for the
divalent rare earths. In this experiment, sufficient
europium was added to liquid Li-Pb (~45-55 at. %) to
produce a solution containing about 15 wt % europium.
Filtered samples taken at temperatures as low as 450°C
indicated that all of the europium was still in solution.
This solubility is far above the rare-earth concentration
(~2500 wt ppm) desired in the divalent strip solution.
Therefore, based solely on solubility considerations,
Li-Pb alloys could be used as strip solutions for the
divalent rare earths.

163

15.3 REDUCTIVE EXTRACTION OF
'TITANIUM AND CESIUM

L. M. Ferris  J. F. Land

Hastelloy N was used for the containment vessel and
other metallic components in the MSRE; however, use
of a modified Hastelloy containing up to 2 wt %
titanium is being considered'® for MSBRs because an
alloy of this type is expected to be more resistant to
radiation embrittlement than Hastelloy N. We would
expect some titanium to be leached from such an alloy
into MSBR fuel salt;'® consequently, the behavior of
titanium in the chemical processing systems was of
interest. Based on thermodynamic estimates,'® we
would expect a large fraction of the titanium to be
removed from MSBR fuel salt as TiF, during fluorina-
tion. The remaining titanium would enter the reductive-
extraction system where, according to our estimates,'®
it would be easily reduced from the salt into liquid
bismuth containing lithium. An experiment was con-
ducted to verify the predicted behavior. Data on the
extraction of cesium were also obtained in this experi-
ment.

The experiment was initiated by simultaneously
hydrofluorinating bismuth and LiF-BeF,-ThF,-CsF
(71.7-15.9-12-0.4 mole %) contained in a molybdenum
crucible at 600°C to remove oxides from the system.
After sparging with purified argon, sufficient titanium
was added to make the titanium concentration in the
bismuth 2500 wt ppm. Analysis of a sample of the
metal phase confirmed that all the titanium was present
in this phase. The two-phase system was then sparged at
600°C for about 20 hr with HF-H, (50-50 mole %).
Subsequent analyses of the respective phases showed
that all of the titanium had been transferred to the salt
phase during the hydrofluorination period. (This result
confirmed our predictions'® that titanium could be
easily transferred from liquid bismuth to a molten
fluoride salt by hydrofluorination.) With the system at
600°C, thorium was added in increments to effect the
reductive extraction of titanium and cesium from the
salt into the bismuth phase. At least 24 hr was allowed
for equilibration after each addition of thorium before
filtered samples of the respective phases were removed
for analysis. We first obtained isotherms (expressed as
usual'' as log Dy, = nlog Dy ; +log K),) for titanium,
cesium, and thorium at 600°C; then, after the bismuth

15. H. E. McCoy et al.,, Nucl Appl. Technol 8(2), 156
(1970).

16. L. M. Ferris, Estimated Behavior of Titanium in MSBR
Chemical Processing Systems, ORNL-TM-3763 (April 1972).
164

Table 15.4. Values of log wa for titanjum, cesium, and thorium obtained
from measurements of the equilibrium distribution of these elements
between molten LiF-BeF,-ThF4-CsF (71.7-15.9-12-0.4 mole %)
and liquid bismuth solutions

Element n Temperature (°C) log Ky Literature values

Ti 3 600 13.6 + 0.4

Cs 1 1.478 £ 0.12

Th 4 8.936x 0.14 9.093 + 0.1 (ref. 11)

Cs 1 650 1.369 = 0.02 1.177 + 0.02 (ref. 18)

Th 4 8.418 + 0.08 8.404 + (.1 (ref. 11)

Cs 1 705 1.235 £ 0.01

Th 4 7.712 £ 0.01 7.726 £ 0.1 (ref. 11)

Cs 1 750 1.193 + 0.02

Th 4 7.309 + 0.07 7.226 £ 0.1 (ref. 11)
phase had been saturated with thorium at 600°C, the 20 ORNL-DWG 72-10398A
temperature of the system was varied between 600 and ' \
750°C. Several sets of samples were taken at three 0 /
temperatures above 600°C. y

/
&
The isotherms obtained at 600°C are shown in Fig. 5 K /e
15.4 as log-log plots of Dy vs Dy ;, and values of log [’Ti3"' y Ad
K}, derived from the data are presented in Table 15.4. 2 :: / /
Vallies of log K¢ and log K7y, at temperatures above . o . /€S+
600 C were calculated using the analytical results for 3 1 A A/
the samples taken at these temperatures. The titanium & s A
data do not clearly show that n = 3; however,  z05 J —~
thermodynamic considerations'® indicate that this is Q;,
3

the proper value for n. The value obtained at 600°C for 9 02 PgS
log K:rif“ confirms that titanium is very easily reduced <~ / I /°/ .;
from MSBR fuel salt and is consistent with the results o1 by e L/ .
of an experiment by Moulton et al.,! 7 which indicated i - v
that log K:riy, should be greater than log K'U 3+ The 0.05 [£ / n
data obtained for cesium can be expressed as log K&s = * o™
—0.5531 + 1771/T(°K), with an estimated uncertainty /
in log Kog of £0.05. The present values of log Ko, 902
indicate that the reductive extraction of cesium will be oo
slightly easier than expected from the prior experi- Ty 2 5§ 40 20 50 400 200 500
mental data obtained at 650°C by Richardson and 10° 0,

Shaffer.'® As seen in Table 15.4, the values of log K'py,
obtained in this experiment are in good agreement with
those reported earlier by Ferris et al.! !

17. D. M. Moulton, J. H. Shaffer, and W. R. Grimes, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1970, ORNL-4548, p.
176. .

18. D. M. Richardson and J. H. Shaffer, MSR Program
Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676, p. 129.

Fig. 15.4. Equilibrium distribution of titanium, cesium, and
thorium between molten LiF-BeF,-ThF4-CsF (71.7-15.9-120.4
mole %) and liquid bismuth solutions at 600°C.

15.4 CHEMISTRY OF FUEL RECONSTITUTION
M. R. Bennett L. M. Ferris

The fuel reconstitution step in the reference flow-
sheet'*? involves (1) absorbing the UF¢ and excess
fluorine from the fluorination step in salt containing
dissolved UF,; and (2) reducing the resultant higher-
valent uranium species to UF; with gaseous hydrogen.
The expected reactions are:

UF¢(g) + UF,4(d) = 2UF5(d) , (7
"hF,(g) + UF4(d) = UF4(d) (8)
2UF;5(d) + Hy(g) = 2UF,(d) + 2HF(g) . 9)

Presently, we are studying the reaction of gaseous UF
with UF, dissolved in LiF-BeF,-ThF, (72-16-12 mole
%), using the gold apparatus and the procedure de-
scribed previously.'?»2° Studies involving UF4-F, gas
mixtures will be made in the near future.

The results of our initial studies’® showed that at
600°C UF, reacted rapidly with UF, dissolved in the
salt. Chemical analyses indicated that UFs; was the
reaction product and that gold was inert both to
gaseous UFg and to dissolved UF5. Recently, in
collaboration with J. T. Bell of the Chemical Tech-
nology Division, we confirmed spectrophotometrically
that U®* was present after adding UF, to a salt
containing dissolved UF,;. Small translucent flakes of
quenched samples of an oxide-free salt that, by chem-
ical analysis, contained about 0.6% U** and 4% US*
were positioned for maximum exposure in the sample
beam of a Cary 14 spectrophotometer, and the spec-
trum was recorded at a scan rate of 25 Afsec. No
evidence for species other than U** was obtained in the
visible region. However, scanning of the near-infrared
revealed three intense and relatively sharp lines at about
1.33, 1.36, and 1.38 u, of which the band at 1.36 1 was
the strongest. The bands cannot be attributed to U%*,
The spectrum obtained was almost identical to the type
Il spectra obtained by Penneman, Sturgeon, and
Asprey?! for pentavalent uranium compounds such as
LiUF,, NaUF¢, and CsUFg4. Thus, the present results
provide independent evidence that dissolved UFs is
indeed the product of the reaction of gaseous UF4 with
UF, dissolved in LiF-BeF,-ThF, (72-16-12 mole %).

All of our previous work had been conducted at
600°C;'?'2% however, we recently conducted two
preliminary experiments at 550°C to determine

19. M. R. Bennett, MSR Program Semiannu. Progr. Rep. Aug.
31,1971, ORNL4728, p. 190.

20. M. R. Bennett and L. M. Ferris, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 212-15.

21. R. A. Penneman, G. D. Sturgeon, and L. B. Asprey,
Inorg. Chem. 3,126 (1964).

165

whether temperature had a marked effect on the rate of
reaction of gaseous UF, with dissolved UF,. The other
experimental conditions were the same as those used in
the work at 600°C. Each experiment was initiated by
dissolving UF, in 200 g of LiF-BeF,-ThF, (72-16-12
mole %) that was contained in a 1.9-in.-diam gold
crucible at 600°C. The system was then treated with
HF-H, (50-50 mole %) at 600°C to remove oxide
impurities. After the system had been cooled to 550°C,
a UF¢-Ar (50-50 mole %) gas mixture was bubbled into
the salt through a Y,-in.-diam gold sparge tube. The
amount of UF¢ (which was metered into the system at
the rate of 0.11 g of uranium per minute) added to the
system was that required by reaction (7) to convert all
of the uranium in the salt to UF;. Samples of the salt
were taken immediately after addition of the UF and
at 24-hr intervals for the following week.

Preliminary analytical results from the experiments at
550°C showed that when the initial UF, concentration
in the salt was about 0.7 wt % only about 50% of the
UF, reacted to form UFs. Since practically quanti-
tative reaction of the UF, was obtained under the same
conditions at 600°C, obviously the rate of reaction (7)
was somewhat lower at 550 than at 600°C. In addition,
the rate of disproportionation of UFs [the reverse of
reaction (7)] appeared to be much lower at 550 than at
600°C.

15.5 PROTACTINIUM OXIDE PRECIPITATION
STUDIES

0. K. Tallent L. M. Ferris

Studies in support of the development of processes
for the selective precipitation of protactinium oxide
from MSBR fuel salt have been continued. The method
currently being considered involves the systematic
precipitation of protactinium from molten LiF-BeF,-
ThF, (72-16-12 mole %) by equilibrating the salt with
HF-H,O-Ar gas mixtures. The apparatus and experi-
mental procedure were described previously.?? The
data obtained are being considered in terms of the
equilibrium

PaFs(d) + %, H, O(g) = %, Pa; O5(s) + SHF(g) (10)

22. 0. K. Tallent and F. J. Smith, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 196.
for which the equilibrium quotient at a given tempera-
ture can be written as
5
PHF

0, = (11)

5/2
PHzONPaFS

In the above expressions (d), (g), (s), V, and p denote
dissolved species, gas, solid, mole fraction, and partial
pressure respectively. If the ratio pl_ho/pH p is fixed at
some value 4, Eq. (11) can be written in logarithmic
form as '

log Npag, = 2.5 log(pyp/4) —log O, . (12)
Thus, if the equilibrium is actually that indicated by
Eq. (10), a log-log plot of protactinium concentration
in the salt vs py; /A should be linear at each tempera-
ture and have a slope of 2.5.

Previous data®?2% were taken at temperatures re-
ported to be 600 and 650°C respectively. However,
during renovation and recalibration of the system, it
was found that these temperatures were in error due to
a faulty constant-voltage unit in one of the Brown
recorders. We estimate that the true temperatures were
25 to 40°C higher than those reported and, conse-
quently, that the previously reported values of O, were
high by a factor of 2 to 3. Unfortunately, a more
accurate compensation cannot be made.

New measurements of , have been made over the
temperature range 535 to 650°C, using LiF-BeF,-ThF,
(72-16-12 mole %) as the solvent. Preliminary data from
these experiments are summarized in Table 15.5. The
protactinium concentrations in the salt were obtained
by gamma-spectrometric analysis for 233Pa and may
require minor revision once the alpha-pulse-height
analyses for 23'Pa are available. The estimated un-
certainty in each value of Q, is £15%. Log-log plots of
the equilibrium protactinium concentrations in the salt
vs py /A, using data obtained at 560 + 3 and 593 #
2°C, are shown in Fig. 15.5. The fact that these plots
are linear with slopes of 2.5 supports the assumption

23. O. K. Tallent and L. M..Ferris, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL4782, pp. 210—12.

24. D. E. Ferguson et al., Chem. Technol, Div. Annu. Progr.
Rep. Mar, 31, 1972, ORNL4794, p. 5.

166

Table 15.5. Equilibrium protactinium concentrations
obtained by sparging LiF-BeF,ThF4 (72-16-12 mole %)
with HF-H,O-Ar gas mixtures under various conditions

Protactinium
Temperature  Pyp/A concentration
Sample (°C) (;{t[:n) in salt ¢
(wt ppm)
1 535 0.0038 19.5 0.17
2 540 0.0040 18.8 0.20
3 540 0.0070 814 0.18
4 546 0.0050 239 0.27
5 546 0.0058 27.8 0.34
6 549 0.0079 48.5 042
7 549 0.0080 58.0 0.36
8 550 0.0063 40.0 0.29
9 560 0.0037 5.3 0.57
10 560 0.0047 11.0 0.52
11 560 0.0056 17.6 0.49
12 560 0.0064 22,0 0.55
13 560 0.0004 29.5 0.40
14 560 0.0068 36.9 0.38
15 560 0.0070 36.9 0.41
16 560 0.0074 35.5 0.49
17 570 0.0080 22.0 0.95
18 573 0.0107 62.7 0.69
19 589 0.0095 24.0 1.34
20 591 0.0113 36.5 1.36
21 592 0.0109 37.0 1.23
22 592 0.0136 60.5 1.30
23 593 0.0093 23.6 1.29
24 593 0.0134 48.0 1.58
25 595 0.0159 71.0 1.64
26 602 0.0167 44 .5 2.96
27 611 0.0175 48.8 3.04
28 6217 0.0155 35.1 3.12
29 638 0.0174 25.8 5.66
30 638 0.0174 26.7 5.47
31 643 0.0342 108.2 7.32
32 646 0.0254 58.1 6.48
33 649 0.0238 308 10.4
34 651 0.0232 31.5 9.46

that Pa, O; is the solid phase at equilibrium as indicated
by Eq. (10). A plot of log @, vs 1/T(°K), using the
recently determined values of Q,, is shown in Fig. 15.6.
The line is defined by the equation log O, = 12.74 —
10,880/T(°K). These values of @, should be regarded as
preliminary; a more complete analysis of the data will
be presented in the next progress report.
167

ORNL~-DWG 72 — 103974 ORNL-DWG 72-8782RA

. 100 TEMPERATURE (°C)
€ 650 600 550
a /
& A | | |
= ]/ 20 N
S 50 / \L
w / AN
T 7/ 7
: Wl R —
z / / -~ AN
= 7 + 20 -5 AW
S / 593 + 2°C 8
p& ) ° ,. ms_/ »
£ 20 7/ o e
3 / : s | N
S—
‘—é " -
O / S hAN
. , 0.5 ~$
Z / ’ 5’(“—.
— .Q
[ L J
a 560 + 3°C 0.2 A\
2 [ - "N
& /
5 e | 0.1
0.002 0.005 0.04 0.02 10.5 .0 1.5 12.0 2.5
pHF/.d {atm) 1o.ooo/r(°K)
Fig. 15.5. Precipitation of protactinium from LiF-BeF,-ThF, Fig. 15.6. Effect of temperature on the equilibrium quotient

(72-16-12 mole %) according to the reaction PaFs(d) +  for the reaction PaFs(d) + 5H,0(g) = 4, Pa,05(s) + SHF(g).
%H,0(g) = %,Pa,05(s) + SHF(g). The slope of each line is 2.5.
168

16. 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. The purification and charging of the salt and the
metal phases to metal transfer experiment MTE-3 were
completed, and operation of the experiment was
initiated. Mass transfer coefficients measured for ra-
dium during the first two runs are in good agreement
with a literature correlation that is based on mass
transfer between aqueous and organic phases in stirred-
interface contactors. The mass transfer coefficient
values measured for lanthanum were considerably lower
than expected, while those for europium were about

10% of those expected. It is believed that these low

values are caused by the presence of an oxide layer at
one or more of the salt-bismuth interfaces in the
system.

Work on the design of the metal transfer process
facility, in which the fourth metal transfer experiment
will be carried out, was continued. Discussions with a
graphite manufacturer indicate that the required graph-
ite components are within the manufacturing capability
of the graphite industry. Studies associated with the
measurement of mass transfer rates in simulated me-
chanically agitated salt-metal contactors were also con-
tinued. The extraction of lead from an aqueous phase
containing Pb(NQO;), into mercury containing zinc at
low concentrations was used for determining mass
transfer coefficients under conditions that are believed
to be similar to those with molten salt and bismuth.
Results of the tests indicate that mass transfer coeffi-
cients in a mercury-water system, and likely in a
salt-bismuth system, are predicted reasonably well by
an existing correlation that is based on mass transfer in
aqueous-organic systems.

Further progress was made in determining the rates of
transfer of ?7Zr and 237U from molten salt to bismuth
in an 0.82-in.-ID by 2-ftlong packed column. A
correlation developed from our data shows that the
height of an overall transfer unit based on the bismuth
phase is constant and has a value of about 4.3 ft for
extraction factors near unity. Developmental work
relative to the design of the reductive-extraction process
facility was continued. The feasibility of flaring com-
mercially available molybdenum tubing in %-, %-, and
'h-in. sizes was demonstrated. Molybdenum fittings
that are compensated for differential thermal expansion
were designed and tested. The design of the molyb-

denum reductive-extraction equipment and the process-
ing materials test stand is essentially complete.

Bismuth analyses of portions of fluoride salt taken
from several different engineering experiments indicate
that an improved sampler design is required in order to
avoid contamination of the samples. Samples with-
drawn from the fluoride salt surge tank in metal
transfer experiment MTE-3 have consistently shown
low bismuth concentrations (0.45 to 1.7 ppm).

Studies were completed in which we demonstrated
the feasibility of using autoresistance heating of molten
salt in continuous fluorinators protected against corro-
sion by means of a frozen wall. Operation of the first
series of experiments in the uranium oxide precipitation
facility was completed, and the precipitator vessel was
disassembled. Corrosion of the equipment and agglom-
eration of oxide on the equipment surfaces were found
to be negligible. A series of tests was completed with an
eddy-current-type salt-bismuth interface detector using
the amplitude measurement technique at 550 and
700°C. The detector output was found to be linear and
reproducible over most of the probe length.

16.1 OPERATION OF METAL TRANSFER
EXPERIMENT MTE-3

E. L. Youngblood  W.L. Carter
H. O. Weeren L. E. McNeese

The purification and charging of the salt and metal
phases to metal transfer experiment MTE-3 were
completed, and operation of the experiment was
initiated in order to measure the rate at which rare
earths transfer through the system. A flow diagram for
the experiment is shown in Fig. 16.1. The equipment
used in the experiment consists of a fluoride salt
reservoir containing about 30 liters of fluoride salt
(72-16-12 mole % LiF-BeF,-ThF,), a salt-metal con-
tactor, and a rare-earth stripper containing 4.6 liters of
4.9 mole % Li-Bi solution. The 10-in.-diam contactor is
divided into two compartments that are interconnected
at the bottom by a pool of bismuth that initially
contained 0.13 mole % thorium. During operation of
the experiment, salt is circulated from the fluoride salt
reservoir to one compartment of the contactor at the
rate of about 33 ml/min, and LiCl is circulated between
the other compartment and the stripper at about 1.25
liters/min; these flow rates are 1% of those expected for
169

ORNL-DWG 71-{47RA

LEVEL AGITATORS
VENT { ELECTRODES LEVEL
h ELECTRODES
| _.— FLUORIDE VENT
SALT PUMP
ARGON
SUPPLY
ARGON
33 em¥min SUPPLY
="
-
Ot te ettt et et e Tt b
72-16-12 mole %
LiF-BeF,-ThF,
FLUORIDE SALT-METAL RARE EARTH
SALT CONTACTOR STRIPPER
RESERVOIR

Fig. 16.1. Flow diagram for metal transfer experiment MTE-3.

processing a 1000-MW(e) MSBR. Mechanical agitators
having 2%-in.-diam blades are used in each phase to
contact the salt and metal without causing dispersion of
the phases. The system operates at about 650°C. A
more detailed description of the equipment was pre-
sented previously.!2

To begin operation of the system, the rare earths were
added to the fluoride salt reservoir; then the rate of
transfer of the rare earths across the three salt-metal
interfaces was determined from samples of each phase.
Overall mass transfer coefficients were calculated from
the resulting data by using the general equation shown

below:
Cm
“5 3

j=K(CS—

(1)
where
j = transfer rate, gcm ™2 sec ™',
K = overall mass transfer coefficient, g cm ™2 sec !
(g/em’),
C, =rare-earth concentration in the salt phase,
glem?,
C,, = rare-earth concentration in the metal phase,
g/em?,

1. E. L. Youngblood, H. O. Weeren, and L. E. McNeese, MSR
Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782,
pp- 221-24.

2. E. L. Nicholson et al., MSR Program Semignnu. Progr.
Rep. Aug. 31, 1971, ORNL-4728, pp. 209-12.

D = distribution coefficient, C,,/C;, at equilibrium.

The first two runs in the MTE-3 equipment were
made using only the 228Ra (which distributes similarly
to a divalent rare earth) that was present in the fluoride
salt as a decay product of thorium. After these runs had
been completed, 510 g of LaF; was added to the
fluoride salt reservoir to produce a lanthanum concen-
tration of 0.15 mole %, and two additional runs were
made. A quantity of ' $*Eu (14 mCi) was then charged
to the fluoride salt reservoir. After addition of the
'54Eu, transfer rates for lanthanum and europium
could be determined for each run; however, the rate for
radium could no longer be measured because of
counting interference. Agitator speeds not exceeding
200 rpm have been used in all of the runs made thus
far, since it is desirable to obtain all of the data needed
at the lower speeds before using higher speeds that
could result in transfer of salt between the two
compartments of the contactor via entrainment in the
circulating bismuth.

The overall mass transfer coefficients that have been
determined for radium, lanthanum, and europium
during a total of six runs at agitator speeds of 100 or
200 rpm are summarized in Table 16.1. Mass transfer
coefficients measured in this experiment have been
compared with values predicted by a mass transfer
correlation developed by Lewis,® who studied mass
transfer rates in aqueous-organic systems. Based on a

3. ). B. Lewis, Chem. Eng. Sci. 3,248-59 (1954).
literature survey of available mass transfer correlations
and a limited amount of mass transfer data obtained
using aqueous solutions and mercury, the Lewis corre-
lation appears to be the best available correlation for
predicting mass transfer coefficients for experiment
MTE-3, even though the correlation was developed for
aqueous-organic systems having considerably different
physical properties from those of salt-bismuth systems.
The Lewis correlation allows calculation of the indi-
vidual mass transfer coefficients for each of the two
phases present at a mechanically agitated interface from
the physical properties of the two phases in contact and
from the Reynolds numbers based on the agitator
diameters. Overall mass transfer coefficients can be
calculated from the individual coefficients and the
distribution coefficient by using the following equa-
tion:

- )

where

K = overall mass transfer coefficient, g cm ™2

sec ! (g/em?),
k., k, = individual mass transfer coefficients, gcm™
sec ! (g/cm?),

D = distribution coefficient, C,, /C..

2

Using the overall mass transfer coefficients measured
for radium, lanthanum, and europium in MTE-3, the
individual mass transfer coefficients were calculated by
using Eq. (2) and the Lewis correlation. 1n the
calculation, it was assumed that the ratio of the two
individual coefficients at an interface was the same as
that predicted by the Lewis correlation. It was neces-
sary to use estimated values for the coefficient for the
distribution of radium between the fluoride salt and the

Table 16.1. Mass-transfer coefficients? measured
in metal transfer experiment MTE-3

. Agitator speed Ky K, K3
Material (rpm) {cm/fsec) (cmfsec) (cm/sec)
Radium 100 0.00007 0.03 0.003
200 0.00032 0.12 0.013
Europium 100 0.00005 0.002 0.0009
200 0.00026 0.009 0.003
Lanthanum 200 0.00005 0.00003 0.008

@K, measured at fluoride salt—Bi-Th interface; K, measured
at lithium chloride—Bi-Th interface; K3 measured at lithium
chloride—Li-Bi interface.

170

bismuth-thorium phase since values have not been
determined experimentally. A comparison of the indi-
vidual mass transfer coefficients from experiment
MTE-3 with values predicted by the Lewis correlation is

shown in Fig. 16.2, where the term of the Lewis
correlation containing the individual mass transfer
coefficient is plotted against the term containing the
Reynolds numbers and the viscosities of the two phases.
The dashed line represents values predicted by the
Lewis correlation. The lines labeled k| and k, designate
the individual coefficients between the fluoride salt and
Bi-Th solution, and the lines labeled k3 and k4
designate the individual coefficients between the LiCl
and Bi-Th solution.

The values for the individual coefficients show the
same general dependence on Reynolds number as that
suggested by the Lewis correlation; however, the
experimentally determined values of mass transfer
coefficients obtained for lanthanum are considerably
lower than predicted by the Lewis correlation, while

“the values for europium are only about 10% of those

ORNL-DWG 72-784BRA

‘05 ! ! 1 :ETng T 1 T'I'[!'[ll 1 %_ é%;;}!!!
LEWIS CORRELATION
60k _,_ H2y165 -6
=7 —1=(Re(+Rep F1) 6.76X40 =
Y
104 oAt
ko i 1]
A #3111
[}
,Z’/
103 k4ol »
7 ~ko 1
/; |2| i
yAmy
" Qly 2/\k3
T S e
=, 102 P /d L, MY H
o III ‘k 1 o 11
© o7 4flE‘k2 H
V4
[ ]
[ )
A
A
10! At
7 “
rd -I
o o RADIUM s
107 E="e EUROPIUM ;i
— a LANTHANUM i £,
!
101 i
102 103 109 109 108
Re.+ Re, 2
g1 e o

‘Fig. 16.2. Correlation of mass transfer coefficient data for
radium, europium, and lanthanum from metal transfer experi-
ment MTE-3.
predicted. It is thought that the low transfer rates
observed for europium and lanthanum may be caused
by the presence of an oxide film at one or more of the
interfaces (as the result of impurities in the system). We
believe that the rate of transfer of lanthanum and
europium c¢an be increased considerably by either
cleaning the system to remove these films or by
increasing the agitator speed sufficiently to break them
up. '

16.2 DESIGN OF THE METAL TRANSFER
PROCESS FACILITY

W.L.Carter E.L.Youngblood L. E. McNeese

We have previously® described the preliminary design
for the metal transfer process facility (MTPF), in which
the fourth metal transfer experiment (MTE-4) will be
carried out. This experiment will use salt and bismuth
flow rates that are 5 to 10% of those required for
processing a 1000-MW(e) MSBR. The principal equip-
ment items are: (1) a fluoride salt surge tank, which has
a volume of about 300 liters and will consist of a
carbon steel liner in a stainless steel vessel; (2) a
three-stage salt-metal contactor made of graphite and
enclosed in a stainless steel containment vessel; (3) a
stainless steel vessel using a graphite or carbon steel
liner in which rare earths will be accumulated in a
lithium-bismuth solution having a volume of about 100
liters; and (4) a hydrofluorinator that consists of a
graphite crucible enclosed in a stainless steel vessel and
has a volume of about 150 liters.

During this report period, discussions were held with
a graphite manufacturer in order to ensure preparation
of the optimum design for the graphite portions of the
salt-metal contactor with regard to ease of fabrication
and accepted design technology for vessels made of this
material. Minor design changes suggested by the manu-
facturer have been incorporated in the design, and
samples of the suggested grade of graphite have been
obtained for compatibility testing with molten salts and
bismuth containing lithium and thorium at concen-
trations of about 0.002 and 0.0025 mole fraction,
respectively. Fabrication of the graphite parts of the
contactor appears to be within the manufacturing
capability of the graphite industry, and the preliminary
quotation of $8500 (f.o.b., the manufacturing site) for
the completed assembly is considered to be acceptable.

4. W. L. Carter et al., MSR Program Semiannu. Progr. Rep.
Feb. 29, 1972, ORNL4782, pp. 224-25.

171

No further design work for the facility will be carried
out until the compatibility studies have been completed
and additional data have been obtained on mass transfer
rates in stirred-interface salt-metal contactors.

16.3 DEVELOPMENT OF MECHANICALLY
AGITATED SALT-METAL CONTACTORS

J.A.Klein H.O.Weeren L. E. McNeese

Mechanically agitated nondispersive salt-metal con-
tactors are being considered as alternatives to packed
columns for MSBR processing systems. Such contactors
are 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. In addition, it is
believed that contactors of this design may be more
easily fabricated than packed columns. Another poten-
tial advantage is that adequate mass transfer rates may
be obtained without dispersing the salt or metal phases.
This mode of operation should eliminate the problem
of entrainment of bismuth in salt streams and is highly
important in the case of the fluoride salt stream, since
an MSBR will be constructed of a nickel-base alloy that
is subject to attack by metallic bismuth.

Since the important physical properties of molten
salt-bismuth systems (viscosities, densities, and density
difference) are quite similar to those for the mercury-
water system, studies made with simulated contactors
using mercury and water will allow examination of the
hydrodynamic and mass transfer characteristics of this
type of contactor under highly desirable conditions for
experimentation. We have previously summarized re-
sults of hydrodynamic studies,® and more recently we
have reported preliminary results on the measurement
of mass transfer coefficients for the extraction of silver
from an aqueous phase containing AgNO; at low
concentrations into the mercury phase.® During this
report period, our work was aimed at finding materials
that distribute between aqueous and mercury phases in
a manner suitable for quantitative study of factors
affecting mass transfer rates in mechanically agitated
contactors. After a number of candidate materials had
been examined, the extraction of lead from an aqueous
phase into mercury containing zinc at low concen-

5. H. O. Weeren and L. E. McNeese, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 207-9.

6. H. O. Weeren and L. E. McNeese, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 225-27.
trations, according to the reaction

Pb(NO,), [H, 0] + Zn[Hg] - Pb[Hg]
+Zn(NOs), [H,0], (3)

was found to be quite satisfactory. Problems associated
with formation of interfacial solids were encountered
initially; however, they were circumvented by the
addition of sufficient HCl to the aqueous phase to
produce an HCl concentration of about 4 X 107> N.
This quantity of acid prevents formation of solids at the
interface without causing precipitation of PbCl, or
formation of significant quantities of ZnCl,.

Four series of transient batch-extraction experiments
were performed to determine the variation of mass
transfer rate with changes in agitator speed and cell
geometry. In all cases the cell was circular, had a
diameter of 6 in., and contained about 1 liter of each of
the phases. The ‘initial concentrations of zinc in the
mercury and Pb(NO;), in the aqueous phase were each

172

0.1 M. Both baffled and nonbaffled cells were used.
Agitator diameters of 3 and 5 in. were employed. The
agitator blades were straight in three series of experi-
ments; in the fourth, they were canted at a 45° angle.
Samples of the aqueous phase were obtained at intervals
during each run and were analyzed for both zinc and
lead by the atomic absorption technique. The overall
and individual mass transfer coefficients for the transfer
of lead and zinc were then determined.

Results of these studies are presented in Fig. 16.3.
The range of literature values for aqueous-organic
systems and a correlation derived by Lewis’ using data
on the rate of mass transfer between various aqueous
and organic liquids are also included in the figure, The
correlation, an empirical equation, is represented by the
following expression:

60k,

6.76 0°¢ / =2 o 1
=6.76 X 10~ + +1, 4
o, 1 \Rel Re, 5 (4)

7. J. B. Lewis, Chem. Eng. Sci. 3, 248—59 (1954).

ORNL-DWG 72-8808A

1000 ~
7
/
/
500 /
//.
/
1/2
/|
//‘ .
Vi 8
200 LEWIS f
CORRELATION/ A .
_ / /
] / )
S 100 A=A 1q/4 /3
o y/d /
xja ) ;________‘
yi
A~
/
50 /' P
/ v _
RANGE OF EXPERIMENTAL VALUES
FROM SOLVENT - WATER SYSTEM
20 -
1 3in. STRAIGHT BLADES NO BAFFLES
2 3in. STRAIGHT BLADES 4-1in. BAFFLES
3 5in. STRAIGHT BLADES NO BAFFLES
4 3in. CANTED BLADES NO BAFFLES
10 1 1 1 1 [ | i A | 1 1 1
109 2 5 10° 2 5 108

n

Fig. 16.3. Correlation of mass transfer coefficient data from a 6-in.-diam simulated salt-metal contactor. The mass transfer
coefficients were measured during the extraction of lead from a Pb(NQj3), solution into mercury containing zinc.
where the subscripts refer to the two phases and

k = individual mass transfer coefficient in indicated
phase, cm/sec,

v = kinematic viscosity of indicated phase, cm?/sec,

Re = Reynolds number based on agitator in indicated
phase = D*N/v,

D = diameter of agitator in indicated phase, cm,

N = agitator speeds, rps,

1

sec !,

n = viscosity of indicated phase, gcm ™
This relationship indicates that the rate of mass transfer
in a mechanically agitated nondispersive contactor is
dependent only on eddy transport and is independent
of the molecular diffusivity of the transferring material
in either of the phases. It is seen that the data obtained
in the present study using 3-in.-diam straight blades in a
baffled cell are in good agreement with the Lewis
correlation, as would be expected. It is also seen that:

1. For the same value of the Reynolds number func-
tion, both straight and canted agitator blades pro-
duce approximately the same value for the indi-
vidual mass transfer coefficient. However, the use of
canted blades allows operation at slightly higher

agitator speeds.

The mass transfer rate is increased by the use of four
1-in. vertical baffles. The mass transfer coefficients
obtained with the baffled cell have values that are
two to three times those obtained with an unbaffled
cell. The use of baffles also tends to reduce the
formation of a central vortex in the cell and permits
operation at higher agitator speeds.

. As the agitator diameter is increased, the mass
transfer rate decreases at a fixed value for the
Reynolds number function; however, the use of
agitators with larger diameters results in a "higher
value for the Reynolds number function at a fixed
agitator speed. Consequently, the individual mass
transfer coefficient may remain approximately con-
stant as the cell and agitator sizes are increased.

The main conclusions reached thus far in the study are:

1. The lead-zinc system is quite satisfactory for deter-
mining mass transfer rates in simulated stirred-
interface salt-metal contactors using mercury and
water.

2. The individual mass transfer coefficients in a water-
mercury system (and probably in a salt-bismuth
system) are predicted reasonably well by the Lewis

correlation.

173

3. A baffled contactor or a contactor configuration
that. performs as a baffled system (such as a
semicircular or square cell) is highly desirable.

16.4 REDUCTIVE-EXTRACTION
ENGINEERING STUDIES

B. A. Hannaford W.M.Woods L. E. McNeese

Three mass transfer experiments were performed in
which the rates of transfer of *’Zr and *3*7U from
molten salt to bismuth were measured by adding the
respective tracers to the salt phase prior to contacting
the salt with bismuth containing reductant in an
0.82-in.-ID by 2-ft-long packed column. The fractions
of ®7Zr and 237U transferred ranged from 36 to 80% at
extraction factors ranging from 1.3 to 6.3. The overall
HTU values based on the bismuth phase ranged from
4.7 to 7.3 ft. The overall HTU based on the bismuth
phase appears to be constant and equal to about 4.3 ft
for extraction factors near unity.

Mass transfer experiments UZTR-2, -3, and 4. Three
mass transfer experiments (UZTR-2, -3, and -4) were
made in which the rates of transfer of both 2*”U and
®7Zr were measured. These experiments were separated
by suitable time intervals to permit the decay of
residual 237U (half-life, 6.75 days) activity and to
permit the adjustment of the uranium and zirconium
distribution coefficients to the desired values before
each experiment.

Experiment UZTR-2 was carried out with a nominal
volumetric flow rate ratio of unity. Control of the
bismuth flow rate was not satisfactory throughout the
entire run because of a malfunctioning controller;
however, a 20-min period of steady flow was obtained
with bismuth and salt flow rates of 172 and 187
ml/min respectively (see Table 16.2). Three sets of
flowing stream salt and bismuth samples (i.e., a total of
six samples) were taken during this period; the reported
237U and ® 7 Zr concentrations were constant to within
+6% for these samples. As noted in footnote a of the
table, the results for uranium transfer were calculated
by using correction factors (determined by analyses of
samples from run UZTR-3) that take into account the
attenuation of the weak 0.208-MeV gamma in the solid
bismuth and solid salt samples. The data reported
previously for experiment UZTR-1 were corrected
similarly. The results for runs UZTR-1 and -2 indicated
a relatively narrow range of calculated HTU values (i.e.,
4.1 to 5.1 ft).

Following run UZTR-2, 1 gequiv of beryllium was
electrolytically dissolved in the salt and bismuth in the
treatment vessel in order to increase the distribution
coefficients for zirconium and uranium significantly.
Analyses of the equilibrated salt and bismuth phases for
uranium (by the wet analysis method) and 237U (by
gamma spectrometry) showed a substantial disparity;
including the results from samples taken after both
UZTR-1 and -2, the value of the Di34y/Dy ratio
ranged from 0.23 to 0.42 rather than being unity (the
expected value).

- In order to obtain accurate data conceming the
inventory of >37U tracer in the system, the bulk of the
salt was transferred from the treatment vessel to the salt
feed tank, and a nominal 10-mCi charge of ?*7U was
added to the salt. After a uniform concentration of
tracer had been obtained, the salt was transferred to the
treatment vessel where the bismuth and salt were
equilibrated and sampled. The combined %37 U activity
in the bismuth and salt was found to be only about 63%
of the initial activity in the salt phase; this result is in
good agreement with measurements made at the con-
clusion of experiment UZTR-2, where the corre-
sponding 237U balance was 68%. This confirmed our
suspicion that significant attenuation of the 0.208-MeV
gamma was occurring during counting of the solid salt
and bismuth samples in the stainless steel sampler
capsules. The bismuth and salt samples (that had been
counted previously) were drilled out using a small lathe

174

and were dissolved for total uranium and for 237U
analyses. Counting results for 237U on the resulting
solutions produced a tracer balance of 101 to 106%
based on analyses of samples from the salt feed tank to
which the tracer was initialy added. The resulting
attenuation of the weak 0.208-MeV gamma was about
63 and 29% for the bismuth and salt samples respec-
tively. Despite the good tracer balance, the counting
results for the dissolvent solutions were still somewhat
biased; the 37U activity per microgram of uranium in
the dissolved bismuth samples was only about 83% of
the activity per microgram of uranium in the dissoived
salt samples.

The bismuth and salt were transferred through the
system to check out the operation of the bismuth flow
control system after modifications (new argon control
valve trim and addition of capacitance to the pneumatic
signal line to the valve) had been made to correct the
previous difficulty with the control system. Significant
improvement in the operation of the system was noted.
Experiment UZTR-3 was then carried out at bismuth
and salt flow rates of 239 and 61 ml/min respectively;
the bismuth flow rate deviated from the programmed
rate by a maximum of about 7% during a 7-min period
midway during the run. All samples taken during the
experiment were dissolved for determination of 237U
and total uranium. On the basis of the actual weight of

Table 16.2. Summary of results from %7Zr and 237U tracer experiments

Bismuth flow

Salt flow  Distribution

Fraction of

Run Tracer rate rate ratio Ex;;:tc;ion Zr (or 1) P:’;;;J
(ml/min) (ml/min) (volumetric) transferred
ZTR-2 Zr 232 70 0.282 0.93 0.30 3-5
ZTR-3 Zr 93 168 1.67 0.92 0.30 4.6
ZTR-5 Zr 147 158 0.21 0.195 0.14 0.5-1.6
ZTR-7 Zr 181 105 0.852 1.47 0.44 4.3
ZTR-9 Zr 45 277 3.40 0.55 0.17 4.6
ZTR-10 Zr 46 283 3.78 0.61 0.22 4.7
UZTR-1 Zr 143 161 0.40 0.36 0.12 4.5
e 0.49 0.43 0.16 4.1
UZTR-2 Zr 172 187 1.37 1.26 0.36 4.8
ue 1.43 1.31 0.36 5.1
UZTR-3  Zr 239 63 1.58 6.01 0.80 6.9
ub 1.67 6.35 0.80 7.3
UZTR4  Zr 233 77 0.964 2.93 0.66 4.7
ub 1.25 . 3.79 0.67 6.2
237

@Uranium transfer results were calculated from

U concentrations after they had been corrected

for attenuation of the 0.208-MeV gamma in the solid samples.

bUranium transfer results were calculated from

using dissolved samples of bismuth and salt.

37y concentrations, which were determined
material dissolved for solution counting of the
0.208-MeV gamma, the calculated attenuation was 59%
for the solid bismuth samples and 26% for the solid salt
samples. Analyses of ten pairs of equilibrated post-run
samples of bismuth and salt for uranium by wet
chemical means indicated a Dy; of 1.89, which is in
excellent agreement with the 1.93 value inferred from
237U counting of the same samples. The HTU values
for zirconium and uranium were 6.9 and 7.3 ft
respectively (see Table 16.2).

Following the usual time interval for decay of ?*7U
activity in the system, mass transfer experiment
UZTR-4 was carried out. Somewhat lower values for
the zirconium and uranium distribution coefficients
were obtained in this run as the result of a slow loss of
reductant from the bismuth in the treatment vessel
between experiments. The rate of reductant loss prior
to run UZTR-3 was about 0.58 meq/hr; the rate of loss
prior to run UZTR-4 was about 0.26 meq/hr. Measure-
ments made thus far of the rate of loss of reductant in
the system generally fall within this range. The results
for run UZTR-4 were calculated from four pairs of
flowing stream samples that were not affected by a
mid-run deviation in bismuth flow rate of about 12%.
This deviation in flow rate was the result of a
malfunction of the ramp generator-transducer.

Following run UZTR-4, 10 g of beryllium metal was
electrolytically dissolved in the treatment vessel in
order to increase Dy  to about 300 prior to the next
mass transfer experiment with ®7Zr and 2?7U. This
experiment was postponed, however, when a bismuth
line failure occurred that resulted in the release of
about 1 liter of fluid (bismuth plus salt) from the
system. The appearance of the failure suggested that it
was caused by the expansion of freezing bismuth. The
affected line had been subjected to about 18 freeze-
thaw cycles. The line was replaced, and salt and
bismuth containing reductant were then circulated
throughout the system to remove trace oxides that
might have been introduced to the system. The com-
bined salt and bismuth phases were then contacted with
a 10-liter/min 30% HF-H, stream for 15 hr to remove
oxide from the salt.

Correlation of mass transfer results. Mass transfer data
from all tracer experiments made to date® are sum-
marized in Table 16.2 and are plotted (except data for
run ZTR-5) in Fig. 16.4 as suggested by a correlation
that was devised to predict the relationship between the

8. B. A. Hannaford, D. D. Sood, W. M. Woods, and L. E.
McNeese, MSR Program Semiannu. Progr. Rep. Aug. 31, 1971,
ORNL4728, pp. 212—-14.

175

ORNL-DWG 72-8784RA

10
5
o A
\I'_’._‘
—_ A
]
lul“" ) /9
+ /
})
Ok
1 S
Kl o %7zr
- 237, -
7 AU
| L
05 [
-1 0 | 2 3 4 5 6

Fig. 16.4. Correlation of data on fraction of tracer extracted
() in terms of the extraction factor (£) for the transfer of °77:
and 227U from MSBR fuel salt to bismuth. The line corre-
sponds to an HTU value of 4.3 ft.

extraction factor (E) and the fraction of material
extracted (f). The overall HTU based on the bismuth
phase, which is inversely proportional to the slope of
the line through the points in Fig. 16.4, was calculated
from the data in this figure to be 4.3 ft. Data from runs
UZTR-3 and -4 diverge from the other data in the
expected direction since the fraction of a material
extracted should approach a constant value as the
extraction factor is increased.

16.5 REDUCTIVE-EXTRACTION PROCESSING
FACILITY DEVELOPMENT

W.M.Woods W. F.Schaffer,Jr. L. E. McNeese

Construction of the reductive-extraction processing
facility (REPF) described previously® will require joints
in molybdenum tubing, some of which must be made in
the field. While it is feasible to weld molybdenum, the
weld-affected region is embrittled, and the weld region
must be supported in order to ensure that stresses
remain acceptably low. Such welds are expensive; thus
some near-standard-type tubing fitting would simplify
construction of equipment for engineering development
studies.

During this report period, we found that it is feasible
to flare arc-melted, low-carbon, low-oxygen molyb-

9. W. M. Woods, W. F. Schaffer, Jr., and L. E. McNeese, MSR
Program Semiannu. Progr. Rep. Feb. 29, 1972, ORNL4782,
pp. 230-31.
denum tubing (available commercially) in %-, %-, and
', -in. sizes by using a standard flaring tool. The tool is
assembled to the tubing, and the tool-and-tubing
assembly is heated to 275°C. Subsequently, the flaring
cone is run in until the force required begins to increase
sharply; then the parts are reheated. Three or four
reheats have usually been sufficient for completing a
flare. It is anticipated that a tool with a built-in heater
can be developed to facilitate the flaring.

We have not determined which type of flare fitting is
the most effective. One could, of course, use an
all-molybdenum fitting. However, molybdenum-on-
molybdenum threads tend to gall, and it is difficult to
tap threads in a molybdenum nut. Thus, it would be
desirable to fabricate the nut from some material other
than molybdenum. On the other hand, because of the
very low coefficient of thermal expansion of molyb-
denum (0.36% from room temperature to 600°C),
almost any other potential material would expand more
than the molybdenum, and the joint would loosen upon
being heated. ~

W. F. Schaffer, Jr., suggested that if a ferrule of
suitable length (i.e., a ferrule that expanded more than
either molybdenum or the material of the nut) were
placed underneath the nut so as to be in compression, it
should be possible to compensate for the differential
axial thermal expansion. Analysis of this suggestion
leads to the following relationship for the relative
lengths of the parts needed for compensation:

Le=[(py =P Of =PI L)y, (5)

where Ly and L, are the lengths of the ferrule and the
body of the molybdenum fitting, respectively, and p is
the percent expansion from the reference temperature
to 600°C for nut, molybdenum, and ferrule, as indi-
cated by the subscripts. While the suggested method
would compensate for differential axial expansion,
there are two further sources of loosening, both of
which are caused by differential radial expansion. If the
ferrule were made to fit the conical surface of the
reverse side of the tubing flare, differential radial
expansion of the ferrule would lead to an axial
loosening (assuming that the surfaces slid smoothly on
each other). Similarly, because of the 60° angle of the
threads, differential expansion of the nut in the radial
direction would cause a loosening in the axial direction.

Both of the expansion effects can be compensated for
by making the ferrule sufficiently long. The relationship
for compensation of each of these two effects is:

AL;= [, — )l Bp - PIDID coter,  (6)

176

where AL is the additional length of ferrule needed; D .
is the appropriate diameter (the pitch diameter in the
case of the nut, the average diameter in the case of the
ferrule); and « is one-half the included angle of the flare
for the ferrule and 30° for the nut, assuming a 60°
thread. Compensation for radial expansion of the
ferrule may be obviated by employing a molybdenum
washer shaped to fit the reverse side of the tubing flare.
Because cot 90° = 0 in this case, no compensation is
needed. |

Since it is not at all obvious that surfaces that are
clamped together, as in a tightened flare fitting, will
slide freely on each other, a set of fittings compensated
only for axial expansion was designed. The bodies of
the fittings were made of molybdenum (p,, = 0.36%
from room temperature to 600°C). The ferrules were
fabricated of 304 stainless steel (py = 1.12%), and the
nuts were made of low-carbon steel (p,, = 0.84%). Test
assemblies were prepared in -, ¥%-, and '%-in. sizes,
using a 45° flare. Figure 16.5 shows the unassembled
parts for the test pieces; Fig. 16.6 shows the test pieces
after assembly. Each assembly was placed inside a
stainless steel enclosure, and the interiors of the
assemblies were pressured to 65 psig with helium. The
containers for the assemblies were connected to a
vacuum pump, vacuum pressure gages, and a helium
leak detector; then the units were placed in a furnace
and subjected to thermal cycling between 300 and
600°C.

Helium leak tests showed that all of the units were
leak-tight at room temperature before the start of the
test and remained leak-tight at 300°C. However, the
%-in. unit leaked grossly after two cycles to 600°C. It
should be noted that the %-in. tubing was less ductile
than the tubing in the other two sizes. For example, it
was not possible to obtain a full flare without cracking
the ¥%-in. tubing. The "4-in. unit did not leak during the
two thermal cycles; although the '“-in. unit had a
measurable leak to helium, it would not have leaked
molten salt or bismuth.

After all the fittings had been retorqued at 275°C,
thermal cycling was continued. Gross leakage was
observed for the ¥%-in. unit after two additional cycles
to 600°C and for the %-in. unit after four additional
cycles to 600°C. No leakage was observed for the ' -in.
unit before the test was terminated.

From the results of these tests we conclude that the
general idea of temperature-compensated fittings is
sound but that compensation for radial differential
expansion is necessary. Consequently, a set of test units
has been designed using a 37° flare, a molybdenum
washer under a type 304 stainless steel ferrule, and a
PHOTO 1800-72

Fig. 16.5. Unassembled test pieces for testing molybdenum fittings with 45°-flared molybdenum tubing in sizes of "4, %, and % in.

PHOTO 1801-72

Fig. 16.6. Assembled test pieces for testing molybdenum fittings with 45°-flared molybdenum tubing in sizes of %, %, and % in.

type 410 stainless steel nut. The constants for this 16.6 DESIGN OF A PROCESSING MATERIALS
design are: p,, = 0.36% from.room temperature to TEST STAND AND THE MOLYBDENUM
600°C; py = 1.12%: p,, = 0.74%. The fittings have been REDUCTIVE-EXTRACTION EQUIPMENT
designed to be fully compensated for both axial and

radial expansion. When fabrication has been completed, WogSchalier Ir

they will undergo thermal cycling between 300 and As the design phase for the all-molybdenum reduc-

600°C until failure occurs. tive-extraction system nears completion and the as-
sembly phase begins (a cooperative effort between the
Chemical Technology and the Metals and Ceramics
Divisions), we feel that a review of the status of the
program is apropos.

We have completed the design drawings (including
revisions necessary to simplify or improve assembly,
fabrication, and welding and brazing techniques) for the
following program phases: (1) all-molybdenum system
components, {2) connecting molybdenum piping and
fittings, (3) freeze valve assembly, (4) structural sup-

178

porting components, brackets, and braces for the

molybdenum system; (5) assembly fixture and trans-
port jig for the molybdenum system, (6) containment
vessel, including the top flange with the transition
nozzles for connecting the molybdenum system to the
required service and instrument lines, (7) internal
insulation to ensure that the top flange will not exceed
the 750°F design temperature, (8) fixtures for lifting
and lowering the molybdenum system into the contain-
ment vessel, and (9) containment and operating cell
structural modifications and containment vessel sup-
port. We have also examined the design and stress
calculations for all critical regions to ensure success of
the system. The design of the containment vessel has
been reviewed and approved by the Pressure Vessel
Committee.

The final machining of all the molybdenum com-
ponents has been completed, and these components
have been inspected and stress-relieved by the Metals
and Ceramics Division. All other parts required for the
reductive-extraction system assembly, including the
freeze valve, vessel supports, and associated hardware,
nozzle inserts, flange assembly, and assembly jig, have
been fabricated and delivered to the Metals and
Ceramics Division. A trial assembly of the components
on the assembly fixture with the connecting piping and
supporting structural system was made and checked
against the design drawings. No significant problems
were found; all line dimensions were within the
specified tolerances. The remaining design work pri-
marily involves the electrical wiring for the containment
vessel and transfer line heaters, the instrumentation and
control panels, and the services.

Following completion of the assembly jig, the frame
was test loaded to measure torsional twist and was
found to be in close agreement with the calculated
values. We prepared procedures and checklists to cover
the movement of the completed molybdenum system
from the assembly area (on the second floor of Bldg.
4508) to the operating area (on the third floor of Bldg.
4505). The procedures were reviewed with the field
engineer, the rigger foreman, and technical personnel.

The wood-and-steel mockup of the system was con-
nected to the assembly jig, along with welded test
specimens of molybdenum tubing supported only by
the ends to the loop support pipe in several strategic
locations. Strain gages for measuring both bending and
torsion and two accelerometers for measuring hori-
zontal and vertical acceleration were installed on the
frame. A transfer of the mockup between the two
working areas was made successfully without damage to
the mockup or molybdenum specimens. Most of the
shock loads were less than ', g. One shock load of
about 2 g was recorded when the assembly frame was
inadvertently mismatched with the trunnion support. A
minor adjustment of the parts and procedures will
prevent a recurrence of this problem.

The present schedule calls for completion of assembly
of the molybdenum system during the early part of
November 1972. Preparation of the cell where the
equipment will be installed is scheduled to begin in
September. The containment vessel which is currently
being fabricated will be installed during October.

16.7 REMOVAL OF BISMUTH
FROM MSBR FUEL SALT

R. B. Lindauer

A program has been initiated to determine the
concentration of bismuth in fuel salt after the salt has
contacted bismuth in the various engineering develop-
ment experiments. The objective of this program is to
determine the probable concentration of bismuth in salt
leaving an MSBR processing plant. Later, methods for
preventing entrainment of bismuth in salt will be
studied, and means for removing bismuth from salt will
be evaluated.

An important part of this program is the development
of effective techniques for obtaining salt samples from
experimental systems. Table 16.3 summarizes the re-
sults of analyses of samples taken from six vessels, using
various types of samplers. The lowest reported bismuth
concentrations are believed to be the most representa-
tive because of difficulty with contamination of the
sample, either during the sampling procedure or during
the sampler cleaning operation. The highest reported
bismuth concentrations were obtained when the sam-
plers were withdrawn through sample ports used for
obtaining both bismuth and salt samples, as in experi-
ment MTE-2B and in the salt-metal contactor for
experiment MTE-3.

Our experience with the various types of samplers
used to date can be summarized as follows:
179

1. Stainless steel vacuum sampler (filtered). This
sampler, which is the standard one used in the
engineering experiments, is disadvantageous in two
respects: difficulty has been experienced in crushing the
salt from the % -in.-OD tubing, and there is a possibility
that bismuth will be removed from the sample by
filtration or adsorption on the filter as the salt passes
through the porous metal.

2. Quartz tubing sampler (unfiltered). The sample is
withdrawn into the tubing (approximately '4 in. ID)
with a vacuum syringe. Although this sampler resulted
in the lowest reported bismuth concentrations for
samples taken from experiment MTE-2B, it is too
fragile for routine use.

3. Mild-steel dip sampler (unfiltered). Several sub-
mergences, as well as rapid movement, of the sampler
are usually required to fill the %-in.-OD cup. We were
unable to obtain samples from experiment MTE-2B. It
is feared that the rapid movement of the sampler might
cause entrainment of bismuth in the salt in systems
where a sample must be taken in the vicinity of a
salt-bismuth interface.

4. Mild-steel vacuum sampler (unfiltered). This %-
in.-OD sampler proved to be reliable but was difficult to
clean. Evidently, bismuth adhered to the outside of the
sampler during passage through the sampler port. This
material is not readily removed by use of a file or emery
cloth. Some decrease in the reported bismuth concen-
tration was observed after the sampler cleaning tech-

nique was modified. In the new technique, the salt-
containing portion of the sampler is not clamped in the
lathe, and new tools are used for cutting open each
sampler.

5. Steel-sheathed vacuum sampler (unfiltered). A
removable steel sheath is placed over the sampler to
prevent contact of the sampler with the sample tube.
This sheath increases the sampler diameter by ' in. and
prevents the %-in.-diam sampler from being used
except in the treatment vessel for the mild-steel
reductive-extraction system; the values for the bismuth
concentration obtained with the new type of sampler
were much lower than had been obtained with previous
samplers. Salt from the Y,-in.-diam sampler could not
be removed by crushing, and the required drilling
operation resulted in the introduction of iron particles
in the sample. Dissolution of these particles with the
salt produced a yellow color which reduced the
sensitivity of the analytical method for detection of
bismuth.

6. Copper vacuum sampler (filtered). This is the most
recently evaluated sampler, and the initial results are
very encouraging. This sampler has produced the lowest
reported bismuth concentrations obtained to date, it is
easy to clean, and the salt is readily removed by
crushing. The halfwave potential for copper is suffi-
ciently separated from that of bismuth to eliminate
interference -in the analysis for bismuth by the inverse
polarographic method (which had been reported ini-

Table 16.3. Average? concentrations of bismuth in fluoride salt samples taken from engineering experiments

Average bismuth concentration (ppm)

Reductive extraction MTE-3 Sampling
Sampler )
MTE-2B Treatment Salt from Fluoride Contactor experiment
tank column tank 4 (no Bi phase)
Stainless steel vacuum (filtered) 58 (6)
Quartz (unfiltered) 11.5(7)
Mild steel dip (unfiltered) 60 (5 0.6 (1) 355 (2)
Mitd steel vacuum (unfiltered); 25 (1) 74 (1) 1.7 (1) 60 (5) 2.8(2)
old cleaning technique
Mild steel vacuum (unfiltered); 775 (2) 23(2) 36 (6) 0.45 (4) 29 (2) 1.0 (2)
new cleaning technique
Steel-sheathed vacuum (unfiltered)
Y, in. OD (drilled out) 77.5 (2) 200 (2)
% in. OD (crushed) 7.3(3)
Copper, vacuum, filtered <0.1 (3)
Averages 120 (18) 40 (1D 36 (6) 0.7 (2) 133 (11 0.8(N

2The number of samples on which an average is based is indicated in parentheses after the value.
tially) unless the concentration of copper is much
higher than the concentration of bismuth.

Relatively high bismuth concentrations (7.3 to 775
ppm) were obtained for all samples that were with-
drawn through a sample port also being used for
obtaining bismuth samples. Only salt samples are
withdrawn through the port on the fluoride salt vessel
of experiment MTE-3, and the reported bismuth con-
centrations in these samples have been consistently low
(0.45 to 1.7 ppm) and comparable to those taken from
the sampling experiment that does not contain a
bismuth phase. It is believed that the actual concen-
tration of bismuth in the fluoride salt in the various
experiments is considerably lower than the reported
values obtained thus far.

16.8 FROZEN-WALL FLUORINATOR
DEVELOPMENT

J. R. Hightower, Jr.

When an open-column continuous frozen-wall fluori-
nator is operated with nonradioactive salt, a heat source
that is not subject to corrosion by the combined action
of molten salt and fluorine must be present in the salt.
Two methods for internal heat generation have been
considered: radio-frequency induction heating and elec-
trolytic or autoresistance heating using 60-Hz power.
The induction heating method was judged to be inferior
to autoresistance heating because of a narrow range of
acceptable operating conditions and because of the
complexity of the required power supply, power
controls, and power transmission equipment.10

In initial tests on autoresistance heating in a 2.5-in.-
diam simulated fluorinator, the desired mode of opera-
tion could not be achieved because the frozen salt layer
near the upper electrode was not electrically insulat-
ing.10 A study of the requirements for forming
electrically insulating frozen salt layers was then ini-
tiated in a straight 6-in.-diam cylindrical vessel; in these
tests, the high-voltage electrode was placed just below
the salt surface at the top of the vessel rather thanin a
side arm (as is planned in an actual fluorinator). These
tests have been successful in defining conditions under
which electrically insulating salt films can be formed.

Equipment and procedures. As described pre-
viously,!® the equipment consisted of a vessel made
from 6-in. sched 40 nickel pipe. The high-voltage
electrode entered the vessel through a nozzle in the top

10. J. R. Hightower, J1., “Frozen Wall Fluorinator Develop-
ment,” MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL-4782, pp. 230-34.

180

flange and protruded about 9 in. into the salt. The
fluorinator vessel, which was grounded, served as the
other electrode. The test section consisted of a 48-in.-
long section of the pipe that had been cooled by
removing the thermal insulation after the desired initial
operating conditions had been achieved; this allowed
heat to be transferred to the surroundings by natural
convection and radiation.

In a typical experiment, the thermal insulation was
removed from the test section, and the wall tempera-
tures in this region were allowed to drop to about
360°C (which is below the salt solidus temperature) in
order to form a layer of frozen salt on the vessel wall in
the test section. A 60-Hz voltage applied across the
electrodes was then adjusted to produce the desired
current through the molten salt (and, hence, the desired
heat generation rate). After steady-state conditions had
been achieved, the molten salt was drained from the
vessel, which was then allowed to cool to room
temperature. The thickness of the frozen salt layer was
subsequently determined from measurements made
through the top flange and from radiographs of the test
section.

Results. We have made six runs in which a layer of
frozen salt was successfully maintained on the vessel
wall in the test section under steady-state conditions.
The resulting heat generation rates have varied from 9,2
to 16.7 kW/ft® of molten salt. These rates are close to
the rate expected in the primary fluorinator in an
MSBR processing plant (i.e., 12.6 kW/ft®) and are
adequate for operation of a nonradioactive continuous

fluorinator protected against corrosion by means of a

frozen wall. The measured thickness of the frozen salt
layer has agreed well with the value predicted by heat
transfer calculations and can be inferred with reason-
able accuracy by measuring the total resistance of the
conducting salt path.

It was found that an electrically insulating frozen salt
layer can be formed reliably if the temperature of the
wall is maintained below that of the salt solidus. In the
initial experiment in the first test vessel,10 wall
temperatures that were only about 25°C below the
liquidus temperature (458°C) were used. Under these
conditions the frozen material on the wall contained
occluded liquid, which caused the layer to be electri-
cally conductive. The salt mixture that was initially
charged to the present fluorinator simulation had the
composition 65-35 mole % LiF-BeF,, which has a
solidus temperature of 363°C. An electrically insulating
layer of frozen salt could be maintained on the walls of
the test vessel only when the temperature of the wall in
the test vessel was maintained below 363°C. During the
181

Table 16.4. Comparison of calculated and experimentally measured values for the thickness
of molten-salt layers produced by autoresistance heating in a simulated fluorinator

Heating Heat Calculated film thickness (in.) Measured film
Run current generation Based on electrical Based on thickness
(A) (W) (kw/ft?) resistance heat transfer (in.)
14 44 2250 16.7 1.33 1.35 1.36
15 55 2760 10.34 1.11 1.15 b
16 60 2890 9,24 0.97 1.04 b
17¢ 60 3150 11.7 1.09 ' 0.98 1.12
18 55 2830 11.5 1.11 1.10 1.19
20 50 2600 13.5 1.30 1.27 1.439
22 55 2600 11.3 1.16 1.16 1.18

9Based on an average of calculated film thickness values.
byessel could not be drained after the experiment; no measurement was made.

heating.

dMeasurements from radiograph are questionable.

course of the experiment the composition of the salt
was changed to 67.4-32.6 mole % LiF-BeF, as the
result of additions of lithium fluoride and the removal
of BeF, by the inadvertent introduction of small
amounts of water and the subsequent precipitation of
BeO. This salt mixture has solidus and liquidus tempera-
tures of 458 and 464°C respectively. During the later
experiments, wall temperatures in the test section as
high as 380°C were employed, and no difficulty was
encountered in maintaining an electrically insulating
frozen salt layer on the vessel wall.

Experimentally determined values for the thickness of
the frozen salt layer during the autoresistance heating
tests are compared with calculated values in Table 16.4.
In each of the runs, steady-state conditions were
" maintained for 1%, to 4 hr before the salt was drained
from the test vessel. The calculated values given in
Table 16.4 are based on (1) the electrical resistance of
the molten-salt core and (2) the rate of heat transfer
through the frozen salt layer on the wall of the test
section. Literature values for the electrical resistivity of
the molten saltl! and the thermal conductivity of the
frozen salt!'? were used in the calculations, and the
agreement between the calculated and measured thick-
ness values is quite satisfactory. Specific heat generation
rates (kilowatts per cubic foot of molten salt in the test
section) bracket the specitic heat generation rate

11, G. D. Robbins, “Electrical Conductivity,” p. 14 in
Physical Properties of Molten-Salt Reactor Fuel, Coolant, and
Flush Salts, ed. by S. Cantor, ORNL-TM-2316 (August 1968).

12. J. W. Cooke, MSR Program Semiannu. Progr. Rep. Aug.
31,1971, ORNL4728, p. 41.

Steady-state condition was not reached; the frozen salt melted near the electrode after 11/2 hr of

expected in the primary fluorinator in an MSBR

-processing plant (12.6 kW/ft*).

Operation of the fluorinator simulation was quite
stable at steady-state conditions, and only minimal
control of the current (applied manually) was required.
No evidence of cracking of the electrically insulating
salt layer was noted. In most of the runs the frozen salt
layer at the top of the test section was much thicker
than that at points lower in the test section because of
radiative heat transfer from the salt surface to the cold
flange of the test vessel. The layer just below the salt
surface was usually thinner than in other parts of the
test section because of a high heat generation rate (500
W in a 3-in. length) produced by a small cartridge heater
in the upper electrode. Addition of a heat shield below
the flange and replacement of the electrode heater with
one having a lower heat generation rate (200 W over a
12-in. length) resulted in a specific power in the vicinity
of the electrode which closely matched that in the test
section and resulted in more nearly uniform thickness
values adjacent to the upper electrode.

Additional studies of autoresistance heating are re-
quired in a fluorinator simulator containing a side arm
in which the high-voltage electrode will be placed. We
have designed equipment for carrying out such studies
in a fluorinator simulator having a 6-in.-diam side arm
and a 36-in.-long by 6-in.-diam vertical section. This
series of experiments will use MSBR fuel carrier salt
(72-16-12 mole % LiF-BeF,-ThF,) and should consti-
tute the final series of experiments before design of the
facility for studying continuous fluorination of molten
salt in a frozen-wall system is initiated.
16.9 URANIUM OXIDE PRECIPITATION STUDIES
M.J.Bell D.D.Sood L.E.McNeese

The first series of uranium oxide precipitation experi-
ments was concluded during this period. Two experi-
ments, OP-12 and OP-14, were performed at 600°C.
Before the start of experiment OP-12, 89 g of LiF
238UF, eutectic was added to the system to compen:
sate for the decrease in the uranium concentration in
the salt that had been observed during experiments
OP-7 through OP-11. In experiment OP-12, more than
907 of the uranium initially in the salt was precipitated
as oxide, and a final uranium concentration of 830 ppm
was attained. Analyses of salt samples obtained from
inside and outside the draft tube during the experiment
exhibited a large disparity in uranium concentration,
indicating that the draft tube may have been restricted.

In experiment OP-14, 98% of the uranium was
precipitated as UO,, and a final uranium concentration
of 230 ppm in the salt was achieved. A salt sample
obtained from inside the draft tube during the experi-
ment contained an unusually low concentration of
uranium, again suggesting that the draft tube was
restricted. Also, during these two experiments, the
concentration of uranium in the salt at the end of each
of the hydrofluorination steps continued to decrease.
As a result, it was decided to terminate the series of
experiments in order to determine the cause of the
experimental difficulties. Accordingly, the salt was
transferred to the feed tank, and the heaters were
deactivated. The precipitator vessel was removed from
the system and disassembled. A salt heel (Fig. 16.7)
containing a large quantity of undissolved oxide was
found in the precipitator vessel. The dark material
around the top edge of the heel was oxide, and contains

182

35 wt % uranium oxide and 30 wt % thorium oxide.
This rim extended along the sloped section at the
bottom of the vessel to a height which was above the
bottom of the draft tube. Apparently, the 45° slope in
the bottom of the precipitator was not entirely effec-
tive in directing the oxide to a point beneath the draft
tube. The dark material in the bottom of the heel was
about % in. thick and had a volume of approximately
60 cm®. The material was heterogeneous, ranging in
color from yellow-green, characteristic of solid solu-
tions with high ThO, content, to dark brown, charac-
teristic of UO,. The light material on top of the sample
was salt. The depression in the interface between the
salt and oxide layers indicated the location of the line
used for transferring salt between the precipitator vessel
and the receiver tank.

A photograph of the gas inlet and draft tube assembly
is shown in Fig. 16.8. The metal surfaces are bright and
clean, and machining marks made during fabrication are
still visible. A small quantity of oxide powder was
present on some of the draft tube surfaces, but there
was no evidence of oxide agglomeration or large oxide
deposits. No evidence of corrosion or HF attack was
observed despite exposure of the equipment to fuel salt
and HF-water mixtures at temperatures ranging from
550 10 630°C over a period of nine months.

Experience gained in the operation of the facility
continues to indicate that oxide precipitation is an
attractive alternative to fluorination for removing ura-
nium from protactinium-free MSBR fuel salt. The solids
formed in the experiments had a higher UO, content
than was calculated to be in equilibrium with the salt,
indicating that only a single stage may be required to
perform the separation. The high-uranium solids that
are formed in the present equipment have a diameter of

PHoTo 199772

INCHES

Fig. 16.7. Photograph of salt and oxide heel removed from precipitator vessel at termination of operation.

183

Fig. 16.8. Photograph of gas inlet and draft tube assembly

50 + 10 u. Particles of this size have a terminal velocity
of 12 + 5 ft/hr, which is sufficiently large to permit
rapid settling, even in full-scale equipment. Corrosion of
the nickel equipment by HF was not observed under
the present experimental conditions. Problems en-
countered with inability to circulate the oxide in the
present equipment are believed to be the result of using
the same gas inlet for the precipitation and hydrofluori-
nation operations, and should be resolved by making
slight modifications in the design of the equipment.
Experience gained with this equipment also indicates
that mechanical agitation could be employed as an
alternative circulation measure. Further experiments
with larger-scale equipment are required to determine
the effect of scale on these results.

16.10 DEVELOPMENT OF A BISMUTH-SALT
INTERFACE DETECTOR

H.0.Weeren  C.V.Dodd!?

An eddy-current-type 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

13. Metals and Ceramics Division.
14. H. O. Weeren et al,, MSR Program Semiannu. Progr. Rep.
Aug. 31,1971, ORNL4728, pp. 222-25.

removed from precipitator vessel at termination of operation.

of a ceramic form on which bifilar primary and
secondary coils are wound. Contact of these coils with
molten salt or bismuth is prevented by enclosing them
in a molybdenum tube. In operation, a high-frequency
alternating current, which is passed through the primary
coil, induces a current in the secondary coil. The
magnitude and relative phase of the induced current are
dependent on the conductivities of the materials lo-
cated adjacent to the primary and secondary coils; since
the conductivities of salt and bismuth are quite dif-
ferent, the induced current reflects the presence or the
absence of bismuth. The principal problem associated
with this type of detector stems from the high electrical
conductivity of molybdenum, the fabrication material
of the protective sheath. Two approaches 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.

The completed probe has been installed in a three-
chambered test vessel made of carbon steel; both the
probe design and the test vessel design were described
previously.'# The upper chamber is a reservoir for
molten bismuth; the middle chamber contains the
sheathed probe; and the lower chamber, which simu-
lates the interior of the high-temperature containment
vessel for the molybdenum reductive-extraction equip-
ment (see Sect. 16.6), contains 13 ft of high-tempera-
ture electrical cable in an inert atmosphere. Bismuth
can be transferred between the upper and middle
chambers to vary the level around the probe; this level
can be measured with a bubbler—mercury manometer
system and compared with probe readings.

Initial results from phase shift measurements were
reported previously.!> A series of level determinations
was made at two temperatures (550 and 700°C) by the
amplitude measurement technique to calibrate the
temperature compensation circuit. The results (see Fig.
16.9) show that at each temperature the probe readings
were linear and reproducible for bismuth levels between
4 and 12 in. However, both the manometer and the
probe gave erratic readings at bismuth levels below 41in.;
this behavior may have resulted from the accumulation
of some sort of material at the interface during the
period of probe operation.

The amount of temperature compensation required to
make the probe and manometer readings coincide at
700°C was found to vary almost linearly with bismuth
level. This indicates that level readings using amplitude
measurement can best be made by determining calibra-
tion curves for the operating temperatures of interest
and using these curves. rather than a temperature
compensation circuit.

15. H. O. Weeren et al., MSR Program Semiannu. Progr. Rep.
Feb. 29, 1972, ORNL-4782, pp. 239-40.

184

ORNL - DWG 72 - 40399A

14 s
//.
2 14
prt 700 °C 2 K4
w
z
o / 550°C
u<[J (¢} ' —
m S /
8 /]
2 8 -
T / /]
> [ ]
o /d/
E 6 v
o / Ve
o A— .
z <
E 4 P —
2 BISMUTH LEVEL READINGS
@ BY AMPLITUDE MEASUREMENT
2
o)
o) 2 4 6 8 10 12 14

BISMUTH DEPTH BY MANOMETER READINGS (in.)

Fig. 16.9. Variation of bismuth depth as indicated by level
probe vs bismuth depth as determined from manometer
readings.

The probe was tested at a temperature of 550°C for a
period of five days. During this time, the bismuth level
did not change; however, the probe reading varied
slowly from an initial value of 9.6 in. to a low of 9.4 in.
and a high of 10.1 in. The reason for this fluctuation is -
not known.
17.

185

Continuous Salt Purification

R. B. Lindauer

We previously! described a facility in which studies
relative to the continuous purification of molten
fluoride mixtures are carried out. In the past the system
has been operated in a semicontinuous manner, with
about 14 liters of salt being fed through a packed-
column gas-salt contactor at flow rates ranging from 50
to 500 cm?3/min. During this report period, the system
was modified to provide for the continuous circulation
of a small volume (about 4 liters) of salt through the
packed column. The new equipment consists of a check
valve pump that has tungsten ball checks and an orifice
head pot having liquid level instrumentation for meas-
uring the salt flow rate. The pump is actuated by a
cyclic variation of the argon pressure in the pressure
vent line between the check valves by the use of
solenoid valves and electrical timers.

Initial tests with the modified system disclosed the
following:

1. Instrumentation is required to indicate the position
of the liquid-gas interface in the pressure vent line of
the pump between the two check valves. Probes will
be installed which, in addition to indicating the

interface location, will actuate relays to close the
vent valve on high level and to open the pressure
valve on low level.

. An erratic pumping rate, which could have been
caused by the presence of oxide in the salt in the
pressure vent line, was noted. The oxide was
probably formed during periods when the system
was not being operated. A new pump that has
double suction and discharge check valves is being
fabricated. This pump should continue to operate
even when the salt contains some particulate matter.
A bismuth phase will also be used to provide a seal
between the pressurizing argon and the salt in the
pressure vent line in order to reduce the possibility
of oxide formation. This will increase the maximum
salt displacement from 20 to 80 cm® per cycle.

. The equipment for determining the salt flow rate
performed quite satisfactorily.

1. R. B. Lindauer, MSR Program Semiannu. Progr. Rep. Feb.
29, 1972, ORNL-4782, pp. 241—-43.
3

R 12

: .
l |
i ‘ 1
. [ |
h 1
187 {
. fi
LK
e
AUGUST 31, 1972
M.W.ROSENTHAL, DIRECTOR A
R. B.BRIGGS, ASSOCIATE DIRECTOR D
P.N. HAUBENREICH, ASSOCIATE DIRECTOR R
H.R. BEATTY,” BUDGET B
DESIGN STUDIES
COMPONENTS & SYSTEMS DEVELOPMENT PHYSICS MATERIALS MSBR PROCESSING DEVELOPMENT CHEMISTRY
E.5.BETTIS R DUNLAP SCOTT"* R A.M.PERRY" R . L. E. McNEESE™* cT .
S 1 DITTO" \&C -M. H. E. McCOY M&C W. R, GRIMES RC
M. I, LUNDIN® R .
HASTELLOY N STUDIES CHEMICAL DEVELOPMENT
MSR PHYSICS ANALYSIS -
H. E. McCOY** M&C L. M. FERRIS cT
J. R. ENGEL R
R. E, CLAUSING* M&C M. R.BENNETT cT
PUMP DEVELOPMENT
MSBR DESIGN STUDIES D.$. EASTON® M&C F. J.SMITH* cT REACTOR CHEMISTRY
A.G.GRINDELL® R R.E. GEMLBACH M&C 0. K. TALLENT cT C.F.BAES, JR.* RC
L.G. ALEXANDER R W. R HUNTLEY" a MSR PHYSICS METHODS W. R HUNTLEY** ] J.F_LAND cT C. F.BAMBERGER RC
0.W.BURKE* 1&C L. V. WILSON® R o. L.SMITH R J.W. KOGER"* ME&C C.T. THOMPSON cT E.G. BOHLMANN' ) RC
H. A. McLAIN®® R 1 E. WOLFE® R B. McNABB M&C S. CANTOR RC
J. A, TALLACKSON 1&C o J. W.wOo0DSs " M&C E. L. COMPERE RC
H.L.WATTS R p— ENGINEERING DEVELOPMENT R G. ROSS RC
MSBR CORE DESIGN . HITE STUDIES L. E. McNEESE** - J pusr?niegs AC
. . . JENNINGS, JR. RC
H. F. BAUMAN"® R W.P. EATHERLY M&C M. J. BELL cT W.P. TEICHERT G
COMPONENTS & SYSTEMS 0.B.CAVIN®* M&C W. L.CARTER cT R.M. WALLER RC
HEAT TRANSFER AND PHYSICAL PROPERTIES W.H. COOK* M&C B.A. HANNAFORD cr
.. D. M. HEWETTE* M&C J. R. HIGHTOWER, JA. cT
. . DUNLAP SCOTT R
H. W. HOFFMAN R 0 MSBR EXPERIMENTAL PHYSICS ‘ C. A. KENNEDY" M&C C.W. KEE cT
A.H. ANDERSON" 14C G.L RAGAN R T.G.KOLLIE® M&C 4. A. KLEIN cr PHYSICAL AND INOR
- . N ISTRY
3. W. COOKE R C. BRASHEAR® 18C R. W. McCLUNG* MEC R.B.LINDAUER cr c GANIC CHEMISTR
J.J.KEYES® R C.J.CLAFFEY® R D. L. McELROY* M&C W.F. SCHAFFER, JR cr J.BRAUNSTEIN® RC
T.5.KREsS R J. L. CROWLEY R J.P. MOORE® MEC D. B. 50001 cr G.D. BRUNTON® RC
S. .. CLAIBORNE" R JOSE A.D. DIEGUES""" R C.B. POLLOCK MEC H.O. WEEREN" cT F.A.0QOSS RC
B.J. SUTTON" R C. H.GABBARD R CHEICAL FROCESSING FATERIALS E. L. YOUNGBLOOD cT L. 0. GILPATRICK RC
G. M. WINN R W. M. WOODS cT A.P. MALINAUSKAS® RC
R.H.GUYMON"* R FUNDAMENTAL RADIATION DAMAGE J. R. DISTEFANO M&C J. BEAMS s cr J. D. REDMAN RC
R. REID R IN GRAPHITE 0.B.CAVIN** MEC €. H. BROWN CcT D. M. RICHARDSOM RC
P.G. HERNDON 1&C N.C. COLE® M&C V. L. FOWLER cT R.A. STREHLOW RC
R.J. KEDL® R T.S. NOGGLE* sS 1. W. KOGER** MEC R. O. PAYNE CT L. M. TOTH RC
A. 1. KRAKOVIAK R A E. McDONALD® MEC T.0.ROGERS? cT C. F. WEAVER RC
E.L.BIDDLE R A 1 MOORHEAD® MEG J.S8. TAYLOR cT
T.5.KRESS®® ] n .
A.C.SCHAFFHAUSER M
H. A McLAIN®® R S.M.CHR" ss &c
J.P. SANDERS® R TECHNICAL SUPPORT
F.C. ZAPP® R M.D. ALLEN® M&C
NSULTA N.M. ATCHLEY*® M&C
co NT E.R.BOYD" MaC ANALYTICAL CHEMISTRY DEVELOPMENT
F.N.PEEBLES ut B. R. CHILCOAT MA&C
J.W. CHUMLEY M&C
S.W. COOK M&C
C. E. DUNN* M&C RESEARCH AND DEVELOPMENT
J.C. FELTNER u:g A.S.MEYER® AC
J.W.GREER R, F. APPLE AC
MSRE SURVEILLANCE -F.
J. L. GRIFFITH M&c J. M. DALE AC
A. H. GUYMON"" R cw HOU%I;' :zg D. L. MANNING AC
4. D. HUDS J. P, YOUNG
F.P. JEFFERS® M&C - Ac
H. KEATING® M&C ANALYSES
E.J. LAWRENCE M&C
W. J. MASON® M&C L. T.CORBIN® AC
B. W. McCOLLUM® M&C C.E. LAMB" AC
R.5.PULLIAM"® Mac
L.G.RARDON" M&C
W. H.SMITH, JR.* M&C
G.D. STOHLER® M&C
H. R. TINCH M&C
AC ANALYTICAL CHEMISTRY DIVISION L.R. TROTTER" M&C
B BUDGET AND PROGRAM PLANNING QFFICE J. F, WILLMERING™ M&C
C CHEMISTRY DIVISION 3. J. WOODHOUSE* MRC SPECIAL ASSISTANCE

CT CHEMICAL TECHNOLOGY OIVISION

D DIRECTORS DIVISION

1&C  INSTRUMENTATION AND CONTROLS DIVISION

M&C METALS ANO CERAMICS DIVISION

A REACTOR DIVISION

AC REACTOR CHEMISTRY DIVISION

S§S SOLID STATE DIVISION
UT UNIVERSITY OF TENNESSEE

* PART TIME ON MSRP
°* DUAL CAPACITY

e PARTICIPANT FROM INSTITUTE OF ATOMIC
ENERGY, SAQ PAULO UNIVERSITY, BRAZIL

11T GUEST SCIENTIST FROM INDIA
CO-0P

M. A. BREOIG"
H.R.BRONSTEIN"
A.S. DWORKIN®

OO0