ORNL-5011.txt

£

ORNL-5011

3 445k 0515101 1 /.

Molten-Salt

Reactor Program

\

\

I
Semiannual Progress Report

jor Deriod Snding August 31, 1974

LIBRARY LOAR PY

OAK RIDGE NATIONAL LABORATORY

OPERATED BY UNION CARBIDE CORPORATION e 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 22161
Price: Printed Copy $7.60; Microfiche $2.26

This report was prepared as an account of work sponsored by the United States
Government. Neither the United States nor the Energy Research and Development
Administration, 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-5011
UC-76 — Molten-Salt Reactor Technology

Contract No. W-7405-eng-26

1 MOLTEN-SALT REACTOR PROGRAM
B SEMIANNUAL PROGRESS REPORT
FOR PERIOD ENDING AUGUST 31, 1974

. L. E. McNeese
: Program Director

- JUNE 1975

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
for the

ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION
ARTIN ENERGY RESEARCH LIBRARIES

DT

3 445k 0515101 9

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
ORNL-4037
ORNL-4119
ORNL-4191
ORNL-4254
ORNL-4344
ORNIL.-4396
ORNL-4449
ORNL-4548
ORNL-4622
ORNL-4676
ORNL-4728
ORNL-4782
ORNL-4832

This report is one of a series of periodic reports that 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
Period Ending August 31, 1972

Contents

INTRODUCTION ... ... . i, e vii
SUMM A R Y e e e e e o ix
~ PART 1. MSBR DESIGN, DEVELOPMENT, AND SAFETY
1. DESIGN AND SYSTEMS ANALY SIS . ..o e e e e e e 2
1.1 ORNL Study of MSR Designs .. ....ouuiiit ittt e e e e e 2
1.1.1 Plant and Equipment Studies ... .......... ... e 2
1.1.2 Neutronic Analysis . . ... ov vttt et e et ettt e 2
1.1.3 Xenon Behavior ... ... ... . e 3.
1.2 Tritium Behavior in Molten-Salt Systems . . ... ... e 3
1.2.1 Reactor Calculations .. ... ... ... .. e 3.
1.2.2 Studies of Loop Experiments . . ...... .. ... . . . . 4
1.3 Molten-Salt Reactor Information System . . ... .. .. .. i 5
2. SYSTEMS AND COMPONENTS DEVELOPMENT .. ... ... .. i 6
2.1 Gaseous Fission Product Removal .. ... ... .. . .. . . . . . e 6
2.1.1 Bubble Generator ........ ... ... e 6
2.1.2 Bubble Separator ... .. ... e e 6
2.1.3 Bubble Formation and Coalescence Test . .............. .ttt 6
2.1.4 Bubble Size Measurement . . . ... .. e 7
2.1.5 Mass Transfer to Circulating Bubbles . ......... ... ... . .. ¥/
2.1.6 Bubble Generation Analysis . ... ........ ... .. . .. ... 8
2.2 Molten-Salt Steam Generator Industrial Program ....... e e 9
2.2.1 Resumption of the Conceptual Design Study ......... ... ... . ... ... ... ... ...... 9
2.2.2 Preliminary Results ... ..o 9
2.3 Gas Systems Technology Facility .. ... .. i e e et i n. 10
2.4 Coolant-Salt Technology Facility ....... e e e AP 10
2.5 Forced Convection Loop (MSR-FCL-2and 2b) ....... ... ., 14
2.5.1 Introduction . ... ... ... e P, 14
2.5.2 Operation of MSR-FCL-2 with Fluoroborate ............ ... ... .. .. ... . . ... ... 14
2.5.3 Post Run Inspections ............... e e e 14
2.5.4 Cleanup and Modifications . . ... .. ... . e 15
2.5.5 Purging the System and Adding Salt ......... ... ... . . . . 15
2.5.6 Operation of MSR-FCL-2b with Fuel Salt ........... ... ... . .. 16
3. REACTOR SAFETY . ...ttt e e e e e S 18
3.1 Safety Eventsin MSBR ... .. e e e e 18
3.2 MSBR Neutronic EXCUISIONS . .. ..o e e e e e e 18

1ii

iv

PART 2. CHEMISTRY

4. FUEL SALT CHEMIST RY ... .. i e e e e et e e 22
4.1 The Chemistry of Tellurium in Molten Li,BeF, ....... ... ... . . . i 22
4.2 Exposure of Metallurgical Samples to Tellurium Vapor .......... .. ... . . i, 22

4.3 Removal of lodide from LiF-BeF, Melts by HF-H, Sparging: Application to Iodine

Removal from MSBR Fuel ............ P e 23
4.4 Protactinium Oxide Precipitation Studies . . .. ... ... . . i 24
4.5 Solubility of BF; in Salts of Molten-Salt Reactor Interest ............ .. .. ... .. .. .. 24
5. COOLANT SALT CHEMIST RY .. ... e e et aas 25
5.1 Oxide and Hydroxide Chemistry of Fluoroborate Melts . . ......... ... ... ... ... iiaat. 25
5.2 Corrosion of Structural Alloys by Fluoroborate Melts . ......... ... .. ... ... i, 25
5.3 Evaluation of Fluoroborate and Alternate COOlants . . . ... vvvuuuror e 26
5.4 Density and Viscosity of Several Molten Fluoride Mixtures . ............ ... .. ... an. 28
6. TRITIUM BEHAVIOR .. ... . e i e e 30
6.1 Permeation of Hydrogen Isotopes Through Structural Metals .............. e 30
6.2 The Solubilities of Hydrogen, Deuterium, and Helium in Molten Li,BeFs .............. ... .. 30
6.3 Chemisorption of Tritium on Graphite . .. ... ... ... . ... . 30
7. OTHER RESEARCH ... .. . e e et i 34
7.1 The Wetting of Graphite by Bismuth and Bismuth-Lithium Alloys as Determined by
the Sessile Drop Method . . .. ..o ..o i e e e 34
8. DEVELOPMENT AND EVALUATION OF ANALYTICAL METHODS .............. e 35
8.1 In-Line Analysis of Breeder Fuel ...................... e e e e b et 35
8.2 Determination of Trivalent Uranium in Simulated MSRE Fuel ..... e e 36
8.3 In-Line and Other Support Analyses for Circulating NaF-NaBF; Coolant . .................... 37
8.4 Studies on the Electroreduction of Uranium(IV) in Molten LiF-BeF,-ZrF, ............ ... ... 42
8.5 Electroanalytical Studies of Tellurium in Molten LiF-BeF,-Z1F, _
(65.6-29.4-5.0M1018 T6) « « -+« ottt et e e e e e 43
8.6 Electroanalytical Studies of Bismuth in Molten LiF-BeF,-ZrF,
(65.6-29.4-5.0mMO0lE T5) . .« ottt 43
8.7 Electroanalytical Studiesin Molten NaBF, . ....... ... ... ... .. .. . . . 44
8.8 Spectrophotometric Research ........ .. ... .. i 45
8. 9l Spectral Studies of f-d and f-f Transitions of Pa(IV) in Molten LiF-BeF,-ThF, ................. 45
8.10 Spectrophotometric Developments for Improved Analytlcal Methods for Fluoroborate
Coolant SYSteIMS . . oottt e e e 46
9. INORGANIC AND PHYSICAL CHEMISTRY ... ... e 47
9.1 The Equilibrium of Dilute UF; Solutions Contained in Graphite . . ............ ... ... ... .. 47
9.2 Effects of Oxygen on the UF; /UF, Equilibrium . ........... ... .. ... ... o ... 48

9.3 Porous Electrode Studies . . ..o oot et e e et et e e e e e et 49
9.4 Formation Free Energies and Activity Coefficients in Fuel Salt Mixtures ..................... 51
9.5 The Chemistry and Thermodynamics of Molten-Salt Reactor Fuels . . ............. ... ........ 51

9.6 Oxide Chemistry of Niobium(V) in Molten LiF-BeF, Mixtures ................... S 53

PART 3. MATERIALS DEVELOPMENT

10. DEVELOPMENT OF MODIFIED HASTELLOY N ... ..o\ 57

10.1 Procurement of New Test Materials ............ ... ... .. ... ... ......... e .. 57
10.2  Weldability of Commercial Alloys of Modified Hastelloy N . .. .......... ... ... ... . ..... 57
10.3  Mechanical Properties of Modified Hastelloy N . ............ e e e 61
10.4 Transmission Electron Microscopy of Ti-Modified Hastelloy N Alloys .. ... ... 62
10.5  Salt Corrosion Studies . ....... .o 69
10.5.1 Fuel Salt Thermal Convection Loop Results . .. .... P 69

10.5.2 Coolant Salt Thermal Convection Loop Results ... ......... e e 74

10.5.3 Blanket Salt Thermal Convection Loop Results ................. [ 75

10.5.4  Status of Thermal Convection Loop Program with Fuel and Coolant Salts ............. 75

10.5.5 Forced Circulation Loop Results ....................... e e Y

10.5.6  Surveillance Specimens from the Coolant-Salt Technology Facility .................... 77

10.6 Corrosion of Hastelloy N and Other Alloysin Steam .......... e 77
10.7 Intergranular Cracking of Alloys Exposed to Tellurium . ... .. e e e 79
10.8 Chemistry of Salt-Metal-Tellurium System .. ... ... ... ...t 79
10.9  Auger Observations Relative to Intergranular Cracking ........................ e 80
10.9.1 Fracture Surface Analysis.................. S 81

10.9.2  High Temperature Surface Studies . .............. ... . 81

10.10 In-Reactor Fueled Experiments .......... e 81
TO10.1 Design . ... e e 82
10.10.2 Calculated Operation and Test Conditions ............. ... .. .. .. .. .. v u. .. .. 82
10.10.3 Preliminary Operating Data . .......... .. .. . . 84

11. FUEL PROCESSING MATERIALS DEVELOPMENT .. ... ... ... . 86
11.1  Static Capsule Tests of Graphite with Bismuth and Bismuth-Lithium Solutions .. ............... 86
11.2 Thermal Convection Loop Tests of Molybdenum, Tantalum, and Graphite in :
Bismuth-Lithium Solution ... ... ... .. 89

12. GRAPHITE STUDIES .. ..............uv... P i 92

12.1  Property Changes in Near-Isotropic Grades of Graphite Irradiated at 715°C to Fluences

as High as 4.2 X 1022 neutrons/cm® .. .. ... ... 00 92

12.1.1 Materials . ... .o 92

12.1.2 Testing . .o 94

12.1.3  Property Changes as Functions of Fluence ... ......... ... ... .. .. ... . 0 .. ... 95

12.1.4 Conclusions . ............ ... ciiio.... e e 95

PART 4. FUEL PROCESSING FOR MOLTEN-SALT REACTORS

13. PROCESSING CHEMISTRY . . .. e e 100
13.1 Chemistry of Fluorination and Fuel Reconstitution . . ............. .. .. ... ... .. .. .. .. .. ... 100

14. ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS

14.1
14.2
14.3
14.4

14.5

14.6

14.7
14.8

14.9

vi

.............................

Examination of Equipment from Metal Transfer Experiment MTE-3 ... ....... ... ... ........
Installation of Metal Transfer Experiment MTE-3B .................... ... e
Design of the Metal Transfer Process Facility

...........................................

Salt-Metal Contactor Development: Experiments with a Mechanically Agitated Nondispersing .
Contactor in the Salt-Bismuth Flow-through Facility ................. . oot

Salt-Metal Contactor Development: Experiments with a Mechanically Agitated

Nondispersing Contactor Using Water and Mercury . ...... ... .. ... i,
14.5.1 THEOTY .ot v ittt e e e e et e et e et e e e
14.5.2 Experimental Apparatus . . ... ... ...ttt e
14.5.3 Results and Conclusions . . ... .v it r v i e e e

Continuous Fluorinator Development: Autoresistance Heating Test AHT-3
14.6.1 Experimental Equipment and Procedure
14.6.2 Experimental Results .. ... ... .. .
14.6.3 Distribution of Current Densities in the Autoresistance Heating Equipment

-------------------

.......................................

............

Fuel Reconstitution Development: Design of a Fuel Reconstitution Engineering Experiment

......

Performance of Open Bubble Columns . . ... ... . i e
14.8.1 Axial Dispersion in Open Bubble Columns .. ....... ... ... . oo
14.8.2 Mass Transfer Rates in Open Bubble Columns . ........ ... .. ... ..

Conceptual Design of an MSBR Processing Engineering Laboratory

..........................

PART 5. SALT PRODUCTION

15. PRODUCTION OF FLUORIDE-SALT MIXTURES FOR RESEARCH AND DEVELOPMENT

PROGRAM

.........................................................................
Introduction

The objective of the Molten-Salt Reactor (MSR)
Program 1is the development of nuclear reactors
that use fluid fuels that are solutions of fissile and
fertile materials in suitable carrier salts. The pro-
gram is an outgrowth of the effort begun over 20
years ago in the Aircraft Nuclear Propulsion
(ANP) program to make a molten-salt reactor
power plant for aircraft. A molten-salt reactor, the
Aircraft Reactor Experiment, was operated at Oak
Ridge National Laboratory (ORNL) in 1954 as part
of the ANP program.

The 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 *’UF, dissolved in a salt that is a mix-
ture of LiF and BeF,, but *’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
(MSRE). This reactor was built to test the types of
fuels and materials that would be used in thermal
breeder and converter reactors; it also provided op-
eration and maintenance experience. The MSRE
operated at 1200°F and produced 7.3 MW of heat.
The initial fuel contained 0.9 mole % UF., 5%
ZrF4, 29% BeF,, and 65% 'LiF; the uranium was

about 339 **’U. The fuel circulated through a reac-

tor vessel and an external pump and heat exchange
system. Heat produced in the reactor was trans-
ferred 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 as-
sembly of graphite moderator bars that were in di-
rect contact with the fuel.

Design of the MSRE was started in 1960, fabri-
cation of equipment began in 1962, and the reactor
was taken critical on June 1, 1965. Operation at low

vil

power began in January 1966, and sustained power
operation was begun in December 1966. One run

-continued for six months, until stopped on schedule

in March 1968.

Completion of this six-month run ended the first
phase of MSRE operation, in which the objective
was to show on a small scale the attractive features
and technical feasibility of these systems for civilian
power reactors. The conclusion was that this objec-
tive 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; also the
fuel is stable; the reactor equipment can operate
satisfactorily at these conditions; xenon can be re-
moved rapidly from molten salts; and when neces-
sary, the radioactive equipment can be repaired or
replaced.

The second phase of MSRE operation began in
August 1968, when a small facility in the MSRE
building was used to remove the original uranium
charge from the fuel salt by treatment with gaseous
Fa. In six days of fluorination, 221 kg of uranium
was removed from the molten salt and loaded into
absorbers filled with sodium fluoride pellets. The
decontamination of the equipment and recovery of
the uranium were completed.

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

In September 1969, small amounts of PuF; werz
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.

Because of limitations on the chemical processing
methods available in the past, most of our work on
breeder reactors was aimed at two-fluid systems in
which graphite tubes would be used to separate

233
{
uranium-bearing fuel salts from thorium-bearing
fertile salts. In late 1967, however, a one-fluid
breeder became feasible because of the develop-
ment of processes that use liquid bismuth to isolate
protactinium and remove rare earths from a salt
that also contains thorium. Our studies showed that
a one-fluid breeder based on these processes can
have fuel utilization characteristics approaching
those of our two-fluid designs. Since the graphite
serves only as moderator, the one-fluid reactor is
more nearly a scale-up of the MSRE. These advan-

tages caused a change in the emphasis of the pro-

gram from the two-fluid to the one-fluid breeder;
most of the 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 appraised so that a decision could be

viii

made about its continuation and the level of fund-

ing appropriate for it. Consequently, a thorough re-
view of molten-salt technology was undertaken to
provide information for an appraisal. A significant
result of the review was the preparation of ORNL-
4812, “The Development Status of Molten-Salt
Breeder Reactors.” A subsequent decision was
made by the AEC to terminate work on molten-salt

reactors for budgetary reasons; in January 1973
ORNL was directed to conclude MSR development
work. A progress report covering work carried out
subsequent to August 31, 1972, was not prepared
because of attention required for closing out the de-
velopment program; for this reason, the present re-
port summarizes work carried out after that date.

In January 1974, the AEC program for molten-
salt reactor development was reinstated. A consid-
erable effort since that time has been concerned
with assembling a program staff, making opera-
tional a number of development facilities used pre-
viously, and replacing a number of key develop-
mental facilities that had been reassigned to other
reactor programs. A significant undertaking was the
formulation of detailed plans for the development of
molten-salt breeder reactors and the preparation of
ORNL-5018, “Program Plan for Development of
Molten-Salt Breeder Reactors.”

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

PART 1. MSBR DESIGN, DEVELOPMENT,
' AND SAFETY

1. Design and Systems Analysis

Additional reviews of the ORNL reference design
MSBR indicated a number of areas in which addi-
tional effort would be required to produce a system
with acceptable performance properties. The core
layout should be modified to permit graphite re-
placement in small units rather than as a single as-
sembly and stresses should be reduced in some por-
tions of the vessel; tube plugging, rather than
tube-bundle replacement, should be provided for in
the primary heat exchanger; the fuel letdown and
return system and the cell containment and cooling
system should be developed in more detail.

Studies were made of long-term fission-product
poisoning in molten-salt converter reactors in-
tended to operate for as long as six years without
fission-product removal. Calculations that treated
fission-product poisoning in detail showed that the
less rigorous treatment provided by the computer
code ROD accurately predicted the effect.

Calculations made with the computer code
MSRXEP and revised nuclear properties for the
mass-135 nuclides indicated a poison fraction due
to '**Xe of 0.0047 for the reference system with low
permeability graphite.

A simplified computer program for calculating
tritium behavior in molten-salt systems was made
operational. Calculations made for the reference de-
sign MSBR suggest that significant reductions in

tritium transport to the steam system can be .

achieved with oxide coatings on the steam side of
steam-generator tubes. Additional parameter stud-
ies are being made for the MSBR, and the code is
being adapted to represent the Coolant-Salt Tech-
nology Facility to help define and analyze experi-
ments involving deuterium injection.

The Molten-Salt Reactor Information System, a
data file containing 373 abstracts of MSR-related
documents, was restored to operational status in the
ORNL computer system. The entries may be

ix

searched and examined via established conversa-
tional-mode computer methods.

2. Systems and Components Development

Development work was completed in 1972 for an
axial-flow centrifugal bubble separator to be in-
stalled in the Gas Systems Technology Facility. A
500 gpm separator with a gas capacity of about 1.3
scfm at 1300°F was tested and shown to have a sep-
aration efficiency of 80 to 95%, depending on bub-
ble diameter and liquid properties. A venturi bub-
ble generator was also developed and empirical re-
lationships were derived from tests with aqueous
solutions to predict overall head loss, gas injection
pressure, and bubble diameter.

Tests were performed in an effort to determine
characteristic bubble sizes in simulated MSBR fuel
salt. Technical difficulties prevented definite mea-
surements, but very small bubbles, as well as larger
ones near 1.25 mm in diameter appeared to form in
experiments where salt was agitated at 1000 strokes
per min at 650°C. No further tests have been per-
formed since the MSR Program was reactivated.

Preliminary studies were made of devices that
might be suitable for measuring bubble size distri-
butions in operating molten-salt systems. Gamma
densitometer methods, acoustic absorption meth-
ods, and laser techniques were considered, but no
firm conclusions were reached. This work has been
discontinued for the present.

Additional tests were performed to evaluate
liquid-to-bubble mass transfer in a 1 1/2-in.-diam
conduit using 25 and 37.5% mixtures of glycerine
in water. The results were consistent with data for a
2-in.-diam conduit when compared on the basis of
power dissipation.

A dimensional analysis was performed to facili-
tate the prediction of bubble diameters produced by
bubble generators. When the analysis was adjusted
to properly account for power dissipation in the
fluid, relationships were obtained which agreed
with the obsérved dependence of bubble diameter
on the -0.8 power of the bubble-generator-throat

Reynolds number and the 3/5 power of a dimen-
sionless grouping of fluid properties.

Prior to the interruption of the MSR Program in
1973, the Foster Wheeler Corporation was working,
under subcontract to ORNL, on the conceptual de-
sign of MSBR steam generators. This contract was
resumed to allow completion of the design for the
ORNL 1000-MWe MSBR reference steam cycle
and to permit an assessment of Hastelloy N-to-
steam corrosion. Refinements made in earlier cal-
culations and an allowance for a 15% margin in
surface area indicate that an effective tube length of
140 ft 1s required for the unit. Allowance for foul-
ing on the salt side of the tubes lowered the outlet
steam temperature by 33 to 41°F. Part-load studies
with coolant salt bypass around the steam genera-
tor showed that flow rates slightly greater than the
full-load design value were required for some inter-
mediate loads.

Most of the construction work required for water
testing of the Gas Systems Technology Facility was
completed. Instrument installation is continuing in
preparation for detailed checkout of the loop. Fabri-
cation and construction required for salt operation
have been temporarily deferred.

The Coolant-Salt Technology Facility was oper-
ated with sodium fluoroborate salt for a total of
1063 hr in two runs prior to program interruption
in 1973. Some difficulties were encountered with
the salt cold trap in both runs, and minor modifica-
tions were made to the system. Operating experi-
ence was generally favorable, although an apparent
cavitation problem prevented salt circulation at the
design flow rate. The facility is currently being re-
commissioned for further development work. Test
plans and operating procedures are being formu-
lated to study the behavior of deuterium (tritium)
in the loop; instruments are being checked and cal-
ibrated; and a revision of the load-orifice arrange-
ment is being studied as a possible cure for the cav-
itation problem.

The forced convection loop MSR-FCL-2 was
shut down in October 1972, after 7000 hr operation
with fluoroborate salt. The salt was subsequently
drained and the loop was cleaned and modified for
operation with fuel salt. A new salt mixture, LiF-
BeF>-ThF.-UF. (68-20-11.7-0.3 mole %) was
loaded into the system (now designated MSR-
FCL-2b) in March 1974. The system has since
been operated isothermally for 2700 hr at tempera-
tures from 560 to 730°C, primarily to study the
oxidation-reduction behavior of the salt. Operation
of the loop has been generally favorable.

3. Reactor Safety

Events with potential safety significance in
MSBRs are being identified, categorized, and eval-
uated. The primary criterion for establishing safety
significance is the existence of a potential threat to
the health and safety of the public and/or reactor
operating staff. Three classes of events that must be
considered for MSBRs are events that directly
cause mixing or spilling of salt, events that cause
major temperature excursions in the primary sys-
tem, and events that cause major gaseous releases
from or within the primary system.

General failure modes and effects analyses were
performed for a number of events that could lead to
reactivity perturbations and hence, to neutronic ex-
cursions. None of the events examined appear to be
capable of overriding the inherent shutdown mech-
anisms of the reactor. However, a reliable poison-
rod shutdown system would be required to prevent
reattainment of criticality after the initial ex-
cursion.

PART 2. CHEMISTRY
4. Fuel Salt Chemistry

The behavior of tellurium and Li;Te was studied
in molten Li;BeF, under various redox conditions
by absorption spectrometry. Researchers found that
a colored soluble tellurium species, tentatively iden-
tified as LiTes, can exist in the melt.

An experimental test stand has been constructed
to permit exposure of metallurgical specimens to
tellurium vapor at a defined temperature. The tel-
lurium is delivered to the specimen at a predeter-
mined rate by diffusion through argon. A theoreti-
cal model of the diffusion process was developed
and is being tested.

The results of earlier transpiration experiments,
in which dissolved iodide was removed from melts
by sparging with mixtures of HF-H:, were ex-

. plained by means of a kinetic model that assumes

that the diffusion of the iodide ion to the surface of
the melt is the rate-controlling step. The removal of
iodine from MSBR fuel was analyzed in terms of
the redox potential required to accomplish the re-
moval efficiently while preventing undesirable reac-
tions in the fuel.

The precipitation of protactinium oxide, Pa,Os,
from MSBR fuel salt by reaction of dissolved PaFs
with H,O-HF-Ar gas mixtures was investigated
and the results were published.
PART 3. MATERIALS DEVELOPMENT
10. Development of Modified Hastelloy N

New test facilities are being developed for the
thermal convection loops and for mechanical prop-
erty testing. Work is progressing well on both facili-
ties and they should be in partial operation by the
end of calendar year 1974.

Procurement of a 10,000-1b heat of the 2% Ti-
modified Hastelloy N has been initiated. The heat
has been melted and partially forged. Tests per-
formed on small commercial melts of this composi-
tion continue to show that the alloy has adequate
resistance to irradiation damage. Production of the
10,000-1b heat, as various product forms, is an im-
portant step in the scale-up process.

Screening tests indicate that the 2% Ti-modified

Hastelloy N has better resistance to intergranular
cracking than standard Hastelloy N, but the test
conditions must be made to coincide more closely
with those of an MSBR to determine whether the
resistance is adequate. Meanwhile, additions of nio-
bium and the rare earths (cerium, lanthanum) were
noted to impart additional resistance to cracking,
and small heats of several alloy compositions are
being prepared for evaluation. Test techniques are
being developed for basic phase-equilibria studies
involving tellurium and Hastelloy N.
. A fueled irradiation experiment containing fuel
pins made of standard Hastelloy N, type 304 stain-
less steel, and Inconel 601 is now in process. These
materials will be strained and examined for inter-
granular cracks following irradiation.

The thermal convection loop program has re-
ceived considerable attention from the planning
standpoint. Some of the loops in operation previ-
ously will be restarted with fuel salt, and several of
the coolant loops will be destructively examined.
Several additional loops will be fabricated and
started during the next few months. Forced circula-
tion loop FCL-2b, initially operated with coolant
salt, has been modified for operation with fuel salt.
The loop has completed about 3000 hr of operation
during which techniques were demonstrated for
measuring and controlling the redox potential of
the salt. The steam corrosion test chamber was re-
assembled with both new and previously exposed
specimens and was returned to the Bull Run Steam
Plant for continued exposure.

11. Fuel Processing Materials Development

Capsule studies of graphite in bismuth-lithium
solutions have shown penetration by bismuth at

xi

650°C but no changes in the lattice parameters of
the graphite that would indicate chemical inter-
action with the solutions. The concentrations of car-
bon in the bismuth-lithium solutions after exposure
to graphite were found to increase with the lithium
concentration of the solution.

Examination of samples from a molybdenum
thermal convection loop which had operated for
8700 hr at a maximum temperature of 700°C
showed that slight dissolution (<0.5 mil) and depo-
sition of molybdenum had occurred in hot and cold
leg specimens, respectively, Maximum weight loss
was 3.62 mg/cm’, and room temperature tensile
tests showed few changes as compared with data
taken before the test.

12. Graphite Studies

Graphite samples irradiated to neutron fluences
up to 4.2 X 10%? neutrons/cm® (£ > 50 keV) at
715°C were subjected to several types of postirradi-
ation tests including thermal expansivity, brittle-
ring fracture strength and strain, shear modulus,
and Young’'s modulus. The property changes due to
irradiation were large in some cases but were not
thought to limit the performance of graphite in an
MSBR.

PART 4. FUEL PROCESSING
FOR MOLTEN-SALT REACTORS

13. Processing Chemistry

Studies of certain aspects of the chemistry of fuel
reconstitution and of the metal transfer process
were conducted between September 1972 and the
temporary termination of the MSR Program in
January 1973. The investigation of the reaction of
gaseous UFs with UFs dissolved in LiF-BeF:-
ThF4 (72-16-12 mole %) to form dissolved UFs was
partially completed and the results were published.
Studies of the distribution of Li;Bi between molten
LiCl and liquid Li-Bi alloys were completed and
the results were published. After the MSR Program
was reactivated in February 1974, apparatus was as-
sembled to conduct studies on the chemistry of
fluorination and reconstitution of MSBR fuel salt.
These studies are now in progress.

14. Engineering Development
of Processing Operations

" The equipment used in metal transfer experiment
MTE-3 was opened and examined. No significant
products. This was a highly practical application,
because it eliminated a number of steps in the fuel
preparation and loading procedure, and it materi-
ally reduced the time required for the operation.

In-line voltammetric measurements were also
performed in NaF-NaBF., a proposed coolant for
molten-salt reactors, in support of the operation of
the Coolant-Salt Technology Facility (CSTF). Re-
producible waves for the reduction of 100 ppm of
Fe** and 30 ppm of Cr’* were recorded at gold elec-
trodes inserted in the melt in a specially designed
salt monitoring vessel that was fed by a side stream
from the CSTF. Reduction waves were also ob-
served for Ni**, Fe”', and an unknown species that
was probably Mo™".

Traces of an electroactive proton species in the
coolant were observed as an increase in the pres-
sure within an evacuated palladium electrode held
cathodic to the melt. From such measurements, the
half-life of this species in the CSTF was estimated
to be about 10 hr. Demonstrations showed that by a
standard addition of water, concentrations of this
species in the parts-per-billion range could be
measured. These observations, together with elec-
troanalytical and spectrophotometric research, in-
dicate that the active proton species is distinct from
BF;OH and highly mobile (diffusion coefficient
about 8 X 107° cm’/sec). Since the active proton spe-
cies is in equilibrium with a condensable species,
probably BF;-H:O, in the cover gas, it may serve as
an agent for the containment of tritium. A large
distribution coefficient, (Ceondensate/Cmet) = ~10°,
was observed for tritium leached from the CSTF
components.

Square wave voitammetry, a technique that is
expected to provide general improvement of in-line
electroanalytical methods, was tested for
applications to the reversible U¥— U*" couple in
MSRE fuel solvent. Voltammetric studies have
provided information on the reduction and oxidation
potentials of tellurium metal in fluoride melts and
have shown thata soluble reduced tellurium species is
adsorbed at electrode surfaces. Electroanalytical

studies of Bi’* are directed toward establishing the

limit of voltammetric detection and the mechanism
for the previously observed loss of bismuth from
analytical research melts. Evidence of loss through
volatilization has been found. Voltammetric studies
of combinations of corrosion product ions were
performed in fluoroborate melts. An anodic wave at
gold electrodes was found to correspond to the
oxidation of BF;OH . Calibration studies are being

Xii

performed for application to experiments in the
CSTF. Currently, no spectrophotometric research is
being funded by the MSR Program; however, the
analytical research program in transuranium
chemistry is contributing basic information expected
to be of future value. A spectrum of tetravalent
protactinium in fuel is reported. Preparations are
being made to improve the sensitivity of the
spectrophotometric measurement of BF:OH and
BF;OD' in fluoroborate samples.

9. Inorganic and Physical Chemistry

The low (UF3)*/(UF.)’ equilibrium quotients that
were observed in contaminated studies of the equi-
librium, 3UF4d) + UCz(c) — 4UFi(d) + 2C(c),
have been duplicated repeatedly. Capsule equilibra-
tions of UO; and UC; in contact with UF, in LiF-
BeF solutions have shown the formation of no new
oxycarbide phase and an unexpected increase in
the UC; lattice parameters. A gas manifold for
equilibrating UF3-UF, solutions with controlled at-
mospheres has been constructed and is being used
to determine the effect of parts-per-million CO im-
purities on the equilibrium quotient.

Assembly of equipment for studying the elec-
trode behavior of a packed bed of glassy carbon
spheres in a LiCI-KCl eutectic melt was completed.
The chloride melt system is being studied in a
quartz cell that allows visual observation of the op-
erations; this method will help establish operating
procedures for the development of a metal system
using fluoride fuels.. Initial voltammetric scans indi-
cated very satisfactory behavior of the packed-bed
electrode without complications from non-Faradaic
processes.

Equipment to determine the free energy of for-
mation and activity coefficient in fuel salt mixtures
1s being assembled. Initial studies will involve the
attainment of such data using the equilibrium

ThO:(c) + 2NiFx(d) = ThF«(d) + 2NiO(c).

Summaries of work completed prior to reactiva-
tion of the MSR Program but not previously re-
ported in these progress reports are given. Included
are “The Chemistry and Thermodynamics of
Molten-Salt Reactor Fuels” by C. F. Baes, Jr., and
“Oxide Chemistry of Niobium(V) in Molten LiF-
BeF: Mixtures,” by Gann Ting, C. F. Baes, Jr., C.
E. Bamberger, and G. Mamantov.

Measurements of the solubility of BF; in LiF-
BeF>-ThFs showed that Henry’s law is obeyed. The
enthalpy of solution (-15 kcal/mole) indicated strong
chemical interaction between the solute and molten
fluorides. Solubilities of BF3;, measured in a melt
containing 8 mole % NaF in MSBR fuel salt, sug-
gested that even in the event of large coolant in-
leakage, very high partial pressures of BF; are un-
likely, provided equilibrium or near-equilibrium
solubility of BF3; occurs.

5. Coolant Salt Chemistry

Individual compounds are being synthesized and
characterized to aid in identification and under-
standing of the hydrolysis products encountered in
actual NaBF, systems. Knowledge of the reactions
of these compounds may lead to methods for trap-
ping tritium in the coolant. The following com-
pounds have been prepared: H;OBF., HBF2(OH);,
H3;OBF;0OH, and NaBF;OH.

To help predict corrosion behavior of Ni-Cr-Fe
alloys in molten NaBF«NaF, a program was initi-
ated to measure the free energies of formation
(AG7) of the known complex corrosion-product
fluorides. The AG} of NaNiFs; and NaFeF; were de-
termined and published. A similar determination
for Na3CrFs was complicated by the formation of
another complex, NasCr3;F4. Experiments are con-
tinuing to investigate the stabilities of both of these
chromium complexes.

Fluoroborate and many other possible MSBR
coolants were evaluated for performance and safety
characteristics required to fulfill the functions of
the intermediate coolant. These characteristics ex-
clude alkali metals, lead, bismuth, and low-melting
molten salts that contain oxide or ions more oxidiz-
ing than Ni’". The MSRE coolant was rejected be-
cause of its high melting point. The primary disad-
vantages of fluoroborate are associated with the
consequences of fuel-coolant interleakage. One of
the low-melting eutectics of NaF-LiF-BeF, would
present minimal effects in the event of leaks.

Densities and viscosities of several molten fluor-
ide mixtures of reactor interest were accurately
measured. Molar volumes obtained from the den-
sity measurements agreed within 2% with molar
volumes calculated from additive contributions of
the components. The viscosities of fluoroborate
coolant were found to be less than two centipoise
at all normal coolant temperatures (850 to 1150°F).
The viscosity of two melts composed of LiF, BeF,
and ThF., one being the MSBR fuel-carrier salt,

xiii

were analogous to viscosities reported for similar
mixtures containing UF, instead of ThF,.

6. Tritium Behavior

The permeation of hydrogen isotopes through
structural metals was studied to determine whether
oxide films on steam generator materials would im-
pede tritium transport to the steam generator. It ap-
peared that oxides could decrease tritium flow by
many orders of magnitude under well controlled
conditions. Work is continuing under conditions
more germane to actual applications.

Solubilities of hydrogen, deuterium, and helium
in molten Li;BeF.: have been measured and values
of the Ostwald coefficient have been determined.

Chemisorption of tritium on graphite has been
studied. From the data obtained to date, a loading
of a few milligrams of tritium per kilogram of
graphite would be expected in an MSBR.

7. Other Research

An apparatus was constructed for measuring con-
tact angles between graphite specimens and bis-
muth alloys. Published results suggest that the pres-
ence of additional metallic species in bismuth may
lead to the formation of stable carbides.

8. Development and Evaluation
of Analytical Methods

Onsite measurements, using proposed in-line
electroanalytical methods, were applied to molten
fuel compositions in two facilities. Voltammetric
measurements of U”*/U* ratios and corrosion prod-
ucts in the fuel circulated in a forced convection
loop showed qualitative differences between the
electroanalytical properties of uranium and chrom-
ium ions in an experimental breeder fuel solvent
LiF-BeF,-ThF4 (68-20-12 mole %) and those in
MSRE solvent, LiF-BeF:-ZrFs (66-29-5 mole %).
Sufficient data were taken to provide a comparison
between this fuel and the reference fuel, LiF-BeFa-
ThF4 (72-16-12 mole %), which will later be circu-
lated in the same loop. Measurements recorded
during the stepwise reduction of the fuel with be-
ryllium metal are expected to provide basic infor-
mation on equilibria in the fuel when analyses of
the data are completed. Measurements of U’/ U*
ratios were used to establish the redox condition of
simulated MSRE fuel in capsules prepared for in-
pile studies of intergranular corrosion by fission

corrosion had occurred inside the vessels, but seri-
ous air oxidation of the carbon steel vessels had
occurred on the outside of the vessels because the
nickel aluminide spray failed to protect against air
oxidation. A superior coating for protection against
air oxidation has been found and tested. Analyses
of the bismuth phases indicated that the surfaces in
contact with the salts were enriched in thorium and
iron. Evidence of oxide was found only in the strip-
per vessel. Oxides at the LiCl-Bi-Th interface in
the contactor were suspected of having contributed
to the lower-than-expected mass transfer rates seen
in that experiment. No firm conclusions can be
drawn concerning the relationship of the findings of
the examination and the low mass transfer rates.
Experiment MTE-3 will be repeated with new
equipment to determine the cause of the low mass
transfer rates.

Mass transfer coefficients in the mechanically
agitated nondispersing contactors were measured in
the Salt-Bismuth Flow-through Facility. The mass
transfer coefficients that have been measured are
about 30 to 409 of those predicted by the preferred
literature correlation but were not as low as those
seen in some of the runs in MTE-3. Additional
studies using water-mercury systems to simulate
molten salt-bismuth systems have indicated that the
model used to interpret results from previous mea-
surements in the water-mercury system has signifi-
cant deficiencies. The results must be analyzed dif-
ferently to extract information that would be useful
for designing contactors for molten salt-bismuth
systems.

Autoresistance heating studies were continued to
develop a means of internal heat generation for

xiv

frozen-wall fluorinators. Equipment was built for
testing one design of a side arm for the heating elec-
trode. Results of experiments with this equipment
indicate that for proper operation the wall tempera-
ture must be held much lower than that for which
the equipment was designed. Studies with an elec-
trical analog of the equipment indicate that there
are no regions of abnormally high current density in
the side arm.

Engineering studies of fuel reconstitution are cur-
rently under way. Equipment 1s described for carry-
ing out the reaction of gaseous UFs with UF; dis-
solved in molten salt and the subsequent reduction
with hydrogen of the resultant UFs.

Future development of the fuel processing opera-
tions will require a large facility for containing en-
gineering experiments. Work ‘has been started on a
conceptual design of a processing engineering labo-
ratory where these experiments can be carried out.

PART 5. SALT PRODUCTION

15. Production of Fluoride Salt Mixtures
for Research and Development

Reactivation of the MSR Program required re-
sumption of production of fluoride salt mixtures for
the research and development efforts. The salt pro-
duction facilities used previously were modified to
meet safety and operational requirements, and pro-
duction of fluoride salt mixtures was resumed on a
limited scale.
Part 1. MSBR Design, Development, and Safety

J. R. Engel

" The composite objectives of the MSBR design,
devclopment, and safety activities are to evolve a
conccptual design for an MSBR with adequately
demonstrated performance, safety, and economic
characteristics that will make it attractive for com-
mercial power generation and to develop the associ-

ated reactor technology required for the detailed de- |

sign, construction, and operation of such a system.
Sincc commercial systems likely will be preceded by
one or more intermediate-scale test and dem-
onstration reactors, these objectives include the de-
sign and development work associated with the in-
termediate systems. In addition, molten-salt reactors
may be attractive for applications other than
clectricity generation in large central stations pow-
cred by breeding cores; therefore, these activitiesalso
include the examination of alternate MSR concepts.

A conceptual breeder design has been prepared
and described by ORNL,' and an independent design
study was done by an industrial group’ under
subcontract to ORNL. The industrial study and
other internal reviews have shown that several as-
pects of the basic conceptual design could be modi-
fied and improved to enhance its overall attractive-
ness. A firmer conceptual design ultimately will be
required as a basis for establishing criteria for the
design and construction of test and demonstration
reactors. The current emphasis in the MSR Pro-
gram is on the development of the basic system
technology, and no specific design work is now in
progress. Design-related studies are being per-
formed as required to support technology develop-
ment, particularly with respect to safety analyses,
tritium management, high-temperature design

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

2. Ebasco Services Incorporated, 1000-MW(e) Molien-Salr
Breeder Reacior Conceptual Design Studyv— Final Report— Task
! {(February 1972).

methods, and evaluation of basic design alternatives.
Additional studies to evaluate more accurately the
performance and economies of MSBRs will be
undertaken as the need for such information develops.

The design work on MSBRs has indicated several
areas in which reactor technology must be developed
to properly make use of the inherent features of this
reactor concept. Components and subsystems will be
developed and operated on various scales up to full-
size to contribute to this technology. Currently, two
engineering-scale and some smaller-scale develop-
ment efforts are being pursued. A Gas-Systems
Technology Facility is being constructed to permit
studies of helium bubble injection and stripping, on a
scalc representative of a test reactor, by methods
expected to be applicable to larger systems. The
facility will also be used to study other engineering
phenomena (e.g., xenon and tritium behavior and
heat transport) in a system that contains MSBR fuel
salt.

A Coolant-Salt Technology Facility is being reac-
tivated to permit studies on a similar scale in coolant
salt. The purpose of the initial studies will be to
better understand tritium behavior and manage-
ment and the behavior of corrosion products in so-
dium fluoroborate. The facility may also be used to
study, on a small scale, the generation of steam with
molten salt. Also, these loops will contribute to a
better understanding of the technology of molten-salt
pumps. Smaller facilities are being built and
operated to expose samples of candidate structural
alloys to rapidly flowing molten salt that contains
various chemical additives (e.g., tellurium) and to
static salt in a reactor-like environment (in-pile
capsules). A

Because the fuel in an MSR is molten and is cir-
culated through the primary circuit, safety consid-
erations differ from those of solid-fueled systems.
Detailed studies are being initiated to evaluate the
safety behavior and safety margins of MSRs under
possible accident conditions.
1. Design and Systems Analysis

J. R. Engel

1.1 ORNL STUDY OF MSR DESIGNS

Although the current MSR Program does not
include activities specifically in the area of reactor
design, some work was done by the ORNL design-
study team, prior to the interruption of the program,
which has not been reported in previous semiannual
progress reports. The following subsections provide a
summary of that work and also indicate the general
status of the ORNL reference design.'

1.1.1 Plant and Equipment Studies
E. S. Bettis

Additional reviews and analyses were made of the
equipment and subsystems described as part of the
reference design, primarily to identify areas in which
improvements or modifications could increase the
viability of the concept. No significant changes were
implemented, but the work did indicate the need for
such changes and their possible nature.

The design of the reactor vessel and internal
graphite structure needs revision. Further consider-
ation has raised serious questions as to the practi-
cality of replacing the moderator for the entire core as
a “cartridge” unit. The handling problems and
possible hazards of such a procedure may rule it out
as a practicalapproach. Since graphite replacement is
required for the reference concept, design studies
should be carried out for a moderator assembly that
is replaceable in modules or in subsections. In
addition, studies of the vessel design revealed areas
where stresses are excessive; design modifications will
be required to eliminate the high stresses.

It appcars desirable to adopt a design for the pri-
mary heat exchanger in which plugging of failed
tubes, rather than replacement of an entire tube
bundle, can be used for maintenance. One possible
way to do this would use a U-tube design. The
original cartridge-tube heat exchanger, because of
the double shell, could not adequately reject the after-
heat from deposited fission products if both the
primary and secondary salt were drained.

1. R. C. Robertson (ed.), Conceptual Design Study of a Single-
Fluid Molien-Salt Breeder Reactor, ORNL-4541 (June 1971).

One area in which design changes have been re-
ported previously is the system for transporting re- '
actor off-gas and salt from the primary circuit to the
drain tank and for returning the salt to the circuit.?
Other . changes will be in the cell (primary
containment) structure and cooling.” Although the
modified designs appear workable, some additional
effort will be needed to firmly establish good con-
ceptual designs.

Other general design studies would be required to
evaluate the effects of release of boron trifluoride,
BF., in the fuel circuit in the case of a leak of
fluoroborate coolant into the primary circuit and to
provide appropriate protection in the design. Pres-
surization of the primary system must be limited, and
the BF; must not be permitted to produce adverse
effects in the off-gas system, particularly the charcoal
adsorber beds.

1.1.2 Neutronic Analysis
J. R. Engel

Neutronic performance calculations have been de-
scribed” for a class of molten-salt coverter reactors
(MSCR) that use plutonium and some uranium as
fuel feed material with batch processing of the fuel
salt at intervals of several years. Such reactors, witha
lifetime-average conversion ratio of about 0.9, ap-
pear economically attractive.

The fuel-conversion performance of an MSCR,
especially with infrequent fuel processing, is sensi-
tive to the buildup of fission-product poisons be-
tween processing operations. Since the comprehen-
sive performance studies were made with the ROD
computer program,’ which has limited capability for
treating the build-up and neutronic effects of fission
products, some uncertainty existed in the predicted
conversion ratios. Therefore, a series of com-

2. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971,
ORNL-4728, pp. 4-7.

3. MSR Program Semiannu. Progr. Rep. Feb. 29, 1972,
ORNL-4782, p. 8.

4, MSR Program Semiannu. Progr. Rep. Aug. 31, 1972,
ORNL-4832, pp. 16-22,

5. H. F. Bauman et al., ROD: A Nuclear and Fuel-Cycle
Analvsis Code for Circulating Fuel Reactors, ORNL-TM-3359
(Scpt. 1971).
putations was made to more carefully evaluate fission-
product poisoning so as to determine the validity of
the ROD results. The approach chosen wastoadopta
fission-product treatment® originally developed for
solid-fueled reactors and to adjust the yields toallow
for the chemical behavior of the nuclides in a
circulating-fuel system. The basic treatment deals
explicitly with 26 nuclidesin 11 chains and then adds
two pseudoelements to account for lJumped long-term
effects of slowly saturating and nonsaturating fission
products. The actual treatment used was reduced by
scveral nuclides to allow (as in ROD) for the rapid
disappearance of noble-metal and noble-gas fission
products from the circulating system. The daughters
of noble-gas radionuclides having half-lives longer
than 40 sec were assumed to be removed from the
rcactor fuel salt, whereas daughter-products of noble-
gases having shorter half-lives were assumed to be
produced in the salt at their full fission yield.

The resultant fission-product set was described for
treatment by the CITATION code,’ which evaluates,
explicitly, the concentration of each nuclide as a
function of power history. Input data required for the
CITATION treatment include fission yields (as
functions of fissile nuclides), decay constants, and
cnergy-dependent cross sections. The fission yields
and decay constants for those nuclides treated
explicitly were taken from the tabulation by Bell,”
whereas those for the pseudoelements were based on
Bennett's values.® The cross-section data were
obtained from the 123-energy-group G123 library in
XSDRN.? These were processed with XSDRN to
produce composition-, energy-, and geometry-
weighted cross sections in 87 energy groups for the
particular reactor configuration. The final cross-
scction set included 30 energy groups in the thermal
range £ < 1.86 eV to ensure reasonable accounting
for the large resonance peak in the **°Pu cross section
at 1 eV. The CITATION calculations were thendone
in 87 energy groups with a point model of the reactor
to evaluate the fission-product poison fraction as a
function of time.

6. L. L. Bennett, Recommended Fission-Product Chains for
Use in Reactor Evaluation Studies, ORNL-TM-1658 (Sept. 1966).

7. T. B. Fowler, D. R. Vondy, and G. W. Cunningham, Nu-
clear Reactor Core Analysis Code: CITATION ,ORNL-TM-2496,
Rev. 2 (July 1971).

8. M. J. Bell, Computer Code for the Solution of Large Sys-
iems of Simultaneous Linear Equations: Application to a 2200
MW(th) Single Region Molten-Salt Reactor, ORNL-CF-687-32
(July 1968).

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

Computations were performed for both the initial
and final 6-year fuel cycles for a converter reactor
concept that had been analyzed with ROD. The first -
cyclec was of particular interest, because, for the
postulated fueling strategy, major changes occurred
in the mix of fissile nuclides which affected both the
fission-product yields and the reactor neutron energy
spectrum. The last cycle was used as a representative
cxample for fission-product buildup in a reactor
fucled primarily with **U. For both cases, the more
detalled CITATION calculations essentially
duplicated, in all important respects, the fission-
product poisoning predicted from the ROD
calculations. The results added considerably to the
confidence in ROD predictions of fuel-conversion
performance. In a similar way, the results add con-
fidence to design studies made with ROD for
breeding systems.

1.1.3 Xenon Behavior
J. R. Engel

The program MSRXEP developed by H. A.
McLain and L. W. Gilley was used by McLain to
estimate the xenon-poison fraction in the reference
design MSBR for one set of reactor parameters, but
with revised nuclear data for the mass-135 nuclides.
For coated graphite with a xenon diffusion coefficient
of 2.6 X 10 cm?/sec in the coating and 2.6 X 10™°
cm’/sec in the bulk material, the estimated poison
fraction was revised from 0.0040 to 0.0047. Some
uncertainty still remains for this value; the
uncertainty is due partly to uncertainties in nuclear
data and partly to lack of detailed experimental
verification of the model itself. However, the target
xenon-poison fraction of <0.005 apparently can be
achicved with xenon stripping in a **’U-fueled
system if low-permeability graphite is available.

The mathematical model and the computer
program for this work have been described."

1.2 TRITIUM BEHAVIOR IN
MOLTEN SALT SYSTEMS

G. T. Mays

1.2.1 Reactor Calculations

Prior to the interruption of the MSR Program,
calculations were made and internally documented

10. H. A. McLain and L. W. Gilley, The MSRXEP (Molten-
Salt Rcactor Xcnon Poisoning) Program, ORNL-CF-73-2-49
(Feb. 2, 1973).

by R. B. Briggs concerning the effect of oxide coat-
ings and molton-salt films on the permeation of hy-
drogen through metal walls. The purpose was to
identify conditions and parameter values that could
significantly affect the distribution of tritium in
MSRs.

The permeation rate for the metal was described by
the relation Q/ A4 = P.(p"* — p.”*) and the permea-
tion rate for the coating by the relation Q/ 4= P,(p —

_P,), where p and p, are the higher and lower partial

pressures of hydrogen at the surfaces of the material
in question which . define the driving force for
transport through that material, and p., and P, are
overall permeabilities of the metal and oxide
measured when p = 1 torr. At p > | torr, the metal
provided the major resistance to permeation until
Po/ P < 1. At p < | torr, the coatings became in-
creasingly important and were the major resistance to
permeation when P,/ P, < 100 at p < 107 torr.
Curves of Q/(PaA) vs p for 107 < Po/ Py < % and
10° < p < 10° were generated which have use in
analyzing experimental data and in applying the data
to reactor systems.

Work on a detailed computer program, MSRTRI,

- that describes tritium behavior inan MSBR was done
by H. A. McLain late in 1972, Programming
was largely completed, but no specific results were
generated. Later, this program will be made
operational and used for parametric studies that will
be compared with results of a less detailed program
developed by R. B. Briggs and implemented by C. W.
Nestor. '

~ The latter program is now operational, and revi-
sions have been made to provide a more direct
trcatment of the processes involved. Initial calcula-
tions made to compare results with those of earlier
calculations indicate that further revisions may be
necessary. Subsequent calculations will involve par-
ametric studies that describe tritium behavior in an
MSBR.

1.2.2 Studies of Loop Experiments

Prior to the program interruption in 1973, ana-
lytical work was done by H. A. McLain in connec-
tion with experiments to be performed in the

11. R. B. Briggs and C. W. Nestor, Jr., A Method for Calcu-
lating the Steady-Srate Distribution of Tritium in a Molten-Salt
Breeder Reactor Plant, ORNL-TM-4804 (in press).

Coolant-Salt Technology Facility (CSTF) to study
the behavior of hydrogen isotopes in molten salt.

Experiments now being planned for the CSTF use
this work as a basis and representa continuation of it.
The necessary equipment is under construction
whereby deuterium can be diffused into the CSTF
salt through an injection probe located in the main
circulating salt stream. The purpose of this work is to
simulate the behavior and disposition of tritium that
enters the MSBR coolant-salt system and to
investigate the possibility of using this system to limit
the amount of tritium that passes into the MSBR
steam system. In addition to indicating whether
tritium can be trapped in the MSBR coolant salt, the
CSTF may provide data on the role of the coolant salt
off-gas system in removing tritium.

Deuterium was selected as the gas for this experi-
ment, because it can be identified separately from
hydrogen and tritium that might be present in the
CSTF. Deuterium will be diffused into rather than
bubbled into the salt to simulate the MSBR condi-
tions as closely as possible. The rate of adding deu-
terium to the salt for 1- and 4-atm deuterium pressure
inside the injection probe will be about 1.3 and 2.8
cc(STP)/hr, respectively; the main-loop salt flow rate
is about 2972 liters/min.

The location of the injection probe was judiciously
chosen to ensure that the deuterium could not pass
directly into the off-gas system without first coming
in contact with the circulating salt. The injected
deuterium can permeate through the metal walls of
the loop or simply accumulate in the salt. The
objective of this series of studies is to determine the
disposition of the deuterium injected into the salt. To
do so, various determinations are to be made of the
deuterium dissolved in and combined with the salt.
Also, the off-gas and any condensate collected from
the off-gas will be analyzed for deuterium. The
necessary equipment for these analyses is being
developed. It has not been determined whether an
attempt will be made to measure the deuterium
permeation through the salt boundary layer and the
mectal loop walls in the first tests. If the measurement
is not made, it should be an objective of future
experiments, the details of which will depend on the
results of this experiment.

Futurc experiments will be based on data gener-
ated by a modified tritium-transport computer pro-
gram.'' The program is expected to provide infor-
mation on the importance of various associated -
physical characteristics and on the significant ranges
of thesc characteristics.

1.3 MOLTEN-SALT REACTOR

INFORMATION SYSTEM
D. W. Cardwell J. R. Engel

A substantial body of technical information that
rclatcs to MSRs has accumulated since 1947, when
work In this area was started. This information is
dispcrscd throughout a large number of documents
and ‘publications that are nominally available at
ORNL but are not uniformly distributed. To provide
a morc generally accessible source of information
rclated to MSRs, a Molten-Salt Reactor Information
System (MSRIS) was established within the
program. This system, or data set, contains biblio-
graphic information and brief abstracts of docu-
mcnts that deal with MSR technology. The MSRIS
is not, and will not become, a repository for such
documents; nor is it expected to be a complete file of
all MSR information. However, it does contain, in a
readily accessible and computer-searchable format,
information about many of the more important
publications in all areas of MSR technology.

Information on MSR documents was aceumu-
lated, asan incidental effort, by many members of the

MSRP staff in 1971-1972. The information was
prepared in a standard format and was stored in the
computer system at ORNL as a data set that now
contains 373 “entries” or abstracts. The entries were
designed to be compatible with conversational-mode
data-retrieval programs under development at ORNL
for other similar data sets.

After reactivation of the MSR Program, the exist-
ing data set and the current information-retrieval
system wcre reexamined, and the judgement was
made that the MSRIS could be reactivated with
minor effort. A number of errors in the data set were
identified and were corrected, and- several trial
searches were run to verify the capability of the sys-
tem. A report was completed to describe the system,
to establish methods for updating it, and to provide -
procedures for search and retrieval which are -
applicable to the system; the report will be published
soon. No further updating of the data file is currently
planned.

The MSRIS is generally available to the ORNL-
MSRP staff, and can, in principle, be made avail-
able to other qualified agencies through arrange-
ments with ERDA.

2. Systems and Components Development

A substantial amount of systems and components
development work was in progress in the MSR Pro-
gram when it was interrupted in February 1973.
While some of this work was generic in nature,
much of it was in direct support of activities associ-
ated with the construction or operation of the vari-
ous engineering-scale facilities. Some aspects of
this work have been completed since the last prog-
ress report’ was issued, some have been deferred,
and some are still in progress. However, because of
the. lower level of such activity in the present pro-
gram, program participants anticipate that in future
progress reports developmental accomplishments in
support of particular facilities will be described in
the sections dealing with those facilities. As efforts
are initiated to develop particular reactor compo-
nents (e.g., pumps or valves), sections will be added
under this heading to describe technical progress in
those areas. This chapter contains an initial section
to permit the recording of technical progress, prior
to program interruption, in areas that are not at
present being treated as separate development
areas. :

2.1 GASEOUS FISSION PRODUCT REMOVAL
2.1.1 Bubble Generator
R. H. Guymon

A venturi-type bubble generator was developed’
for use in the xenon removal system proposed for a
molten-salt breeder reactor. Gas injected into the
high-velocity liquid at the venturi throat is formed
into bubbles by the fluid turbulence in the diffuser
conc. Tests were conducted using aqueous solutions
to determine the various pressure drops of the bubble
gencrator as a function of liquid and gas flow rates
and to determine the bubble diameter produced.
Empirical relationships were developed which could
be used in combination with the more conventional
fluid flow equations to predict the overall head loss
and thc gas injection pressure of the bubble
generator. A dimensionless correlation for predicting

I. MSR Program Semiannu. Progr. Rep. Aug. 31, 1972,
ORNI1.4832.

2. C. H. Gabbard, Development of a Venturi-Tvpe Bubble
Generator for Use in the Molien-Salt Reactor Xenon Removal
Svstem, ORNL-TM-4122 (December 1972).

the bubble diameter was developed for bubble
generators of similar geometry.

2.1.2 Bubble Separator
R. H. Guymon

Work associated with the development of a bubble
separator for installation and testing in the Gas-
Systems Technology Facility (GSTF) was concluded
in 1972, and an internal report’ describing the
development was published. The nature and scope of
the publication are indicated in the following abstract
of the report.

An axial-flow centrifugal-type bubble separator
was developed for application in the xenon
removal system proposed for a molten-salt breeder
recactor. The gas bubbles in this type separator
would be driven toward the center line by the radial
pressure gradient developed in the liquid vortex.
The gas would be collected in a void along the
center line and would be removed along with some
entrained liquid through take-off ports on both the
upstream and downstream ends of the separator.
This report covers the theory of operation as well
as the design and development testing ofa 500 gpm
scparator having a gas capacity of about 1.3 scfmat
1300°F. The separation efficiency at design flow
rates varied between 80 and 95%, depending on the
bubble diameter and test fluid.

2.1.3 Bubble Formation and Coalescence Test
R. H. Guymon

In October 1972, attempts were made to obtain
information regarding characteristic sizes of gas
bubblcs in simulated MSBR fuel salt. This work was
subscquently discontinued, and studies were made of
methods that might be used in circulating salt loops.
The results of this work, as originally reported by C.
H. Gabbard, are presented below.

Still photographs and 1000 frame/sec movies were
taken of a capsule containing 72-16-12 mole % LiF-
BeF:-ThFs salt at 1200°F (650° C) for comparison

3. C. H. Gabbard, Development of an Axial-Flow Cenirifugal
Gas Bubble Separator for Use in MSR Xenon Removal Systems,
ORNL-CF-72-12-42 (December 1972).
with photographs made earlier with other liquids.
The still photographs taken after agitation at 1000
strokes/min indicated a bubble size between that of
2LiF-BeF: salt and of CaCl: solution. The high speed
movies were too dark to be of value for observing the
formation and coalescence of the bubbles. However,
bubbles could be observed breaking the surface of the
salt fora period ofabout 0.8 sec (real time), indicating
a population of bubbles of about 0.050 in. (1.25 mm)
in diameter in addition to the population of smaller
bubbles which made the salt essentially opaque. This
same double population was observed in some of the
still photos and in the high speed movies.of a
demineralized water capsule. However, the popula-
tion density of small bubbles in the water capsule was
very low.

The salt capsule cracked and leaked salt during the
tcst program, but exactly when this occurred is
unknown. Evidence of salt leakage was observed in
the first movie. Oxide contamination resulting from
this lcak could have affected the bubble size distribu-
tion obscrved in the tests.

2.1.4 Bubble Size Measurement
R. H. Guymon

Mecthods of measuring the bubble size distribution

in a molten salt were investigated by C. H. Gabbard

and T. S. Kress in 1972 and early 1973 and are
rcported below.

Prcliminary studies were made on two devices
bascd on bubble rise velocity. The first device
consisted of a continuous flow U-tube which depends
on slip velocity to produce a void fraction between
the two vertical legs. This work wasdiscontinued due
to the difficulty of adjustment and measurement of
liquid flow rates to cover various size ranges. The
sccond dcvice consisted essentially of a vertical
column through which the bubble-liquid mixture
would flow. The flow would be stopped and the
transicnt void fraction due to the bubbles would be
mecasured using gamma densitometers. A theoretical
analysis’ was made to derive the relations needed to
cxtract the bubble size distribution from the
mcasurcd void fraction transient. This analysis in-
dicated that the technique is theoretically feasible,
but there may be overriding practical considerations
(c.g., thermal convection currents) which are impor-
tant for small bubble diameters.

4. T, S. Kress, personal communication.

An alternate method studied consisted of passing
acoustic waves of different frequency through the
mixture and measuring the absorption. The amount
of acoustic energy absorbed by a bubble strongly
pcaks at the resonant frequency, and it should be
possible to extract the size distribution from these
measurements. This method appears to be most
attractive if sufficient sensitivity can be obtained by
the use of wave guides as conduit penetrations which
would serve to couple acoustic generators and
receivers with the molten salt. ,

A size spectrometer based on the scattering of a
laser beam was dropped from consideration because
of the very small sampling volume, defined by the
intersection of the laser beam and the field of view of
the detector, which would be required for our range
of sizc and void fraction. This sampling volume was
estimated to be 6 X 10°° em’ for our range of interest.

2.1.5 Mass Transfer to Circulating Bubbles
T. S. Kress

Liquid-to-bubble mass transfer data were obtained
in 1972 and early 1973 in the 1 1/2-in.-diam test
scction using both 25 and 37.5% mixtures of glycerine
in watcr, .

The original correlation established using 2-in.
conduit data’ was

Sh/Sc"? = 0.34 Re™™ (dus/ D),

where Sh is the Sherwood number, Sc is the Schmidt
number, D is the conduit diameter, and d.; is the
bubble Sauter-mean diameter.

The mass transfer Sherwood numbers measured
for the 25% mixture generally fell only about 109
higher than the correlation developed from data
taken in a 2-in.diam conduit. However; the
mecasurcd Sherwood numbers for the 37.5% mixture
in the | 1/2-in. conduit were consistently about twice
the values obtained using the correlation obtained
with the 2-in.-diam conduit. However, the data
appcared to be consistent with the previous 2-in.
conduit data when compared onanequivalent power
dissipation basis. On plots of log Sherwood number
vs the log of the power dissipation parameter,

(€m ge dus/v")"?

5. T. S. Kress, Mass Transfer Berween Small Bubbles and
Liquids in Cocurrent Turbulent Pipeline Flow, ORNL-TM-3718,
April 1972, p. 71.
where

ém = power dissipation per unit mass,

gc = proportionality constant,

d.s = Sauter-mean bubble diameter,

v = liquid kinematic viscosity,

both the 25 and 37.5% data for the 1 1/2-in. conduit
fcll on straight line extensions of the 2-in. conduit
data.

2.1.6 Bubble Generation Analysis
T. S. Kress

As a guide for taking experimental data to
characterize the bubble generator, a dimensional
analysis was made. Based on data already available,
the important variables controlling the bubble size
were assumed to be

d=F(D’O7p5#5V)9

whcre
d = mean bubble diameter,
D = throat diameter of bubble generator,
gc = proportionality constant,
IV =liquid velocity in the bubble generator
throat,
u = liquid viscosity,
p = liquid density,
o = surface tension.
Using the I theorem, it was assumed that a correla-
tion can be found of the form,

(d/ D)= C (VDp/p)" (u*/ogDp)’. (1)
In an earlier report,’ the relationship
dvs ~ (g.0] ) (pleg)” . (2)

where ¢, = power dissipation per unit volume, was
obtained from a balance of surface tension and
inertial forces. It was further noted that use of the
power dissipation for well-developed conduit flow
gave

dvs/D ~ (gc 0 p D’y R, (3)

where
D = bubble generator throat diameter,
u = liquid density,
Re =bubble generator throat Reynolds number.
Bubble size measurements on the bubble generator
used in the mass transfer facility appeared to confirm
Eq. (3). However, tests on the generator proposéd for

the Gas-Systems Technology System (GSTF)
appeared to confirm only the 3/5 power on the fluid
properties term, but the bubble size was better
corrclated by a power of —0.8 on the Reynolds
number.

A fundamental difference exists between the two
generators. In the former, gas is injected through a
centrally located probe and the bubbles are formed
ncar midstream. In the latter, gas is injected at the
wall, creating a gas annulus. It is conceivable that the
local power dissipation could be quite different in the
two cases. Consequently, Eq. (2) may be vahd for
both only if the proper choice is made for the power
dissipation.

In the case of the mass transfer facility generator,
the usc of a power dissipation applicable for flow in
conduits seems reasonable, thus giving Eq. (3). For
the GSTF generator, however, the presence of the gas
annulus creates a condition quite different from that
of liquid flow in a conduit. The pressure gradient in
this device should be given by,

dPldx ~ Tw|D ~ (ug dV/dy|wall)/g.D, (4)

where

V' = local axial velocity,

P = local pressure,

x = axial coordinate,

y = coordinate normal to the wall,

7w = wall shear stress.
Assuming the velocity boundary layer lies within a
gas annulus of thickness 8, then the velocity gradient
could be approximated by

AV /dWwan ~ Ve,

where ¥ = bulk average axial velocity, and the
pressure gradient becomes

dP|dx ~ ugV/g.Dé |
where u; = gas viscosity. Consequently, the local
power dissipation per unit volume is given by

¢, = V dP|dx ~ (ug/g6D) V* , (5)

where ¢, = power dissipation per unit volume.
Substituting Eq. (5) into Eq. (2) gives for the bubble
size,

d”/D "‘(chle/3 62/3/#2/3#4/3)3/5Re -0.8 ) (6)

Equation (6) predicts the observed —0.8 power
dependence on Reynolds number and still retains the
3/5 power on the physical properties grouping.
2.2 MOLTEN-SALT STEAM GENERATOR
INDUSTRIAL PROGRAM

2.2.1 Resumption of the Conceptual Design Study
J. L. Crowley

The Molten-Salt Steam Generator Design Study
by Foster-Wheeler Corporation has been reactivated.
The previous study subcontract was to consist of four
tasks and two critical reviews entitled as follows:

I. Task I—Conceptual Design of a Steam Generator
for the ORNL 1000 MWe MSBR “Reference
Stcam Cycle.”

2. Task Ill—Feasibility Study and Conceptual
Design  of Steam Generators Using Lower
Feedwater Temperature.

3. Task III—Conceptual Design of a Steam
Generator for a Molten-Salt Reactorofabout 150
MWt, .

4. Task IV—Description of a Research and
Development Program for Task Il Steam
Generator.

5. Critical reviews: Past and present Hastelloy N-
to-stecam corrosion work; and ORNL’s steam
generator technology loop.

Foster-Wheeler personnel performing this study
had progressed to the final stages of Task | when the
subcontract was terminated in January 1973, A large
portion of the reactivation work consisted of the
examination and reevaluation of the raw data and
notes stored since the cancellation. The documenta-
tion of this work in a Task [ final report and the
review of past and present Hastelloy N-to-steam
corrosion work are to be performed during the first
and second quarters of FY 1975. The remainder of
the tasks are to be further postponed.

2.2.2 Preliminary Results

Some of the preliminary results as reported by
Foster-Wheeler’s monthly progress reports follow.

Unit size. Further refinements of design calcu-
lations by Foster-Wheeler have resulted in an
increase in the steam tube length and thus the overall
unit size. With the application of a 15% surface area
margin and a necessary increase in tube wall
thickness, the effective tube length is now calculated
to be 140 ft (from 114 ft), making an overall unit
height of about 145 ft nozzle-to-nozzle. The single
unit per coolant loop with a thermal capacity of 483

MW would have 1014 tubes and 13 tie rods located in
a 39 1/2-in.-ID shell. The tubes would be 0.482 in.
minimum ID, with a maximum wall thickness of
0.134 in.

Salt-side fouling factor. Data are unavailable
concerning the salt-side fouling factor that should be
applied to a sodium fluoroborate heated steam
generator. ORNL designs in the past have assumed
zero salt-side fouling for other salt compositions,
and, indeed, this assumption seemed to be the case in
the experience with MSRE heat exchangers. The
sodium fluoroborate coolant salt of the present
reference system is likely to lead to some tube fouling
duc to the deposition of corrosion products. Foster-
Wheeler was requested to base their reference design
on zero salt-side fouling but also to determine the
cffect on thermal performance of a reasonable
fouling factor based on their own experience.

Foster-Wheeler Corporation calculated thermal
performance of the reference design (1014 tubes, 140
ft long) using two different salt-side fouling
resistances of 0.000086 (ref. 6) and 0.005 hr ft’-
°F/Btu.’

The thermal performance calculations, based on
these resistances, resulted in outlet salt temperature
increases of about 2 and 14° F (1 and 8° C), respective-
ly, and outlet steam temperature decreases of about
33 and 41°F (18 and 23°C), respectively.

Part load performance if coolant salt is bypassed
around primary heat exchanger. Due to the
restraints of the fuel salt liquidus temperature (about
500° C for the reference salt), special precautions
must be taken at startup and at part load to maintain
both the fuel system and coolant system temperatures
above their respective liquidus temperatures. Of the
scveral possible means of accomplishing this goal,
only two are attractive, and both require that the
coolant salt flow be varied with (but not proportional
to) load. These methods are: (1) to vary coolant salt
flow as neccessary to maintain desirable salt
temperatures while allowing the steam temperature
to increase above 337°C (1000° F), or (2) to vary the
total coolant salt flow through the steam generator
and bypass part of the flow around the primary heat
cxchanger. The bypass of the second method is more
desirable from the standpoint of the steam generator

6. W. E. Ray et al., “The Structure of Sodium Corrosion
Dcposits and Their Effect on Heat Transfer Coefficients.” Nuel.
Tech. 16, 249-62 (1972).

7. A. C. Mucller, “New Charts for True Mean Temperature
Diffcrence in Heat Exchangers,” Ninth National Heat Transfer
Conf., AIChE-ASME, Seattle, Washington, August 6-9, 1967,
Table 2.1. Molten-salt part load performance, method 2
(Outlet salt temperature = 850° F)

Salt flow Load Inlet salt

ratc (%) (%%) temperature (° F)
100.00 100.00 1150.0
101.75 89.64 1114.3
103.00 79.62 1081.9
101.66 69.60 10554
96.50 59.62 1035.5
87.25 49.62 1020.6
73.90 19.68 1011.1
57.60 29.74 1004.9
39.15 19.82 1001.8

since this bypass not only reduces the maximum
temperature difference between the coolant-salt cold
lcg and the inlet feedwater at part load but also
climinates the necessity for outlet steam attempera-
tion and the resulting hazard to the turbine. The
disadvantage of this method is that a salt diversion
valve is necessary.

Foster-Wheeler was asked to design the steam
gencerator for the first method above, which requires
stcam attemperation but no salt bypass valve. As a
point of interest Foster-Wheeler was asked to also
perform a thermal performance analysis of the
resulting design under certain specified conditions of
the sccond method above (the details of which are not
included here).

Table 2.1 lists salt flows and temperatures in the
stcam generator which result from the part-load
control method 2, which maintains constant outlet
stcam temperature 537°C (1000°F) and constant
outlet salt temperature of 455° C (850° F). Note that
the maximum salt flow of about 103% of full-load
flow 1s required at about 80% load.

2.3 GAS SYSTEMS TECHNOLOGY FACILITY
R. H. Guymon

Most of the construction necessary for the water
tests on the Gas-Systems Technology Facility® has
been completed. The critical path item is the refur-
bishing of the salt pump rotor, and work on this has
been restarted. The electric drive motor for the salt
pump was reconditioned. The motor was cleaned and
checked out electrically, and new bearings were

8. MSR Program Semiannu. Progr. Rep. Aug. 31, 1971,
ORNL-4728, pp. 27-28.

10

installed. Machining and dynamic. balancing was
donc on a new 11 3/4-in.-OD coolant-salt impeller,
which will be used on the Mark Il salt pump.
Fabrication was also completed on the temporary
spool picce to replace the shield plug during the initial
water runs. This is to allow measurement of the shaft
deflection.

Two instrument panels and parts of a third were
removed during the project shutdown for use by
another project. Replacement instruments have been .
located, but construction of the panels is not com-
plete, which prevents detailed checkout of the loop.

The swirl and recovery vane assemblies for the
bubblc separator were tested by C. H. Gabbard in the
watcr test loop. The performance was satisfactory in
regard to void stability, gas outlet flow rates, and gas
capacity. The head difference between the separator
inlet and gas outlet from the swirland recovery vanes
was 89 ft. On the development model this difference,
which had been 89 ft for the swirl vane and 77 ft for
thc recovery vane, was probably caused by a reduc-
tion in the removal port size from 0.5 to 0.413 in. and
will probably necessitate changing the size of the
orifices in the gas outlet lines. The bubble generator
and bubble separator casings were being fabricated
by an outside vendor at the time the project was
terminated. After reactivation of the project, the
vendor declined to complete the job, and it was
finished at ORNL.

The vertical variable flow restrictor (FE-104A) was
tested by C. H. Gabbard in the water test loops. There
was some hydraulic noise at 500 gpm at the predicted
stem positions (2.1 in. for FE-104A and 2.3 in. for
FE-102A). However, there was no evidence of wear
arcas which would indicate vibration after 16 hr of
operation at the 2.1 in. setting.

Due to limited funds in this fiscal year, fabrication
and construction which are not needed for the water
runs have been stopped.

2.4 COOLANT-SALT TECHNOLOGY
FACILITY

A. N. Smith’
The last report issued was for the period ending

August 31, 1972;10 therefore, this report covers
activities from September 1972 through August 1974,

9. A.l. Krakoviak was the responsible engineer forall activities
through February 1973,

10. MSR Program Semiannu. Progr.
ORNL4832.

Rep. Aug. 31, 1972,
Checkout of the instruments and equipment and
purging of the salt circulating system were completed.
Salt was circulated for the first time on October 5,
1972, and operation of the loop was continued (Run
No. 1) for a period of 603 hr. The following
observations were made during Run No. 1.

1. Cavitation was observed at the load orifice (Fig.
2.1), The cavitation was ascribed to formation of BF;
bubbles at the load orifice, because the vena contrac-
ta pressure at the load orifice was too low relative to
the BF: partial pressure of the salt. Based on the
calculated pressure profile (Fig. 2.2), a pump tank
(pump suction) pressure of 35 psia should have been
adequate to suppress cavitation at a pump speed of
1760 rpm. However, experimental data that were
takcn to define points of incipient cavitation as a
function of pump speed and pump tank pressure (Fig.
2.3) indicated by extrapolation that a pump tank
prcssure of 75 psi would be required to completely
suppress cavitation at maximum pump speed (1760
rpm).

2. The salt level in the salt cold trap rose excessive-
ly whenever the salt pump was operated in the upper
cnd of the speed range. Because of this problem and
the cavitation problem mentioned above, circulation
was normally done at reduced speed using the motor-
gencrator set.

11

3. After about 100 hr of operation, a restriction
developed in the cold trap circuit. The flow decreased
gradually from an initial value of 0.53 liter/ min to
zero after about 400 hr. _

4. The BF: content of the BFs:-He off-gas stream
(1.5 liters/min) was reduced from 2.7% BF:
(cquivalent to a salt temperature of 508°C and a
pump bowl pressure of 5 psig)to 1.19% BF: by volume
in passing through the BF; economizer. Thisreduced
volume compares with a theoretical exit concentra-
tion of 0.59% based on the economizer operating
temperature of 440°C.

5. A calibration was made for the salt flow
through the salt monitoring vessel as a function of
pump speed (Fig. 2.4).

On November 1, 1972, the Coolant-Salt
Technology Facility (CSTF) was shut down and
drained to examine the surveillance specimens. The
top coupon was missing, the center one was de-
formed, and the bottom one was in good condition It
is believed that the damage occurred because of mis-
alignment of the specimen holder during insertion
prior to the start of Run No. . The missing coupon
(1.2-g std Hastelloy N, 1/32in. by3/16in.by 1 11/16
in.) is probably lodged at the bottom of the pipe
immediately upstream of the load orifice. While the
loop was shut down, the cold finger was removed

ORNL-DWG 74-8978

SALT
MONITORING
_ 785 GPM - 96 FT, HEAD
COVER GAS VESSEL e====m— OFF-GAS 800°-1150° F.
5-25 PSIG -~ BF; RECOMBINER SURVEILLANCE
AR SPECIMEN
7 coLp l AIR-COOLED |
LEVEL PUMP TRAP l/ / HEAT EXCHANGER
ELEMENT T T \,
— u[u 17 e —
MOTOR
FLOWMETER

ECONOMIZER

—

CONTROLLED
MANUALLY DAMPER /
OPERATED —t_| AIR

DAMPER ~ oy —» ) EXHAUST
—(F—T— TO BLOWER
GRAVITY it 1
CONNECTIONS FOR /_ DAMPER
FUTURE EXPERIMENTS I~
nomR R MR ‘/L___——’————’_"
—— S _ _______’//—' f
|
: ORIFICE

FREEZE _
VALVE

v

o e
GAS SUPPLY

\—EQUALIZER LINE

FILL &

DRAIN TANK

Fig. 2.1. Coolant-Salt Technology Facility schematic.
12

ORNL-DWG 74-11872

OPERATING CONDITIONS

SALT FLOW 785 gpm
SALT TEMPERATURE 1150 °F
SALT DENSITY 113 16/ft3
SALT PUMP SPEED 1760 rpm
140 ! I ! !
PUMP DISCHARGE
120 — = —
E j
2100 |— — —
x
Z 80 |- B _
v
v
@ 60 — - -
a
- MINIMUM PUMP SUCTION
g 40 — ORIFICE - =
v PRESSURE :
20 — \ — = ]
BFy PARTIAL PRESSURE AT 1150°F
P St st —— b R o ——— ————— S —— —
0 10 20 30 0 10 20 30
FLOW PATH (f1) FLOW PATH (ft)

CSTF PRESSURE PROFILE USING
EXISTING 2.400-in.-DIAM ORIFICE

CSTF PRESSURE PROFILE USING TWO
ORIFICES: 2.466-in. AND 3.270-in.-CIAM

Fig. 2.2, Calculated pressure profiles for the Coolant-Salt Technology Facility.

ORNL-DWG 74 -11873

80 , T
SALT TEMP 940 -980°F 4
DATA TAKEN 10-18-72 e
. A
s
- Vo |
o 60 L4 g
2 NO CAVITAT!ON yd |
e |
ul 7
z I
2 CAVITATION BUT
o No vieraTion /)
& P
5 40 / ¥
x i
N
-~ |
$ -/ |
2
z 20 e CAVITATION
PLUS I
HEAVY VIBRATION I
LINE OF INCIPIENT |
CAVITATION 60 Hz —|
o 1
1000 1200 1300 1600 1800

INDICATED PUMP SPEED {rpm)

Fig. 2.3. Experimental data showing line of incipient cavitation
as a function of pump speed and pump tank pressure.

from the cold trap, and the 0.104-in. orifice at the cold
trap inlet was probed with a wire rod in an attempt to
clear the flow restriction.

Operation of the CSTF was resumed on December
1, 1972, and salt circulation was continued for «
period of 460 hr (Run No. 2). The following obser-
vations were made during Run No. 2:

ORNL-DWG 74-14874

12
<
E
3 r
E 8 L / t
= |
w b / |
2 / |
T |

P

?;: 4 — }
o ~ |
— P l
3 P 60 Hz — iy
N e | \'

o |

0 400 800 1200 1600 2000

PUMP SPEED INDICATOR (UNCORRECTED) {rpm)

Fig. 2.4. Flow rate through salt monitoring vessel as a function
of pump speed.

1. Flow through the cold trap circuit was low at
the outset and dropped to zero at the end of 42 hr.

2. Measurements were made of cold trap salt level
asa function of pump speed, in order to determine the
corrective action needed to cure the cold trap high-
salt-level problem.

3. One gram of water was injected 5 in. below the
surface of the salt in the pump bowl. The purpose of
this test was to use the analytical probes in the salt
monitoring vessel to observe the effect of a water
addition on the proton activity in the salt.

4, A 3-in.<diam by 22-in.-long NaOH trap was
inserted in the off-gas line to test the line’s effec-
tiveness in removing BFs from the off-gas stream.
The experience during Run No. 2 was mostly nega-
tive because of plugging of the trap.

Circulation was stopped, and the loop was drained
on Dccember 20, 1972. The 0.104-in. orifice at the
cold trap inlet was removed from the system. The
orifice was found to be plugged with a spongy-
metallic material, whose major constituent was
shown to be nickel by spectrographic analysis. Visual
cxamination failed to provide a definite answer as to
whether the plug had been formed from particles
cntrained in the salt stream or by precipitation as a
result of the reduced salt temperature at the cold trap
inlet. A new orifice, fabricated from 0.030-in.
Hastelloy N sheet and using the same 0.104-in. hole
diamcter, was installed about 1 in. upstream from the
point where the old orifice had been located. The
purpose of the new location was to move the orifice to
a place where the salt temperature is higher in case the
precipitation mechanism is controlling. Concurrent-
ly, in case settling out of entrained particles is the
controlling mechanism, the pipe immediately up-
strcam of the orifice was packed with nickel wool to
serve as a filter. While this work was being carried
out, the clevation of the cold trap was raised by 4 1/2
in. This change was made to correct the cold trap
high-salt-level problem.

During Runs Nos. | and 2, analyses were obtained
on |5 salt samples—the first ten taken from the pump
bowl and the last five from the salt monitoring vessel
(Fig. 2.5). None of the constitutents show any
significant increase or decrease with circulating time.
The coneentrations of Fe, Cr, Ni, and O appear to be
consistent with priorexperience in the FCL.-2and PK
fluoroborate loops.'"" Comparable data are not
available for the other constituents.

While the CSTF was in operation, all of the off-gas
was passcd through at 0°C cold trap. A brown fluid
was collected in the trap at a rate that decreased with
time. The average collection rates for Runs Nos. 1 and
2 were 6 X 107 and 3 X 107 g of fluid per liter of off-
gas, respectively.

On February 26, 1973, the CSTF was placed in
standby because of cancellation of the MSR
program. From February 1973 through March 1974,
the system was maintained in standby, namely, room
tcmperature with a static pressure of 5 psig helium

1. J. W. Koger, A Forced-Circulation Loop for Corrosion
Studies: Hastelloy N Compatibility with NaBFs-NaF (928 mole
%), ORNL-TM-4221 (December 1972).

12. A. N. Smith, Experience with Fluoroborate Circulation in
an MSRE-Scale Facility, ORNL-TM-3344 (September 1972).

13

ORNL-DWG 74-14875

RUN NO. 1 RUN NO. 2
2.0
FREE F (%) |— o5 2} —
°xm o 6
1.0
{000
0, (ppm) 500 — o —
2
0 Qoo® 0 0 o _ ]
0
50
+
H (ppm) — —
PP oop %Poo o a o
0
100
M — i —
o (ppm) o oV W ¢ vve 9
500
. a ]
Fe (me} AQAAAA A AA A
0
100
Cr (ppm) — © o < oo ©
F o
0
100
Ni (ppm) — alg © o T
o (80 Q o 09 o
OCT NOV DEC
1972

Fig. 2.5. Coolant-Salt Technology Facility sample results.

blankcting the saltcontaining portions. In April
1973, work was begun on recommissioning the
CSTF,and this work wasstillin progressat theend of
the report period. The following are some of the
activitics that are under way.

Preparation of test plans and operating
procedures. In 1972, plans were formulated fora test

~ to study the behavior of tritium in fluoroborate

systems. A known quantity of deuterium will be
allowed to diffuse into the circulating salt stream, and
the distribution of the deuterium will be checked by
sampling the salt and the off-gas stream. The work
necessary to complete the planning phase of this test
IS NOW In progress.

Checkout of instruments and systems. Instrument
calibrations are being checked and flow, pressure,
and temperature regulating systems are being tested
to makc sure they are functioning properly,

Load orifice cavitation problem. This problem is
being cvaluated to determine if operation will be
possiblc at moderate pump tank pressure either
without cavitation or with only moderate cavitation.
Calculations indicate minimum salt pressure can be
incrcased by 15 psi by using two load orifices in series
(Fig. 2.2), and plans are being made to make this
change to the system.

2.5 FORCED CONVECTION LOOP
(MSR-FCL-2 AND 2b)

W. R. Huntley
2.5.1 Introduction

Forced convection loop MSR-FCL-2 is con-
structed of 1/2-in. OD and 0.042-in. wall commercial
Hastelloy N tubing, and it contains three corrosion
tcst speceimen assemblies exposed to circulatingsalt at
thrcc different temperatures with bulk flow velocities
of 10 and 20 fps."’ The loop was operated with
sodium fluoroborate until it was shut down on
October 23, 1972. After the molten-salt program was
rcactivated, circulation with a fuel-type salt was
started on April 18, 1974, to evaluate the tellurium
cracking problem. The test facility will be designated
as MSR-FCL-2b during operation with fuel salt.

2.5.2 Operation of MSR-FCL-2 with Fluoroborate

The MSR-FCL.-2 facility operated satisfactorily
for about 7000 hr until operation was terminated on
October 23, 1972. A report summarizing the loop
design and corrosion results during this period has
been issued.'* Loop operation was stopped because
of a rough power trace and a loss of purge gas flow to
the ALPHA pump, which were later found to be
caused by salt deposits in the shaft annulus and
helium purge inlet line. This failure was probably due
to level surging and trapping which had occurred in
earlier operations when plugged filling lines between
the dump tank and system piping resulted in im-
proper salt filling.

2.5.3 Post Run Inspections

The ALPHA pump'’ was removed from corrosion
test facility MSR-FCL-2 and disassembled to deter-

13. MSR Program Semiannu. Progr. Rep. Aug. 31, 1970,
ORNL-4622, pp. 176-178.

14. J. W. Koger, A Forced-Circulation Loop for Corrosion
Studies: Hastelloy N Compatibility with NaBF-NaF (928
mole %), ORNL-TM-4221 (December 1972).

15. MSR Program Semiannu. Progr. Rep. Aug. 31, 1972,
ORNL-4832, p. 33.

mine the cause of a loss of purge gas flow through the
lower oil seal catch basin and an intermittent increase
in friction in the drive. Inspection showed that the
liquid salt level had been about 3 in. too high in the
pump bowl. Frozen salt was found in the pump shaft
annulus and in the purge gas inlet line, which can
account for both the loss of purge and the increased
drive friction. The rough power trace for the pump
drive: motor was caused by the frozen salt rubbing
against the rotating shaft in the narrow (0.010-in.
radial clearance) annulus region, which normally
contains only gases.

The pump shaft was not worn seriously and was
rcused. Slight scoring occurred where the wedging
action of the frozen salt caused the shaft to rub
against the stainless steel labyrinth threads on the
outer diameter of the annulus. The shaft deflection in
the annulus region also caused the impeller hub to
rub against the lower shroud assembly, but again,
only minor scratching occurred.

The cause of the plugged fill lines was found to be
faulty installation of tubular electric heaters, which
had left two different sections of drain line unheated
along lengthsofabout 5in. The 1/4-in.-ODdrain line
was also removed and examined and no internal
restrictions were found. This examination indicated
that the plugged line observed previously was due to
the heater installation and not corrosion product
accumulation or salt segregation as had been
speculated.'® The heaters on the two other drain lines
were also inspected and found to need replacement
due to short unheated areas 1 to 2 in. long.
Replacement heaters were x-rayed to ensure that the
exact location of the heating element was known
before installing the heaters on the salt lines.

The pump bow! was clean and appeared to be in
excellent condition after 6800 hr of operation. No
carbonaceous deposits were visible, which indicates
that no significant oil contamination occurred during
operation. Only minute amounts of green corrosion
products were visible above the normal liquid level.
Color photographs were taken of both the pump
bowl and the rotary assembly.

Post-test inspection showed that the metal-to-
metal static seal between the two halves of the
impeller casing performed well. This Hastelloy N
toroidal seal is U-shaped in cross section and has a
(0.002- to 0.003-in.-thick nickel plating. During 6800
hr of operation, the seal was immersed in 850°F

16, MSR Program Semiannu. Progr. Rep. Aug. 31, 1972,
ORNL-4832, p. 136.
sodium fluoroborate salt at 850° Fand had a pressure
difference of 143 psiacross its sealing surfaces. There
was no evidence of leakage on either the seal or the
matching Hastelloy N surfaces.

The shaft bearings and oil seals were in good
condition when removed from the pump, which
suggests that they could have continued in service at
4800 rpm for much longer than 6800 hr.

Examination of the spark plug probes from MSR-
FCL-2 resulted in observations that pertain to long-
term operation of sodium fluoroborate systems. The
3/ 16-in.-diam Hastelloy N spark plug probes in both
the dump tank and auxiliary pump tank were badly
corroded after being in use for about 9000 hr. The
probes were immediately checked and confirmed to
have been fabricated of Hastelloy N. Record
photographs were also taken. The upper dump tank
probe had lost about | in. of length,and the probes in
the auxiliary pump tank had lostup to 1/8 in. of their
original length. Corrosion on the cylindrical surfaces
of the probes varied greatly. Possible causes of the
sevcre corrosion were inleakage of impurities, the
active nature of sodium fluoroborate, or the newly
designed “Triac” power supply which energizes these
probes. Studies of the probes indicated that the
accclerated corrosion was due to the current that
passcs from the probes through the salt to the
container wall.

2.5.4 Cleanup and Modifications

The sodium fluoroborate salt was effectively
rcmoved from the piping and dump tank of MSR-
FCL-2 by recirculation of hot, distilled water. Fifty-
five gallons of water was used in removing 1.6 gal of
frozen salt from the system. The majority of the salt
was located in the 5-in.-diam dump tank, where it had
been drained during loop shutdown. During the
entire cleaning operation, water was kept in the
Hastclloy N piping system for a total time of about 4
weeks. Complete removal of all salt in the dump tank
and auxiliary pump tank was verified by use of a
borescope after all water was removed.

The ALPHA pump was cleaned and reassembled
with ncw bearings and oil seals and successfully
opcrated in the cold shakedown test stand for 450 hr
at 5000 rpm. Seal oil leakage during the cold
shakedown test was similar to that observed during
previous operation of the ALPHA pump in corrosion
test facility, MSR-FCL-2; that is, the lower seal
lcakagc rate was less than | ec/day and the upper seal
lcaked about 25 cc/day. The reason for greater oil
Icakage at the upper seal is not known, but it might be

15

related to shaft deflections or vibrations from the V-
belt drive. In any event, an oil leak rate of 25 cc/day is
not considered to be excessive for the pump in its new
application in the MSR-FCL-2b test facility. The
cold shakedown test was terminated, and the pump
was prepared for installation in the facility during
January 1973.

A new 7 1/2-hp adjustable-speed electric motor
was installed to replace the previously used S-hp
motor because the fuel salt is more dense than the
previously pumped sodium fluoroborate salt. The
ncw drive motor also features a brushless design and
a transistorized control system that should increase
reliability. A recording wattmeter was installed in the
main control panel of MSR-FCL-2b to assist the
facility operators to monitor pump and drive motor
pcrformance. New Hastelloy N corrosion specimens
werce installed at all three locations.

During fluoroborate operation, batch-analyzed
bottled helium from plant stores had been used.
However, due to the greater sensitivity of fuel salt to
oxygen contamination, a connection was made to the
GSTF-CSTF helium supply, which is processed by
the MSRE helium treatment station. On-line
analytical instruments show that the purified helium
contains less than | ppm of oxygen and less than 1
ppm of water vapor.

The loop temperature profile was calculated, and
the loop design was reviewed to assure adequacy for
the higher temperatures and higher pressures re-
quired for operation with fuel salt. A revised set of
operating procedures was prepared. All instruments
were recalibrated and new alarm set points were
cstablished for operation with the fuel salt mixture.
Lcak testing of the cover gas-piping system and gas
control panels was done with a mass spectrometer
helium-leak detector to ensure that all fittings and
valves were tightly sealed. Three small leaks were
found and corrected.

2.5.5 Purging the System and Adding Salt

The dump tank and the piping system were heated
to 260°C and then evacuated and backfilled with
helium about ten times to outgas the metal surfaces
and remove water vapor prior to charging the system
with salt. The fuel salt mixture for MSR-FCL-2b was
taken from a full transfer pot containing 8.7 liters of
LiF-Bet2-ThFs-UF. (68-20-11.7-0.3 mole %) which
had been reserved for use in MSR-FCL-2b in 1973,
Six liters of the fuel salt at 670° C were transferred
into the loop dump tank on March 15, 1974.

A series of salt samples taken from the dump tank
and from the salt remaining in the transfer pot
suggested thorium segregation may have occurred
during the freezing-thawing cycles that preceded the
salt transfer. These salt analyses are shown in Table
2.2. The first sample (IA and 1B) taken from the
dump tank on March 18 contained only 28.28 wt %
thorium instead of the expected 42.4 wt %. A special
sample from the heel of salt remaining in the transfer
pot scemed to confirm that some segregation had
occurred, since this salt contained about 52%
thorium. Surprisingly, a second salt sample (3A and
3B) from the dump tank indicated a thorium content
of 39.8%, even though nothing intentional had been
done to the salt in the dump tank since the preceding
sample 1A and 1B. The salt in the dump tank was
transferred back into the transfer potand helium was
bubbled through the transfer pot dip tube for 2 hr to
mix the entire salt charge. After bubbling of helium
was stopped, the salt was immediately transferred
back to thedump tank on April l. Asaltsample taken
from the dump tank on April 2 showed a thorium
content of 39.4% and an acceptable oxygen level of
135 ppm. Based on sample 2A and 2B, the salt
composition appears to be 67.4-22-10.3-0.37 mole %
in lieu of the nominal composition of 68-20-11.7-0.3
mole %. 1t was decided to start operation with this
charge of salt and to replace it after new fuel salt is

16

Possible errors in sampling and analytical tech-
niques preclude exact definition of the degree of salt
scgregation which may have occurred. The unusual
analytical results of sample 1 A may have been due to
the near freezing that occurred shortly before sam-
pling since a second sample, taken one week later, had
more normal analysis. However, the incident does
suggest that mixing of the salt with helium bubbles is
advisable before any salt transferoperation. Also, the
salt should be maintained at least 300° F above the
liquidus temperature of 898°F for at least 24 hr
before any salt sampling or salt transferring
operations are made. :

2.5.6 Operation of MSR-FCL-2b with Fuel Salt

The MSR-FCL-2b piping system was filled with
4.4 liters of fuel salt on April 18, 1974. An elec-
trochemical probe has been in use to monitor the
redox potential and Cr’*, Ni¥’, and Fe** concen-
trations in the salt since the loop was first filled.
Collection of electrochemical data is hampered by
vibrations from the belt-driven pump or perhaps by
liquid movements within the auxiliary tank which
severely affect the output signal from the probe at
pump speeds above 1500 rpm. Therefore, it has been
necessary to reduce the normal pump speed or
preferably, stop the pump when readings are taken.

available from the reactivated salt production Future loops will require modifications to overcome
facilities. this problem.
Table 2.2. MSR-FCL-2b salt analysis (wt %)
Samples
Elements IA & 1B° 3A & 3B° Special® 2A & 2B Theoretical’
Li 9.1 7.66 ~6.0 7.8 7.37
Be 4.14 3.21 ~1.4 3.28 2.81
Th 28.28 39.8 ~520 394 42.4
U 1.66 - 1.46 ~1.35 1.46 1.11
F 53.21 48.8 ~40.0 489 46.31
Totals 96.39 100.93 100.84 100.00
Fe 185 (ppm)
Cr 85(ppm)
Ni 30 (ppm)
O 240 (ppm) 150 (ppm) 135 (ppm)

“Taken 3/18/74 from dump tank at 648°C.
*Taken 3/27/74 from dump tank at 537°C.
‘Average of three analyses along 1/4-in. OD tube.
“Taken 4/2/74 from dump tank at 626°C.

‘Expected composition for 68-20-11.7-0.3 mole % LiF-BeF>-ThF-UF,.

The loop was shut down and drained on May 7,
1974, for repairs and for examination of the 18
corrosion specimens after 418 hr of operation.
Photographs recorded the bright, clean appearance
of the specimens. Very few droplets of salt clung to
the specimens and stringer assembly, which again
demonstrated the excellent drainability of the fuel
salt mixture. Average weight loss of the specimens
during operation was equivalent to a corrosion rate
of about 0.2 mils/year. Metals and Ceramics Divi-
sion weighed the specimens; results of their detailed
cxamination are reported in Part 1. All specimens,
cxccpt one, were reinstalled in the loop for further
testing. Specimen number 45 was kept for possible
futurc examination because it had a slight burnished
appearance. Loop operation was resumed on May
21, 1974,

The major effort during June and July 1974 was
the addition of small increments of beryllium at the
auxiliary pump tank to lower the oxidation potential
of the fuel salt mixture. Beryllium additions were
made by lowering small pieces of 0.023-in.-thick
beryllium sheet below the salt surface on a platinum-
tipped push rod where they completely dissolved. A
total of 208 mg of beryllium was added to the 4.4-liter
salt inventory in eight increments ranging in weight
from 8 to 68 mg. The beryllium additions resulted in
expected increases in the U/ U* ratio as monitored
with the electrochemical probe. This work is dis-
cussed in greater detail in Part 111,

17

Fifteen salt samples have been taken at varying
intervals since isothermal salt circulation was started
in mid-April 1974. The metallic impurity level of
chromium has averaged about 100 ppm throughout
the operation. The iron content has decreased from
the initial level of 185 ppm to a recent average of
about 55 ppm. Nickel content has ranged from 20 to
35 ppm in most samples with no trends observable.

The oxygen analyses of the salt have ranged from
80 to 200 ppm. These relatively high oxygen levels
might be expected to exceed the solubility limits of
the salt, but no oxide has been observed in
pctrographic microscope examinations, which have
becen made on every sample. To further evaluate
oxide precipitation, a test was run from August 21 to
23 by reducing the loop operating temperature from
its normal value of 650 to 565°C. Two salt samples
taken at the lowered operating temperature had no
oxide precipitates when examined petrographically,
and no significant change in oxide level was observed
in the chemical analyses.

At the end of August 1974, corrosion loop MSR-
FCL-2b had accumulated 2700 hr of isothermal
operation circulating fuel salt at temperatures from
560 to 730°C and at a flow rate of about 3.2 gpm.
Detailed results of the electrochemical probe perfor-
mance are reported in Part II. Operation is con-
tinuing with plans for another corrosion specimen
examination followed by additions of NiF> to the
salt.

18

3. Reactor Safety

J. R. Engel

Studies have been started to systematically iden-
tify, categorize, and evaluate events of safety
significance that can be postulated for molten-salt
reactor systems. Initially, a broad spectrum of events
is being qualitatively examined for safety
significance. The events thus identified will be sub-
jected to transient analyses to determine safety
margins and/or to establish requirements for ad-
ditional lines of defense.

3.1 SAFETY EVENTS IN MSBR

E. S. Bettis J. R. Engel

The primary criterion being used, at least initially,
to determine safety significance 1s whether or not a
given event or its consequences representsa potential
threat to the health and safety of the public and/or
the reactor operating staff. In this context, it is
presumed that events which reduce to one the
number of barriers between a major fraction of the
plant’s radioactive inventory and the environment
are safety significant, even though the resultant
hazard to the public health and safety is minimal.
Since the MSBR conceptual design requires double
containment,’ of special safety significance are those
events that cause or threaten to cause a breach of the
primary containment with an attendant major release
of radioactivity from the primary or secondary salt
system. While breaches of the primary system boun-
dary which release radioactivity into the primary
containment appear to present no immediate threat
to the integrity of the primary containment, it is
conceivable that long-term exposure of the contain-
ment liner to spilled salt could result in liner attack,
particularly if an off-normal (wet) containment
atmosphere is present. Therefore, events that may
breach the primary system boundary and spillsaltare
currently being treated as hazards. However, heat-
exchanger tube failures, which lead to some mixing of
the primary and secondary salts, are notautomatical-

1. Secondary containment is defined as the metal-lined concrete
reactor building that surrounds and completely encloses the equip-
ment cells that house the primary and secondary salt systems.
Primary containment is defined as the set of hermetically-sealed,
concrete-shielded equipment cells. The primary-system boundary
is the reactor equipment and associated piping that contains the
fuel salt.

ly included in this class of primary system failures,
because the secondary salt system may also be doubly
contained.

Studies performed to date indicate that three
general classes of events must be considered in the
safety of MSBRs: (1) events that directly cause
mixing or spilling of salts; (2) events that cause major
temperature excursions in the primary system; and
(3) events that cause major gaseous releases within or
from the primary system. Each of these key events
may be caused by a variety of primary events and
may, in turn, lead to a variety of consequences. Only.
those combinations that fall within the primary safety
criterion discussed above will be analyzed initially.

Since event sequences depend, to a degree, on the
particular MSR system under study, current work is
being limited to the ORNL reference design MSBR
with sodium fluoroborate as the secondary coolant.
This study requires, forexample, consideration of the
consequences of the release of BF3 gas when fueland
coolant salt are mixed; however, this consideration
would not be valid for some other coolant salts. Thus
far, no single events have been identified which could
lead to unacceptable consequences.

3.2 MSBR NEUTRONIC EXCURSIONS
J. R. Engel

Neutronic excursions, as a class, are the primary
events that have the greatest potential capability for
causing positive excursions in fuel salt temperature
that could damage the primary system boundary.
(Salt freezing as a result of negative temperature
excursions can, in principle, also cause damage, but
in practice, such freezing is limited to the tubes of the
primary heat exchanger in the reference design.)
Several varieties of nuclear excursions have been
examined in the past, primarily by O. L. Smith, from
the standpoint of the source of the reactivity pertur-
bation that causes the excursion. Events that have
been considered, but which must be studied in more
detail include the following:

1. Control or safety-rod malfunctions

2. Externally caused temperature perturbations

3. Salt flow perturbations

4. Redistribution of graphite moderator
5. Core voiding
6. Introduction of excess fuel

The principal factors that tend to limit the conse-
quences of power excursions resulting from these
cvents are the following:

. The small amount of reactivity that must be
controlled by the rods during normal operation, as a
consequence of on-line fuel reprocessing and concen-
tration adjustment (about 8 dollars).’

2. The prompt negative temperature coefficient of
reactivity of the fuel salt (—2.7 cents/° C).

3. The lower limit on fuel-salt temperature im-
posed by its liquidus temperature (about 500° C).

4. The small void coefficient of the core (33 cents
per percent-voids), coupled with the high boiling
temperature of the salt (about 1430°C vs normal
tempcerature of about 630°C).

19

5. The small amount of excess fuel available for
introduction from the chemical processing plant
(about 9 dollars) coupled with its slow rate of
addition.

Nonc of the events examined appear to be capable
of adding enough reactivity sufficiently fast to
override the inherent shutdown capability of the core
and thus endanger the integrity of the primary system
boundary. However, a reliable poison-rod shutdown
system might be required to prevent reattainment of
criticality after the initial excursion and subsequent
excessive temperatures. Such a shutdown system
would have several seconds in which to act.

2. One dollar of reactivity for this reactorisdefined as the effec-
tive delayed-neutron fraction under normal operation. Because of
the lower vield of delayed neutrons from *'U and losses associated
with fuel circulation, $1 = 0.12% Ak/k.

Part 2. Chemistry

L. M. Ferris

The research and development activities described
below deal with the chemical problems related to
design and ultimate operation of molten-salt reactor
systems. Much of the information reported here was
obtained between September 1972 and March 1973,
at which time the MSR Program was temporarily
terminated. Much of the effort since February 1974
(the time at which the program was reinstituted) has
been devoted to program planning and reactivation
of the experimental work.

Tellurium chemistry in fuel salt is under investiga-
tion, since tellurium apparently is responsible for the
intergranular cracking of Hastelloy N observed in the
MSRE. An experimental test stand has been con-
structed to expose metallurgical test specimens to Te:
vapor at defined temperatures and deposition rates.
This work is being conducted in close cooperation
with members of the Metals and Ceramics Division.
Work previously completed, but not reported, in-
cludes an investigation of the chemistry of tellurium
in molten LizBeFs and the formation of the ion Tes.
Another aspect of fuel salt chemistry 1s the removal of
iodine by HF-H: sparging. The results indicate that
this may be a practical approach if the U/ U* ratio
can be adjusted to the desired range.

Other previously unreported studies are sum-
marized here. These include work on the precipita-
tion of protactinium oxides from MSBR fuelsalt and
on the solubility of BF; in several molten fluoride
mixtures.

To better define the chemistry of fluoroborate
coolant, several aspects are being investigated. The
behavior of hydroxy and oxy compounds in molten
NaBF: is being investigated to define reactions and
compounds that may be involved in corrosionand/or
could be involved in methods for trapping tritium. A
systematic approach to synthesize and characterize
compounds is under way; this information will then
permit an understanding of the hydrolysis products
in actual systems. The chemistry of chromium(IiI)
compounds in fluoroborate melts is of importance
since chromium is the most easily oxidized compo-

20

nent of the Hastelloy N. Two corrosion products,
Na;CrFs and NasCr;Fi4, have been identified from
fluoroborate systems. The simultaneous appearance
of two compounds has complicated efforts to deter-
mine free energies of formation. Work is continuing
to define the stability of these compounds with regard
to temperature, melt composition, etc., and to
determine their solubilities. Studies performed before
the interruption of the program, but not previously
reported in a semiannual progress report, include a
determination of the free energies of formation of
NaNiF: and NaFeF; and measurements of the
density and viscosity of several molten fluoride
mixtures.

A significant problem area is the evaluation of
fluoroborate and alternate coolants. Although the
present candidate coolant, NaBFs-NaF (92-8 mole
%), has a number of advantages, it is less than ideal
with regard to corrosivity to the Hastelloy N contain-
ment, and the consequences of fuel-coolant mixingas
a result of leaks in the primary heat exchangers are
unfavorable and could be unacceptable from safety
considerations. Other potential coolants are being
evaluated and compared.

The behavior of hydrogen and its isotopes con-
stitutes an important area of chemical research,
because approximately 2400 Ci/day of tritium will be
produced in a 1000-MWe MSBR and probably little
of this amount is safe enough to reach the environ-
ment. Research on tritium behavior is currently
funded by the Division of Physical Research. Some
results not previously reported are summarized here.
The solubilities of hydrogen, deuterium, and helium
in Li;BeF4 are very low. The sorption of tritium on
graphite is significant (a few milligrams of tritium per
kilogram of graphite), possibly providing a means of
sequestering a portion of the tritium produced. Initial
work on the permeation of tritium through structural
metals coated with an oxide film offered encouraging
results for the control of tritium. In some cases, the
permeation rate was 100 times lower than that for
clean metals.
Other research reported here includes an investiga-
tion of the wetting of graphite by bismuth and
bismuth-lithium alloys. The work was terminated
before much data was accumulated, but a review of
the literature indicated that rare earths dissolved in
bismuth may react with graphite.

Development of analytical methods has continued
with emphasis on voltammetric and
spectrophotometric techniques for the in-line
analysis of corrosion products such as Fe’" and Cr’*
and the determination of the U**/U*" ratio in MSBR
fuel salt. Similar studies were conducted with the
NaBF:-NaF coolant salt. Information developed
during the previous operation of the CSTF has
been assessed and used to formulate plans for
evaluation of in-line analytical methods in future
CSTF operations. Electroanalytical and spec-
trophotometric research suggests that an elec-
troactive protonic species is present in molten
NaBF.:-NaF, and that this species rapidly equilibrates
with a volatile proton-containing species. Data ob-
tained from the CSTF indicated that tritium was

21

concentrated in the volatile species. Other research
activities include studies of the electroreduction of
uranium(IV) in fluoride melts, redox chemistry of
tellurium species in fuel salt, electroanalytical
behavior of bismuth in fuel salt, and voltammetric
and spectrophotometric analysis of corrosion
products in fluoroborate. Previously conducted
spectrophotometric studies on the f/~d and f~f tran-
sitions of protactinium(IV) in fuel salt are sum-
marized.

In other research, no evidence was found for the
formation of uranium oxycarbides when LiF-BeF:
melts containing both UF;and UF4 were equilibrated
with UO; and UC.. Equipmenthas been assembled in
preparation fora systematic study of the potential use
of packed-bed, glassy-carbon electrodes for con--
tinuous on-line monitors for trace elements (Bi-Te-0)
in MSBR fuelsalt. Toenhance the generalknowledge
regarding molten fluoride mixtures, a program to
determine formation free energies and activity coef-
ficients in ternary systems has been started.

22

4. Fuel Salt Chemistry

A. D. Kelmers

4.1 THE CHEMISTRY OF TELLURIUM
IN MOLTEN Li:BeF,
C. E. Bamberger J. P. Young R. G. Ross
. A description of the work done on the chemistry of
tellurium in molten Li;BeF. (ref. 1) was published
and a summary follows.

Because tellurium seems to play a role in the
intergranular cracking of Hastelloy N and other
alloys, its behavior in the presence of molten Li.BeF4
was investigated. Absorption spectrophotometry was
the main experimental technique used; silica was
chosen for containment to avoid undesirable reac-
tions with metals. The “as received” Li,Te dissolved
in LizBeFa giving a strongly colored solution that
absorbed light with a maximum at 478 nm. The same
absorption was noted when Tex(g) reacted with
purified LixTe in the presence of Li:BeFa. This was
probably the result of a reaction such as

(m/2)LizTe(c) +{(2n—m)/4] Te2(g) = LinTen(d) (1)

leading to the formation of a colored species LinTex
(where m and n represent values that are as yet
unknown). It was also learned that the same species
could be formed in solution by partially reducing Te:
with silicon or by partially oxidizing Li»Te with air or
Fe’*. When LiTe; (prepared at Argonne National
Laboratory) was contacted with molten Li:BeFa,
color developed immediately in the melt, but no
significant Te: was observed in the gas phase. This
suggested that m/»n was equal to, orvery close to 1/3.
From other limited experiments it was estimated that
the solubility of Te: in LizBeFu is of the order of 107
wt % or less at 655°C.

42 EXPOSURE OF METALLURGICAL
SAMPLES TO TELLURIUM VAPOR

A. D. Kelmers * D. Y. Valentine

An experiment is now in progress to expose
metallurgical samples to tellurium vapor at con-
trolled rates to aid in the evaluation of grain
boundary attack and cracking observed in the
MSRE. Tensile specimens of Hastelloy N, Incoloy
600, and modified Hastelloy N will be exposed to
tellurium vapor. This procedure will test whether

alloying additives provide increased resistance to
grain boundary cracking. The experimental
parameters partially simulate the exposure con-
ditions of the Hastelloy N reactor vessel used in the
MSRE.

An exposure time of 1000 hr has been chosen, with
the tensile specimens at 700°C and the tellurium ata
temperature calculated to deposit by concentration
diffusion a specified amount of tellurium in the
specimens. Six tensile specimens are positioned near
the top of each of four quartz tubes,24 mm ID by 42
in. long. Tellurium is loaded into the bottom of the
tubes (which are positioned in a rack inside a flanged
stainless steel vessel); the tubes are then sealed to
contain | atm argon at operating temperatures. The
specimens and the tellurium are maintained at
separate constant temperatures with a linear
temperature gradient between them.

Two extra tensile specimens were sealed in a quartz
tube under these conditions and used in a trial run of
100 hr. These specimens will be weighed after the run,
and the 100-hr experiment will be repeated twice.
This procedure serves two purposes: (1) any
necessary improvements in the experimental system
can be identified and completed before the actual
samples are used in the 1000-hr run; and (2) the three
100-hr runs will provide data to test the theoretical
model of the diffusion process.

Results from the first 100-hr trial run showed that
several modifications to the existing apparatus were
necessary. The reaction vessel was lengthened to
modify the seal; the vessel purge gas was changed to
provide a reducing atmosphere for protection of the
copper heat distribution blocks; one heater was
replaced; and various valves, meters, controllers and
other parts were replaced to improve the control and
reproducibility of the experiment.

The theoretical model of the process was improved
to better define the experimental parameters. The
diffusion coefficient, D, estimated by the Chapman-
Enskog kinetic theory method, as a function of
absolute temperature, is given by

_0.0018583 [T°(1/ M.+ 1/ Mar))'”*

TC:AI_ 2
P (O' [esAr QT)

1. J. Inorg. Nucl. Chem. 36, 1158 (1974).

where M 1s the mass; (1t is a tabulated function of
kT/e; P is the total gas pressure; and o and € are the
Lennard-Jones potential parameters. For Te; in |
atm Ar, the diffusion coefficient varies from 0.3643
cm’/sec at 440°C to 0.6466 cm?/sec at 700°C.

Values of the diffusion coefficient were used in the
solution of Fick’s first law of diffusion

Jo=cDVx.— g (1)

— (2— f) [( T‘ts/ TTc)— ]]DTe,ArMTeApTc ;
IRT [ (T,,/Tz)? "¢ —1)

where |

T:s = temperature at the tensile specimens,
Tre = tellurium temperature,
T'ls > TTC,

Dy = diffusion coefficient at the specimen tem-
perature,

Dt = diffusion coefficient at the solid tellurium
temperature,

f =In (Dts/DTe)/ln(Tts/TTe) 3
[ = diffusion path length,

A = cross-sectional area of the diffusion path,

P1e € par (p denotes partial pressure of designated
gas).

Therefore, to deposit 10,000 ppm of tellurium to a
depth of 5 mils (about 8 mg per specimen) on the
tensile specimens in 1000 hr, the temperature of the
solid tellurium will have to be maintained at 440°C.
Since the deposition rate is a sensitive function of
temperature, approximately doubling for every 20°
increase, care must be taken to accurately control this
temperature to £1°C or better. These calculations
were made using the assumptions of complete getter-
ing of the Tez at the specimen, and that the reaction of
Te: vapor with the specimens is not limited by
diffusion of tellurium into the solid specimen.

In addition, the Metals and Ceramics Division has
also supplied Hastelloy N sheet samples and foils for
x-ray diffraction and electron microscopy analysis.
These samples will be run along with the tensile
specimens in the next 100-hr trial runs.

23

4.3 REMOVAL OF IODIDE FROM LiF-BeF;
MELTS BY HF-H: SPARGING:
APPLICATION TO IODINE REMOVAL
FROM MSBR FUEL

R. P. Wichner
B. F. Freasier’

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

A paper describing this work has been submitted
for publication;’ it presents experimental data
reported in refs. 4 and 5, which were subse-
quently interpreted by a model developed by R. P.
Wichner.® An abstract of the paper follows.

The results of experiments in which iodine dis-
solved as I' in LiF-BeF, melts was stripped as HI by
sparging with HF-H; mixtures have indicated that
the use of such a treatment to remove iodine from the
molten fluoride mixtures used in MSR fuels may be
possible. This concept is particularly significant to
MSR technology, because it provides an indirect
means for removing a large fraction of '**Xe, a decay
daughter of "**I

Data obtained from transpiration experiments
indicated a linear decrease in the logarithm of the
iodine concentration of the melt with the number of
HF moles passed, and a linear increase of the
reciprocal of the apparent equilibrium quotient Qhpp =
Pui/[ Pu(1)] with the partial pressure of HF in the
sparge gas. The iodine removal mechanism is ex-
plained by a model that assumes that the rate-
controlling step is the transport of I” from the bulk of
the melt to the surface, and that the rates of the other
steps are rapid.

The removal of iodine from an MSBR fuel was
analyzed in terms of the redox potential required to
efficiently remove the iodine while preventing un-
desirable reactions in the fuel or between the fuel and
its environment, '

The relative abundances of different iodine species
present in the off-gas during sparging of an MSBR

2. Present address, Louisiana
Ruston, La. 71270.

3. Paper submitted for publication to Nucl. Sci Eng.

4. B. F. Freasier, C. F. Baes, Jr., and H. H. Stone, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1965, ORNL-3872, p.
127.

5. C. E. Bamberger and C. F. Baes, Jr., Reactor Chem. Div.
Annu. Progr. Rep. Dec. 31, 1966, ORNL-4076, p. 32.

6. R. P. Wichnerand C. F. Baes, Ji., Sidestream Processing for
Continuous lodine and Xenon Removal from the MSBR Fuel,
ORNL-CF-72-6-12 (June 30, 1972).

Technological  University,

fuel were estimated and, as expected, the results
indicated a strong dependence on the temperature
and hydrogen partial pressure. Low hydrogen
pressures and low temperatures favor the formation
of molecular iodine. High temperatures and low
hydrogen pressures favor the formation of atomic
iodine, while HI is formed at high temperatures and
relatively higher hydrogen pressures.

44 PROTACTINIUM OXIDE
PRECIPITATION STUDIES

O. K. Tallent L. M. Ferris

Studies of the selective precipitation of protac-
tinium oxide from MSBR fuel salt were concluded
and a publication’ was issued, an abstract of which
follows. : _

Molten LiF-BeF:-ThFs-UF4 (71.9-16-12-0.1 mole
%) containing dissolved PaFs was equilibrated with
various H0-HF-Ar gas mixtures at 535 to 670°C.
The data obtained indicate that the protactinium was
precipitated as Pa;Os and, hence, that the reaction
involved was PaFs(d) + 5/2H,0(g) = 1/2Pa,0s(c) +
SHF(g), for which the equilibrium quotientis h = Pir
Pfisz% NEL:S. The values of Q1 derived from the data
can be expressed as log Q1 =[12.50 — 10,690/ T{° K)]
+ 0.2. In these expressions (d), (¢), (g), P, and N
denote dissolved component, solid, gas, partial
pressure (atm), and mole fraction, respectively. The
point at which uranium oxide began to precipitate
along with protactinium oxide was determined at
several temperatures, using salt mixtures with
different UF4 concentrations. These data were con-
sidered in terms of the equilibria UF4(d)+2H20(g)=
UOa(ss) + 4HF(g) and 1/2Pa;0s(c) + 5/4 UF4(d) =
PaFs(d) + 5/4UQOx(ss), in which (ss) denotes UO»-
ThO: solid solution. The respective equilibrium
quotients can be expressed as log Q» = [9.27 -
8966/ T(°K)] £ 0.08 and log Qs=[-0.827-590/ T{° K)]
+ 0.1. These results are consistent with the estimated
thermodynamic properties of protactinium com-
pounds.

24

4.5 SOLUBILITY OF BF; IN SALTS
OF MOLTEN-SALT REACTOR INTEREST

S. Cantor

A paper, which has been published, gives detailed
measurements in four molten fluoride mixtures,each
containing LiF and BeF2;® an abstract follows.

Using a volumetric method, solubilities of BF;
were measured in four molten solvents, LiF-BeF (66-
34 mole %), LiF-BeF.-ThF:-UF, (71.7-16-12-0.23
mole %), LiF-BeF:-ThF; (76-12-12 mole %), and
LiF-BeF:-ThFs-NaF (66-15-11-8 mole %). The pur-
poses of the study were to investigate the interaction
of BF; with fluoride ions in these melts and to aid in
evaluating BF; in reactor control and in reactor
safety. Within the pressure and concentration ranges
studied, solubilities obeyed Henry’s law. Enthalpies
of solution, derived from the temperature
dependence of Henry’s law constant, varied from -15
to -18 kcal/mole; these large negative values indicate
strong chemical interaction between dissolved BF;
and molten fluorides, most likely the formation of
tetrafluoroborate ion, BF;. Entropies of solution
were interpreted as the loss of most of the rotational
entropy of BF;. From the observed solubilities,
estimates were made of the partial pressures of BF;
that would occur following accidental mixing of

‘reactor fuel salt with coolant liquid composed of

NaBF.-NaF (92-8 mole %). Barring massive In-
leakage of coolant into fuel salt, the high solubility of
BF; in fluoride melts would prevent very high partial
pressures of BFi. The rate of solution of BFs was
measured in quiescent reactor fuel salt. The results
indicate that BF; is useful in a control system for
“scramming” a molten-salt reactor.

7. O. K. Tallent and L. M. Ferris, “Equilibrium Precipitation
of Protactinium Oxide from Molten LiF-BeF,-ThF.-UF;-PaFs
Mixtures between 535° and 670°C.,” J. Inorg. Nucl. Chem. 36,
1277-83 (1974).

8. S. Cantor, “Solubility of BF1 in Salts of Molten-Salt Reactor
Interest,” J. Nucl. Mater, 47, 177 (1973).

25

5. Coolant Salt Chemistry

A. D. Kelmers

5.1 OXIDE AND HYDROXIDE CHEMISTRY

OF FLUOROBORATE MELTS
[.. Maya

. The work in thisarea isdirected toward developing
knowledge of the behavior of hydroxy and oxy
compounds in NaBF; that may be involved in
corrosion reactions or in reactions that lead to the
trapping of tritium as THO, TF, or tritiated com-
pounds in the fluoroborate coolant salt. Individual
compounds are being synthesized and characterized
to simplify the identification and understanding of
the behavior of hydrolysis products in actual systems.
The following compounds were prepared' ™ in the
period covered by this report: H;OBFs, HBF>,(OH):,
BF;-2H,0 [H3;OBF;OH], and NaBF;OH.

The first three compounds might be components
of the volatile fraction above the melt. Dihydroxy-
fluoboric acid [HBF2(OH),] appears to be a likely
candidate as one of the volatile components, since it is

thermally stable and can be distilled (bp 159°C). The .

proton nuclear magnetic resonance (NMR) (kindly
run by B. Benjamin, Chemistry Division) of all three
compounds consists of a single line. The respective
chemical shifts from TMS (tetramethylsilane) are
-10.8, -8.55, and -8.66 ppm. The presence of a single
line indicates a rapid exchange of protons between
hydronium ions, undissociated acid, and the anion.
This fast exchange could be important in providing a
mechanism to trap tritium in these acids if they are
present in actual systems.

An experiment was run whereby dihydroxy-
fluoboric acid (DHFBA) and NaBF:; were

simultaneously produced according to the reaction

4BF; + 2NaHF; + 3H;0 + B,0;
—~ 2NaBF, + 4 HBF(OH), (1)

The resulting slurry was distilled, ytelding DHFBA
and leaving NaBF. as the residue. Both products were
identified by analysis and x-ray powder diffraction
for the salt. This experiment demonstrates that under
certain conditions these species are compatible, and it
lends support to the hypothesis that DHFBA could
be volatilized out of partially hydrolyzed coolant
melts,

The fluorine NMR (run by A. Rutenberg,
Chemistry Division) for DHFBA and BF:-H:O
consisted of single lines in both cases; the shifts from

CsF¢ were -20.2 and -18.8 ppm, respectively. Thus,
the fluorine NMR could be of value as an identifica-
tion tool for these species.

The compound NaBF;OH was prepared according
to the reactions

H.O
BF3 + NaHC03 - NaBF;OH + COz
(ref. 3) (2)
2NaHF; + H3:BO;— NaF + NaBF;0H + 2H,0
(ref. 4) (3)

Reaction (2) gave a product of 80% purity; the
main contaminant was NaBFi. Reaction (3) gave a
product of greater than 90% purity. It was slightly
contaminated with NaF but was free of NaBF.. This
is an advantage since the BF ion can interfere with
infrared characterization of this product as well as
with that of Na:B.FsO, which will be prepared from
the NaBF;0OH.

The compounds described in this report, as well as
otners such as Na:B:FsO and Na;B3;FsOs, will be
completely characterized by chemical and physical
means (IR, NMR, DTA, etc.) to generate data that
will be useful in the identification of compounds
formed in actual NaBFs-NaF systems. Research will
be initiated to study the chemical interactions among
these compounds.

5.2 CORROSION OF STRUCTURAL ALLOYS
BY FLUOROBORATE MELTS

B. F. Hitch S. Cantor C. E. Bamberger
C. F. Baes, Jr.

When oxidants are introduced into NaBF,-NaF
melts in contact with nickel alloys containing iron
and chromium, complex fluorides are formed from
these elements. The chemical stabilities of these

1. F. J. Sowa, J. W. Kroeger, and J. A. Nieuwland, J. Amer.
Chem. Soc. 57, 454 (1935).

2. J.S. McGrath, G. G. Stack, and P. A. McCusker, J. Amer.
Chem. Soc. 66, 1263 (1944).

3. L. O. Gilpatrick, Synthesis of Sodium Hydroxy-
trifluoroborate, U. S. Patent 3,809,762 (May 7, 1974).

4. G. Ryss and M. M, Slutskaya, J. Gen. Chem. USSR 22,45
(1952).
elements are being determined to help predict the cor-
rosion behavior of Hastelloy N and other nickel
alloys in NaBFs-NaF. Previously, the free energies of
formation of NaNiF; and NaFeF; were determined
and the results were published, an abstract of
which follows.’

The free cnergies of formation of NaNiFs and
NaFeF; were estimated from measurements of the
following equilibria in molten NaBFa:

NaMF;(c) = NaF(d) + MFy(d)

where M represents either nickel or iron. The
necessary activities of NaF were estimated from
measurements of BF; pressures corresponding to the
dissociation pressures of the solvent, NaBFi. The
data obtained permit the calculation of corrosion
equilibria such as

2HF(g) + M(in a Ni alloy) + NaF(d)
= NaMF;(c) + Ha(g).

Corrosion by NaBF: of alloys containing
chromium usually results in the formation of
Na;CrFs, a green salt with a cryolite structure® which
is only slightly soluble in molten fluoroborate.” The
chemical stability of this compound has not been
quantitatively determined. Accordingly, the present
study is intended to determine the free energy of
formation (AG’) of Na;CrFs so that equilibria can be
predicted for reactions in which this compound is a
product. Such prediction will lead to a better
understanding of potential corrosion problems with
fluoroborate as well as alternative MSBR coolants
that contain NaF.

We attempted to use the following equilibrium to
yield AG' of Na;CrFe(c):

CrFi(c) + 3NaBFa(d)

= NaiCrFe(c) + 3BF;s(g). (4)
In the experiment, vapor pressures of BF3; were
measured in a vessel charged with CrFs, NaBF,, and
Na;CrFs. The latter compound was synthesized by
heating a 3:1 solid mixture of NaF and CrF; with a
small quantity of NaBF.; x-ray diffraction analysis
indicated only Na;CrFsafter heating. During the first
experimental run, CrF3; was added until successive
increments gave the same logarithmic plot of BF3
pressure versus reciprocal temperature. Following
the vapor pressure measurements, the vessel was
quenched, cut open, and samples removed for
petrographic and x-ray analyses. Petrographic ex-
amination revealed another NaF-CrF; crystalline
phase in addition to NaiCrFe. The x-ray powder
pattern of this phase showed lines and intensities
matching those reported® for the compound,

26

NasCr;F 2. No CrF; was present in the quenched salt.
It is quite possible that, instead of reaction (4), we
have measured the equilibrium

1/3NasCr3sFis(c) + 4/3NaBFa(d)
= Na;CrFe(c) + 4/3BF;(g). (5

The additions of CrF; to the reaction vessel may have
altered the proportions of the two complex com-
pounds; i.c.,
5/3 Na3CrFe(c) + 4/3 CrFs(c) _
- NaSCr3F14(s). (6)

Chromium(III) chemistry in molten NaBF; is
apparently more complicated than originally an-
ticipated. Conceivably, NasCr:Fi4 may be a corro-
sion product in fluoroborate coolant. This com-
pound is green and, without a microscope, cannot be
visually differentiated from NaiCrFes. However, x-
ray analysis of green crystals, selected from
fluoroborate that circulated in an Inconel loop,’
indicated only Na3CrFe.

Another series of BF3; vapor pressure
measurements are being carried out to investigate the
effects of NaF concentration on a mixture of CrFs,
NaiCrFs, and NaBF.i. These measurements are being
supplemented by quench studies in which encap-
sulated salt of the same composition as that present in
the vapor pressure vessel 1s examined to help
determine what conditions favor the formation of
NasCr3Fi.. In addition, other methods for deter-
mining the AG’ of Na;CrFs are under consideration.

5.3 EVALUATION OF FLUOROBORATE
AND ALTERNATE COOLANTS"

S. Cantor
The objective of this study is to evaluate the

performance and safety characteristics of
fluoroborate coolant NaBF4NaF (92-8 mole %) and

5. C. E. Bamberger, B. F. Hitch, and C. F. Baes, Jr., J. lnorg.
Nucl. Chem. 36, 543 (1974).

6. G. Brunton, Mater. Res. Bull. 4, 621 (1969).

7. C. J. Barton, J. Inorg. Nucl. Chem. 3, 1946 (1971).

8. A. DeKozak, Compt. Rend. 268C, 416 (1969).

9. A. N. Smith, Experience and Sodium Fluoroborate Circula-
tion in a MSRE-Scale Facility, ORNL-TM-3344 (September
1972).

10. The work reported here covers the period ending August 31,
1974. Soon thereafter, a committee was appointed to further
examine and review the factors that govern the choice of coolants
and to recommend action that should be taken with regard to the
MSBR coolant.
other coolants as these characteristics relate to the
functions of the MSBR intermediate coolant. In an
MSBR, the coolant serves two basic functions: (1) to
bridge the temperature difference between the freez-
ing point (500°C) of the fuel salt and the feedwater
temperature (about 300°C) of the steam system, and
(2) to act as a protective barrier between the highly
radioactive fuel circuit and the steam system.

To fulfill these functions, the following nine
criteria have been identified.

(1) The chemical composition of the coolant is
required to be limited to compounds or elements that
are tolerable in the fuel or which may be removed
from the fuel; the limits are necessary to minimize or
reverse the effects of leaks between fuel and coolant.

(2) If leaks occur in the steam-generating system,
mixing of coolant and water must not cause violent
chemical reactions or cause failure of major plant
components.

(3) The coolant should have heat-transferand fluid
characteristics compatible with the relatively low fuel
inventories characteristic of molten-salt reactors.

(4) The degree of corrosion by the coolant must be
acceptable.

(5) Normal radiation levels within primary heat
exchangers should not greatly increase the corrosivi-
ty or vapor pressure of the coolant.

(6) The coolant should have a freezing point below
330°C (626°F) to permit a 304°C (580° F) feedwater
temperature for conventional supercritical steam
cycles; higher coolant freezing points entail cost
penalties, but if the feedwater must be preheated to
427°C (800°F) to prevent the coolant from freezing,
the increased cost penalties are probably too great.'’

(7) The vapor pressure of the coolant should be low
at the highest normal operating temperature; if the
vapor pressure is not low, it is advantageous if the
vapor does not condense as a solid.

(8) Since the coolant will probably have to play a
significant role in preventing tritium from entering
the steam system, it is advantageous if the coolant or
tolerable additives in the coolant can chemically
scquester tritium entering the coolant circuit.

(9) The coolant should be low in price and available
in high purity.

The nine general attributes stated above can serve
to eliminate many possible coolants. Sodium and
other alkali metals, although possessing superb heat-
transfcr properties, are unacceptable because leaks
into the fuel will chemically reduce and precipitate
the fuel. Fluids that contain a high concentration of
oxide would also precipitate the fuel in case of leaks.
Low-melting molten salts that contain ions more

27

oxidizing than Ni** would be too corrosive. The
heavy liquid metals, lead and bismuth, are too
corrosive toward nickel-base structural alloys. The
MSRE coolant, 'LiF-BeF; (66-34 mole %) is un- -
acceptable because of its high melting point (860° F)
and perhaps because it is too costly.

The present candidate coolant NaBF:-NaF (92-§
mole %) has the advantages of low cost, minimal
radiation decomposition,'’ and low viscosity. The
thermal conductivity and heat capacity are adequate.
The freezing point of 384°C (723°F) necessitates
some modification to a conventional supercritical
steam system. The decomposition pressure of BF;
from this coolant requires measures to protect pump
parts and lubricating oils.

A disadvantage of fluoroborate coolant is the
increased corrosion and mass transfer that follows
inleakage of steam or moisture.”” An associated
complication is the formation of the relatively
insoluble corrosion products, NasCrFs and NaNiF;,
which may foul steam-generator surfaces if they are
not removed from the coolant circuit. Although the
inherent corrosivity of fluoroborate is probably
tolerable, corrosion reactions with minor alloy con-
stituents (Cr, Ti, Mn, Si, Hf) in the primary heat
exchangers may cause radiation embrittlement; this
embrittlement would be started by boron deposited
on tube surfaces as metal borides, followed by
diffusion of boron into tube walls; the helium
produced within the tube walls by delayed neutrons,
"“B(n, *He)iLi, would probably shorten the design life
of the heat exchangers. ,

The consequences of fuel-coolant interleakage are,
at the very least, unpleasant and may represent a
serious shortcoming of fluoroborate. Small leaks of
coolant into fuel will lead to the release of some BF3;
that might damage the charcoal beds; BF; dissolved
in the fuel can probably be removed by inert-gas
sparging; but sodium ions, which cannot be removed
by present chemical treatments, will slightly decrease
the breeding performance of the fuel. A large
inleakage of coolant will have more deleterious
effects; besides possible pressure surges due to BF;
release, damage to the charcoal beds could also
involve release of krypton and xenon, and, although
most of the coolant will probably be immiscible,

11. R.C.Robertson (ed.), Conceptual Design Study ofa Single-
Fluid Molten-Salt Breeder Reactor, ORNL-4541 (1971), p. 80.

12. E. L. Compere et al., J. Nuci. Mater. 34, 97 (1970).

13. J. W. Koger, Corrosion and Mass Transfer Characteristics
of NaBFy-NaF (92-8 mole %) in Hastelloy N, ORNL-TM-3866
(October 1972).
enough NaF may dissolve in the fuel salt to
necessitate discarding part of the fuel salt. Any
inleakage of fuel into the coolant will be followed by
precipitation of the fuel; immiscible droplets of fuel
salt (whose freezing point is 500° C) will either freeze
in the cold leg (454°C) of the coolant loop or, more
likely, the LiF will dissolve in the coolant and the
remaining fuel salt components will precipitate out.
Sequestering of tritium in fluoroborate coolant is,
as yet, undemonstrated. Although an ionic species,
BF;OH , can be maintained in molten NaBFs-NakF,
the protium in this species does not readily exchange
with other hydrogen isotopes.' There is evidence " of
another dissolved protonic species that may ex-
change more readily with tritium. Oxidation within
the coolant may be another chemical method for
trapping tritium (e.g., T2 + Ni** — Ni°® + 2T"). The
oxidation may be aided by the presence of dissolved
oxide: T2 + Ni** +20° = Ni° +20T". Both methods,
isotope exchange and oxidation, will increase the
corrosivity of the coolant. The increased corrosivity
must be tolerable to materials of construction.
Fluoride mixtures composed of NaF and BeF;
with lesser amounts of LiF are possible alternate
coolants. Examples are two eutectic mixtures, NaF-
BeF»,-LiF (53-42-5 and 43-36-21 mole %), freezing
points of either mixture permit conventional
feedwater temperatures. Other advantages are
minimal effects in case of leaks and negligible vapor
pressure and radiation decomposition. Volumetric
heat capacities and thermal conductivities are greater
than those of fluoroborates;, viscosities are also
greater (by factors of 10 to 20) and may require a
significant increase in pumping requirements. These
melts are less corrosive than fluoroborate; however,
in the event of steam leaks, the insoluble corrosion
products (BeO and perhaps Na3CrFe) may also foul
steam generator surfaces. A disadvantage of these
mixtures is their high price; 8500 ft’ of NaF-BeF-
"LiF (43-36-21 mole %) costs about $8 million, about
twenty times more than an equal volume of
fluoroborate coolant.'® Cost and viscosity disadvan-

28

tages may be alleviated by substituting Mgkz, AlF;,
or ZrF, for part of the BeFz; optimizing the coolant
circuit may also reduce the volume and cost of the
coolant inventory. As is currently the case with
fluoroborate, these mixtures may also be capable of
trapping tritium by means of protonic or oxidizing
additives. _

Other single intermediate coolants (e.g., helium or
molten chlorides) as well as double-coolant com-
binations (e.g., a molten salt in a first coolant loop
followed by a second molten salt or gaseous coolant
in another loop) are currently under consideration.

5.4 DENSITY AND VISCOSITY OF
SEVERAL MOLTEN FLUORIDE MIXTURES

S. Cantor

The measurements of density and viscosity of
several molten fluoride mixtures have been
reported.17 The objectives of this work were (1) to
determine, with high accuracy, densities and expan-
sivities of several molten fluoride mixtures of
significance to molten-salt reactors, (2) to derive
additive molar volume contributions that can serve to
predict densities in LiF-BeF,-(Th,U)F4 melts, (3) to
estimate density changes upon melting of the fuel-
carrier and coolant salts and, (4) to determine the
viscosity of fluoroborate coolant and of fuel-carrier
salt. An abstract of the report, modified to include
least-squares equations of the data, follows.

Table 5.1 gives densities, which were obtained by
using a dilatometric method, for several molten-salt

14. J. B. Bates et al., MSR Program Semiannu. Progr. Rep.
Feb. 29, 1972, ORNL-4782, pp. 59-62.

I5. D.L. Manningand A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL4782, pp. 83-84.

16. J. P. Sanders, A Review of Possible Choices for Secondary
Coolants for Molten Salt Reactors, ORNL-CF-71-8-10, (Avg. 6,
1971), p. 12.

17. S. Cantor, Density and Viscosity of Several Molten
Fluoride Mixtures, ORNL-TM-4308 (March 1973).

Table 5.1. Densities of specified molten-salt compositions

Density (g/cm’) at

Composition Mole % temperature t (°C)
Li-BeF: 66-34 2.280 — 4.88 X 107t
LiF-BeF:-ThF. 70.1-23.9-6.0 30112 -6.71 X 107"
70-18-12 3.824 — 8.06 X 107"t
70-15-15 4.181 —9.53 X 107%
LiF-BeF;-ZrF; 64.7-30.1-5.2 2.539 - 577 X 107
LiF-BeF:-ZrFs-UF. 64.79-29.96-4.99-0.26 2.553 — 5.62 X 107
NaBF.-NaF 92-8 2.252 — 7.1l X 107%
KNO; Pure 2.125 - 743 X 107

compositions. The last salt was measured to assure
the accuracy of the method; the densities measured
for KNO; agreed within 0.15% with critically
evaluated densities obtained by Archimedean
methods.

For the fluorides, molar volumes obtained from
the density measurements agreed within 2% with
volumes calculated from additive contributions of
the components. The expansivities of three LiF-
BeF»>-ThF. mixtures were practically identical, 2.5 X
107 per °C. :

Density-temperature curves from 25 to 700°C for
LiF-BeF:-ThF, (72-16-12 mole %) and for NaBF,-
NaF (92-8 mole %) were derived from room-
temperature pycnometric determinations and from
estimated expansivities of the solid salts. The
calculated expansion upon melting 1s 7% for LiF-
BeF,;-ThF, and is 89 for NaBFs-NaF.

29

The following viscosities of three salt mixtures
were determined by oscillating-cup methods.

NaBF. -NaF (92-8 mole %) _
n(cP) = 0.0877 exp[2240/T(°K))
LiF-BeF.-ThF. (72.7-15.7-11.6 mole %)
n = 0.1094 exp[4092/T (°K)]

(70.1-23.9-6.0 mole %)
n = 0.06602 exp[(4380/T(°K)]

Viscosity measurements were conducted at Mound
Laboratory, Miamisburg, Ohio, using capsules and
samples prepared at ORNL. The viscosities of the
two melts composed of LiF, BeF2, and ThF, were
analogous to viscosities reported for similar mixtures
containing UF, instead of ThF..
30

6. Trittum Behavior

A. D. Kelmers

6.1 PERMEATION OF HYDROGEN ISOTOPES
THROUGH STRUCTURAL METALS
R. A. Strehlow H. C. Savage

One means of impeding the permeation of tritium
from the reactor fuel through the heat exchangers
and coolant into the water in the steam system is to
develop an oxide film on the steam side of the steam-
raising system. An initial investigation of oxide films
was completed and the results were published, of
which the following is an abstract.'

The permeation and the pressure dependence of
the permeation of hydrogen isotopes through metals
and oxidized metals were studied at temperatures
from 300 to 800°C and at pressures of 107 torr to 1
atm. Such knowledge is important to tritium
management in both fusion and fission nuclear
reactors. An adequate basis for predicting the
permeation of hydrogen at very low pressures has not
previously been established; therefore, the two com-
plementary objectives of this study were (1) to
determine the pressure dependence of hydrogen
permeation through materials of which steam
generators might be built, and (2) to determine
whether an oxide film might serve as a tritium
permeation barrier.

The metals studied included nickel, Type-403 L
‘stainless steel, Hastelloy N, Incoloy 800, Croloy T9,
Croloy T22, and Type-406 stainless steel. Deuterium,
rather than normal hydrogen, was used as the
permeating gas to achieve high sensitivity in the mass
spectrometric analyses. At a given temperature, the
permeation rate of deuterium through metals thatare
substantially free of oxide films was found to proceed
with a half-power pressure dependence in accordance
with the relationship

J= K(P!"* - Py},
where J is the permeation flow rate, K is a constant,
and P, and P; are the upstream and downstream gas
pressures, respectively.

The rates of the permeation of deuterium through
oxidized metals were usually lower than through
unoxidized metals, and the observed pressure
dependence was frequently greater than the half-
power. For some alloys, a reduction in permeation
rate by a factor of 100 or more at l-atm pressure was
observed; the reduction at low pressures was even

greater. The observations made are consistent with
considerations of the chemical stability of the oxide
and of the presence of cracks or other imperfections
in the oxide film.

Additional research on this subject has been
initiated and results will be reported in the future.

6.2 THE SOLUBILITIES OF HYDROGEN,
DEUTERIUM, AND HELIUM
IN MOLTEN Li;BeF,

A. P. Malinauskas D. M. Richardson

The results of this experimental work have been
published and an abstract follows.”

Solubilities of hydrogen, deuterium, and helium in
molten Li;BeF,; have been measured. Values of the
Ostwald coefficient K. (equilibrium ratio of gas con-
centration in the dissolved state to the gas-phase con-
centration), which were determined for helium, can
be described over the temperature range 773 to
1073°K by the expression log (I0'K.) = log T —
(1177/T) — 0.7954, whereas the combined data for
hydrogen and deuterium can be represented over the
temperature range 773 to 973 °K by the expression
log (10°K.) = log T — (1535/T) — 0.7684.

6.3 CHEMISORPTION OF TRITIUM
ON GRAPHITE

R. A. Strehlow H. E. Robertson

The discovery of significant amounts of tritium on
the graphite from the MSRE prompted a study of the
chemisorption of tritium on graphites. The MSRE
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 was constructed and operated to
permit exposure of graphite at elevated temperatures
to a gaseous mixture of tritium in helium. The
apparatus is shown schematically in Fig. 6.1. The

I. R. A. Strehlow and H. C. Savage, “The Permeation of
Hydrogen lsotopes Through Structural Metals at Low Pressures
and Through Metals with Oxide Film Barriers,” Nucl. Technol. 22,
127 (1974).

2. A. P. Malinauskas and D. M. Richardson, Ind. Eng. Chem.
Fundam. 13, 242 (1974),
31

ORNL—-OWG 727515

P1 P2
| bttt v 2T ’
) ] 1i r H
l ' | VACUUM [1 T i
| PuMP § —TC  1|ANALYZER
________ | Rnp——— :
— N
] TC
o 1
UH Ra
F3,Cu0
 ————
— }
-DG%:T:,DQ' o
o ! F1, SAMPLES

|
I

|

! F = FURNACE
} R = ROTAMETER
|

]

TC = THERMOCOUPLE VACUUM GAGE

j" OUT OF HOOD COMPONENTS

Fig. 6.1. Apparatus for study of tritium sorption on graphite.

Table 6.1. Tritium sorption on commercial graphites

Exposure conditions

Type of Concentration Temperature Amount of tritium sorbed
graphite  (uCi of T2/cm’ He) ©C) (ng T2/ g)
POCO 8 750 0.67 (surface)
0.07 (internal)
CGB 8 750 > 1.0 (surface)
0.03 (internal)
POCO 0.77 780 0.3 (surface)
0.04 (internal)
CGB 0.77 780 0.4 (surface)
0.016 (internal)
H-327 6.4 790 0.18 (surface)

0.068 (internal)

furnaces containing copper oxide (F2 and F3 in Fig.
6.1) were operated in conjunction with the apparatus
for water analyses as well as continuous tritium
analyses. These furnaces converted tritium to the
oxide for use in some studies of the interaction of T.0
with graphite. The exposure of l-cm® samples of
graphite to the tritium-helium mixtures was carried
out in the sample furnace (F1 in Fig. 6.1) in a quartz
reactor. The tritium concentration was in the range of
| to 10 ppm by volume, partial pressures of about 0.1
Pa (107 torr). Data were obtained on four types of

graphite: CGB (MSRE-type), POCO, H-327, and a
carbon-black—pitch-bonded graphite A-681 (madeat
ORNL). After exposure conditions of about 4 hr at
750° C, the graphite pieces were sampled by removing
successive 0.0152-cm (0.006-in.) samples that were
analyzed for tritium. The tritium contents of the
sample were substantially the same as those found in
the examination of MSRE graphite, with values of
several times 10° dis min™" g”' for the outermost cut,
decreasing to values below 10° dis min™ g™ for
interior samples. Tritium contents of 6 to 8 X 10'° dis

min ' g ' were found for some samples of high surface
area (1 g tritium = 2.2 X 10'® dis min™").

The gradient of tritium concentration within the
graphite qualitatively reflected the microstructure of
the specimens (Table 6.1). Graphite sample H-327,
with a ratio of 2 between external and internal
tritium, appeared under the microscope to be most
porous, whereas type CGB, which was highly im-
pregnated, had a ratio of over 300. The third cut was
found in several tests to be near the mean value at
great depths. The data in Table 6.1 indicate a square-
root dependence of sorption amount with T: in
pressure, and the dissociative process suggested is

1/2Tx(g) + C(s) = T-C(s) .

Rinsing with water and alcohol did not remove
significant amounts of tritium; hence, the chemical
form of the tritium is uncertain. However, the large
amount adsorbed at these very high temperatures is
surprising.

The effect of surface area isshown by data in Table
6.2 obtained with several varieties of graphite that

32

had been oxidized with steam to increase their surface
areas and were simultaneously exposed. All of the
oxidized and unoxidized samples sorbed tritium to
the extent of about 5 X 10'? atoms/cm” with some
differences attributable to graphite types. A similar
experiment with T>O as the sorbent resulted in about
5 X 10" atoms/cm® except for the extensively
impregnated CGB type. The tritium sorbed in this
latter experiment possibly was sorbed as T; rather
than T;0. These data are shown in Table 6.3.

The amount of tritium in the moderator of an
MSBR cannot be estimated with precision because of
the uncertain effect of neutron irradiation damage to
the graphite and the achieved tritium pressure. From
the data obtained so far in this work, a few milligrams
of tritium per kilogram of graphite would be ex-
pected. If the kinetics prove to be adequate, carbon
might be useful in coolant processing for the removal
of tritium. The daily production rate of tritium in an
MSBR is about 250 mg. Thus, a processing method
using perhaps a few hundred kilograms of graphite
could possibly be effective for tritium removal.

Table 6.2. Tritium chemisorbed on three graphite types after exposure
to tritium at 750°C for 6.5 hr
Pr, = 0.14 Pa (1.1 X 10 torr)

Cut No. (ug T2/g ©) BET
Graphite type surface area Tritium”
and sample number | 2 3 (m*/g) (atoms/cmz)
A681
17 1.3 0.86 0.86 1.43 1.21x 10"
18° 3.4 2.7 2.3 2.25 2.1 X 10"
21 0.31 0.11 0.981 0.203 g x10"
CGB
9° 2.1 0.79 0.52 1.68 6.2 X 10"
10° 3.0 0.88 0.59 2.31 5.1 % 10"
13 0.24 0.011 0.0076 0.218 7 % 10"
i4 0.29 0.011 0.0077 0.319 5 x 10"
AXF-5QBG
1° 0.59 0.21 0.151 0.72 4 % 10"
2? 0.32 0.131 0.081 0.757 2.1 X 10"
5 0.40 0.026 0.027 0.180 3 X% 10"
6 0.30 0.030 0.028 2.6 X10"

0.213

“Based on cut No. 3.
*Oxidized.
33

Table 6.3. Tritium chemisorbed on three graphite types after exposure to T:0 for 4 hr
Py,0-=0.04 Pa (3 X 107 torr) and at several temperatures

Cut number (u Tz/g C) BET
Graphite type Temperature surface area Tritium*
and sample number °C) 1 : 2 3 (m?/g) (atoms/cm’?)
A681
3? 728 0.21 0.144 0.117 1.51 1.6 X 10"
12° 194 6x 107 6x 107 5% 107 2.08 s x10°
8 626 0.19 0.15 0.13 0.506 5 X 10"
CGB
2° 726 0.17 0.095 0.074 1.66 9 x 10"
1 725 0.019 5x 107 2x10° 0.292 24 % 10
AXF-5QBG (POCO) '
5° 728 0.11 0.0086 0.0054 1.04 1 xX10"°
4 728 0.070 0.0029 0.0019 0.205 1.9 X 10"

“Based on cut No. 3.
*Oxidized.
34

7. Other Research

7.1 THE WETTING OF GRAPHITE
BY BISMUTH AND BISMUTH-LITHIUM
ALLOYS AS DETERMINED BY
THE SESSILE DROP METHOD

G. E. Creek C. E. Bamberger

Alloys of bismuth and lithium are used in process-
ing the fluoride salts of a molten-salt reactor, and,
since graphite is being considered as a material of
construction for the processing equipment, it was of
interest to determine whether graphite would be
wetted by bismuth-lithium alloys. If the contact angle
(measured through the liquid) of a sessile drop resting
on a smooth surface of a solid is less than 90°, the
liquid is said to wet the solid.'

The objective of these experiments was to measure
the contact angles between sessile drops of bismuth
and various bismuth-lithium alloys and surfaces of
various types of graphite. After determination of the
contact angles, values of the surface tension and
density of the various alloys could be computed as
additional information.

To perform these measurements, a silica furnace
with a flat window and appropriate gas delivery and
exit lines was constructed.” To test the procedure, the
contact angle of bismuth on ATJ graphite was
measured at several temperatures. The correspond-
ing surface tension was estimated to be 298 dyne-cm
at 564°C and 327 dyne-cm at 704°C; the values
predicted from literature values’ were 359 and 348
dyne-cm, respectively. The disparity between the
measured and predicted values could be due to
adsorbed ‘gases,' or the fact that the contact angles
were all greater than 150°, making the contact points
difficult to measure.” At this time, the project was
halted and the equipment was dismantled.

The only conclusions we are able to draw concern-
ing the applicability of graphite as a container for
bismuth and bismuth-lithium alloys are those based
on a review of the literature. In the experiments
described above, using molten bismuth on type ATJ
graphite, the minimum value obtained for a contact
angle was 155° at 704°C. Since the criterion for
wetting is a contact angle of less than 90°, no wetting
occurred. The surface tension (7) of bismuth at 470°C
is 363 dyne-cm, with a temperature coefficient
(dt/dt) of -0.069 dyne-cm/° C.* The surface tension
of lithium at its melting point' is 398 dyne<m with a
temperature coefficient of -0.14 dyne<cm/°C. At

700° C, the surface tensions of bismuth and lithium
would be 348 and 325 dyne-cm, respectively; thus,
one would expect the surface tension of alloys of
these two metals, particularly those of low lithium
concentration,to differ by a negligible amount from
that of pure bismuth at this temperature.

However, wetting is not the only important aspect

" to consider when studying compatibility. The reac-

tions of lithium with graphite to form intercalates and
carbides change the geometry of the componentsand
very probably change the nature of the contact
surface.

If the lithium-bismuth alloys are used as proposed,
the presence of other metals, such as uranium,
cerium, and other fission products, must be expected.
Klamut et al.’ reported that CeC: was identified on
graphite contacted with 25 ppm cerium in bismuth at
700°C for 110 hr, and that 180 ppm neodymium in
bismuth reacts with graphite, with the neodymium
concentrated at the graphite—liquid-metal interface.
From diffusivity studies and because bismuth seems
to permeate the pores of reactor grade graphite,
Klamut et al.” indicated that uranium, magnesium,
zirconium and fission products may diffuse into
graphite.

Cavin et al.’ report that pure lithium reacted with
graphite at 700°C to form LixCy; however, they
estimate that lithium-bismuth (50-50 at. %) would
not react. They also report that tests with graphite
crucibles indicate that the major problem is penetra-
tion of liquid metal into the pores. Although the
project was terminated at a stage too early to obtain
the data sought, the results quoted lead us to infer
that graphite might not be a good choice for
components exposed to alloys of such elements as Bi-
Li-Re.

1. J. W. Taylor, “The Wetting by Liquid Metals,” Frogr. Nucl.
Energy Ser. 5 2, 398416, (1960).

2. A full description of the experimental details of the
measurements and proposed alloy preparation and transferring
apparatus can be found in ORNL-CF-73-12.

3. D. N. Staicopolus, J. Colloid Sci. 17, 43947 (1962).

4. B.R.T. Frostetal,“Liquid Metal Fuel Technology,” Progr.
Nucl. Energy Ser. 4 Technol. Eng. Safety 2, 411-16 (1960).

5. C. J. Klamut et al., “Material and Fuel Technology for an
LMFBR,” Progr. Nucl. Energy Ser. 4 Technol. Eng. Safety 2,
463—64, (1960).

6. O.B. Cavin, J. L. Griffith, and L. R. Trotter, “Compatibility
of Materials with Bismuth,” Molten-Salt Reactor Program
Monthly Report for November 1972, MSR-72-82, p. 27
35

8. Development and Evaluation of Analytical Methods

A. S. Meyer

8.1 IN-LINE ANALYSIS OF BREEDER FUEL
R. F. Apple J. M. Dale
D. L. Manning A. S. Meyer

In-line electrochemical analyses of the fuel are
planned for every major installation and many minor
facilities of the MSR Program. Such measurements
will determine the redox condition of the fuel
(U*/U" ratios) and, at present, allow estimates of
the concentration of corrosion product ions. The
performance of such analyses in a variety of systems
will also provide practical experience that will be
useful in perfecting in-line analytical devices for
ultimate applications to reactor salt streams.

The first installation to resume operation was
FCL-2b, a forced convection loop in which
fluoroborate coolant had been circulated. This loop,
which is designed to study corrosion of structural
materials under conditions simulating flow velocities
and temperature gradients in an MSBR, had been
modified to provide a single penetration for elec-
trodes. An eclectrode assembly consisting of two
iridium electrodes in a 3/4-in. shield tube has
recorded several thousand voltammograms from the
initial filling of the loop and throughout controlled
reduction of the fuel by beryllium additions. These
measurements have provided initial experience in
electrochemical measurements in breeder fuel com-
positions. The solvent for the initial runs, LiF-BeF:-
ThFs (68-20-12 mole %), will later be replaced with
the reference fuel composition (72-16-12 mole %).
Measurements in FCL-2b will, therefore, provide a
comparison of the electroanalytical properties of the
reference fuel with those of the more familiar MSRE
compositions (LiF-BeF:-ZrF:) and, at least
qualitatively, the effects of variations in composition
in the region of the reference fuel. Also, because the
conditions of measurement are far from ideal (the
electrodes are located in an auxiliary vessel that is
affected by pump-induced vibrations), experience is
gained in making measurements under unfavorable
conditions. Currently, diffusion-limited voltam-
mograms can be obtained only when the pump is
stopped and measures are taken to disrupt convective
flow patterns within the electrode shield. A novel
clectrode assembly is being fabricated which contains
five assorted electrodes within a 1/2-in. shield tube

that can be heated in the region above the melt level to
minimize convection.

Measurements have been made of several hundred
electroanalytical waves including peak heights and
potentials and parameters for the assessment of
ideality of wave shapes. Quantitative interpretation
of these measurements is now in progress but is
somewhat involved because both the temperature of
the melt and the area of the exposed electrode
changes continually after the pump is turned off.
These values are expected to be sufficient to permit a
comparison of the electroanalytical properties of the
present and reference fuels, and certain qualitative
observations can be made.

Because of the absence of zirconium, the thorium-
bearing fuel exhibits a higher cathodic limit than
MSRE compositions; this allows the observation of
the U — U" reduction wave as well as the useful U*'
— U™ wave. The measurements confirm the validity
of the voltammetric U**/ U™ ratio method in this fuel.
As expected, the initial charge was quite oxidizing,
U*/U* ~ 10°. Additions of beryllium metal in
increments of from 8 to 60 mg reduced the uranium in
an expected manner. In the mid-range reductions, the
yield of trivalent uranium reached a value equivalent
to about 90% of the added beryllium metal. Inearlier
additions, a substantial fraction of the beryllium
metal was consumed by FeF: and other oxidants,
while in the more reducing melts some CrF: was
reduced. Reductions were continued until about 2%
of the uranium was converted to U** and a significant
reduction in the height of the Cr** — Cr’ wave was
observed. The data will allow an estimate of the
equilibrium quotient for the reaction

20% + Cr'— 2U™ + Cr™.

The measurements have also pointed out problems
in the application of stripping techniques to practical
systems—even small variations in flow must be
eliminated—and have revealed interesting anomalies
in the behavior of Cr’” in this melt. The chromium
reduction wave is generally better separated from the
uranium wave than in LiF-BeF:-ZrFs melts. It has
been observed, from stripping measurements, that
the plating of chromium virtually ceases when
reductions are made at potentials several hundred
millivolts more negative than the reduction wave
(Fig. 8.1). Earlier it had been postulated that such an
36

ORNL-DWG. 74-10657

o
g ol 0 onowowno o 1 RIPPING
52 PLATING = QUEOY¥ete %  WAVES FOR
2 f Fe-GCr ALLOYS
0 - _—
'_" L CONDITIONS :
E : L Melt : see below
x 2 Plating : 40 seconds
xra at indicated
2.2 21— potentials ]
o - Stripping Scans :
- 0.1 Vsec N
4 [~ —
I ] T
6 . —
CONDITIONS : VOLTAMMETRIC WAVE
o - Fuel Solvent : LiF-BeFp-ThFy N
E (68-20-12mol %)
" 4 Melt Temperature : ~845°C
- " Uranium Concentration : 1.4 wt percent ]
= U4+/ U3* Ratio: ~10%
@ o [ Total Chromium Concentration: ~100 ppm
xs Electrodes: 18ga Iridium, ~1mm diameter
3 2 21— SconRate: 01V sec X ~Tem insertion —
S . DERIVATIVE OF VOLTAMMETRIC
o | WAVE (y-scale arbitrary) _
r 1 START OF VOLTAMMETRIC SCAN
Y 0
=
s |
e
=5
Seo»
5 cro— cr2
| 1 [ ] ! ] I | ] l 1 | [
+0.4 +0.2 0 -0.2 -0.4 -0.6 -0.8 -1.0

POTENTIAL vs IRIDIUM QUASI REFERENCE ELECTRODE, volts

Fig. 8.1. Typical electroanalytical measurements of MSBR fuel in a forced convection loop (FCL-2b).

effect could occur as a result of homogeneous

reduction of Cr** by U** diffusing from the working
electrode, but this effect had not been demonstrated
in MSRE fuel, and there was no literature reference
to similar phenomena, at least for inorganic systems.
When the temperature of the fuel is increased, the
height of the chromium waves decreases, followed by
slow recovery. This behavior is unexpected because
higher temperatures tend to favor increased dissolu-
tion of chromium and reaction rates have generally
been assumed to be quite rapid in the molten fuels.
Work is now in progress to examine the fuel during
reoxidation by nickel fluoride additions.

8.2 DETERMINATION OF TRIVALENT
URANIUM IN SIMULATED MSRE FUEL
J. M. Dale D.L.Manning A.S. Meyer
The equipment for the automated in-line deter-
mination of U™/U" ratios in molten-salt reactor
fuels' was used to measure the reduction of the

synthetic fuel charged to fuel pins (metal capsules
fabricated from proposed structural alloys) for an in-

l. J. M. Dale and A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 69.
pile experiment, TeGen-1. This experiment was the
first of a series to investigate the effects of fission
products, particularly tellurium, on the intergranular
corrosion of alloys for molten-salt reactor fuel
containment. To simulate the conditions of reactor
operations in which the fuel would be maintained ina
reducing condition, the U’/ U ratio in the charge
was adjusted to a level sufficient to maintain a
reducing fuel (approximately 1% of the uranium as
U’") despite the oxidation resulting from the fission
process. Moreover, calculations showed that oxide
on the inner surface of the capsules might be
sufficient to consume a substantial fraction of the
U™, and therefore, a pre-equilibration contact of the
pin with the fuel was required to assure a known
concentration of U™ in the capsules. One proposal
was to drain the pins after the equilibration period
and recharge them with reduced fuel. This operation
would require cooling the system and breaking and
reconnecting numerous fittings, a procedure that
might )jeopardize the experiment. It would also
consume time from an exceedingly tight schedule.
Accordingly J. H. Shaffer, Chemical Technology
Division, designed the following procedure to permit
analyses of the equilibrated fuel and requested that
we adapt our analytical techniques to it. _

The fuel (MSRE solvent with ***UFs) was mixed in
a reaction vessel and partially reduced by hydrogen
sparging. After analysis, the U’'/U" ratio was
adjusted to the desired range by successive treatments
with beryllium metal. A portion of the fuel was then
transferred to a prefill vessel connected to the fuel pin
assemblies; this portion was again analyzed. The pins
(three sets of four pins, each set fabricated from a
different alloy) were filled and held at temperature for
four days. Following this equilibration, the excess
fuel from each set of pins, about 4 cm’, was
sequentially blown back into a small cavity at the
bottom of the prefill vessel and again analyzed. 1f the
ratios remained within the desired reducing range,
the filling operation would be terminated and the
costly refilling operation would be eliminated.

Electrode systems to fulfill these unusually exac-
ting requirements were designed and installed.
Because of the hazards from the ***U in the fuel, the
operations were done in a dry box in an alpha
containment facility. To protect the equipment from
contamination, the electronic  components,
voltammeter and PDP-8/1 computer were located in
an adjacent room. Thus, the operation fulfilled many
of the criteria of the proposed use for in-line analysis,
including remote operation and the rapid acquisition
of data without the hazard of sample withdrawal.

37

Table 8.1. MSR-ORR capsule TeGen-1 fuel analysis

Operation U*wt%
Preparation container
After fuel hydrogenation 1.88
Fuel frozen—remelted 1.76
Prefill container
After transfer from prep container 1.83
After pin equilibration
Stainless steel 304-L 1.75
Inconel 601 1.61
Hastelloy N 1.50

Table 8.1 summarizes results of the analyses. The
agreement among the first three values reflects both
the precision of the analyses and the care in the
performance of the operation by the personnel
involved. The ratios found in the samples from the
fuel pins follow a trend that would be expected if the
fuel were partially oxidized during the transfer, then
reduced during equilibration in proportion to the
activity of chromium in these different alloys. In view
of the limited amount of data, this correspondence
may well be fortuitous.

The first ratio in the table, 1.88%, was substantially
higher than expected. On the basis of the rate of
evolution of HF at the termination of the hydrogena-
tion step, a ratio below 0.7% had been calculated
from equilibrium coefficients for the reduction of
UFs in LiF-BeF: melts with hydrogen. An investiga-
tion to resolve this discrepancy revealed that the
thermocouple used to measure the melt temperature
was malfunctioning. The ratio recalculated at the
corrected temperature was in reasonable agreement
with the analytical results. Had beryllium reduction
been performed on the basis of the initially calculated
ratios, the upper concentration limit of 2% would
most surely have been exceeded and the results of the
experiment would have been compromised. Thus,
this operation serves as an example of the practical
value of the in-line analytical methods that are being
developed for reactor salt streams.

8.3 IN-LINE AND OTHER SUPPORT
ANALYSES FOR CIRCULATING

NaF-NaBF,; COOLANT
J. M. Dale D. L.Manning A.S. Meyer
During the last quarter of 1972, a large fraction of

the analytical chemistry research and development
effort was expended in support of the Coolant-Salt

Technology Facility (CSTF), a pumped loop in
which about 500 kg of the NaBFi-NaF cutectic
mixture 1s circulated at a flow rate of about 785 gpm
through the loop once every 10 sec. The facility
contains a salt monitoring vessel to provide access to
the salt forexperiments carried out by various project
members.

The salt monitoring vessel 1s fed by a side stream
that enters the bottom of the vessel through a
throttling valve that permits control of salt flow from
0.2 to 2 gpm at maximum loop pumping speed. The
salt flows upward through a funnel-shaped vessel and
overflows to produce a relatively stable surface. The
top of the salt monitoring vessel is penetrated by six
risers, each fitted with 1.5-in. ball valves. Two of these
risers are fitted with Pyrex windows and angled so
that their axes intersect the axis of one of the vertical
risers at the surface of the melt and thus provide for
illumination and observation of the melt. For the
initial operations, three of these vertical risers,
including the one equipped for observation, were
assigned to analytical measurements. Thus, an un-
usual opportunity was provided to study more
carefully the proposed in-line electroanalytical
methods. Experience at the MSR E has demonstrated
that the composition of large quantities of salt can be
more readily controlled than that of the much smaller
research melts.

On the basis of these research investigations, we
expected to perform voltammetric measurements of
certain corrosion products and electroactive
protonated species, and to measure the melt potential
vs a lanthanum-fluoride-membrane e¢lectrode that
uses the Fe-FeF, couple as a reference. Before
starting the loop, we made voltammetric
measurements in a quiescent melt on the material
used to charge the CSTF to determine whether the
material presented any unusual features.

At 500°C, the melt exhibited the anticipated
reduction waves for iron and a rather large proton
wave (despite low BF3:OH analysis and careful
thorough evacuation before remelting), but it ex-
hibited no chromium wave. Subsequent analyses of
samples from the CSTF indicated that the total
chromium content of the salt was 30 ppm or less
initially. On heating to 600°C, the cathodic limit of
the melt decreased several hundred millivolts, so that
observation of the reduction of 200 ppm of added
Cr’" at a platinum electrode was not possible.
Previously, at 500°C, a Cr’* — Cr’ reduction wave
had been observed at about -1.0 V, which is near the
cathodic limit of the melt.” An even greater decrease
in the cathodic limit was seen at the palladium

38

electrode. This change in limit is attributed to the
alloying of the boron reduction product with the
electrode material.

We screened a number of alternative electrode
materials and found that iridium, pyrolytic graphite,
and copper yield resolved Cr'* waves at 600°C
comparable to those obtained from platinum at
500° C. However, on a gold electrode the chromium
wave appeared about 100 mV earlier, giving better
resolution from the melt limit, apparently as a result
of alloying of the deposited chromium. Silver offers
no advantages, and rhenium and tantalum are too
active for use in NaBFs melts.

On the basis of these studies, we selected electrode
systems which included the evacuable palladium
electrode for the measurement of the protonated
species and pyrolytic graphite, and copper, iridium,
and gold electrodes for the measurement of corrosion
products. Some of these electrodes were installed ina
shield tube that was inserted in one of the vertical
risers. The shield is a 1-in. tube of thin-walled Inconel
which penetrates the melt to below the level of the
electrodes. It can be purged with He-BF; cover gas
and then connected to the melt blanket gas to provide
fresh melt surfaces and an essentially quiescent melt
around the electrodes. The remaining electrodes were
installed without shields beneath the viewing ports so
that the effects of surface films on the measurements
could be evaluated. From these preliminary
measurements, we expected to observe milliampere
reduction waves in the salt monitoring vessel of the
protonated species at the palladium electrode and of
Fe' — Fe*' atabout 0 V, Fe*" — Fe’atabout 0.4 V,
and Cr’' — Cr’ at -0.7 to 0.8 V at the gold, pyrolytic
graphite, iridium, and copper electrodes.

Actually, our first measurements revealed a large
previously unobserved wave of about 100 mA near 0
V. The wave and its attendant stripping wave were
quite persistent and were obtained on all electrodes.
After several hours of scanning, the wave began to
decrease with each successive scan (Fig. 8.2). This
wave Is attributed to deposition of nickel on the
surface of the electrode, and indeed a plating of
metallic nickel was later observed on electrodes
removed from the salt monitoring vessel. When the
electrode is set at the initialanodic potential, a film of
sparingly soluble NiF: i1s formed, which blocks
further current flow. With cathodic scanning, as the
sharp peak is observed, slightly negative of zero volt,
the film is reduced rapidly. On the return scan, the

2. A.S.Meyeretal., Anal. Chem. Div. Annu. Progr. Rep . Sept.
30, 1972, ORNL-4838, p. 25.
ORNL -DWG. 73-166A

CONTINUQUS SCANS IN

GCSTF FLUOROBORATE SALT: — | ==t f
Ir Electrode ke
0.1 volt/sec e HAL
500°C .
T —
2ma
1
———— I 'é
-n_»\\ Do I i :
74 ;
................ | 7 ! Lo ‘ :
i ! 1
............................ ‘;\ i Nio—’Nin(s) ' :
| ! l | | . 1 | | |
+0.49 +0.2 0 -0.2 -0.4

POTENTIAL vs MELT EQUILIBRIUM POTENTIAL, voits

Fig. 8.2. Decrease in NiF: — Ni° wave during scanning.

NiF: blocking layer is re-formed when the potential
becomes anodic. The nickel film forms even on an
clectrically isolated electrode, apparently by mass
transfer to the electrodes, which are slightly cooled by

conduction of heat up the support rod. At present,:

this nickel wave does not appear to offer any practical
analytical application and is primarily a nuisance. It
can be substantially eliminated by holding the
clectrodes at anodic potentials (Z+0.3 V) when not in
use.

By using this technique and by using faster scan
rates (0.5 V sec’') to minimize the effects of flow
variations, we were able to obtain corrosion product
voltammograms of the quality shown in Fig. 8.3,
which 1s an actual recording of a series of voltage
scans. Accurate interpretation of these scans in terms
of concentrations is not yet possible, because the
baseline at this scan rate isnotaccurately established.
Tentatively, the curves have been estimated to
correspond to values of 100 ppm of iron and 15 to 30
ppm of chromium. These concentrations are
somewhat lower than those measured by wet
chemical analyses of samples withdrawn from the
loop (about 150 ppm of iron and 40 ppm of
chromium). This result is to be expected, because the
voltammetric method is sensitive only to soluble ionic
species while the chemical analyses include insoluble
species such as free metals. The small peak between
the Fe'' and Fe™ reduction waves has not been

observed previously. Mamantov has suggested that
its potential (about 0.2 V) might correspond to the
reduction of Mo™ to the metal.’ About 25 ppm of
molybdenum was reported by chemical analysis.
The precision of the measurement was quite
satisfactory when parameters were held constant.
Although during the early stages of operation many
experimental changes were made, sufficient
measurements were made at the gold electrode at a
0.5-V-sec”' scan rate to permit an estimate of the
precision of the voltammetric measurements. For 25
measurements over a six-weeck operating period, the
wave heights for the Cr’" and Fe’" waves were
reproducible to better than + 15%. These values were
calculated by including statistically rejectable values
and without attempting to correct for possible
systematic variations in the actual concentration of
Cr’ and Fe” in the salt. Reproducibility would be
expected to improve with additional data and more
careful evaluation. The precision of measurement of
the Fe'' wave, which is subject to interference from
the residual nickel wave, and the smaller unidentified
wave, 1s of the order of £ 209%. Therefore, the

3. G. Mamantov, personal communication, 1972.

ORNL-DWG. 73-167A

SCANS IN CSTF FLUOROBORATE MELT /
Au Electrode ;
0.5 volt / sec
500°C

I T R R

+0.8 +0.4 0 -04 -0.8
POTENTIAL vs MELT EQUILIBRIUM POTENTIAL, volts

Fig. 8.3, Voltammograms of corrosion products in CSTF.

voltammetric method, although requiring much ad-
ditional experimental work, should offer a practical
method for the in-line monitoring of corrosion
products in reactor coolant streams.

Electroanalytical’ and  spectrophotometric’
research suggests that the clectroactive protonic
spccies in fluoroborate melts is present in very small
concentration and is highly mobile, in contrast to
infrared-active  BFsOH , which appears to be
relatively stable and perhapscovalent. Moreover, the
electroactive species rapidly equilibrates with a
volatile species that apparently corresponds to the
condensate found in downstream sections of the off-
gas systems of engineering test loops. Based on
evidence from various sources, the phenomena
observed are best explained by postulated equations
of the following types:

slow
2NaBF:OH(d) + NaBFs(d) ~ (NaBF:):0(d)

BF;-H:0(g)
!
+ NaF(d) + [HBF;OH(d)]
1)
H*(d) +BF:OH (d)

The actual composition of the compounds on the
right of the above equations is not established; the
intermediate, HBF3:OH, is expected to be of only
transient stability, and the free proton is surely
solvated, probably as HF: . Attempts to establish
stoichiometry and equilibrium constants by
variational methods in small research melts were
inconclusive because of the ad ventitious introduction
of traces of protons during additions made to modify
gross melt composition. The large salt volumes in the
CSTF offered promise for more precise studies of
these protonated species.

During the initial operations of the CSTF, the
nickel wave obscured any voltammetric
measurements of the electroactive protonic species.
The relative concentration of active protons could be
measured by observing the pressure rise within the
hollow palladium electrode when held at a potential
cathodic to the wave (about 0.5 V). The rate of
pressure rise decreased rapidly to approach the limit
of detection with a thermocouple vacuum gage
during the first week’s operation. After the loop was

4. A. S. Meyer et al., Anal. Chem. Div. Annu. Progr. Rep.
Sept. 30, 1972, ORNL-4838, p. 23.
5. Ref. 4, p. 24.

40

drained for a maintenance shutdown, some uncir-
culated salt was brought up from the drain tank, and
a significant pressure rise was again noted. The rate of
pressure rise over a two-day period is plotted as the
upper curve in Fig. 8.4. The data are in good
agreement with a first-order reaction of 10 hr half-
life. After the protons had been removed substantial-
ly from the melt, 1 cm’ of water was injected via a
stream of helium into the pump bowl. The data
plotted in the lower curve show a similar rate of
decay.

If all the hydrogen in the water had been converted
to active protons, a concentration of 200 ppb would
have been obtained. Since infrared measurements

6. J. P. Young and A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 52.

ORNL-DWG. 70-5985A
T I T

600

500¢

400

300

MOLAR ABSORPTIVITY

200

"BLANK" \_

100

N

400 500
WAVELENGTH, nm

300 600

Fig. 8.4. Decay of active protons in CSTF fluoroborate melt.

have demonstrated that some of the water absorbed
on NaBF; is converted to BF;OH™ on melting, and
since some of the injected water may not have been
absorbed, the initial rate of pressure rise apparently
corresponds to a concentration of no more than 100
ppb of protons. By combining these values with wave
heights and pressure rise rates observed at similar
electrodes in a research melt, we have calculated a
diffusion coefficient of 8 x 107 ¢cm’ sec”' at 500°C.
This calculation confirmed our earlier conclusions
that the active protonated species, probably a loosely
solvated proton, must be more mobile than other ions
in fluoroborate melts, whose diffusion coefficients
are typically 10°° cm? sec™.

The above experiments supported our hypothesis
regarding the nature of proton reactions in molten
fluoroborate. We, therefore, suggested that the active
proton species might be exploited to minimize the
amount of tritium that could be released to the
environment from MSBRs by diffusion through heat
exchangers and steam generators. Earlier suggestions
that the tritium could be held up via exchange with
BF;OH appear to be of dubious value, since no
significant exchange between BF3:OH ™ and diffused
deuterium has been demonstrated.” Exchange with
the active protons would appear much more likely
and would offer an avenue for stripping the tritium
through the off-gas system via either condensation or
collection in a proton-depleted salt. ' :

Dunlap Scott of the Reactor Division suggested
that if this hypothesis were valid, residual tritium
diffusing from the MSRE components from which
the CSTF was fabricated would have collected in the
condensate. The values reported in Table 8.2 clearly
demonstrate a high concentration factor of about 10’
in the collected condensate.

7. J. P. Young, J. B. Bates, and G. E. Boyd, MSR Program
Seniannu. Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 52.

41

These data are indeed promising but are not
sufficient to establish the mechanism of the tritium
transport or its overall effectiveness. We have no
reliable information as to the quantities of tritium
that could have diffused into the melt, the efficiency
of trapping of the condensate and its chemical nature,
or the nature of tritium bonding to the black film on
the surface of the fluoroborate samples. We have
performed analyses on the limited quantities of
condensate available and on synthetic condensate
generated by the reaction of BF; with water. Typical
results are shown in Table 8.3.

Attempts to perform material and ionic balances
on these results and to use nuclear magnetic
resonance and infrared spectra of such material
indicate that it is a mixture rather than a simple
compound. The synthetic mixture corresponds
reasonably well to a mixture of BF;-H:0 with HF
and H3;BOs. The condensate from the CSTF contains
much less free fluoride and has a lower density than
the synthetic products. Possibly, its composition had
been modified during the extended sweeping periods
and by reaction with the glass traps used for
collection. The density of pure BF3-H:O is reported
to be 1.82.

Table 8.2. Tritium content of NaF-NaBF,
coolant loop exposed to tritium

Sample Range of tritium

values (nCi g'')

0.8X10°t0 3.0 X 10°
4 to 10
>200°

Off-gas condcnsatc
Salt
Black film concentrate

“A sample of black film was impossible to
obtain without including some salt substrate. This
valuc is certainly a minimum value; the tritium
content of the black [ilm only may be an order of
magnitude higher. ‘

Table 8.3. Analyses of BFs condensates

Component

(millimoles/ g) Melt condensate®

Product from
8F;-H,O vapor”

Product from
BF;-liquid H,0"

H* 12.0
Free F~ 0.22
Bound F~ 226
Tota! B 9.3
B as H;BO; ¢

12.1 12.9
7.7 7.1
19.5 20.1
11.9 12.0
c 4.6

“Density, 1.52 g/cm',
"Density. 1.72 gfcm".
‘Not determined.

We have assisted Reactor Division personnel in
planning additional experiments that involve the
controlled diffusion of deuterium into the CSTF.?
Such experiments should answer fundamental and
practical questions about the effectiveness of NaBF.
as a tritium containment medium, and establish
whether the interdiction of tritium can be achieved
through the exchange with active protons or through
a proposed redox mechanism such as

1/2T: + O + 1/3Cr"— OT + 1/3Cr° °
Calculations® based on various proposed
mechanisms have indicated that, in some cases,
analytical methods of improved sensitivity will be
required to provide definitive data for the interpreta-
tion of deuterium injection experiments. The needed
improvements are being investigated under the areas
of electroanalytical and spectrophotometric research
and development as discussed in the following
sections.

84 STUDIESON THEELECTROREDUCTION
OF URANIUM(IV) IN MOLTEN LiF-BeF;-ZrFs

D. L. Manning Gleb Mamantov

We applied square-wave voltammetry to a study of
the reduction of U*" in molten LiF-BeF:-ZrFi. The
low resistance of molten-salt solutions, which results
in a low resistive capacitance time constant for small-
area electrodes, should render them attractive for
square-wave voltammetric techniques. Other attrac-
tive features of square-wave voltammetry are the
ability to measure Faradaic current in the virtual
absence of charging current and the increased sen-
sitivity afforded over conventional voltammetry.
However, until now there has been only one applica-
tion of square-wave voltammetry to the study of
solute species in molten salts."

At 500°C under our conditions, the equation for
the peak current of a square-wave voltammogram is
given as

imax = 3.496 X 10° n” AD'? CF'* AE, (1)
where inax = peak current (amps), n = number of

electrons, A = electrode area (cmz), D = diffusion
coefficient (cm® sec”'), C = concentration of diffusing

8. H. A. McLain, “The CSTF Deuterium Experiment for the
MSR Program,” MSR-73-16 (Jan. 26, 1973).

9. S. Cantor and R. M. Waller, MSR Program Semiannu.
Progr. Rep. Feb. 29, 1972, ORNL-4782, pp. 57-59.

10. A. D. Graves, G. J. Mills, and D. Inman, Advan.
Electrochem. Electrochem. Eng. 4, 117-83 (1966).

42

species (moles cm’’), F = square-wave frequency
(Hz), and AE = magnitude of the square-wave step
(V).

A melt was prepared which contained 0.09 M U™,
the concentration of interest in the MSRE fuel salt.
We recorded current-voltage curves at small
platinum and iridium electrodes by superimposing a
square-wave voltage on the controlled potential ramp
of a controlled-potential, controlled-current cyclic
voltammeter. Oscillographic recordings were made .
of both cathodic and anodic branches of the cell
current simultaneously. We observed that i maxvaried
with frequency and amplitude of the square wave, as
predicted from Eq. (1). The frequency and amplitude
were varied from 100 to 1600 Hz and 0.01 to 0.05 V,
respectively. Linear voltage scan rates that were used
varied from about 0.02 to 0.1 V sec’’. From these
observations, square-wave voltammetry appears well
suited for the study of reversible electrode processes
in molten salts. The apparent rapid decay of double-
layer charging renders feasible the use of higher
square-wave frequencies than normally used for
aqueous systems. This work is reported in more detail
in Electrochimica Acta."

8.5 ELECTROANALYTICAL STUDIES
OF TELLURIUM
IN MOLTEN LiF-BeF:-ZrF4 (65.6-29.4-5.0 mole %)

D. L. Manning Gleb Mamantov

While awaiting delivery of the thorium-containing
MSR fuel salt, we set up two cells for electrochemical
studies on tellurium and bismuth in molten LiF-
BeF:-ZrF: remaining from the earlier program.
Electroanalytical studies of tellurium are of interest,
since tellurium occurs in nuclear reactors as a fission
product and recently has been considered a prime
cause of shallow intergranular cracking in structural
metals and alloys.'” For the tellurium studies, the
melt was contained in a pyrolytic boron nitride cup,
and the holder for a small tellurium pool electrode
(1/8-in.-diam) was fabricated from hot pressed boron
nitride. Contact to the molten pool was made with
iridium wire.

Immediately after dipping the electrodes into the
melt at 500°C, cathodic pglarization curves on the

1. D. L. Manning and G. Mamantov, Electrochimica Acta 19,
177 (1974).

12. H. E. McCoy, “Matenals for Salt Containing Vessels and
Piping,” The Development and Status of Molten Salt Reactors,
ORNL-4812, p. 207.

tellurium pool electrode revealed a prewave at about
-0.6 V vs an indium quasi-reference electrode, fol-
lowed by the limit at about 0.75 V. The anodic limit
occurs at about 0.9 V. After about 24 hr, the anodic
limit remained the same; however, the prewave
shifted to about 0.8 V, and the cathodic limit was
seen at about -0.9 V. After this initial shift, the
potentials of the prewave and cathodic limit re-
mained stable.

The anodic limit is believed to be due to the
oxidation of tellurium, and the cathodic limit cor-
responds to the 2Li" 4+ Te + 2e — Li:Te reaction. The
voltage span from anodic to cathodic limit of about
1.6 V (fresh tellurium surface) and 1.8 V (aged
surface) can be compared to 1.2 V in LiCI-KCl at
450°C reported by Bodewig and Plambeck."”
However, these workers did not report a prewave
prior to the cathodic limit as observed in the molten
LiF-BeF:-ZrFa.

The reaction responsible for this prewave has not
been established. It was not seen on background
scans, and the wave height has not changed
significantly with time after the tellurium electrode
was dipped into the melt. One possibility is the
reduction to a lower valent tellurium species, such as
LiTes, by some impurity route. Bamberger, Young,
and Ross'* have reported the existence of such Li.Te,
species in molten fluorides based on spectral studies.

Chronopotentiograms recorded for this wave at
the tellurium pool electrode, although not well
defined in all cases, did reveal a ratio of forward-to-
reverse transition time (7) ofabout 3:1. Adsorption of
the tellurium species at the electrode was also
indicated from the chronopotentiometric data since
the i'”* product consistently increased with increased
current density.

Adsorption of the supposed tellurium species was
also noted at gold and iridiumelectrodes as indicated
from voltammetric and chronopotentiometric data.
The chronopotentiometric results best fit the equa-
tion that pertains to the simultaneous reduction of
diffusing and adsorbed species'’ which is given as

ir=1/2nFa'"" AD"* Cr'"? + nAFT . (2)
The symbol I' represents surface excess (moles cm ’);
the other terms have their usual significance. By
assuming an » value of 2 (probably the hmiting case),

3. F. G. Bodewig and J. A. Plambeck, J. Electrochem. Soc.
117, 618 (1970).

4. C. E. Bamberger. J. P. Young, and R. G. Ross, J. fnorg.
Nuct. Chenr. 36, 1158 (1974).

15. 8. V. Tatwawadi and A. ). Bard, Anal. Chem . 36, 2 (1964).

43

calculation of the I' term gives a value of about 4 X
10 moles cm *, which can be compared to about
107" moles cm * typical for aqueous systems."”

These observations support other work'’ that
points to intergranular cracking by tellurium. Ad-
sorption of tellurium compounds would provide a
reservoir of tellurium at the surface. The adsorbed
LinTes species may disproportionate to more stable
tellurides and elemental tellurium. Elemental tel-
lurium in molten LiF-BeF:-ZrF, is more difficult to
reduce than Ni** or Fe2+, because it is comparable to
Cr*" and more easily reduced than U*. As known
LinTe, species become available, standard additions
to the fuel salt are planned for electroanalytical
studies. If the electrode reactions are reversible, the
limits of detection by pulsed voltammetric techniques
may be extended.

8.6 ELECTROANALYTICAL STUDIES
OF BISMUTH IN MOLTEN
LiF-BeF,-ZrF, (65.6-29.4-5.0 mole %)

Gleb Mamantov D. L.. Manning

Bismuth, because of its use in fuel processing
systems,'® is a possible impurity in MSBR fuel salt.
This study is directed toward the characterization of
the electroanalytical chemistry of BiF; in molten
fluoride salts to assess the potential of the in-line
measurement of traces of bismuth in the return
stream from processing systems. For our previous
work on bismuth,'” the melt was contained in
graphite or copper cells. Bismuth (added as BiF;) was
slowly lost from the molten LiF-BeF2-ZrFs at 500° C.
The two main routes for the loss would appear to be
the reduction of the Bi’" by the container material or
volatilization of the BiFs. To further investigate this
problem, a cell containing Bi’* was set up; the melt
was contained in a glassy carbon cell.

Well defined voltammograms and chronopoten-
tiograms were obtained in a melt ofapproximately 10
millimoles in BiF; at gold, iridium, and glassy carbon
electrodes. The peak potential for the Bi’* — Bi’
reduction occurs at about 0.1V vsan iridium quasi-
reference electrode (Ir QRE). (Prior results'” showed
that this reduction occurs at about + 0.1 V vs Ni**/ Ni
reference electrode.) For the chronopotentiograms,

16. L. E. McNecse, “Fuel Reprocessing,” Development and
Status of Molien-Salt Breeder Reacrors, ORNL-4812, p. 331.

17. J.S. Hammond and D. L. Manning, High Temnp. Sci. §, 50
(1973).
the ratio of the forward-to-reverse transition time
was essentially unity at the three indicator electrodes.
Again, however, the bismuth wasslowly lost from the
melt as revealed by a gradual decrease in the peak
height of the voltammograms. To check for
volatilization, a cold finger was placed in the cell fora
few days and an x-ray fluorescence analysis of the
deposited film indicated that the major constituent
was indeed bismuth. Thus, bismuth appears to be
slowly lost from molten LiF-BeF:-ZrFi at least in
part by volatilization; this conclusion is in agreement
with the work of Cubicciotti."

These studies to determine the lower limits of
detection by voltammetry and stripping techniques
will continue. We believe meaningful analytical
results from on-line measurements can still be ob-
tained, because the rate of bismuth loss is slow.
Nickel constitutes a primary interference. Therefore,
studies will be made on melts that contain both
bismuth and nickel. Transpiration techniques will be
investigated as methods to separate the volatile BiF3
for determination by high-sensitivity gas analytical
methods, for example, flameless atomic absorption.

8.7 ELECTROANALYTICAL STUDIES
IN MOLTEN NaBF,

D. L. Manning B. R. Clark’
E. R. Malone'” R. V. Buh!®

We have resumed electroanalytical studies in a
fluoroborate coolant melt prepared from materials
purified during the earlier program. Planned studies
of the free proton wave have been delayed by the
failure of our only remaining hollow palladium
electrode. While awaiting materials for the fabrica-
tion of new electrodes, we have started studies of
corrosion products. Observations reveal that the Fe®’
— Fe’ reduction conforms to a soluble species
reduction at a gold electrode and to a partially soluble
species reduction at an iridium electrode. This
solubility was demonstrated by chronopoten-
tiometric measurements that serve as useful in-
dicators of the nature of the electroreductions. For
the soluble species case, the ratio of the forward-to-
reverse transition time is 3:1, while for the insoluble
case it 1s 1:1. For both electrode materials, the

18. D. Cubicciotti, J. Electrochem, Soc. 15, 1138 (1968).

19, Student guest.

20. A, S. Meyer et al., Anal. Chem. Div. Annu. Progr. Rep.
Sept. 30, 1973, ORNL-4930, p. 26.

44

observed ratio is approximately 3:1. It was noted
earlier,” on the basis of wave positions and slopes,
that chromium alloys with gold but not with indium
or pyrolytic graphite electrodes.

Establishment of the nature of the electrodeposi-
tion reactions i1s of practical analytical interest.
Generally, the alloying is favorable for direct
voltammetric measurements, because the position of
the waves does not change with concentration and
such soluble systems are amenable to square-wave
voltammetry. Square-wave voltammetry offers in-
creased sensitivity and stability.'' This technique
(typically a 400-Hz, 50-mV square wave superim-
posed ona0.1-V-sec ' linear ramp) has produced well
defined waves with current enhancement in close
agreement with theoretical predictions.

After adding chromium as Na3CrFs to the melt, the
iron and chromium waves were resolved without
difficulty. The peak potentials are separated by about
350 mV, which is adequate for analytical purposes.
However, at slow scan rates (<0.5 V sec ') the
chromium wave in the presence of iron is not as well
defined as in the absence of iron. This effect is
probably caused in part by the predeposition of iron
on the gold electrode which alters the surface
characteristics of the electrode by the time chromium
is deposited. At faster scan rates, this distortion for
the most part disappears, and the waves are well
defined. Linear i, vs v'/? plots were obtained for both
species. A calculation of the iron concentration from
the slope of the line agreed well with the value from
chemical analysis (about 300 ppm). A similar calcula-
tion for the concentration of chromium revealed a
value of 130 ppm. This value is low when compared to
the amount of chromium (about 260 ppm) added to
the melt, but it is 1n agreement with the value of
chromium (about 125 ppm) obtained from chemical
analysis. The melt, therefore, is probably saturated in
chromium atabout 125 ppm at 500° C. We will make
measurements and take a sample at 600° C to observe
the effect of temperature on the chromium solubility.
Concentration values from square-wave voltam-
mograms recorded at 400 and 900 Hz were in
agreement with the linear scan results. As expected,
the addition of chromium exhibited no effect on the
chronopotentiograms for the iron wave, which con-
forms to a soluble species reduction at a gold
electrode.

Voltammetric studies were also carried out for an
anodic wave seen at a yold electrode atabout+ 1.5V
vs an iridium quasi-reference electrode. This wave
seems to be related to the NaBF:OH content of the
melt, the electrode reaction probably being -OH™ —
H"+ 1/20:+ 2e. The wave is reasonably well defined;
noise is encountered on the diffusion current plateau
which is indicative of gas formation at the electrode
surface. However, the wave is irreversible; no reverse
current is seen even at relatively fast (20 V sec”') scan
rates. In nonrecrystallized “as received” NaBF4 which
was found to be 16 millimoles NaBF3;OH by infrared,
the wave height was about 1.5 mA. Ina recrystallized
batch of NaBF4 now analyzing about 120 millimoles
NaBF;OH, the wave height increased to about 30
mA. The wave height varied in proportion to the ratio
of known mixtures of the nonrecrystallized and
recrystallized salt used for melt preparation, and it
appears to be stable with time. Additional calibration
will be required, but this anodic wave will probably
prove valuable for the investigation of the effect of
BF;OH on corrosion and on tritium containment
by coolant salt.

A review of the analysis of the deuterium injection
experiment at the CSTF® has indicated that improve-
ment will be needed in the sensitivity of the palladium
electrode, particularly for the measurement of dis-
solved elemental deuterium that may be present at
about 10 torr. Calculations are being made to
optimize the design of the electrode-vacuum
assembly (palladium probe) with respect to
parameters such as active electrode area, maximized
conductance of connecting tube consistent with small
volume requirements, and the combination of
vacuum gages to accommodate the range of elemen-
tal deuterium and active proton concentrations that
are predicted under various assumptions. Highly
sensitive thermocouple gages (range 10~ to 107 torr)
are being obtained for improved measurement of the
collected hydrogen isotopes. Use of the more sen-
sitive ionization gages is precluded by their high
pumping rates for hydrogen.

8.8 SPECTROPHOTOMETRIC RESEARCH
J. P. Young

No spectral measurements are currently being
made in MSBR salts, because we have assigned our
resources to more urgentneeds of the MSR Program.
Since the interruption of the program, we have been
conducting basic spectrophotometric research in the
analytical chemistry of transuranic elements which,
while not of direct and immediate application to
molten-salt reactors, 1sexpected to provide basicdata
for later development of in-line methods. Such
investigations include a study to relate absorption

45

spectra and coordination of rare earth halides,”' the
development of a new class of isotopic light sources,**
and preparations for the characterization of unusual
oxidation states of transuranic elements in molten
salt solvents.”” Isotope pairs of widely divergent
activity have been requested for two transuranic
elements, curium and plutonium, to measure their
relative solubilities in fuel solvents (by
spectrophotometry) and thus to test an hypothesis
that oxide is removed from fuel by radiolytic
processes. Experimental work will begin when
1sotopic pairs of either element are made available.

Results of work completed since the interruption
of the MSR Program are summarized below,
together with a program of spectrophotometric
studies for the improved analysis of fluoroborate
coolant salt.

8.9 SPECTRAL STUDIES
OF f-d AND f-f TRANSITIONS
OF Pa(IV) IN MOLTEN LiF-BeF;-ThF.
J. P. Young C. E. Bamberger R. G. Ross

The following paragraph is a condensed version of
a note published in J. Inorg. Nucl. Chem.”

The transparency of the solvent for MSBR fuel,
LiF-BeF2-ThFs (72-16-12 mole %), in the range 200
nm to 2.5 u permits detailed examination of the
absorption spectra of dissolved Pa(1V), a species of
particular interest to the MSR Program. Solutions of
600 to 2000 ppm Pa(lV) were prepared by hydrogen
reduction of Pa(V) solutions in oxide-free fuel
solvent. The dissolved Pa(1V) exhibited two areas of
light-absorption, one in the ultraviolet region at-
tributable to f-d transitions and one in the near-
infrared attributable to f~f transitions (Figs. 8.5 and
8.6).

The ultraviolet spectrum of Pa(lV) observed in
molten LiF-BeF:-ThF: is similar to that seen for
Pa(1V) in aqueous 15 M NH4F.* The spectrum we
observe in the melt is likewise similar to the spectrum
reported for solid PaFs asa Nujolmull but shows less
detail than that seen for solid 7RbF-6PaFs.”” Asis the

21 A S. Meyer ctal, Anal. Chem. Div. Annu. Progr. Rep.
Sept. 30, 1974, ORNL-5006, p. 28.

22, Ref. 21, p. 29.

23. J. P. Young, C. E. Bamberger, and R. G. Ross. J. fnore.
Nucl. Chem ., in press,

24. M. Humssinsky, R. Muxart, and H. Arapaki, Bull. Soc.
Chem., France, 2248 (1961).

25. L. B. Asprey, F. H. Kruse, and R. A. Penneman, fnorg.,
Chem. 6, 544 (1967).

ORNL-DWG, 71- 7614

06 l T l T I T I T

o
»
I
|

0.2+

ABSORBANCE

]
1400

1600 1800

WAVELENGTH, nm

Fig. 8.5. f-d spectrum of Pa(IV) in molten LiF-BeF.-ThF, at
570°C.

2000

ORNL- DWG. 73-163
100 T T T T T m
< 50~_ DECAY OF ACTIVE PROTONS IN CSTF ]
E FLUCROBORATE MELT 4
N 4
W
ud i
<
w
o
o
z —
= ]
tw ]
x -
)]
@ J
[
x o After Pump Start-up |
B s After Water Injection -
: | l | 1

20 30
ELAPSED TIME,

40
haurs

Fig. 8.6. f+f spectrum of Pa(IV) in molten LiF-BeF:-ThF.: at
570°C.

60

case with Pa(IV) in aqueous NH4F, no absorption
peaks are observed in the visible region of the
spectrum. There have been no previous solution
spectra of the complete f-f transition range for
Pa(1V). The molten fluoride solvent system used
permits unambiguous measurements in this region.

Since Pa(V) dissolved in the same solvent exhibits
no interfering absorptions, these spectra may prove

46

useful in developing an analytical technique for
determining the redox status of protactinium in
fluoride melts.

8.10 SPECTROPHOTOMETRIC
DEVELOPMENTS FOR IMPROVED
ANALYTICAL METHODS FOR
FLUOROBORATE COOLANT SYSTEMS

B. R. Clark R. F. Apple J. P. Young

At least one order of magnitude improvement in
the sensitivity of the determination of BF3OD by the
measurement of its infrared absorption in pressed-
pellet samples of NaBFai-NaF eutectic will be needed
to assure that the disposition of deuterium can be
followed during its injection into the CSTF. No
calibrations for BF;OD have been made; but we are
assured”® that its molar absorbancy will be indis-
tinguishable from that of BF3;OH . Therefore,
calibrations (by solid state standard additions of the
sodium salt) with either isotope should suffice.
Extension of sensitivity to needed levels (typically 50
ppb as hydrogen) should be possible by performing
absorption measurements on thicker pellets. The
technology of pressing thick pellets of NaBF: has not
vet been developed; in fact, it cannot currently be
experimentally investigated for BFiOH' because its
residual concentration in any existing NaBF4 1s too
high for accurate measurement in thick pellets. The
inorganic preparations group of the Analytical
Chemistry Division is examining the possibility of
preparing isotopically pure NaBF;0D, and we are
considering routes for producing highly purified
NaBFi. This material would also be useful for
electroanalytical studies. We are also using spec-
trophotometry to support our chemical studies on the
nature of the condensate from cover gas and on
practical techniques for collecting it.

26. J. B. Bates, oral communication, September 1974.
47

9. Inorganic and Physical Chemistry

9.1 THE EQUILIBRIUM
OF DILUTE UF; SOLUTIONS
CONTAINED IN GRAPHITE
L. M. Toth L. O. Gilpatrick

The equilibrium of dilute UF;-UF: molten-
fluoride solutions in contact with graphite and UG,

3UFa(d) + UCi(c) = 4UF3(d) + 2C(c), (1)

has been studied as a function of temperature (370 to
700°C), melt composition, and atmospheric con-
tamination. Equilibrium quotients Q
(UF3)’/(UF4)® for reaction (1) were determined by
measuring the UF; and UF: concentrations
spectrophotometrically. The solvents used were
primarily LiF-BeF: mixtures. Results from this
solvent system were related to the reactor solvents
LiF-BeF>-ZrFa (65.6-29.4-5 mole %) and LiF-BeF;-
ThFs (72-16-12 mole %). The equilibrium quotient
has been found to be very sensitive to both
temperature and solvent changes, increasingas either
the temperature increases or as the alkali-metal
fluoride content of the solvent decreases.’

Figure 9.1 shows equilibrium quotients in LiF-
BeF:-ThF. (72-16-12 mole %). These data, which are
similar to those in LiF-BeF: (66-34 mole %), show
that more than 10% of the total uranium in solution
can existas UFs at temperatures greater than 550° C.

Experiments also showed that the large
temperature and solvent effects measured for reac-
tion (1) could be explained on the basis of a difference
in the heats of solution between UFs and UF.:.” By
assuming AHson for UFs is equivalent to the AHson of
CeFs obtained from solubility studies,3 and by using
the difference in the enthalpies of reaction deter-
mined from our data and from the respective heats of
formation, AH . for UFs was found to be -16
kcal/mole. No previous determinations of solution
enthalpies for M jons in fluoride melts are known,
but this value is in agreement with previous
preductions” based on the greater ionic potential of
the U jon.

Equilibrium quotient data' for all the solvent
systems studied are summarized in Fig. 9.2. In
addition to the four solvent mixtures (plots Ato D),
plots E and F identified as “oxide contaminated” are
shown. These data were obtained in the early stages
of the UF:;-UFs equlibrium study when the
cquilibrium of reaction (1) was approached from the

right-hand side. Because these values were much
lower than the ones measured for plots A and D and
because they were accompanied by the formation of
UQgz, trace contamination by some oxygen-bearing
impurity had presumably lowered the equilibrium
quotient and a more stable oxycarbide phase was
formed in place of the pure carbide. Studies ultimate-
ly showed that more stringent control of the ex-
perimental conditions could hold the UF;
disproportionation equilibrium at the values in-
dicated in plots A to D.'

I. L. M. Toth and L. O. Gilpatrick, The Equilibrium of Dilute
UFs  Solutions Comtained in Graphite, ORNL-TM=056
(December 1972)

2. L. M. Toth and L. O. Gilpatrick, J. Phvs. Chem. 717, 2799
(1973).

3. C. J. Barton, M. A. Bredig, L. O. Gilpatrick, and J. A,
Fredricksen, fnorg. Chem. 9, 307 (1970).

4. G. Long and F. F. Blankenship, The Stability of UF;, Part |
and I, ORNL-TM-2065 (1969).

ORNL-DWG 72-42321
TEMPERATURE (°C)

500 550 600 650 700
-1 T I T /
_2 /
- 086
.,
- os
& /l 4
/e
L}
// -+ 04
<
_4 / 5
+
Q '/ —403
o $ 3
g / <
o Thid
-5
| / Jdo2 7
T
LS
1) /
-8
130 125 120 145 110 105 100
1000/T (°K)

Fig. 9.1. Equilibrium quotients @ (UF3)*/(UF)’ vs
temperature for UF;/UF, in LiF-BeF:-ThF. (72-16-12 mole %).

48

ORNL-DWG 72-12322

TEMPERATURE (°C)

370 400 450 500 550 600 650 700
| i [ | | [ I [
B C
7 D
/ / — 0.6
) /
- — 0.5 —~
/ 4 :
/ / — 0.4 +I")
-4
/ 7 5
/ ] <
/ y, 0.3 LDLn
]
o

log o @

/, / — 0.2
A /// —

/|

- / /
8 v / /
-9 / /F
-10 —— (A) LiF-BeF, (48-52 mole %)
(B} LiF-BeFp-Zr F4 (65.6 -29.4~5.0 mole %) /
1 (C) LiF-BeFp=ThF (72-16-12 mole %) /
(D) LiF-BeF, (66- 34 mole %)
(E) LiF-BeF, (48-52 mole %)
2 OXIDE CONTAMINATED
(F) LiF-8eF; (66 - 34 mole %)
OXIDE CONTAMINATED
1.6 1.5 1.4 1.3 1.2 1.1 1.0
10097 (oK)

Fig. 9.2. A comparison of equilibrium quotients vs temperature for UC: + 3UF«(d) = 4UFis(d) + 2C in various solvent systems.

9.2 EFFECTS OF OXYGEN

ON THE UF;/ UFs EQUILIBRIUM

L. O. Gilpatrick R. M. Waller L. M. Toth
In view of the previous results suggesting that the
UF;-UF4 equilibria in molten fluorides contained in
graphite were different from those measured with a
pure UC: phase, a study was initiated during this
period to determine (1) if the “oxide contaminated”
equilibria were real and reproducible, (2) if any U-C-
O phases were formed at temperatures below 1000° C
through the action of the UF3;-UFs reaction with
graphite and oxygen-containing impurities, and (3) if
the impurity and its partial pressure at equilibrium

could be identified. The LiF-BeF: (48-52 mole %)
solvent system, (LB), was selected for this study,
because the equilibrium quotient is higher (Fig. 9.2,
plots E and F) than in the 66-34 mole % (L:B)
solution and, consequently, the UF; concentrations
are easier to measure spectrophotometrically. Ul-
timately, these results will be compared with the LB
solvent, The experimental procedure involves both
equilibrium studies in sealed graphite capsules and
spectrophotometric experiments with controlled at-
mospheres over melt mixtures.

The capsule experiments were intended to
equilibrate UO: and U(C: in a graphite container by
means of the UF3;-UFs reaction in a LiF-Bek:
solution. The reaction of UF: with graphite to form
uranium carbides would then occurin the presence of
an oxide-saturated melt. If any U-C-O phase existed
which was more stable than UCa, it should form by
the overall reaction of oxide ion with UG,. Capsules
were equilibrated for periods of 30 and 90 days with
solvents of LB and LB composition. The formation
of an oxycarbide phase was sought by x-ray analysis
and was expected to be identified as either a new
phase ora noticeable decrease in the lattice parameter
of the existing uranium carbide phases.” Neither
expectation has been verified at this time. Instead, an
increase in the UC: tetragonal structure lattice
parameter was shown by the following results:

Before:

a= 35225+ 0.0002 A, ¢ =5.9946 + 0.0007 A

After:
a = 35352+ 0.0009 &, ¢ =6.0240 + 0.0047 A

At present the increase cannot be explained.
However, these results do suggest that no U-C-O
phase forms when UO: and UC: are equilibrated
during the above time intervals.

The spectrophotometric study was intended to
verify the previous “oxide contaminated” observa-
tion and to shift the equilibrium accordingly by
control of the atmosphere over the melt. The previous
observations have been repeatedly reproduced by
allowing UFs to react with the graphite
spectrophotometric cell. Furthermore, while at
equilibrium, the furnace atmosphere was sampled
directly through a variable leak valve into a time-of-
flight mass spectrometer to identify any impurity
gases over the melt. Hydrogen fluoride at an es-
timated pressure of 107° to 10™” torr was the only gas
seen at 600 and 700° C while the furnace waseitherat
I atm argon or under vacuum. It was possible that
CO was also present, but it could not be distinguished
from the background mass 28 of the spectrometer..

Experiments currently in progress and continuing
into the next reporting period involve contact of the
melt solutions with dilute gas mixtures of CO in
argon. These experiments will test the effect of mildly
oxidizing mixtures on the UF3-UFs equilibria in
graphite. Quantitative results will be obtained as a
function of gas composition in the ppm range of CO.

5. The inclusion of oxygen in the carbide lattices is reported to
decrease the lattice parameters by measurable amounts. For
example sce (a) H. Tagawa and K. Fujii, J. Nuel. Mater. 39, 109
(1971), (b) J. Henny, United Kingdom Atomic Energy Authority
(Harwell) Report AERE-R-4661 (1966).

49

9.3 POROUS ELECTRODE STUDIES

H. R. Bronstein F. A, Posey

Studies were initiated on the feasibility of using a
porous or packed-bed electrode as a continuous,
on-line monitor of the concentration of dissolved
bismuth in MSBR fuelsalt. Part-per-million concen-
trations of various chemical species (e.g., oxygen,
tellurium, bismuth) can profoundly affect molten-
salt reactor systems, and thus a device is needed to
monitor, and perhaps remove, dissolved bismuth
and/or other deleterious species from fuel salt.
Porous electrodes have provided excellent means of
monitoring and eliminating similarly low-level im-
purities in aqueous systems™ and should have
applicability in molten salts.

Design and fabrication of a prototype quartz cell
for study of electrolytic monitoring and removing
dissolved bismuth was completed during the report
period. The cell is a part of a bench-scale test
assembly that was used in a preliminary study of the .
LiCI-KCl eutectic system; this study was carried out
in a furnace that permitted visual observation of the
quartz cell assembly during operation. In this way,
proper operating procedures can be established for
later studies on MSBR fuel salt, for which a metal cell
assembly will be required. Figure 9.3 is a photograph
of the quartz cell assembly contained in a tantalum
holder that could be lowered into the LiCl-KCl
eutectic melt. The electrode itself wasa packed bed of
glassycarbon spheres (about 100 microns in diameter)
supported on a fritted quartz disk. Another fritted
quartz disk was pressed onto the bed from above,

providing compaction and allowing use of a glassy

carbon rod for electrical contact with the bed. A long
stainless-steel rod (not shown in Fig. 9.3) contacted
the top of the glassy carbon rod and led through the
top of the test vessel, which could be evacuated or
pressurized with inert gas, as required. The melt
flowed up through the central tube (Fig. 9.3), through
the packed bed, and out a number of vertical slots
into an overflow chamber. In this way, a reproducible
volume of melt could be obtained inside the packed-
bed electrode. The area-to-volume ratio with this
type of electrode system is quite high and permits
rapid exhaustion of electroactive species from the

6. F. A. Poscy and A. A. Palko, Ecologr and Analvsis of Trace
Contantinants Progress Report January 1973 —September 1973 .
ORNL-NSF-EATC-6 (1974), pp. 360-89,

7. H. R. Bronstein and F. A. Poscy, Cheni. Div. Annu. Progr.
Rep. May 20. 1974, ORNL-4976, pp. 109-11.

melt. We estimate that the area-to- volume ratio is
approximately 1.2X 10" cm®/ em’ of solution mlumc.
the total bed volume is approximately 2.5 cm’. Other
components used in the bench-scale test assembly
were conventional items for temperature control,
instrumentation for accurate measurement of the
temperature of the melt, and a manifold for vacuum
and inert pressure control over the melt.

Some results of the first experiment carried out in
the LiCI-KCl (58.841.2 mole %) eutectic melt at
414° C are shown in Fig. 9.4. Small quantities of iron

PHOTO 2259-74

Fig. 9.3. Packed-bed clectrode assembly with glassy carbon
spheres for use in molten salts.

50

ORNL-DWG. 74-10495

—15—
10
Q
-~ 5l B
I
i
=3
-
z 0 By
o o
= o
5]
3 5)-2 e
= F&'Fe
10—
15— {c1—=ci® Al
| |- | | |
1.5 1.0 05 E0,0- =05 - L0~ =LE

ELECTRODE POTENTIAL vs Ag/AgCl (V)

Fig. 9.4. Linear-sweep voltammetry with packed bed of glassy
carbon spheres. Melt. LiCI-KCl (58.841.2 mole ) + about 1 X
107 M Fe" +about 4 X 10 * M Cd™': temperature, 414° C: sweep
rate, [0 mV/ s eference electrode, Ag/ AgCl(0.1 m)in LICI-KCl
eutectic/Pyrex gl

and cadmium salts were added to the melt to provide
species with known electroactive properties. The
reference electrode was a Pyrex glass tube containing
0.1 m AgCl dissolved in LiCI-KCl eutectic melt into
which a silver wire was inserted; the conductivity of
the Pyrex glass was sufficient to allow sa factorily
low impedance between reference and test solutions
at the temperature of the melt. The counter electrode
a cylindrical graphite sheath surrounding a
thermocouple well. Figure 9.4 showsa voltammetric
current-voltage curve obtained at linearly varying
potential (scan rate = 10 mV/sec). The peaks
observed during the cathodic sweep correspond to
reduction of ferrous ions to metallic iron (about -0.37
V) and of cadmium ions to metallic cadmium (about
20.75 V) on the electrode surfaces; the limit for
cathodic decomposition of the melt (deposition of
ssium) is seen at still lower potentials. The return
anodic sweep also shows the expected behavior and
peak sequence, that is, redissolution of potassium
(about -1.1 V), cadmium (about -0.45 V), and then
iron (about -0.20 V), followed by further oxidation of
Fe(11) to Fe(111) (about +0.85 V) and finally the limit
for anodic decomposition of the melt (chlorine

evolution). Estimation of the number of coulombs
under the iron and cadmium peaks (with a
planimeter) shows that approximately 5 ug of iron
and 38 ug of cadmium were contained within the bed.
These results show the inherent sensitivity of this
method of analysis and suggest that use of a
coulometric technique with electronic integration’
could enhance sensitivity still further. We have
recently acquired instrumentation that will allow us
to record simultaneously current-voltage and charge-
voltage curves for these electrode systems. The results
suggest also that the background current due to non-
Faradaic processes (such as charging of the double-
layer capacity) is small with little variation over a
wide potential range (approximately +1.0 Vto-1.0V
vs Ag/AgCl). Successful operation of the cell,
instrumentation, and auxiliary systems will allow us
to proceed with studies of bismuth species in the LiCl-
KCl system as soon as fabrication is completed on a
modified cell container assembly that will allow easy
addition of various substances to the melt. Subse-
quently, we expect to design and construct a similar
apparatus foruse in molten fluoride melts, which will
facilitate studies of the applicability of the packed-
bed electrode to monitoring and/or removal of
bismuth from MSBR fuel salt. '

9.4 FORMATION FREE ENERGIES
AND ACTIVITY COEFFICIENTS
IN FUEL SALT MIXTURES

C. F. Baes, Jr. W. C. Waggener

We arec resuming a program' in which
measurements of heterogeneous equilibria betwcen
solid oxides and molten fluoride mixtures will be
used to derive formation free energies and activity
coefficients of the components in the MSBR fuel.
Staffing was completed, and the assembly of equip-
ment was begun during August 1974.

The apparatus will consist primarily of a vigorous-
ly stirred, scaled nickel vessel that can be maintained
at a controlled temperature (£2°C) in the range of
500 to 700° C. Provision will be made for intermittent
salt addition, sampling, and for insertion of an
electrode assembly for voltammetric anaylsis—all
under a purified argon atmosphere. The voltammeter
needed for the analyses is also under construction.

8. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr., MSR
Program Semiannu. Progr. Rep. Aug. 31, 1972, ORNL-4832, p.
57,

51

The first reaction to be studied will be

ThO,(c) + 2NiF,(d) = ThF4(d) + 2NiO(c) . (2)
At a given temperature, the variation of the equilibrium
quotient, @ = Xtp5,/XNir, , with melt composition
should reflect the changes in the activity coefficients,
YNiF, and yrpF,, 50 that @ = Kk, /YThr, - Here K
is the equilibrium constant, defined equal to @ in
Li;BeF;. From the value of K we obtain the free
energy change

~RTInK = AG° = Gy,
+ 2AGYi0 — 28GNir, — AGTho, -

The formation free energies of the oxides are quite well
known, as is that for NiF, in Li, BeF4(AC_;fNiF'2).
Hence, from AG® we can obtain an improved value for
the formation free energy of ThF,; Li, BeF,.

The activity coefficient of NiF:2 in the fuel salt
mixture [ NiF2(c) = NiF2(d)] can be derived from a
measurement of the solubility of NiF: where

K = XNiF, YNIF, - (3)
From vywir, and the quotient ')/ZNiFZ_/’)/Tth, obtained
from a measurement of reaction (2) in the same LiF-
BeF:-ThF: salt mixture, we can obtain the activity
coefficient of ThF. in this fuel salt mixture.

9.5 THE CHEMISTRY
AND THERMODYNAMICS
OF MOLTEN-SALT REACTOR FUELS’

C. F. Baes, Jr.

The fuel of an MSBR, in which fissile **’U is bred
from ***Th by thermal neutrons, will probably consist
of the mixture LiF-BeF:-ThF4 (72-16-12 mole %) at
about 600° Cin which less than | mole 9% of fissile UF,
or PuFs is dissolved. A fairly detailed knowledge of
the chemistry of such a molten salt mixture has been
gained from measurements of heterogeneous
equilibna involving various gaseous (e.g., HF, H20)
and solid (usually oxide) compounds combined with

9. Summary of invited paper presented at the U.S.-Japanese
Conference on the Thermodynamics of Nuclear Materials, Ames,
lowa, July 1973, and included in the published proceedings J.
Nucl, Mater, 51, 149 (1974).

the thermochemical data available for the pure
compounds.

The entropy of a fluoride component in the fuel can
be cstimated with useful accuracy from the known
entropy of the corresponding oxide and the charge on
the cation (Fig. 9.5). The activity coefficients of BeF2,
ThF., and UFs decrease strongly as the mole percen-
tage of LiF in an LiF-BeF:-ThFs solventis increased,
but the varnations do not exceed a factor of 10 from
activity coefficient values in the reference composi-
tion LiF-BeF: (67-33 mole %). The activity coef-
ficient changes for CeFs, UF;, PuF;,and NiF; are less
pronounced.

The solubility of oxides in such molten fluoride
mixtures generally decreases with z7/r (the square of
the charge of the cation divided by its radius) for the
cation involved (Fig. 9.6). Thus, BeO, ZrO,, the
actinide dioxides, and especially Pa2Os show a low
solubility in MSBR fuels. The actinide dioxides form
solid solutions that canexchange tetravalent cations
with the molten fluoride and could prove useful in
fuel reprocessing. The oxide Pa:0s (or an addition
compound of it) is so insoluble that it could be used
selectively to remove “7Pa bred in the fuel (Fig. 9.7).

The fuel of the MSBR will be maintained mildly
reducing with the ratio U* /U™ in the range 10 to
100. The other actinides then will be present as Th*,

Pa*, and Pu”. Corrosion reactions with the metals

52

ORNL-DWG 73 - 6976

40
S[MF (g =5[MO, x5 (c)] + 7.

¥ x
:_ 30 / 3
3 3
e 3+

¥ - ?Ce o x
g Beg . oTh 3
e Fe U'“n Q
5 20 30 &
— qu‘H -
s L
o .+ 2+ S
OH Li Ni o o
3 =
Ty / »
L 10 [ ”s 0
= Be ece® )
< g
= =
= NiZt »

FeZ*
o] 10
Li"/
S[MF, (c)]=5[MO,/ (e)]) +7.52
I R .
0 1 2 3 4 5

CATION CHARGE

Fig. 9.5. Correlation of entropies of crystalline fluorides and of

fluorides dissolved in molten LiF-BeF: (67-33 mole %) with
entropies of crystalline oxides.

ORNL-DWG 73- 6977A

30
. 20 —’Rb: Ve MF,(c) + Y2 Lig0(c) == LiF ¢} + Yz MOz, (c)
3 .\,K__Cs+
~ e Nat
—~ 10
X \
@ . 2+
& or LI;_EO ca2*
®
E Sr2+ ) a %03'*
S \ 34
o
L':) Mgz: .Y
b= 20 - TIt— A 24 2+ oy \ Th4+ 34
& Cd?*a Fecq e AP g,
2+ _ £0_ 2+ Bi3+ Be's 4+ @ Zr

W Aqt Mn o Ni . Pa"" e ’
2 -30 S~ B 7net © Fed* T— Pg>* -
E Cu?* cr3t  epu?t *

-40 ': 2+

_ g l A TI3* o T4+

_50 I

0 ) 10 15 20 25 30

Fig. 9.6. Effect of cation charge and radius on the relative stability of oxides and fluorides. @, pre-transition; O, transition; and A, post-

transition elements.

ORNL~DWG 73-6973

10
10 i I T T
Pt /pu3t=1
10° |- _ L~ Pu0,-U0,-ThO, |
xF‘u‘""' xPu'“'= io (ss)
108 -
107 -
6 -4
10° =1 A
Xp 5+, Xpgd+ ™ 10 )
-5
'-3; 10 /POZOS(C)
2 .5
~. 107 ]
$D |
Pa>t/Pc*t = ¢ ~
104 ~ ~ ]
~
.
[ ~ ~ ~
103 — ~ —
~
~
S
~
102 - 3
Pa0,~ThO,- UG,
o' o
100 | | | J | 1

100" 07 107 10 1w0* 10 wE o
MOLE FRACTICN OXIDE

Fig. 9.7. Pourbaix diagram for the actinide elements in LiF-
BeF,-ThF, (72-16-12 mole %) at 600° C. X denotes mole fraction;
ss denotes solid solution.

of the nickel-base container alloy (Hastelloy N) will
not be favored thermodynamically (Fig. 9.8). The
noble-gas fission products are insoluble and can be
removed by continuous stripping with an inert gas.
Group VII, 1, 11, 111, and IV fission products should
appear as dissolved species in their normal valences.
The remaining fission products (principally Nb, Mo,
Te, and Ru) should be reduced to the metallic state.

9.6 OXIDE CHEMISTRY OF NIOBIUM (V)
IN MOLTEN LiF-BeF; MIXTURES!'?

C. E. Bamberger
G. Mamantov'’

Gann Tang“
C. F. Baes, Jr.

The chemistry of niobium in molten fluorides is
relevant to the MSR Program for several reasons: (1)
During the operation of the MSRE, the appearance
of fission product *’Nb in the fuel seemed to be a
sensitive function of the state of oxidation (the
U*/U* ratio) of the fuel.” This effect, which

ORNL-DWG 73-6974

{0
10 - l ] T
10° _
NiO (¢) IC)
o
8 e
10° - —
X2+ =107> i
o= _ | ]
1074 :
-5 *~
10 -={
105 — ( )X e
Fe O, (c
=403 34 .
:; XFe2+ 10 . \\
D 5| - \\_\.
:;\ 10 1o a__ |__ N
2
Cr,0,(c}
a | 2v3
10 \ (o=5
.xCr2+= 0% \
103 \ N —
\
-4
10— \\
~\N
10% |- 10 °>—-— -3
1 _
10" - X j4+=3%10 3 —
UOa-Th02
(ss)
100 ! | ] |

10°% 107 10°% 107 10 107 107°
MOLE FRACTION OXIDE

Fig. 9.8. Pourbaix diagram for structural metals in LiF-BeF-
ThF, (72-16-12 mole %) at 600°C. The horizontal lines denote
equilibrium between the indicated mole fraction of ions and the
pure metal.

presumably involves the oxidation of the metal to a
soluble species, might be a useful indicator of the
state of oxidation of the fuel. (2) Niobium pentoxide
(Nb20s), like protactinium pentoxide (Pa:0s), is
expected to be sparingly soluble in molten fluorides,
and Nb(V) has been proposed for use as a stand-in for
Pa(V) in studies of fuel reprocessing methods that

10. Summary of paper now in preparation, based on portions of
a Ph.D. thesis “Thermodynamic and Electrochemical Studies of
Niobium in Molten Fluorides and Chloroaluminates,” by Gann
Ting, University of Tennessee, Knoxville, August 1973.

1. Guest scientist, Institute of Nuclear Energy Research,
Republic of China; present address: Institute of Nuclear Energy
Research, Lung-Tan, Taiwan,

2. Consultant to Analytical Chemistry Division, Department
of Chemistry, University of Tennessee, Knoxville.

13. R. E. Thoma, Chemical Aspects of MSRE Operation,

. ORNL-4658 (December 1971), pp. 94-99.
tnvolve oxide precipitation. However, the small size
of Nb’"ion (0.69 &) compared with Pa** ion (0.89 &)
might produce oxyions of Nb(V), while none were
found for Pa(V)." This would profoundly affect the
chemistry of Nb(V) in molten fluorides containing
oxide. (3) In the system NiO-Nb:Os, at least two
intermediate compounds, NiNb:Os and NisNb2Os,
are known, and since NiQ is also a sparingly soluble
oxide, precipitation of these nickel niobates may be
expected further to complicate the chemistry of
Nb(V) in . an MSBR fuel contained in a nickel-base
alloy.

The results of our equilibrations at 600°C that
involve the sparingly soluble oxides BeQO, NiO,
Nb20s, and NiNb2Os in molten LiF-BeF2 (67-33 mole
" %) are represented by the solid boundaries shown in
Fig. 9.9. The reactions that correspond to each of
these three-phase equilibria (two oxide phases and
the molten fluoride phase) are listed in Table 9.1,
reactions (1), (4), and (5). The dashed boundaries can
be constructed from the stoichiometries of the
derived reactions (2) and (3). Contrary to reports in
the literature,'” the compound NisNb,Os could not be
established as a stable phase in this system.

The boundary between NiNb:Os and NiO in Fig.
9.9 was of special interest, since its slope gives direct
evidence for the formation of an oxyion, NbQ:", in
the molten fluoride phase. This evidence can be seen
from the general reaction (d indicates dissolved
species) ‘

x/2 NiNb, Og(c) + ¥/2 NiF,(d)

= Nbe(sx_Ly)/sz(d) + (x +y)/2 NIO(C)

14. R. G. Ross, C. E. Bamberger, and C. F. Baes, Jr.,J. fnorg.
Nucl. Chem. 35, 433 (1973).

15. E. V. Tkatchenko, F. Abbattista, and A. Burdeze, [zv.
Akad. Nauk 8SSR, Neorg. Mater.5,1671(1969); F. Abbattista, E.
V. Tkatchenko, and C. T. Chiantavetto, Atti dccad. Sci. Torino,
Cl. Sci. Fis., Mat. Nai. 103, 605 (1969).

54

and from its equilibrium quotient (X denotes mole
fraction)

Q= X[Nb, Os,_,y/2F, 1/ [X(NiF)]¥/2 .

From the slope of 1/2 observed for the boundary in
Fig. 9.9 corresponding to this reaction, the value of y
is 1. From this value we conclude that Nb(V) forms
the oxyion NbO:' in solution. The alternative
polynuclear ions that are possible (such as NbsO;*
and NbsO12") seem quite unlikely at the low concen-
trations of niobium involved.

The magnitude of the solubility of Nb2O:s (i.e., the
height of the Nb2Os/BeO boundary’ in Fig. 9.9),
which is much greater than originally expected,
provides additional strong evidence thatan oxyion of
Nb(V) is formed in solution. The stability of Nb*>*
otherwise required to explain this high solubility

'6. For convenience in this reaction and in Table 9.1 we write
the formulas of dissolved species as neutral components. The
number of F™ ions thus shown is not intended to imply the number
F ions actually present in a given species.

o ORNL-DWG. 74 -5839
10 [ T T vlll;r‘ |||; T T T T
F Nb,Og el -
- - NiNB,Og( <)
-5 \
10 ~
F \ A
w r *-\
g |
2 [
pod
-6 . _
107 BeO {c} NiO {c) 3
-7 l | wul Ll
10 111 111
107 107° 107" 0 * 167 102

XniF,

Fig. 9.9. Phase diagram representing the NbO:F and NiF;
concentrations in molten LiF-BeF; (67-33 mole %) at 600°C as a
function of the oxide phases present at equilibrium.

Table 9.1. Equilibria of Nb:20s, NiO, and BeO with LiF-BeF; (67-33 mole %) at 600°C

Reaction Boundary Equilibrium reaction Q°

(1) Nb2Os(c)/ BeO(c) [/2Nb20s(c) + 1/2BeFa(d) = NbO;F(d) + 1/2BeO(c) X(NbO:F)=10"*"

(2) Nb20s(c)/ NiNb20Os(c)  NbzOs(c) + 1/2 NiFx(d) = NbO:F(d) + 1/2NiNb;Os(c) X(NbO:F)/ [ X(NiF2)] =107
(3) NiNb20s(c)/ BeO(c) 1/2NiNb20s(c)+ Be F2(d) = NbO:F(d)+ 1/ 2NiF:(d)+BeO(c)  X(NbO:F) X(NiF2)"*=10""
4) NiO(c)/ BeO(c) NiO(c) + BeFa(d) = NiFa(d) + BeO(c) X(NiF>) = 107"

(5) NiNb:Os(c)/ NiO(c) 1/2NiNb20s(c) + 1/2NiF2(d) = NbO:F(d) + NiO(c) X(NbO:F)/[X(NiF)])* =10+

“In LiF-BeF: (67-33 mole %) the activity of BeF:(d) has been defined as unity.
would be so great that the Nb** found by Weaveretal.
to be formed by the reaction'’

Nb’(c)+4HF(g) — NbFi(d) + 2Ha(g)  (4)
should have disproportionated almost completely as

NbFas(d) < 4/5 NbFs(d) + 1/5 Nb(c).  (5)
This was clearly not the case.

From the measured equilibrium quotients and the
formation free energies of the other oxides and
dissolved fluorides involved, we obtain the following
free energy of formation of NbO:F in the mole
fraction standard state in LiF-BeF: (67-33 mole %):
AG ' [NbO:F(d)] = —250.7+45.9 T/ 10 keal/ mole,
and for NiNb2Os we obtain
AG[NiNb2Os(c)] =—509.0 +112.7 T/ 10" keal/ mole.
With the measurements of the equilibrium quotient
for reaction (4),'” we can also construct the Pourbaix
diagram (Fig. 9.10), wherein the conditions of oxide
concentration and redox potential are shown under

which the various forms of niobium are stable in a
molten-salt breeder reactor fuel.

17. C. F. Weaver, H. A. Friedman, and J. S. Gill, MSR
Program Sentiannu. Progr. Rep. Aug. 31, 1970, ORNL4622, p.
73.

55

ORNL—DWG 73—-6972

9
10 1 | ]
i
08 NbFs(g) [
{ = —
- [0 %am NbBOS (d) |
| =
07 - | o: -]
| Fe)
I =
-6
10
10° |- SN
e
S
. Np*t(d) o' ]
o107 - =
>
=
< “6 - 40—8
2 10% |- 01 o7 —————— g
-6
10 10—8
-7
103 10 NbO, (c) |
-8
10
~
e
NB° >~
10! |- ~d
U0,-ThO, {ss)
10° | | | | | |
1078 407 0% 4075 407 407> 1072 0™

MOLE FRACTION OXIDE

Fig. 9.10. Pourbaix diagram for niobium in LiF-BeF:-ThF.

(72-16-12 mole %) at 600°C.
Part 3. Materials Development

H. E. McCoy

The main thrust of our materials program is
aimed at the development of structural material for
the primary circuit which would have adequate re-
sistance to embrittlement by neutron irradiation
and to intergranular cracking by fission product
penetration. A modified composition of Hastelloy
N containing 2% Ti has good resistance to irradia-
tion embrittlement, but some question remains as
to whether it has sufficient resistance to intergran-
ular cracking. Further modification of the alloy
with higher chromium or rare earth additions to
impart better resistance to intergranular cracking
may be necessary. Laboratory tests are in progress
to answer this very important question.

A 10,000-1b production heat of the 2% Ti-modi-
fied Hastelloy N is being procured, and smaller
heats of several other modified compositions are
being procured. The evaluation program includes
irradiated and unirradiated mechanical property
tests, tellurium-compatibility tests, salt corrosion
studies, microstructural characterization, and steam
corrosion studies.

Much of the effort during this report period has
been to reestablish the test facilities previously used
by this program. Most of these facilities were either
made available to other programs or dismantled. A
new facility 1s being built for the thermal convec-
tion loops and a new mechanical property and gen-
cral test facility is also under construction.

The work on chemical processing materials is
concentrated on graphite. The primary question is
whether chemical reactions occur between the
bismuth-lithium solutions used 1n the processing
flow sheet and graphite. Several bench experiments
are in progress to study this question.

Tests run previously have shown that some exist-
ing commercial graphites will maintain good struc-
tural integrity to target fluences of 3 X 107

-neutrons/cm’. Although our development work has

shown that graphites with higher allowable fluences
can be developed, this is not a high priority item
and 1s not currently being pursued. We are prepar-
ing a status report on our graphite coating efforts
but do not have an active experimental program 1n
this area.
57

10. Development of Modified Hastelloy N

H. E. McCoy

The purpose of this program is to develop metal-
lic structural material for an MSBR. The current
cmphasis is on the development of a material for
thc primary circuit, since this appears to be the
" most important materials problem at the present
time. The material for the primary circuit will be
cxposed to a modest thermal neutron flux and to
fucl salt that contains fission products. Researchers
currcntly believe that a modification of standard
Hastelloy N will be a satisfactory material for this
application. An alloy that contains 2% Ti seems to
adequatcly resist irradiation embrittlement, but
whcther or not this alloy satisfactorily resists inter-
granular embrittlement by the fusion product tel-
lurium remains to be demonstrated. Small amounts
of niobium and rare earths (e.g., cerium and lan-
thanum) also improve the resistance to intergranu-
lar cracking, and likely will not reduce the benefi-
cial cffects of titanium from the neutron embrittle-
ment standpoint. Currently, scale-up of the 2% Ti
alloy 1s being pursued while smaller heats are be-
ing made of alloys that contain both 2% Ti and
additions of niobium and rare earths. These materi-
als arc being evaluated in several ways.

Considerable effort has been expended in devel-
opment and construction of new test facilities since
thc program was restarted. Most of the test facili-
tics previously in use were made available to other
programs when the MSR Program was terminated.
The two major facilities constructed recently con-
sist of a new thermal convection loop test facility
and a mechanical property laboratory. The loop
facility will accommodate 12 loops and will likely
be in operation before the end of CY 1974, The
mcchanical property laboratory will have a total of
30 creep machines, 5 strain cycle machines, a heat
trcating facility, and facilities for bench tests. This
cquipment will gradually become operational, with
at least 50% in use by the end of CY 1974.

10.1 PROCUREMENT
OF NEW TEST MATERIALS

One small (about 50 Ib) commercial heat of 2%
Ti-modified Hastelloy N was received from the In-
ternational Nickel Corporation just as the MSR

Program was terminated. The composition of this
alloy, designated 73-008, 1s given in Table 10.1
along with the analyses of several heats of material
that will be discussed in this chapter. Heat 73-008
was rcceived in the form of 1/2-in.-thick plate and
appeared sound. This material has been included
in all phases of the Hastelloy N development
program.

Four small (~125 Ib) commercial melts chosen to
study the combined effects of titanium and rare
carths have been melted and partially fabricated by
Cabot Corporation. Each of the four heats contains
2% Ti and three of the heats also contain additions
of about 0.01% rare earths. The three rare earth
additions being evaluated are cerium, lanthanum,
and Misch metal (primarily cerium and lantha-
num). These heats were melted by vacuum-induc-
tion melting (VIM) followed by electroslag remelt-
ing (ESR). The material has been fabricated to
1/2-in.-thick plate and is being conditioned for
shipment. :

Procurement has been initiated on a 10,000-1b
heat of 29 Ti-modified Hastelloy N which will be
fabricated into numerous product forms. The mate-
rial has been melted and forged to 19 in. X 10 in. X
length. Fabrication into the various product forms
is In progress. ».

10.2 WELDABILITY OF COMMERCIAL
ALLOYS OF MODIFIED HASTELLOY N
B. McNabb H. E. McCoy

A small heat of 295 Ti-modified Hastelloy N (heat
73-008) was received from the International Nickel
Company in the form of 1/2-in.-thick plate, 4 in.
wide X 18 in. long. The chemical analysis of this
matcrial is given in Table 10.1.

Strips that were 1/2-in. square were sawed from
the plate, and the corners rounded and swaged to a
I/4-in.-diam rod. Some of the rod was used for
mecchanical property specimens and some was
swaged to 3/32-in. weld wire. The remaining por-
tion of the 1/3-in.-thick plate was sawed into two
9-in.-long pieces, and the edges were beveled to
produce a 100°C-included angle for the weld de-
posit. The beveled plates were welded to a 4-in.-
58

Table 10.1. ORNL analysis of several heats of modified Hastelloy N

The heat numbers are often used with a prefix of “4.” This designation
means that the '/, <in.-thick, as-received plate was cut into strips

'/, X '/, in. and swaged cold to '/, -in.-diam rod.

Concentration, wt %

Element Heat Heat Heat Heat Heat Heat
71-1144 71-583b 72-503¢ 72-604b 72-115¢ 73-008%

G 7.4 1.25 6.79 6.69 7.03 7.63
Mo 12.5 12.4 12.9 11.5 11.9 12.4
Fe 0.062 0.13 0.089 0.048 0.02 0.18
C 0.058 0.057 0.066 0.118 0.091 0.078
Si 0.026 0.055 0.089 0.12 0.073 0.06
Mn 0.02 0.03 <0.01 0.03 <0.01 0.45
P 0.007 0.007 0.0008 0.003 0.0028
S 0.01 0.004 0.002 0.002 0.002
Al 0.07 0.2 0.09 0.17 0.08 0.03
B 0.00005 0.00005 0.00007 <0.00002 <0.00002 <0.003
Ti 1.75 1.44 2.16 <0.005 <0.005 2.1
Hf <0.003 0.12 0.62
Zr <0.03 <0.03 0.003 0.009 0.015
H : 0.0022
N 0.0012 0.0029 0.0003 0.0008 0.0004 0.0010
O 0.0003 0.012 0.0017 0.0031 0.0017 <0.001
Co 0.07 0.1 0.05 0.05 0.05
Cu 0.007 0.015 0.01 0.03 0.02
Cb 0.05 <0.005 <0.005
v <0.01 <0.01 <0.01
w <0.05 <0.005 <0.05
Mg 0.01 0.01 0.02

%The melting process used was vacuum induction melt (VIM).

bThe melting process used was consumable electrode vacuum melt (CEVM).

“The melting process used was electroslag remelt (ESR).

thick stecl strongback for a fully restrained weld
test, and weld metal was deposited in the beveled
arca in the horizontal position. The weld was made
in 18 passes, with the first, second, and sixth
passes being made with the base metal at room
temperature. The welding current for the first pass
was 80 A, increasing about 10 A per pass for three
passcs, then 20 A per pass up to 180 A on the sev-
cnth pass. The remainder of the passes were made
at 180 A with the interpass temperature at about
77°C (170°F).

The root pass and the final passes were inspected
by liquid dye penetrant, and the intermediate
passcs were inspected visually. No flaws were de-
tccted. The completed weld was then x-ray in-
spccted and one small spot of porosity was de-
tected. The welding operation was somewhat diffi-
cult duc to dense smoke from the weld puddle, but
this phcnomenon could not be explained from the
rcported chemical analysis or from the cleaning
procedurcs used to prepare the materials before

welding. The same cleaning procedure was used as
with previous heats of modified Hastelloy N which
did not have this problem. The weld wire was elec-
tropolished and both the wire and beveled plates
were cleaned with acetone prior to welding.
Side-bend specimens (1/8-in. thick X 1/2-in,
wide) were sawed from the weld area and bent
around a 3/8-in.-radius mandrel in a guided bend
test. Dye penetrant inspection revealed several fine
shallow cracks in the weld as shown in Fig. 10.1.
Onc specimen appeared to have some cracks in the
basc metal in the bend area adjacent to the weld.
Accordingly, a cection of the base metal from the
same specimen was bend tested, but no cracks were
revealed by dye penetrant inspection. Figure 10.2
shows the shallowness of the weld cracks and their
nonpropagation through the specimen during bend-
ing. Figurc 10.3 is a photomicrograph of the bend
arca of the same specimen shown in Fig. 10.2, and
shows the very shallow cracks in the weld metal
and the heat-affected zone. The bend around a 3/8-
Modified Hastelloy N
Heat 73008 Weld

Fig. 10.1. Photomicrograph of side bend specimens 1/8-in.
thick by 1/2-in. wide bent around a 3/8-in.-radius mandrel. Dye
penctrant has been applied, and dark streaks indicate shallow
cracks

100 200 MICRONS 600
L i00X—ir
0005 0.010 INCHES 0020 0.02!

59

in.-radius mandrel is a very severe test, and crack-
ing to the depth shown could be tolerated in most
applications.

Tensile specimens were machined from the weld
arca of the plate and from the 1 /4-in.-diam rod that
was swaged from this heat for comparison of base
metal and weld properties. Table 10.2 gives the ten-
sile properties of the base metal and weld metal at
temperatures of 25 and 650°C. The weld metal in
the as-welded condition had higher yield stress
than the base metal, but only one half as much
ductility or elongation. The usual stress-relief an-
ncal of 8 hr at 871°C (1600°F) lowered the yield
stress of the weld metal with no recovery of ductil-
ity. The solution anneal of | hr at 1177°C (2150° F)
further lowered the yield stress of the weld metal,
and the ductility recovered to that of the base
metal.

Apparently, heat 73-008 can be satisfactorily
welded and, by proper heat treatment, can attain
properties equivalent to those of the base metal.

¥-125970

Fig. 10.2. Photomicrograph of weld in bend area of side bend specimen. Cracks at surface that was in tension are approximately | mil

deep. Etched with 50% HCI, 109 HNO:. 100 X

Fig. 10.3. Photomicrograph of bend area of side bend specimen at fusion line. Approximately 1-mil-deep cracks o not propagate in
metal. Etched with 50% HCL, 10% HNOx. 100 X

weld or

60

¥-125967

100 200 MICRONS 600 700
Y P e 2 ?

0005 0010 INCHES 0020 0025

Table 10.2. Comparison of base metal and weld metal tensile properties of modified

Hastelloy N heat 73-008 at 25 and 650°C

Stress (psi X 10°)

Test Total Reduction

Specimen temperature Ultimate elongation in area

Number Condition (W] Yield tensile Fracture (%) (%)
14530 Lhrat 1177°C 25 53.7 139.0 104.0 46.20 5284
14589 As-welded 25 802 1202 94.1 2167 3505
14623 Welded—8 hr at 871°C 25 68.4 115 105.4 2139 4280
14621 Welded—1 hrat 1177°C 25 483 1136 104.8 4190 42,60
14531 Lhrat 1I7C 650 39.1 97.3 80.0 2894
14590 As-welded 650 595 78.1 55.4 35.70
14624 Welded—8 hr at 871°C 650 519 75.1 61.7 3180
14622 Welded—1 hr at 1177°C 650 24 789 712 2305

10.3 MECHANICAL PROPERTIES
OF MODIFIED HASTELLOY N

H. E. McCoy

Thc creep. properties of three heats of 2% Ti-
modified Hastelloy N and two heats of Hf-modified
Hastelloy N were reported previously.' The results
for the unirradiated condition showed clearly that
all five heats were as strong as or stronger than
standard Hastelloy N at test temperatures of 650
and 760°C. These data have not been expanded sig-
nificantly since that time. Samples of these heats
have also been irradiated and a few of the postirra-
diation creep tests completed. Many additional
postirradiation tests have been completed since
that time, and these will be discussed in more detail
latcr in the report.

The postirradiation stress-rupture observations at
650° C for the three Ti-modified heats are shown in

I. H. E. McCoy and B. McNabb, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 117.

61

Fig. 10.4. The numbers by the individual points are
the fracture strain. Although all three heats had ex-
ccllent properties after irradiation at 650°C, the
properties of heat 72-503 were superior. Irradiation
at 704°C reduced the rupture life and strain, but
the properties of all heats are acceptable under
these conditions. Irradiation at 760°C caused an
cven larger drop in rupture life and strain. The
properties of heat 72-503 are acceptable, but the
propertics of the other two heats at stress levels
above 30,000 psi are poor. At lower stress levels,
the properties of heats 71-114 and 71-583 improve.
This type of behavior has been noted previously,
suggesting that there may be some critical stress
bclow which irradiation has little effect on the
mecchanical properties.

Our studies have generally shown that optimum
postirradiation properties are obtained when the
material is annealed | hr at 1177°C prior to irradia-
tion. Since the improved resistance to irradiation
damage results from obtaining a specific carbide
morphology, continued attention must be given to

ORNL-DWG 75-153

70
60
16.2 19.8
STANDARD HASTELLOY N 'T
L IRRADIATED AT 650° 12.0
50 DIATED AT 650 °C. 60 gz Jse
_ N : o2 o || raz
> 98 | 39 ‘ 1
O - - ~— : || .
S *0 [TsTrRaINS <05% | | N fskes'“ 1"37 9'4"°¢21.6
= ~-a 422 ‘-n 92w | .19.2
2 RIS T~ L
: T T TR = - ]| A
[72) 7Y M ov-
NUMBERS BY POINTS ~ - 2.5 | (T2
ARE STRAINS N SOl TS
20 = o 71-144 (1.96 % Ti) ~ " 7%
a 71-583 (1.79 % Ti) ™ 571 '
0 72-503 (4.94 % Ti) : -
10 | OPEN POINTS IRRADIATED AT 650°C TS
FILLED POINTS IRRADIATED AT 760 °C STANDARD HASTELLOY N
CROSSED OPEN POINTS IRRADIATED AT 704 °C IRRADIATED AT 760 °C
IS I N TSR VO i 1 1 IlllH
5

0! 2 5 40° 2 5 10

2

5 102 2 t0®> 2 104

RUPTURE TIME (hr}

Fig. 10.4, Postirradiation creep properties of Ti-modified alloys at 650° C after irradiation at indicated temperature to thermal fluence
of 3 x 10® neutrons/ cm®. Numbers by points indicate fracture strains.
62

thc possible importance of heat treatment. Speci-
mens of heat 71-114 were annealed for | hr at tem-
peraturcs from 1038 to 1260°C and irradiated at
760° C. The results of the creep test at 650°C are
shown in Fig. 10.5. These data show that the opti-
mum rupture life and strain occur from annealing
at 1204° C. However, samples annealed at 1177°C
arc “second best” with an indicated trend of the
two anneals resulting in equivalent properties at
stresses below about 20,000 psi.

Weld samples of heat 71-114 and 71-583 were ir-
radiated and the results of postirradiation creep
tests are shown in Fig. 10.6. The test matrix is not
complete, but the data indicate that the properties
of both heats irradiated in the as-welded condition
at 650 and 760°C are very poor. Postweld anneal-
ing at 1177°C greatly improved the properties of
hcat 71-583 after irradiation at 650° C. Postweld an-
ncaling at 1177°C will probably produce markedly
improved properties in both heats after irradiation
at cither 650 or 760° C.

The results of postirradiation tests on the two
Hf-modified heats are shown in Fig. 10.7. After ir-
radiation at 650°C both heats had good properties,
although the properties of heat 72-115 with 0.79% Hf
were superior. Irradiation at 704° C reduced the
rupture life and strain compared with those ob-
scrved at 650°C. The fracture strain of heat 72-604
was reduced to a marginal level. Irradiation at

STANDARD HASTELLOY N

760° C reduced the properties of both heats to about
the levels noted for standard Hastelloy N.

We found previously' that heat 72-115 with 0.5%
Hf formed weld metal cracks during welding. This
problem, coupled with the difficulty of holding a
metal as rcactive as hafnium in the melt, discour-
ages further development of the Hf-modified
Hastelloy N. Future efforts will be concentrated on
the development of the Ti-modified alloy.

10.4 TRANSMISSION ELECTRON
MICROSCOPY OF Ti-MODIFIED
HASTELLOY N ALLOYS

D. N. Braski J. M. Leitnaker G. A. Potter

The near-term objective of this investigation is to
cxplain why two Ti-modified alloys, heat 471-114

‘and 472-503, behaved so differently in the postirra-

diation creep tests described in the previous section.
The 472-503 alloy was vastly superior to the 471-
114 material; in some tests the rupture lives of 472-
503 specimens exceeded those of 471-114 by
greater than two orders of magnitude. These re-
sults are surprising since the two alloys have vir-
tually identical compositions (Table 10.1). To un-
derstand the results of the postirradiation creep
tests, the combined effects of long exposure times at
clevated temperature in the reactor and the radia-

ORNL-DWG 75-154

Or IRRADIATED AT I
650°C — 1 o7
760 °C ?‘ <
. e
40 \.LL N 368
_— " | 2. [~ .
'g 27a | (0247~ fl\ °03 ~~ Nl g 140
S 30 1.0dR NG 3.6 =
O #Nn \h
Q 0.4~ J ™™ ] 8.6
~ ~ | 034 al8 N~0,
" || g 0.8 d
Q 20 1 N e ~ 7.9d-3
5 NUMBERS BY POINTS ARE STRAINS i 42-2
o 119 - 1038°C ™
o L 4120-1093°C e
! o 121~ 1177 °C
® 127-1204 °C
& 123-1260°C
0 IR R | B N B W N 1T R
! 2 5 10° 2 5 10 2 5 102 2 5 10° 2 5 10%

RUPTURE TIME (hr)

Fig. 10.5. Influence of preirradiation annealing on the postirradiation creep behavior of heat 71-114 at 650°C. Samples annealed for |
hr at the indicated temperature and then irradiated at 760°C to a thermal fluence of 3 X 10°° neutrons/cm’. The numbers by each pointare the
7.1 fracture strains,
63

ORNL-DWG 75155

70 ) 1 T 1T TTTTT T T T TTTTET T T
NUMBERS BY POINTS ARE STRAINS ’ I l””
o 7{-114
o 71-583
60 OPEN POINTS IRRADIATED AT 650 °C
FILLED POINTS IRRADIATED AT 760 °C
STANDARD HASTELLOY N CROSSED POINTS FOR SAMPLES ANNEALED
50 'RRAgjj“(‘)TECD AT (M 1hr AT #77°C, ALL OTHERS AS WELDED |
Z 760 °C \I\\\ 11
O \ N '{"h _5¢’6_ »
8 40 / 0.4 N[ 8.6 ";' I ’
= 04| 04[]]I v, B L .
) ~L T~
2 10 BASE METAL
= 0.2 o
~d B
@ *%.2 ~ T\.,
- b ~ o
20 <
o.7¢|[l| 0.2
h.~
10
0

100" 2 5 109 2 5 ' 2 5 102 2 5 0% 2 5 0%
RUPTURE TIME (hr)

Fig. 10.6. Postirradiation creep properties at 650° C of welds after irradiation to a thermal fluence of 3 X 10%° neutrons/cm’. The
numbers by ecach point are the fracture strain.

ORNL-DWG 75-156

T TTIME 7T
STANDARD HASTELLOY N
0 L IRRADIATED AT
650 °C <
760 °CN 184
[ \ \
_ 30 6.7 9.3 Ta71(14.0
- * < < o)
[= 8 -
o / ™ | }
o ~ ~ —o 1.8
e % InlVEx s L 9-B+:ln||
= e 8 2\34»,H ©6.0 8.0410.9
] 1\ 30 | & ~ !
& 30 — \i' e H—; 5k % HHp"3
) 24! | 1.2 ™.
@ NUMBERS BY POINTS ~ > F M~
ARE STRAINS ]_H S
20 - o 72-604 (0.3 % Hf) 19 00 TN
o 72-115 (0.7 % Hf) - |
OPEN POINTS IRRADIATED AT 650°C ~
10 | FILLED POINTS IRRADIATED AT 760 °C N
CROSSED OPEN POINTS IRRADIATED AT 704 °C
Lo v v UL LD
107! 2 5 10° 2 s 10! 2 5 10° 2 5 10° 2 5 40

RUPTURE TIME (hr)

Fig. 10.7. Postirradiation creep properties of Hf-modified alloys at 650° C after irradiation of indicated temperature to thermal fluence
of 3 X 10% neutrons/ cm®. The numbers by each point are the fracture strain.

tion damage itself must be considered separately.
For this reason, duplicate samples of the two al-
loys were anncaled in furnaces for 1000 hr at the
irradiation temperature to assess the effect of the
thermal treatment alone on their respective micro-
structures. Sections of the samples were sub-

sequently examined in a 200-kV transmission elec-

tron microscope, and the results of these observa-
tions arc presented. Electron microscopy studies of
the irradiated samples are currently under way
and will be discussed in a later report.

Before comparing the two allloys, the micro-
structurec of Ti-modified Hastelloy N alloys must
first be considered.” After the alloy is fabricated
into a given form it is usually given a solution an-
ncaling treatment of | hr at 1177 to 1260°C. The
microstructure after this anneal consists of equi-
axcd grains, twins, a low density of dislocations,
and precipitate particles. The precipitates have vari-
ous morphologies (Fig. 10.8a) for a 1.2% Ti-modi-
fied alloy. Large “blocky™ precipitates are often
found in grain boundaries, while more spherical-
shaped ones are located within the grains. Interfa-
cial dislocations are nearly always observed with
the primary spherical precipitates. After the alloy
had becen aged for 200 hr at 760° C, other fine pre-
cipitates were observed on stacking faults which
usually cmanated from the primary precipitates
within the grains (see Fig. 10.86). This interesting
phenomenon has been studied in some detail for
austenite stainless steels by Silcock and Tunstall’
who proposed that (1) solute atoms segregated to
dislocations: (2) the dislocations dissociated into
partials by the reaction a/2 [110] — a/3[111]+a/6
[112]; (3) MC carbides nucleated on the tension
side of the a/3 [111] Frank partial dislocation;
and (4) the stacking fault and MC carbide precipi-
tatcs developed simultaneously. This mechanism
appears to also apply to the titanium modified
Hastelloy N alloys.

Two 29, Ti-modified alloys were aged for 1000 hr
at temperatures of 650, 704, and 760° C, and typical
microstructures of samples of heats 471-114 and
472-503 aged at these temperatures are shown in
Figs. 10.9, 10.10, and 10.11, respectively. All of the
samples contain stacking-fault precipitates which
formed during the thermal aging period. Using

2. C. E. Sessions, Influence of Titanium on the High-
Temperature Deformation of Some Nickel Based Alloys, {The-
sis) ORNL-4561 (July 1970).

3. J. M. Silcock and W. J. Tunstall, “Partial Dislocations As-
sociated with NbC Precipitation in Austenitic Stainless Steels,”
Phil. Mag. 10, 360-89 (1964).

64

sclected-arca electron diffraction and the photo-
micrographs, these stacking-fault precipitates were
found to lie on [111] planes. Selected-area diffrac-
tion and dark-field techniques were used to identify
the stacking-fault precipitates as MC-type carbides.
Similar analyses identified grain boundary and pri-
mary precipitates also as MC. These MC precipi-
tates arc face-centered cubic (fcc) and were simply
rclated to the fcc matrix such that
(100)mc || (100)marix. The lattice parameters of the
MC particles measured ranged from about 4.20 A
to 4.28 A. This range is similar to that reported by
Scssions® using x-ray diffraction on extracted MC
precipitates. The exact composition of the MC par-
ticles in the two 2% Ti-modified alloys has not yet
been determined. However, extrapolation of Ses-
sions data’ on alloys containing from 0.5 to 1.2%
Ti predicts that a 2% Ti alloy would contain
(Moo.s7Tio.33)C. We plan to verify this result by ex-
tracting the MC particles from the alloy and ana-
lyzing them by spectroscopic methods.

The main effect of the 1000-hr aging treatment,
therefore, was to cause precipitation of MC parti-
cles in a stacking fault morphology along [111]
plancs. Lesser amounts of MC were also observed
to dccorate the large grain boundary precipitates
and, in a few instances, along twin boundaries. The
cffect of increasing the aging temperature from 650
to 760°C was to form coarser precipitates at the
higher temperatures for the same aging time. This
point is illustrated by comparing the coarse MC
precipitates along the stacking faults in Fig. 10.8a
with the finer ones observed in Fig. 10.9a. The
amount of precipitation occurring at different tem-
peraturcs and in the different alloys is not known
quantitatively, but extraction experiments are being
initiated to determine these quantities.

After examination of many electron photomicro-
graphs for both the 471-114 and 472-503 alloys, no
obvious qualitative differences in their microstruc-
turcs were noted at the respective aging tempera-
turcs. This result is demonstrated in typical photo-
micrographs (Figs. 10.9-10.11), although admit-
tedly, comparison by the reader is difficult with
only one photomicrograph per condition. Appar-
cntly, thermal aging alone cannot account for the
difference in postirradiation creep behavior. This
tentative conclusion will be strengthened if the pre-
cipitate extraction experiments also indicate that
the respective precipitates of the alloy are similar.
We expect to find significant differences between
the microstructures of the irradiated 471-114 and
472-503 alloys.
YE-11058

S9

lp

Fig. 10.8. Effect of aging at 760°C on microstructure of 1.2% Ti-modified Hastelloy N. (a) Solution annealed 1 hr at 1260°C. (b) Solution annealed 1 hr at
1260°C and aged 200 hr at 760°C.

4 YE-11057

"”LYE-11056

Fig. 10.9. Microstructure of 2% Ti-modified Hastelloy N after solution annealing 1 hr at 1177°C and aging 1000 hr at 650°C. (a) 471-114. (b) 472-503.

99
Fi1g. 1U.1U, Microstructure o1 2% Ti-modified Hastelloy N after solution annealing 1 hr at 1177°C and aging 1000 hr at 704°C. (a) 471-114. (b) 472-503.

Fig. 10.11. Microstructure of 2% Ti-modified Hastelloy N after solution annealing 1 hr at 1177°C and aging 1000 hr at 760°C. (a) 471-114. (b) 472-503.

89
10.5 SALT CORROSION STUDIES

J. R. DiStefano  J. R. Keiser

Hastelloy N exhibits excellent corrosion resis-
tance to fluoride salts containing LiF, BeFz, ThFi,
and UFi. The most reactive constituent of Hastel-
loy N is chromium, which constitutes about 7% of
the alloy. If the fuel salt is pure and the metal 1s
clean, UFu is the strongest oxidant, and corrosion of
Hastelloy N occurs from the reaction

UFa(salt) + 1/2 Cr(alloy)
— UF;(salt) + 1/2 CrFa(salt). (1)

Other reactions can result from impurities 1n the
mclt such as

Cr(alloy) + FeFa(salt)

or
Cr(alloy) + 2HF(d) — CrFz(salt) + Ha(g) . (3)

Oxide films on the metal surface will react accord-
ing to -

Ni1O(alloy) + UF4(salt)

— UOa(s) + 2NiFa(salt), (4)
followed by reaction of NiF: with Cr. Reactions
such as Egs. (2), (3), and (4) proceed essentially to
complction, which results in a high initial corrosion
ratc followed by a decline as the impurities are
consumed.

The cquilibrium constant for Eq. (1) is tempera-
turc dcpendent and thereby affords a driving force
in monisothermal systems for continuous chromium

— CrFz(salt) + Fe(alloy), (2)

69

rcmoval in hot regions and deposition in the cooler
rcgions.

A secondary coolant will link the fuel circuit to
the stcam generator in an MSBR. A mixture of so-
dium fluoride and sodium fluoroborate (NaF-
NaBFi) has a number of attractive properties for
this application. Although thermodynamic data are
not yct well established, predictions and demonstra-
tions thus far reveal that Hastelloy N is very com-
patible with NaBFs-NaF mixtures. However, oxi-
dizing impurities in the fluoroborate' mixture, such
as FeF: or HF, will readily attack chromium (and
perhaps iron and nickel) in the alloy.

The objective of the experiments discussed in this
scction is to quantitatively determine corrosion
rates in fluoride salt-alloy systems. Test variables
include salt purity, salt velocity, system tempera-
turc, cxposure time, and alloy composition.

10.5.1 Fuel Salt Thermal Convection Loop Results

Prior to termination of the MSR Program in
January 1973, a number of thermal convection
loops circulating fuel salt were in operation (Table
10.3). The purposes of these tests were to (1) evalu-
atc electrochemical methods for determining
U*/U* and Cr* in the salt (NCL-21); (2) deter-
mine the corrosion rates of iron-based alloys in fuel
salt (1258 and NCL-22); (3) study the mass transfer
of tellurium in fuel salt (NCL-16A); (4) determine
the cffect of entrained bismuth in salt on the corro-
sion resistance of Hastelloy N (NCL-19A); and (5)
cvaluate the corrosion resistance of several alloys
having improved resistance to cracking by tellur-
ium in fuel salt (NCL-18A).

Table 10.3. Fuel salt thermal convection loop tests in operation January 1973

Maximum Operating
Loop Salt composition temperature time
number Material Specimen {mole %) °QC) {hr)
1258 Type 304L SS Type 304L SS LiF-BeF:-ZrF,-ThFs-UF. 688 80,000
(70-23-5-1-1)
NCL-16A Hastelloy N Te-coated LiF-BeF2-UF.4 704 5,878
Hastelloy N (65.5-34-05)
NCL-I18A Hastelloy N Various alloys LiF-BeF:-ThFi-UFa 704 5.351
(68-20-11.7-0.3)
NCL-19A Hastelloy N Hastelloy N plus LiF-BeF;-ThF:-UF;s 704 24,515
madified Hastel- (68-20-11.7-0.3) plus
loy N Bi in Mo hot finger
NCL-21 Hastelloy N Hastelloy N LiF-BeF2-ZrFi-UF, 650 13.800
(65.4-29.1-5.0-0.5)
NCL-22 Type 316 SS Type 316 SS LiF-BeF:-ThFi-UF, 650 4298

(68-20-11.7-0.3)

Electrochemical analysis of salts. According to
Eq. (1) the oxidizing potential of the melt is deter-
mined by the amounts of UFs; and UF; present.
Loop NCL-21 was equipped with electrochemical
probes® to measure the ratio of uranium (IV) to
uranium (I11) in the salt. Weight losses and gains
in specimens from the hot and cold legs were moni-
torced during loop operation. In general, the ratio
measurements showed excellent reproducibility and
were consistent with test conditions.’

Iron-based alloys. Loop 1258, which was con-
structed of type 304L stainless steel, was shut down
after 1t had accumulated over 80,000 hr of opera-
tion. Two sets of specimens were exposed to fuel
salt during this period. The second set had been
cxposed to 49,057 hr at loop shutdown. Maximum
weight loss of the latter set was 95 mg/cm’ (0.86
mil/year). Weight losses and depth of void forma-
tion were proportional to time to the one-half

4. J. M. Dale and A. S. Meyer, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, pp. 69-70.

5. J. W. Koger, Alloy Compatibility with LiF-BeF, Salts
Containing ThFs and UFs, ORNL-TM-4286, pp. 23-27 (Decem-
ber 1972). '

70

powcr, indicating that solid state diffusion of
chromium was the rate controlling step.’

Loop NCL-22 was constructed of type 316 stain-
less steel and operated for 4491 hr prior to termina-
tion. Wcight changes of the specimens as a function
of position and temperature are shown in Fig.
10.12. After 2842 hr, the maximum weight loss was
6.2 mg/cm’ (0.96 mil/year). At this time, UF; was
added to the salt to make it less oxidizing. In the
next 645-hr period, the maximum corrosion rate
was 0.83 mil/year, but in the final 811-hr period it
was >>| mil/year. The average uniform corrosion
ratc for the hottest specimen was | mil/year for the
entire 4298 hr of exposure to salt. Figure 10.13
shows the microstructure of the hottest and coldest
specimens. The maximum depth of attack in the hot
leg was approximately 3 mils, and the attack area i1s
probably depleted in chromium.’

6. J. W. Koger, Alloy Compatibility with LiF-BeF> Salts
Containing ThFs and UF;, ORNL-TM-4286, pp. 31-33 (Decem-
ber 1972).

7. J. W. Koger, Alloy Compatibility with LiF-BeF: Salis
Containing ThFs and UFs, ORNL-TM-4286, pp. 23-27 (Decem-
ber 1972).

ORNL-DWG 73-9846R

C4 C5 Cé6 C7 cC8 H8 H7 H6 H5 H4
10
| I I i I ] | T
& 5 1200
£ o === 100 &
g N 8
P / Na §
z / %\\ “"
5 -5 — 74 ® 599hr < 1000 &
e © 1440 hr O w
5 a 2842hr -
Y10 p & 3487 hr 900
© 4298 hr
-5 ‘ f
0 10 20 30 40 50 80 70 80 90 100

DISTANCE (in.)

Fig. 10.12. Weight changes of type 316 stainless steel specimens in Loop 22 exposed to LiF-BeF:-ThF-UF, (68-20-11.7-0.3 mole %) asa

function of position and temperature.
71

Fig. 10.13. Microstructures of type 316 stainless steel specimens
in Loop 22 exposed to LiF-BeFs-ThFi-UF4 (68-20-11.7-0.3 mole
9%) after 4298 hr. Hot leg, 650° C: (a) cross section of specimen,
100%; (b) as polished, 500; (c) etched, 500X. Cold leg, 560° C: (d)
as polished, 500X

Tellurium mass transfer. Examination of Hastcl-

loy N samples from the MSR experiment revealed
that all surfaces exposed to fuel salt had intergran-
ular cracks.” Evidence gathered thus far indicates

that the cracking may have been due to intergranu-
lar penctration of the Hastelloy N by the
product tellurium. Loop NCL-16A was operated to
determine  whether  tellurium  would  transfer
through the salt from one part of the system to an-

fission

other. Initially the loop contained standard Hastel-
loy N specimens in hot and cold leg sections plus
one Te-coated Hastelloy N specimen (1.6 mg, cm®).
In the first 1300 hr of operation. the weight loss of
the Te-coated specimen equaled the amount of tel-
lurium that had been added to the specimen. In the
next 2100 hr, the weight loss of the Te-coated speci-
men was the same as the uncoated Hastelloy N
specimen in a neighboring position. Approximately
half of the tellurium removed from the coated sam-
ple was found on a specimen in the cold leg, and
analysis of the salt indicated that it contained the
remainder of the tellurium. These results suggest
that tellurium has some solubility in fuel salt and
will transfer in a temperature-gradient
system.”

Radioactive ('*"Te) Te-coated Hastelloy N spec-
imens were put into the hot and cold leg sections of
loop NCL-16A, replacing the previous specimens.
After the first 42 hr in the loop, all specimens lost
>66; of their initial *""Te activity. By compari
son, a control sample stored in air at room tempera-
ture lost 4% of its '*""Te activity. After 159 hr all
loop specimens had lost 75% of their initial ac-
tivity while the control specimen had lost 8%. Anal-
yses after 396 and 701 hr indicated that hot leg
specimens were continuing to lose '*"Te, but that
a small increase in activity was found in cold leg
specimens. Measurements after 1318 and 1792 hr
indicated no further changes in "Te activity on
specimens in either the hot or cold leg. Weight
change measurements generally corroborated the
dircctions of the changes in '*""Te activity. Overall,
specimens in the hot leg lost weight while those in
the cold leg gained weight. Since the loop operat-
ing temperature was above the melting temperature
of tellurium, and most of the loss in tellurium oc-
curred during the first 42 hr (the first time period
at which measurements were made), the mecha-
nism of tellurium loss into the salt is undetermined.

mass

8. H. E. McCoy and B. McNabb, Intergranular Cracking of
INOR-8 in the MSRE, ORNL-4829 (November 1972).
9. Ref. 7. pp. 27-28

However, the overall pattern of high initial loss on
all specimens, followed by smaller losses in hot leg
specimens and gains in cold leg specimens is con-
sistent with a temperature-gradient mass transfer
model. _

Bismuth-salt-Hastelloy N. Loop NCL-19A was
opcrated to determine whether contact of the salt
with liquid bismuth would affect the corrosion rate
of Hastelloy N immersed in the salt. The bismuth
was contained in a molybdenum “finger” that was
attached to the bottom of the hot leg. The loop
opcrated for 24,515 hr, and weight loss results are
shown in Fig. 10.14. Assuming uniform loss, the
maximum change was —0.02 mil/year. A modified
Hastelloy N alloy (Ni—13%, Mo—8%, Cr—0.17%,
Fc—0.8%, Te—1.6%, Nb) lost slightly less weight
than a standard alloy at the same position. After
18,000 hr, chromium in the salt increased 194 ppm,
iron decreased 70 ppm, and bismuth was not de-
tected. There appeared to be little or no entrain-
ment of bismuth in the salt under these test condi-
tions; therefore, the effect of small quantities of bis-
muth in the salt on corrosion of Hastelloy N cannot
be inferred from this test.

Corrosion resistance of alloys having improved
resistance to tellurium cracking. Several alloys
having improved resistance to tellurium cracking
were exposed to fuel salt in loop NCL-18A. Table
10.4 shows the compositions of the alloys tested

72

Table 10.4, Ndm'inal compositions of test alloys (wt %)
Alloy Ni Cr Fe Mo Cu Al Nb Mn
Moncl 65 35
Hasteclloy N ~70 7 5 16
Inconcl 600 ~75 15 7
Inconcl 606 ~70 20 2 2 3
Inconcl 601 ~60 23 14 1
Inconcl 690 ~50 30 IS5
Incoloy8IIE  ~30 21 47

and Table 10.5 shows the weight changes after an
cxposure of 2776 hr at 690°C. Testing several dif-
fcrent materials together in the same system intro-
duces the possibility of mass transfer of certain
constituents between the alloys because of activity
gradient effects. However, there was a good correla-
tion between weight changes and the chromium
and iron content of the alloy, and the Ni-based al-
loys were more corrosion resistant than the Fe-
based alloys.

Microstructures of the alloys tested are shown in
Fig. 10.15. Hastelloy N and Inconel 600, which had
the smallest weight loss, showed the least attack.
Large areas of void formation can be seen in Inco-
ncl 601 (5 mils deep), Inconel 690 (15 mils deep),
and Incoloy 8I11E (18 mils deep). Monel, which

‘showed a weight gain, showed a surface layer and

ORNL—DWG T71-6990R

2
&y
S T 550°C___ | —o —e
& —e °o— 545°C o .
E /.__—__— @ s ®
W o1 _MODIFIED HASTELLOY N
z e e—
g © —_—— - o 685°C
®

T N705°C ° ——e
1
W —1
z

-2

0 2000 4000 6000 8000 10,000 12,000 18,000

14,000 - 16,000

OPERATING TIME (hr)

Fig. 10.14. Weight changes of Hastelloy N from Loop 19 exposed to LiF-BeF;-ThFs-UF (68-20-11.7-0.3 mole %) for various times.
73

oxr ‘."w-‘.f o

Fig. 10.15. Microstructures of various alloys exposed to LiF-BeF>-ThFi-UF: (68-20-11.7-0.3 mole %) at 690°C for 2776 hr. (a)
Hastelloy N 500%; (b) Monel, 500 (¢) Inconel 600, 500X; (d) Inconel 606, 100X: () Inconel 601, 100 (f) Inconel 690, 100x: (g) Incoloy
S11E, 100

74

Table 10.5 Weight changes of alloys exposed to LiF-BeFz-ThFi-UF4(68-20-11.7-0.3 mole %) at 690° C for 2776 hr

Total active

Content (%) alloving
constituents’ Weight change
Alloy Chromium Iron (%) (mg/cm?)
Monel 0 1 2 +0.78
Hastelloy N 7 5 14 —0.15
Inconel 600 15 7 25 —0.49
Inconcl 606 20 2 27 —6.6
Inconel 601 23 14 38 —16.0
Inconcl 690 30 15 45 —55.5
Incoloy 811E 21 47 70 —116.7

“ Constituents that tend to form rather stable fluorides. All corrosion specimens were annealed 1 hrat 1121°C except Monel, which

was anncaled 1 hr at 800°C.

subsurface void formation. The weight gain was
probably the result of activity gradient mass trans-
fcr from constituents of the other alloys which gave
risc to the surface layer. The amount of subsurface
void formation indicates that Monel would probably
show a greater weight loss than that exhibited by
Hastclloy N and Inconel 600 if it had been tested in
a system where deposition from other constituents
did not occur,

10.5.2 Coolant Salt Thermal Convection
Loop Results

Four Hastelloy N thermal convection loops con-
taining coolant salt were in operation when the pro-
gram was terminated in January 1973 (see Table
10.6). Loop Nos. NCL-13A and NCL-14 had been
operated for over 4 and 5 years, respectively, and
their purpose was to evaluate the compatibility of
Hastelloy N and Ti-modified Hastelloy with
NaBFi-NaF (92-8 mole 7). The maximum uniform
corrosion rate at the highest temperature, 607°C,
was determined to be 0.7 mil/year.'™"

Loop NCL-17 was operated to determine the ef-
fect of steam in leakage in a fluoroborate salt-
Hastelloy N circuit. After 1000 hr of normal opera-
tion, steam was injected into the loop."? The loop
then operated over 30,000 hr following that injec-
tion. Measurements made after 24,000 hr (which in-
cluded 1000 hr of normal operation) showed a maxi-
mum corrosion rate of 1.6 mils/year.

Loop NCL-20C was originally operated as
NCL-20 with standard Hastelloy N specimens to
cvaluate their compatibility with fluoroborate cool-
ant salt under conditions expected for the MSBR
sccondary circuit (687°C maximum, 438°C mini-

mum)."” Since experience indicated that corrosion
in fluoroborate systems is largely due to impurities
(particularly water) in the system, subsequent ex-
pcriments were conducted in this loop to simulate
the cffects of water in the coolant salt, After the
original salt was drained from NCL-20, a fresh
batch was added and the loop was redesignated
NCL-20A. KBF3OH was added to the salt to simu-
late the reactions that occur when water interacts
with the salt, namely,

H-.O + NaBFs; — NaBF;OH + HF
NaBF;OH = NaBF:0O + HF
6HF + 6NaF + 2Cr — 2Nas;CrF¢ + 3H» .

However, leaks which developed in the system
made the salt highly oxidizing, thereby increasing
the obscrved corrosion rate. Because of these opera-
tional difficulties, weight change data were difficult
to interpret. A tentative conclusion was that the ad-
dition of KBF;OH increased the corrosion rate.
New salt was added to the loop (NCL-20B and
NCL-20C), but additional difficulties with cover
gas leaks were encountered and no further data
were obtained.

10. J. W. Koger, M SR Semiannu. Progr. Rep. Aug. 31, 1972,
ORNL-4832, p. 127.

11. J. W. Koger, Corrosion and Mass Transfer Characteris-
tics of NaBFs-NalF (92-8 mole Th) in Hasielloy N, ORNL-TM-
3866, pp. 11-20 (October 1972).

12. J. W. Koger, MSR Program Semiannu. Progr. Rep. Aug.
31, 1970, ORNL-4622, p. 170.

13. J. W. Koger, Mass Transfer Between Hastelloy N and a
Molten Sodium Fluoroborate Mixture in a Thermal Convection
Loop, ORNL-TM-4271 (December 1972).
75

Table 10.6. Coolant salt thermal convection loop tests in operation January 1973

Maximum Operating
Salt composition temperature time
Loop number Material Specimen (mole %) °Q) (hr)
NCL-13A Hastelloy N Hastelloy N, NaBF:-NaF 607 37,338
Ti-modified (92-8) plus tritium
Hastelioy N addition
NCL-14 Hastelloy N Ti-modified NaBFs-NaF 607 46,016
Hastelloy N (92-8)
NCL-17 Hastelloy N Hastelloy N, NaBFs:-NaF 607 31,680
Ti-modified (92-8) plus steam
Hastelloy N additions
NCL-20C Hastelloy N Hastelloy N, NaBFs-NaF 687 2,594
Ti-modified (92-8) plus
Hastelloy N KBF;OH

All of the coolant salt loops were terminated in
January 1973, and they have been stored during the
interim period. None of these loops will be re-
startcd, but they will be sectioned for metallo-
graphic and chemical analyses.

10.5.3 Blanket Salt Thermal Convection
Loop Results

Before the 1968 advances in fuel processing tech-
niques, the design of a molten-salt breeder reactor
called for a two fluid system. In addition to the
uranium-containing fuel salt, there was a thorium-
containing blanket salt. To determine the compati-
bility of Hastelloy N with this blanket salt, thermal
convection loop 15A was operated. Loop 15A was
madc of Hastelloy N, it circulated LiF-BeF,-ThF,
(73-2-25 mole ¢p) salt, and contained Hastelloy N
and modified Hastelloy N specimens. After a total
speeimen exposure time of 37,792 hr, the corrosion
ratc in terms of uniform material removal was 0.05
mil/ycar. Because the current design of an MSBR 15
a one-fluid system, no further work with blanket
salt is anticipated.

10.5.4 Status of Thermal Convection Program
with Fuel and Coolant Salts

Facilities for the operation of ten thermal convec-
tion loops are currently being constructed. Startup
of the first loop is scheduled for October 1974, with
startup of additional loops scheduled for about one
per month. Table 10.7 provides a listing of the
loops to be operated along with a description of the
purposc of each loop.

10.5.5 Forced Circulation Loop Results

Forced circulation loops FCL-1A and FCL-2
wecre operated to determine the compatibility of
Hastelloy N with sodium fluoroborate under high-
vclocity salt flow.

Loop FCL-1A was shut down (November 10,
1972) after 8079 hr of operation at a maximum tem-
pcrature of 620°C. Assuming uniform dissolution,
during the last 1977 hr of operation the maximum
corrosion rate was 0.33 mil/year, and for the entire
8079-hr pcriod the maximum corrosion rate was
about 1.7 mils/year (5.75 fps velocity). g

Forced circulation loop FCL-2 was operated at a
maximum temperature of 620°C for 6806 hr before
being shut down (October 23, 1972). Assuming
uniform dissolution, the maximum corrosion rate
was 0.74 mil/year at a velocity of 10.9 fps and 0.94
mil/ycar at a velocity of 20.8 fps.

Forced circulation loop FCL-2B was operated
previously under the designation FCL-2 for over
6800 hr to evaluate compatibility of standard
Hastclloy N with sodium fluoroborate. As FCL-2B,
it will circulate fuel salt'* to determine (1) the cor-
rosion ratc of standard Hastelloy N as a function of
U*/ U of the salt and (2) the mass transfer of tel-
lurium and its effect on standard Hastelloy N as a
function of U*"/U” ratio, and Cr’" concentration
and other corrosion products in the salt."”

Previously obtained forced circulation loop re-
sults with Hastelloy N have shown that initial cor-

4. Sce Part I, Section 2.5.2 of this report on operation of
FCL-2B by W. R. Huntley, Reactor Division.

I5. Sec Part 2, Section 8.3 of this report on voltammetric
studics of FCL-2B by A. S. Meyer, Analytical Chemistry
Division, '
76

Table 10.7. Thermal convection loop tests

Loop number

Matenal

Samples

Electrochemical
probe

Purpose

2tA

23

18B

24

25

26

27

28

29

30

Hastelloy N

Inconel 601

Hastelloy N

316 SS

Hastelloy N

Hastelloy N

2% Ti-modified
Hastelloy N

31688

Hastelloy N

Incoloy 800

2% Ti modified
Hastelloy N

MSBR fuel salt

Hastelloy N

Inconel 601

2% Ti modified Hastelloy N

Yes

2% Ti'La modified Hastelloy N

316 8S

2% Ti'Ce modified Hastelloy N Yes

Hastelloy N
Modified Hastelloy N alloys

29% Ti modified Hastelloy N
Other modiflied Hastelloy N

316 S8, other Fe-based
alloys

Yes

Yes

MSBR coolant salt

Hastelloy N
Medified Hastelloy N

Incoloy 800

2% Ti modified Hastelloy N
Other modified Hastelloy N
alloys

Yes

Yes

(1) Analytical method development
(2) Effect of graphite on stability of
uranium fluorides in molten salts

(1) Determine if high Cralloy such as
Inconel 601 can be used in fuel
under MSBR conditions; vary tem-
perature and redox potential of
salt

(2) Te mass transfer studies

(1) Screening test loop for modified
alloys

(1) Determine if Fe-based alloy can be
used in fuel under MSBR conditions;
vary temperature and redox poten-
tial of salt

(1) Screening test loop for modified
alloys

(1)Te mass transfer studies
(2) Effect of Te on mechanical proper-
ties of samples

(1) Baseline corrosiondata on modified
alloys

(2) Te mass transfer studies

(3) Effect of Te on mechanical properties
of samples

(1) Replace loop 22A

(2) Te mass transfer studies

(3) Effect of Te on mechanical properties
of samples

(4) Redox potential studies

(1) Analytical method development
(2) Baseline corrosion data
(3) Tritium studies

(1) Baseline corrosion studies of high Cr
alloy that has potential foruse ina
steam circuit

(1) Baseline corrosion studies

(2) Tritium studies

(3) Studies to determine if borides form
in alloy and effect on properties of
alloy

rosion rates are high because, in addition to oxida-
tion of chromium by UF4 in the salt, there are reac-
tions of iron and chromium with impurities in the
salt. As the mimpurity concentrations decrease, the
overall corrosion rate has been found to decrease
until a steady-state condition occurs during which
thc surface concentration of chromium is deter-
mincd by the UM/UJr3 ratio of the salt, and the rate
of mass transfer under temperature gradient condi-
tions 1s controlled by the rate of chromium diffusion
to thc surface from which 1t is being removed.

The Hastelloy N specimens from forced con-
veetion loop FCL-2B were weighed and examined
after time periods of 418 and 2748 hr of near-
isothermal exposure to LiF-BeF>-ThF:-UF: (68-
20-11.7-0.3 mole %) salt. The average weight loss
(0.38 mg) during the initial 418-hr test period was
significantly higher than the average loss (0.06 mg)
during thc ensuing 2330 hr. The decrease in weight
loss with time is due in part to decreasing impurity
concentrations in the salt, but beryllium additions
were also made to the salt. In terms of a uniform
corrosion rate, for the first period of 418 hr the av-
crage loss was 0.09 mil/year and the maximum loss
was 0.14 mil/year. For the second period of 2330 hr,
the average value for the uniform corrosion rate was
<0.01 mil/ycar and the maximum rate was 0.01
mil/ycar,

During the forthcoming isothermal-operation
pcriod, NiF: additions to the salt will be made.
These additions should increase the corrosion rate
initially.

10.5.6 Surveillance Specimens
from the Coolant-Salt Technology Facility

The Coolant-Salt Technology Facility (CSTF)
provides an opportunity for investigations in a
pumped, isothermal coolant salt loop. Three
metallurgical specimens, two standard and one
modified Hastelloy N, were exposed to NaBFs«-NaF
(92-8 mole %) at 500 to 510°C and a flow rate of 520
gpm (~12 fps). During the first 600 hr, one of the
standard Hastelloy N specimens lost 0.3 mg/cm’
while the modified Hastelloy N specimen lost 0.5
mg/cm®’. The third specimen came off the specimen
holder and was lost. After an additional 460-hr
exposure to the salt, two standard and one modified
Hastelloy N specimens all gained weight. These
weight gains suggested that the oxidation potential of
the salt had decreased and that there were surfaces in
the loop system which were somewhat hotter than the
surveillance specimens.

77

10.6 CORROSION OF HASTELLOY N
AND OTHER ALLOYS IN STEAM

B. McNabb H. E. McCoy

A program has been in progress to investigate the
corrosion in steam of a number of iron-, nickel-,
and cobalt-based alloys. Unstressed specimens of
these matenials in the form of 2-in.-long X 1/2-in.-
wide X 0.035-in.-thick coupons were exposed to
flowing steam (approximately 1000 1b/hr) at 538°C
for up to 15,000 hr at TVA’s Bull Run Steam Plant.
A report'® detailing the results of this work has
been i1ssued and the results will only be summa-
rized in this report.

Figure 10.16 compares the corrosion behavior
(wcight change) of some of the alloys. To show the
approximate magnitude of corrosion, plots of calcu-
lated uniform metal reaction rates of 0.1 and 0.25
mil/year are shown in the figure. The range of
weight changes for the alloys studied was from 0.01
mg/em’ for Inconel 718 at 4000 hr to 12 mg/cm” for
the Croloys at 14,000 hr. Hastelloy N continues to
show good compatability with steam at 538°C for
cxposure times up to 15,000 hr. For exposures be-
tween 4000 and 15,000 hr, the weight change can
be described by an equation of the form

AW = K™ |

where AW is the weight change in mg/cm?, 7 is the
time in hours, and K is a constant. Incoloy 800 is
widely used in steam systems, and although its
weight change in 1000 hr was higher than for other
alloys, its corrosion rate for times greater than 2000°
hr was low (in proportion to K:*%),

Low alloy ferritic steels such as Croloy 2 1/4 Cr-1
Mo steel are conventional steam plant construction
matcrials, and several alloys of this type containing
from 1 to 9% Cr were included in the test program.
These materials have the highest corrosion rates of
the alloys studied. Inconel 600 and type 347 stain-
less steel both gained weight more rapidly than
Hastclloy N, and their weight changes were propor-
tional to Kr*°. Hastelloy X and Haynes alloy No.
188 had low corrosion rates, and the rate for Inconel
718 was so low that it was difficult to measure.

Structural materials are often placed in service
in stcam systems with some residual cold work

16. H. L. McCoy and B. McNabb, Corrosion of Several Iron
and Nickel-Base Alloys in Supercritical Steam ar 1000° F,
ORNL-TM-4552 (August 1974).

ORNL-DWG 73-4138

20
10
N -2k %
2 925 mpy
Slli4dEE
— Py
o »
‘E’ ! - I// 0.10m
\UI A o X rad py
E = 1 Z
- // 4 /’
8 os )L > b
2 -, o HASTELLOY N | | ||
5 | LT
[ [ ]
T ) 7 /|
2 02 //,/’ 4 AW =Kt
4 Y
= -// ’ AW =K10:5
/s / 1 o
0.4 ‘/ Vd AW=KitC
. ,[ —
L
‘v
3 v |
0.05 | 7 o INCOLOY 800 1]
|V & HASTELLOY X
o H-188 il
» INCONEL 600
0.02 4 a 347 STAINLESS STEEL ]
« INCONEL 718
Lot E
0.0t
103 2 5 104 2 5 10°
TIME (hr)

Fig. 10.16. Corrosion of several alloys in steam at 538°C and
3500 psi.

present. Since cold working introduces a variable
that can influence the corrosion characteristics,
samples of many of the alloys were exposed in the

509, cold-worked as well as the annealed condi--

tions. Cold working (50%) had a large effect on the
corrosion of type 201 stainless steel (16.55% Cr,
5.23% Ni, 7.28% Mn, 0.076% C, 0.54% Si, 0.034%
P, and 0.059% N). The corrosion rate of annealed
material was much higher than cold-worked mate-
rial, and the difference became progressively greater
as the exposure time increased. The rates for an-
ncaled 201 and 347 stainless steels were approxi-
mately equivalent. Figure 10.17 compares the rates
for annealed and 509, cold-worked 201 stainless
steel and Hastelloy N. Interesting to note is that for
times up to 15,000 hr, the cold work had no effect
on the corrosion rate of Hastelloy N.

78

Various alloy additions have been made to modi-
ficd Hastclloy N for radiation damage resistance,
and some of these alloys are compared to standard
Hastclloy N in Fig. 10.18. Maximum and minimum
weight changes for these alloys differed by less than
a factor of 2, but only the alloy containing 1.03%
Nb gained less weight than the standard Hastelloy
N.

The sample holder for stressed and unstressed
spccimens has been reassembled and delivered to
Bull Run Steam Plant for reinsertion into the
stcam corrosion facility. Several new specimens
were installed in the sample holder. Some un-
stressed specimens removed at the conclusion of the
program had accumulated 15,000 hr, and most of
these specimens were reinstalled. Several alloys of
the Croloy type (Fe 2 1/4 Cr-1 Mo) containing from
[ to 9% Ce in the annealed and cold-worked condi-
tions wcre included in this loading.

Stressed specimens currently in the facility in-
clude some that were previously exposed and some
replacements for ruptured specimens that were re-
moved. The stresses range from 28,000 psi to 76,600

psi.

ORNL-DWG 73-4134

[:Y "
5 .
)/AW= Kt0-77
/l
1=
vl
Fa
PV
N’; 0.5 ‘/U
] >
2 T
= /r
W 0.2 |3 AW=k10-30
S
[e]
T é
= o © 2477 (ANNEALED)
© e 2477 (COLD WORKED) ]
S & 201 (ANNEALED} il
0.05 I a 201 (COLD WORKED)
& 4 N N
S
&
0.02
0.01
103 2 5 ok 2 5 109
TIME (hr}

Fig. 10.17. Effect of cold work on the corrosion of Hastelloy N
(Heat 2477) and type 201 stainless steel in steam at 538° Cand 35'00

psi.
ORNL-DWG 73-4136

v au’ 8 A
0.5 9 4
) 8 A.?.D ]
Y Y
o~ ® X —_‘___-—"" v
= A A Y
e 7 v
on
£ l . /Q v
0.2 e STANDARD
w . 9/
[&]
<
<1 v
I
(&)
T 04
o © 185 {0.91 Ti, 0.98 Zr)
S A 186 (0,88 Ti, 0.84 Al)
0 188 (0.95 Al, 1.4 Hf)
005 @ 231 (1.3 Hf, 1.2 Y)
’ A 232 (1,2 Hf}
® 234 (0.75 Ce)
v 236 (1.0 Al, 0,47 Zr)
v 237 (1.03 Cb)
002 | ¢t [Tl
10° 2 5 10? 2

TIME (hr)

Fig. 10.18. Corrosion of various modified compositions of
Hastelloy N in steam at 538°C and 3500 psi.

10.7 INTERGRANULAR CRACKING
OF ALLOYS EXPOSED TO TELLURIUM

H. E. McCoy

- The status of our test program was given in sev-
eral references' ™" prior to discontinuation of the
program carly in 1973, At the time these status re-
ports wcre written, numerous experiments were in
progress. Many samples from these experiments are
of vital importance, but the first order of priorities
has been to revitalize the program with new lab
facilitics and new personnel.

The sercening tests that we ran previously
showed that copper, Monel, and iron-based alloys
completcly resisted intergranular embrittlement by
tellurium. The experiments with nickel-based alloys
indicated  that the propensity for intergranular
crack formation decreased with increasing concen-
trations of Cr, Ti, Nb, and Ce. However, the screen-
ing experiments were not sufficient to ascertain the
coneentrations of these elements required to prevent
cracking under typical MSBR operating conditions.

79

We had already begun work toward the develop-
ment of a more realistic screening test late in 1972,
and this cffort has been revived. In the tests leading
to the trends described above, the samples were
scaled in a quartz vial with a small amount of tel-
lurium. They were heated slowly to 650 and 700° C,
and the specimens and the tellurium were allowed
to react for times from 200 to 10,000 hr. Initially,
the specimens were exposed to a relatively high flux
of tcllurium compared with that which would be
obscrved in an MSBR where tellurium would be
produced at a constant (after an initial saturation
pcriod), slow rate. To a large extent, the flux of
teflurium controls the tellurium activity which in
turn dctermines the nature of the tellurium-metal
rcaction. For example, high tellurium activities
would cause nickel tellurides to be formed, whereas
only chromium tellurides would be stable at lower
tellurium activities. The tellurium activities re-
quired to form the various tellurides are not known
accurately, so experiments with total assurance of
rclating to MSBR conditions are difficult to design.

Although the information needed to develop a
ncw screening test was not available, we concluded
that the tellurium flux must be of the order of 10"
atoms/cm’/sec. Also, the flux must be known and
held rcasonably constant during the experiment.
Since the screening test must maintain the features
of being relatively quick and cheap, several ideas
arc currently being pursued for screening tests.

10.8 CHEMISTRY OF
SALT-METAL-TELLURIUM SYSTEM

J. Brynestad

The chemistry of tellurium in the MSBR system
is complex and largely unknown. An important
fcature i1s that the system as a whole is never in
chemical equilibrium with regard to the tellurium
that is continually produced in the fission process.
Once tellurtum is formed, it reacts with the melt
cnvironment and assumes the valence states dic-

17. H. E. McCoy, “Intergranular Cracking of Structural Ma-
terials Exposed to Fuel Salt,” MSR Program Semiannu. Progr.
Rep. Aug. 31, 1972, ORNL-4832, p. 63 (March 1973).

18. M. W. Rosenthal et al., The Development Status of
Molten-Salt Breeder Reactors, ORNL-4812, p. 195 (August 1972),

19. H. E. McCoy and B. McNabb, fntergranular Cracking of
INOR-8 in the MSRE, ORNL-4829, p. 165 (November 1972).
tated by the (variable) redox potential in the melt,
by thce chemical nature of the melt, and by the ki-
nctics of the processes. But since the melt is not in
thermodynamic equilibrium with the construction
metals, the tellurium species formed in the melt will
in turn react more or less aggressively with the met-
als, depending upon the redox potential of the melt
at any given time, and upon the nature of the con-
struction metal.

Onc objective is to obtain and maintain a de-
fined redox condition of the melt at which the cor-
rosivity of the formed tellurium species is at a mini-
mum (idcally zero). Another alternate or simultane-
ous objective would be to obtain construction
mctals that are resistant to tellurium attack over a
wide redox potential range.

Towards these objectives, it is important to ob-
tain information about the products of reactions be-
tween tellurium and the construction metals (in-
cluding potential new alloy constituents), namely
their thermodynamic stabilities, crystal structures,
and their rates of formation as a function of the re-
dox potential of the melt. To minimize the tellur-
ium corrosivity, it is important to learn about the
tellurium/salt interactions; namely, the nature and
stabilities of the formed tellurium entities in the re-
dox potential range of interest, as compared to the
stabilities of the construction metal tellurides.

Our present work is focused on the interaction
between Hastelloy N and tellurium at moderate and
low tellurium activities. The phase relationships
and phase stabilities of he metal-rich end of the
Ni-Mo-Cr-Te system are being investigated. Also,
the corrosion of Hastelloy N by tellurium at con-
trolled tellurium activities 1s under study.

The nickel-tellurium phase diagram is largely
known,20 and the most nickel-rich compound in
this system is NisTezx. There 1s no information on
the solid solubility of tellurium in nickel metal, ex-
cept that it is low. The chromium-tellurium phase
diagram is not well known. However, the most
chromium-rich compound seems to be CrsTes (ref.
21), and the most tellurium-rich compound is
Cr;Tes. The molybdenum-tellurium phase diagram
is not well known. The most metal-rich compound
scems to be MoTei.1a (ref. 22). Work by Weaver
and Redman®’ and our own observations indicate
that chromium forms by far the most stable binary
tellurium compound in the metal-rich end of the
Ni-Mo-Cr-Te system. We are currently investigat-
ing the metal-rich end of the Ni-Cr-Te system
where preliminary results indicate that a ternary com-

80

pound exists. Work is in progress to identify this
compound and to assess its stability relative to the
binary chromium tellurides.

It is important to know which tellurium com-
pounds are formed with Hastelloy N as a function
of tellurium activity. We have verified that tellur-
ium attacks Hastelloy N even at tellurium activities
so low that NiiTe; cannot be formed. At higher tel-
lurium activities, NisTe: 1s also formed. However,
whether the formation of chromium telluride(s)
(and/or ternary compounds) causes tellurium em-
brittlement in Hastelloy N is not yet known. We
know that NisTe; causes embrittlement in pure
nickel and probably in Hastelloy N. If the formation
of chromium tellurides does not cause embrittle-
ment, whereas the formation of NisTe: does, a pos-
sible solution would be to keep the redox potential
of the melt low enough to prevent the formation of
NisTez. Higher chromium contents in the alloy
would case the requirements on the redox poten-
tial. Therefore, 1t 1s important to ascertain whether
chromium tellurides cause embrittlement in Hastel-
loy N, and work is in progress to resolve this
qucstion.

10.9 AUGER OBSERVATIONS RELATIVE
TO INTERGRANULAR CRACKING

R. E. Clausing L. Heatherly

Auger clectron spectroscopy and other newly de-
veloped techniques for surface analysis provide
powerful and versatile methods for studying the
role of tellurium or other fission products in pro-
ducing intergranular fracture or cracking. Not only
can wc determine the elemental composition of
suitably prepared fracture surfaces and the distribu-
tion of elements in the immediate vicinity of the
fracture surface, but we can also examine in a non-
destructive way the changes in surface composition
as a function of time and temperature while thermal
trcatments are in progress. This last capability pro-
vides a way of rapidly determining the tendency for
surface segregation and mobility of elements on or

20. K. O. Klepp and K. L. Komarck, Monats. fur Chemie
103, 934-46 (1972).

21. A. F. Andresen, Acta. Chem. Scand. 17, 133542 (1963).

22. M. Spiesser and J. Rouxel, C. R. Acad. Sci. (Paris) Ser.
C. 265(2), 9295 (1967).

23. C. F. Weaver and J. D. Redman, MSR Program Semi-
annu. Progr. Rep. Aug. 31, 1972, ORNL-4832 (March 1973).
near solid surfaces. By using this kind of informa-
tion we may be able to deduce useful information
rclative to the behavior of surfaces and grain
boundaries of Hastelloy N in the molten: salt reac-
tor application.

Our cfforts during the past few months have been
dirccted toward (1) a review of old reports and pre-
vious work, (2) the development of a new program
coordinated with other new programs, (3) the ac-
quisition of existing samples (from previous stud-
ics), (4) the planning for and preparation of new
samples, and (5) the development of techniques for
the ncw program. The new program will make full-
cst possible use of improved surface research tech-
niques; thus, this program will require the acquisi-
tion, adaptation, and development of both equip-
mcnt and techniques. We are upgrading the resolu-
tion of onc of our Auger systems to provide analysis
of arcas as small as 2.5 X 107 ¢m® so that we can
analyze fracture surfaces in great detail. We have
improved our ability to examine samples at high
temperatures and can now examine samples at tem-
pcratures up to 800°C under ultrahigh vacuum
conditions.

10.9.1 Fracture Surface Analysis

The methods used for examination of fracture
surfaces and obtaining information on the ele-
mental composition gradients near the fracture sur-
facc have been described previously;”* these meth-
ods remain unchanged except that we will soon be
able to use a much smaller diameter beam of about
5 u diam FWHM (full width at half~-maximum in-
tensity). Sample preparation was also described
previously and is essentially unchanged.

Progress during this report period consists pri-
marily of making equipment modifications and ac-
quisitions to provide the small diameter electron
beam. Commercially available parts are being used
wherever possible. We expect the initial system
testing to be finished by January 1, 1975. Samples
of Hastclloy N and other alloys that have been ex-
poscd to tellurium vapor at 700° C for 500 hr will be
available, and they will be examined if the equip-
ment and techniques give good results.

10.9.2 High Temperature Surface Studies

Apparatus has been designed, constructed, and
tested to permit heating of Hastelloy N samples to

81

800° C so that Auger analyses may be made contin-
uously while thermal treatments are in progress.
The equipment is operational and several prelimi-
nary runs have been made in which samples of
modificd Hastelloy N alloys containing tellurium
were heated stepwise to 800°C. The results are en-
couraging; strong segregation of carbon and sulfur
to the surfaces occur on Hastelloy N; sulfur, car-
bon, and tellurium segregate to the surface of a
spccial Hastelloy N alloy that does not contain
chromium. These results are preliminary and are
intended only to indicate the extent of our progress.
Carcfully controlled experiments now in progress
will bc required before we attempt to present
quantitative results or draw conclusions.

10.10 IN-REACTOR FUELED EXPERIMENTS
C. R. Hyman

The MSR in-reactor irradiation experiment
TeGen-1 has been resumed after a period of almost
two years. Previously, a capsule had been designed
for the study of the effect of fission products on
prospective MSR containment materials.” The de-
velopment and construction of the experiment had
reached the point that only a small amount of
work had to be performed before insertion of the
capsulc into the Oak Ridge Research Reactor
(ORR). However, the experiment was not done be-
causc of tcrmination of the MSR Program. The
rcactor was dismantled to remove the fuel. Since
revival of the irradiation program, the work has
been in the reorganization and fabrication of
TcGen-1, which is the first in a series of related
tests.

TeGen-1 contains three fuel pins, each filled with
fucl of approximately the MSR experiment compo-
sition. The three materials under test are type 304
stainless steel, standard Hastelloy N, and Inconel
601. The experiment is designed to produce fission
product concentrations arid to operate at tempera-
ture ranges that were found to produce cracking in
standard Hastelloy N in the MSR experiment. Be-

24. H. E. McCoy, “Intergranular Cracking of Structural Ma-
terials Exposed to Fuel Salt,” MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 87 (March 1973).

25. R. L. Senn, “Design of an In-Reactor Experiment to
Study Fission Product Effects on Metals,” MSR Program Senii-
annu. Progr. Rep. Aug. 31, 1972, ORNL-4832 (March 1973).
havior of the different materials will be evaluated by
straining specimens of each material after ir-
radiation.

10.10.1 Design

Each of the three test specimens or fuel pins con-
sists of a 1/2-in.-OD X 0.43-in.-ID tube of the
particular test material with welded end caps to
provide a 3.5-in.-long internal cavity. Each pin 1s
filled with 7.14 cm® (3-in. depth) of fuel salt leaving
a 1/2-in. helium-filled gas plenum at the top. The
wetted surface area in each specimen is 27.1 cm’.
The fuel composition is the same for each pin
[LiF-BeF;-ZrFs-UF: (63.1-29.3-5.1-2.5 mole %)].
Exccpt for a slightly higher UF4 loading, this {uel 1s
practically the same as that used in the MSRE. The
uranium is U rather than **’U since **U pro-
duces more tellurium (the suspected cause of crack-
ing of Hastelloy N in the MSRE). The design
power, 344 W/cm’ of salt (1.2 kW/ft of fueled
length), is expected to produce 5 X 10'® atoms of
stablc tellurium per square centimeter of wetted
surface in the planned 1100-hr irradiation. The
spccimen (fuel pin wall) operating temperature is
700° C.

The capsule experiment itself is typical of ORR
poolside experiments. It consists of a double-walled
containment vessel connected to a lead tube that
carries instrumented gas and heater leads to the
capsule. The three specimens are suspended within
the inner containment and are submerged in Nak,
which serves as a heat transfer medium. Each speci-
men is instrumented with four thermocouples along
its length (Fig. 10.19). Thermocouples in the bot-
tom of the capsule will sense abnormally high tem-
peratures if salt should leak from a pin and react
with NaK to precipitate uranium. The annulus be-
tween the primary and secondary containment ves-
sels is argon-filled and is sized to provide the ther-
mal resistance required to achieve the design speci-
men temperature of 700°C during operation.

Each specimen is provided with a 1/16-in.-OD
stainless steel-sheathed heater coiled around the
ends (Fig. 10.19). The heater serves the dual pur-
posc of minimizing axial temperature variations at
the ends of the specimens by providing heat where
there 1s no fuel, and of maintaining the salt above
150°C at all times when the reactor is down. The
latter condition is specified to prevent the radiolytic
dissociation of the fuel and the release of fluorine
from irradiation of fuel salt at low temperatures.

82

ORNL-DWG 72~ 9654 A

3 HEATER LEADS

N 34 GAS LINES
—3 5 THERMOCOUPLES

BULLKHEAD
ARGON

HELIUM —_ Pn : ] 304 STAINLESS STEEL
'\ o PRIMARY CONTAINMENT
NaK— Y | ===l ] __-304 STAINLESS STEEL
') i SECONDARY CONTAINMENT
TE- 104 Ll [] — FUEL PIN NO.1: 304
- STAINLESS STEEL
TE-102 — U@l ——MSR FUEL SALT
- g
- H
TE-103 el
I ‘E/ HEATER COIL NO. 1
TE-10C4 ()
' —] H
E—105 4 o oy o ,
TE-1 : 4§ i FUEL PIN NO.
Bl i 4 HASTELLOY-N
q — - =
TE-106 Rl [l — MSR FUEL SALT
2 A AT
TE-107 Wi’
L' j/— HEATER COIL NO.2
TE-408 w
7 8 FUEL PIN NO. 3:
TE-109 | 2 INCONEL 601
3 ==
TE-110 RN g h— MSR FUEL SALT
5 i—==*T1
TE-111 e 4
f U _— HEATER COIL NO.3
TE-112 T~ U &
TE-1414

Fig. 10.19. Schematic drawing of TeGen-1 capsule.

The capsule rides in an adjustable track in the
ORR poolside facility. Therefore, the capsule posi-
tion, relative to the reactor, and hence the fission
power and neutron flux, is adjustable.

10.10.2 Calculated Operation and Test Conditions

The capsule is designed to operate at a linear
power density of 1.2 kW/ft in the fueled regions of
the specimens and with a fuel-to-metal interface
temperature of 700°C. Figure 10.20 presents the
calculated one dimensional radial temperature pro-
file through the capsule and the middle test ele-
ment. The upper curve is for the case of 20% over-
power excursion, a most improbable occurrence.
The rather steep radial gradient from 700 to 962°C
83

ORNL-DWG 72-9655A

¢ ( &
CAPSULE S Y\@Q;,O’
&Q/ v_\63\
o/ D /K
AORS)
&V
RN\
v @« \$
QY
N4
o
4;
TEMPERATURE (°C)
1073
Qg2
DESIGN POINT —==700
190
85
RADII (in.)
0.215
0.2155 —
0.250 e —
0.388 — —
]
0.428 B —
0.453 — -
0.498 -

Fig. 10.20. TeGen-1 schematic temperature profile from GENGTC calculations. Temperatures are in °C.

UPPER LINE REPRESENTS TEMPERATURE PROFILE WITH
CAPSULE OPERATING AT 120 % NOMINAL OPERATING POWER.
ASSUMES 1.43 kW/ft FISSION POWER AND 0.2 W/g GAMMA HEAT.

LOWER LINE REPRESENTS TEMPERATURE PROFILE WITH CAP-
SULE OPERATING AT NOMINAL DESIGN POWER. ASSUMES
.19 kW/ft FISSION POWER AND 0.2 W/g GAMMA HEAT.
84

CRNL-DWG 73-839

<
Z g4 NOMINAL DESIGN OPERATION WITH
- = POINT OPERATION HEATER ONLY
7 © =
3 £ ¥8 1.2 kW/ft FISSION HEAT O kW/ft FISSION HEAT
“ g 1.1 kw/ft HEATER POWER 1.1 kW/ft HEATER POWER
a .
FUEL FUEL PIN FUEL FUEL PIN
CENTER LINE INNER WALL CENTER LINE INNER WALL
TEMPERATURE (°C) | TEMPERATURE (°C) | TEMPERATURE (°C) | TEMPERATURE (°C)
T - POSITION A POSITION B POSITION A POSITION B
906 (694) 386 396
FUEL PIN
966 704 328 327
Eog 970 707 286 286
312in ' =~ FUEL SALT '
972 708 267 266
974 707 269 268
L ‘ 968 706 291 290
—’—r . ' 700 340 336
Yy in. g | 259 14 119 130
TYPICAL
- [ ] MAXIMUM TEMPERATURE
MAXIMUM .
AXIAL ") MINIMUM TEMPERATURE
AT(°C)

Fig. 10.21. TeGen-1 capsule two-dimensional temperature profiles at various operating conditions.

through the salt assumes no convection currents
within the salt. Such steep gradients are expected
to cause some convection currents and a reduction
in this temperature difference. Fission power gener-
ation in thc upper and lower specimens will be less,
duc to the axial variation of the ORR neutron flux.
The upper specimen will operate at about 149 less
powcr and the lower specimen will operate at a
power between the other two. The heaters will be
adjusted to compensate for this variation and to
maintain the required temperatures.

Figure 10.21 indicates the location of the heater
coils and gives the temperature distribution calcu-
lated with the 3-dimensional heat conduction code
HEATING 3. (The 1.2 kW/f{t 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 distribu-
tion for normal operation indicates that the spread
in temperature 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 temperatures along the pin wall. The cal-
culated temperatures for the case of zero fission
heat also show that the design power of the heaters
is ample to keep the salt temperature well above
150° C, the threshold for fluorine evolution.

10.10.3 Preliminary Operating Data

Although the capsule irradiation did not begin
until September 1974 (past the August 31, 1974
dcadline for this report), Table 10.8 presents a typi-
cal set of operating data taken during the first week
of .irradiatipn. (See Fig. 10.19 for thermocouple lo-
cations.) Figure 10.20 shows that there is about a
7°C difference between the salt-to-specimen wall
interface and what would be the junction of a 1/16-
in.-diam thermocouple strapped to the outside of
the specimen can. Thus, temperature presented in
Table 10.8 should be increased by about this
amount to represent the salt-wall interface operat-
ing condition. There is a definite pattern in the
temperatures for each fuel pin. In the upper and
lower pins the uppermost thermocouples (adjacent
to the gas void) are hotter than those at the bottom
which tend to be at the lowest temperature. In fu-
turc capsules, the heaters will be redistributed to
achieve more of a uniform temperature distribution

85°

over the length of the specimen. The middle speci-
men shows a different temperature profile with the
temperature at the top of the fuel being the hottest.
This variation is as would be expected since the
middle pin generates the maximum fission power.
The heater power to each pin (Table 10.8) 1s also

~indicative of the difference in fission heat.

Table 10.8. Typical operating conditions for irradiation capsule TeGen-1
Capsule position 7.8 in. retracted from reactor face

Thermocouple Temperature Heater power
Specimen number? °C) (watts)
Top (304 101 705
stainless steel) 102 692
103 678 346
104 688
Middlc (Hastelloy N) 105 694
106 713
107 693 92
108 681
Bottom (Inconel 601) 109 737
) 110 705
111 684 186
112 651

“Sce Fig. 10.19 for thermocouple location,
86

1. Fuel Processing Materials Development

J. R. DiStefano

Materials requirements for molten-salt fuel pro-
cesses include containment of the following en-
vironments:

. molten fluorides plus HF-H: mixtures at 550 to
650° C (hydrofluorination process),

2. molten fluorides plus F>-UFs gaseous mix-
turcs at 500 to 550°C (fuel reconstitution process),
and

3. molten fluorides or molten LiCl in the pres-
ence of bismuth containing S to 50 at. % Li at 550
to 650°C (metal transfer process).

If a frozen salt layer can be maintained on the
container wall, a nickel-base alloy can be used in
the fluorination and fuel reconstitution portion of
the system. The present objectives of the materials
program are to identify materials that will satisfac-
torily contain bismuth-lithium-thorium solutions,
and to develop the technology necessary to con-
struct a processing facility, '

Initial studies showed that iron-, nickel-, or
cobalt-base alloys would not be suitable because of
their rapid rate of mass transfer in bismuth systems
having a temperature gradient. Recent studies have
indicated that graphite, molybdenum, and tantalum
(alloys) are promising, and work is continuing to
cvaluate the compatibility of these materials with
bismuth-lithium solutions and molten salts in cap-
sule and thermal convection loop tests.

11.1 STATIC CAPSULE TESTS
OF GRAPHITE WITH BISMUTH
AND BISMUTH-LITHIUM SOLUTIONS

J. R. DiStefano O. B. Cavin J. L. Griffith
L. R. Trotter

Graphite has a very low solubility in bismuth
(<1 ppm at 600°C) and has been demonstrated to
have excellent resistance to corrosion by molten
fluoride salts. Therefore, it is an attractive candi-
date as a containment material in molten-salt chem-
ical processing applications. Of principal interest is
the compatibility of graphite with bismuth-lithium
solutions at temperatures from 550 to 700°C. Re-
sults of static capsule tests conducted previously are
summarized below and those from thermal convec-
tion loop tests are given later in the report.

H. E. McCoy

[. Bismuth-lithium solutions penetrate the open
porosity of graphite. The extent of penetration is a
function of graphite bulk density, accessible pore
volume, pore diameter, and time.

2. No evidence of chemical interaction has been
found between graphite and bismuth or bismuth-
lithium solutions (containing up to 3 wt % Li) at
tcmperatures as high as 700°C.

3. Lithium reacts with graphite at 700°C; Li,C:
was identified as a corrosion product. Vitreous car-
bon completely disintegrates when exposed to lith-
ium even at temperatures as low as 200°C.

Additional tests have been conducted to deter-
mine more precisely the nature of any interactions
bectween graphite and bismuth-lithium solutions. In
onc scries of tests, several grades of graphite speci-
mcns were exposed to bismuth solutions containing
0,0.01,0.17, and 3 wt 9% Li in three different grades
of graphite crucibles for 3000 and 10,000 hr at
650° C. From these tests, data were obtained on (1)
the extent of penetration of the solutions into the
graphite crucibles, (2) the amount of bismuth in the
graphite samples, and (3) the lattice parameters of
the graphite samples after exposure to the bis-
muth-lithium solutions.

Penetration results are summarized in Table
11.1. After 3000 hr, limited penetration of PGX and
PGXX graphite had occurred, but there was no
penetration of pyrolytic-carbon-coated capsules.
After 10,000 hr, PGX graphite was penetrated com-
pletely by bismuth and the bismuth-lithium solu-
tions. However, PGXX graphite, which is PGX
base stock that has been liquid impregnated,
showed only slight evidence of penetration. Each
capsule contained samples of PGX, PGXX, AXF-
5Q (ref. 1), and H-337 graphite.’

Results of lattice parameter measurements are
summarized (Table 11.2). The lattice parameter in
the a crystallographic direction was determined
from the (1120) diffraction maxima and was found
to be constant at 2.466 *+ 0.001 A for all samples.
The ¢ parameter was determined from the (0004)
diffraction maxima. It varied slightly with the type

[. AXF-5Q graphite: bulk density, 1.9 g/cm’; maximum pore
diameter, | to 2 u; average pore diameter, | to 2 u.

2. H-337 graphite: bulk density, 2.0 g/cm’; maximum pore
diameter, 3 u; average pore diameter, 1 to 2 pu.
87

Table 11.1. Extent of penetration of graphite by bismuth and bismuth-lithium solutions at 650°C

Bismuth Bi0.01% Li Bi—0.17% Li Bi-3% Li
Type of
graphite 3000 10,000 3000 10,000 3000 10,000 3000 10,000
capsule ' hr hr hr hr hr hr hr hr
PGX”’ Slight Complete Slight Complete Slight Complete Heavy Complete
PGXX’ Slight Slight Slight Slight Slight Slight Slight Medium
Pyrocarbon- None d None d Nane d None d

coated graphite®

“PGX graphite: bulk density, 1.8 g/em”; maximum pore diameter, 7 u; average poreidiameter,! to 2 u. .
"PGXX graphite: bulk density, 1.85 g/cm"; maximum pore diameter, 6 u: average porediameter, | to 2 u. This graphite is PGX base

stock that has been liquid impregnated. .

“Pyrocarbon-coated graphite: bulk density, 1.8-2.2 g/cm’; however, this type of graphite is generally impervious to gases because of

the way it is made.
“Not tested for 10,000 hr.

Table 11.2. Two physical characteristics of graphite samples
exposed to bismuth and bismuth-lithium solutions
; for 3000 and 10,000 hr at 650°C

Average value Diffracted
lattice parameters light
. intensity
Type of for indicated crystallographic of bismuth
graphite direction (counts/sec)
a c
PGX 2.466 6.750 260
PGXX 2.466 6.754 ‘ 70
AXF-5Q 2.466 6.781 26
H-337 2.466 6.755 34

of graphite, but was relatively constant for any
grade of graphite and was not affected by length of
cxposure or the solution to which it was exposed.
The concentration of bismuth was measured by the
intensity of the (014) diffraction peak. Values
shown in Table 11.2 have been averaged for all test
conditions. PGX graphite, which has the greatest
amount of accessible porosity, showed the highest
uptake of bismuth. Intensity data (Table 11.3) show
that there was no consistent increase in bismuth
with ‘time of exposure, but values were generally
higher in samples exposed to Bi—0.179% Li and Bi-
3% Li. Lithium was not detected in any of the sam-
ples, but the x-ray techniques used are not sensitive
to low concentrations of lithium.

The solutions from these tests were analyzed for
carbon, and these results showed higher concentra-

tions of carbon in the Bi-3% Li samples than in

solutions containing lower concentrations of lith-

ium. Based on these results, tests were started to
determine the carbon content of a series of bis-
muth-lithium solutions after exposure to graphite at
650°C. Solutions of Bi—100 ppm Li, Bi~19 Li, Bi-
2% L1, and Bi—3% Li were heated in PGX, PGXX,
and ATJ crucibles for 1000 hr at 650°C. The sam-
ples were protected from contamination during
sampling and analysis by an inert gas atmosphere.

Scvere segregation of lithium occurred when the
bismuth-lithium melt solidified; two to three sam-
ples were submitted for analyses from each test.
The values listed (Table 11.4) relate the nominal
lithium content of the melt to the average values
found for carbon. Although the individual data
points were more scattered, there 1s a definite trend
indicating increasing carbon concentrations in
melts having successively higher lithium. No differ-
cnces were noted among the three types of graphite.

Thermal convection loop results reported in the
following section indicate no significant mass trans-
port of carbon in Bi—100 ppm Li after 3000 hr of
operation at 700°C maximum temperature and a
100° C temperature difference. However, the high
carbon concentrations found in the above tests indi-
cate that mass transport could be a significant prob-
lem at higher lithium concentrations. A test is now
being constructed in which graphite will be ex-
posed to a Bi—2.59% Li solution in a molybdenum
thermal convection loop under conditions similar to
those above.

! ' 88

Table 12.3. Additional physical properties of graphite grades AXF, AXF-5QBG-3, H-395, and P-03 as functions of fluences from
0 to 42.1 X 10* neutrons/cm?® > 50 keV) accumulated at 715°C*

* * Coefficient of thermal
Specimen * * * * ‘ Bacon Anisotropy expansion, CTE AL AD™e AV Brittle- Fracture
number Crystallography, A ) Layer anisotropy factor 200 to 600°C X 107° : : ring strain® AGH AY
(orien- Fluence® : height? factor Lo Do Vo strength’™® € — —
tation)’ - . (10" neutrons/cm’) a c L. L. (um) (BAF) R“e R ¢ With grain Across grain (%) (%) (%) (10" psi) (%) G, Yo
| . : .
i Grade AXF
122 0 2.4627 6.7687 500° 2007 2.8° 1.03" 0.673¢ 0.6637 7.56 16.53 0.95
120 5.4 ) : 1.1 7.90 0 0 0 14.93(2) 0.72(2) 1.25 [.14
s 8.9 5.13 0.74 0.75 2.28 15.34 0.57 1.48 1.27
119 34.2 0.51 . 3.16 091 1.02 295 9.963 0.22 3.28 2.78
Grade AXF-5QBG-3’
174 0 2.463° 6.750" 7507 3507 ' 1.05" 0.656° 0.671¢ 6.48 ‘ 14.15 - 1.03
170 7.6 v 1.35 7.60 0.73 0.26 1.25 15.57 0.77 1.48 1.37
173 10.2 1.87 7.95 0.09 0.18 0.44 13.48 0.59 1.69 1.65
171 383 0.32 ‘ . 396 1.28 1.57 4.40 10.35 0.18 2.97 295
177 42.3 0.30 398 5.37 392 12.90 6.950 0.18 2.60 2.34
' Grade H-395 .
835(1) 0 : 1.03” 0.659" 0.670” 691 : _ : . 9460 0.62
832(11) 4.8 8.77 —0.20 -0.12 —0.44 10.55
833(1 20.5 4.15 0.63 0.31 1.25 10.78 0.22 2.84 290
834(11) 23.0 ‘ . 4.14 0.66 0.35 1.37 13.41 0.26 2.96 3.02
840(L) 0 : ‘ 6.60 ' ’ 10.29
839(1) 12.7 4.54 0.13 0.06 0.01 13.48
) . Grade P-03
759(1]) 0 : 1.0017 0666” 0.670 4.22 9.431 054
756(11) 5.7 , . 5.99 —0.27 -0.42 -1.10 12.23 0.36 1.89 1.92
757(11) 37.3 4.18 7.12 731 21.3 6.328 0.21 1.60 [.67
758(11) 42.1 : 4.15 12.57 11.41 33.44 4.457(2) 0.23(2) 0.94 1.00
769(1) ' 0 4.72 _ - 8.877
“All data were taken directly on the actual control or irradiated specimens except in those columns headed with an asterisk "Cfiange in diameter; D, = initial diameter.
{*). In the columns headed with an asterisk (*), the data were taken on other special specimens taken from the same parent "Change in ¥olume; V, = initial volume. ,
stock or they are representative data. ’The small volumes stressed make these values approximately 1.33X higher than those obtained on specimens having a
"The axis of rotation of the specimen is paraliel with the axis of rotation of the stock for the symbol ||;itis perpendicularto geometry approaching the ideal conditions for pure flexural (bending) strength tests.
the axis of rotation of the stock for the symbol L. No symbol means the material is near isotropic. : *Change in shear modulus; G, = initial shear modulus.
‘E > 50 keV., : ‘Change in.Young’s modulus; Y, = initial Young’s modulus.
“Layer height = 2/ BET surface area X helium density. "Porosity that is closed to helium, ‘ ) ,
Ry =§3" [ ¢.B) Sin® ¢ dp dB / J5” fi"l(s.B) Sing dop B "With grain = direction in which most crystallites a-axes are oriented: against grain = direction in which most crystallites
where c-axcs are oriented; near-isotropic graphite data are listed between the with-grain and against-grain columns.
¢ = angle between normals to sample surface and basal planes, “Data in this column were reported by C, R, Kennedy, MSR Progrant Seriannu. Progr. Rep. Aug. 31, 1970, ORNI.~
B = angle of rotation about sample normal, . 4622, p. 147,
[(¢.8) = x-ray intensity at position ¢ and 8, \ ?Private communication from O. B. Cavin of the Metals and Ceramics Division of the Oak Ridge National Laboratory.
R|| measured experimentally and R obtained from R.) =1 — (Ry/2). 0. B. Cavin, MSR Program Semiannu. Progr. Rep. Feb. 28,1960, ORNL-4396. p. 221. '
‘Change in length; L, = initial length. "This grade AXF-5QBG was fired for | hr at 3000° C. ‘

YAl reported values are averages of three test values unless specified otherwise by suffix numerals in parentheses.
Table 11.4. Carbon concentration in bismuth-lithium melts
after equilibration for 1000 hr at 650° C

Carbon concentration (ppm)

Nominal
% Li PGX* PGXX ATJ Average
0.01 15, 15 10 15 15
| 45, 40 50 45 45
2 50, 75 60 90 70
3 155, 140 125 135 140

“Duplicate tcsts.

11.2 THERMAL CONVECTION LOOP TESTS
OF MOLYBDENUM, TANTALUM,
AND GRAPHITE IN BISMUTH-LITHIUM
SOLUTION

J. R. DiStefano L. R. Trotter

Onc important requirement of ~materials. that
must contain bismuth-lithium solutions at 550 to
700° C is resistance to mass transfer in the presence
of a temperature gradient. Graphite, molybdenum,
and tantalum alloys have shown promise as con-
tainment materials, and the weight changes in sam-
ples exposed in quartz thermal convection loops
containing Bi—100 ppm Li are summarized in Table
11.5. These loops were operated for 3000 hr at a
maximum temperature of 700° C with a temperature
gradient of 100°C. In addition, little or no metallo-
graphic cvidence of attack was seen in specimens
from these tests. Maximum attack occurred in unal-
loyed tantalum to depths of | to 1.5 mils, while
<{0.5-mil attack occurred in the other materials. No
significant changes in the mechanical properties of
molybdenum or of Ta—10% W were found. However,
T-111 (Ta—8% W-2% Hf) became embrittied as a
result of small increases in oxygen content (100 to 200
ppm).’

Two additional thermal convection loop tests
have been conducted using Bi—2.5% L1 solutions.
One loop was constructed of T-111 and was oper-
ated for 3000 hr while the other was constructed of

3. 0. B. Cavin and L. R. Trotter, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1971, ORNL-4728, p. 173.

89

Table 11.5. Weight changes in materials tested in
Bi—100 ppm Li for 3000 hr in quartz thermal
convection loop“

Maximum loss

Maximum gain

Material (mg/cm?) (mils{ year) (mg/em?)
Mo 3.2 04 1.1
Ta 45 3 ' )
T-111° 4.5 0.3 0.4
Ta-10 % W 0.1 0.0l 0.2
Graphite 0.1 0.01 3.6

“Maximum temperature 700° C: temperature differential 100° C.
"All samples lost weight.
‘Samples were brittle.

molybdenum and operated for 8700 hr. The maxi-
mum weight loss for T-111 was 2.73 mg/cm® (0.19
mil/year), and no loss in ductility was found for
spccimcns' that were tensile tested at room tem-
perature.’

The molybdenum loop operated for 8700 hr and
was terminated when the MSR Program was dis-
continued i1n January 1[973. Weight changes for
specimens from hot and cold leg sections of the
loop are shown in Fig. [1.1. The maximum weight
loss was 3.62 mg/cm® (~0.15 mil/ year), and occurred
in the specimen located in the maximum tempera-
turc section at the top of the hot leg. Deposition
occurred in all cold leg samples and in the first
three samples in the heated section. The maximum
loss of 3.62 mg/cm’ is about the same as that found
in a previous quartz loop test that circulated Bi—100.
ppm Li for 3000 hr (Table 11.1) and indicates that
mass transport of molybdenum is relatively insensi-
tive to the lithium content of the bismuth. .

Metallographic examination of samples from the
molybdenum loop revealed dissolution and deposi-
tion of molybdenum in the hot and cold leg sam-
ples, respectively. Attack was intergranular (Fig.
11.2) and occurred to depths less than 0.5 mil.

Mecchanical properties of selected specimens
from the hot and cold leg sections are shown in
Table 11.6. Specimens from the hot leg were, on the
average, more ductile and weaker than those from
the cold leg. However, the properties of all speci-
mcns fall within a normal range for this material.
90

Table 11.6. Room temperature mechanical properties of molybdenum exposed
to Bi-2.5% Li for 8700 hr in thermal convection loop test

Ultimate 0.2% Offset
Sample Leg Elongation tensile strength  yield strength
number location (% in 1125 in.) (psi) (psi)
1 Hot 700 15.1 100,800 98,300
3 Hot 680 125 104,100 103,700
5 Hot 650 136 97,700 93,400
7 Hot 630 175 93,900 80,900
Hot 620 152 96,700 87.700
- Hot 14.8 98,600 92,800
11 Cold 600 8.5 120,000 110,900
13 Cold 615 84 117,600 109,100
15 Cold 630 8.1 113,100 108,200
17 Cold 640 9.8 118,900 113,100
19 Cold 650 104 111,100 106,100
Average Cold 9.0 116,100 109,500
Y-126386
10 20 MICRONS - 60 70
L A ———1000X 4
0.001 INCHES
A Bottom
Hot Leg Cold Leg

Fig. 11.1. Photomicrographs of the edges of specimens from a molybdenum thermal convection loop containing B-2.5% Li. The loop
operated for 8700 hr at a maximum temperature of 700°C with a temperature difference of 100°C

WEIGHT CHANGE (mg/cm?)

ORNL-DWG 74-10730
SAMPLE NUMBER
4 2 20 18 16 14 12 8

T ol T

B \ (WEIGHT GAINS)

DEPQSITION

BOTTOM OF
HOT LEG 60Q°C

DISSOLUTION
(WEIGHT LOSSES)

/ BOTTOM OF 7
4 / COLD LEG 600°C —|

TOP OF TOP OF —1
HOT LEG 700°C COLD LEG 670°C

N I O T T T S O T
w

2 16 20 40 44 48 52 56 60 B8O
DISTANCE FROM BOTTOM OF HOT LEG (in.}

91

Fig. 11.2. Weight changes of molybdenum samples in a
molybdenum thermal convection loop that contained Bi-2.5% Li
and operated for 8700 hr ata maximum temperature of 700° C with
a temperature difference of 100°C,
92

12. Graphite Studies

H. E. McCoy

The characterization of property changes in vari-
ous grades of graphite as a function of fluence at
715°C has been a continuing part of the graphite
studics for the Molten-Salt Reactor Program
(MSRP). In these studies, almost all of the speci-
mens, nominally 0.126 in. ID by 0.400 in. OD by
0.500 in. long, have been irradiated in target pin
locations in the High Flux Isotope Reactor
(HFIR). The target pin location dictated the size
and gcometry of the specimens. The specimens
were machined so that they and their parent stock
had parallel axes of rotation. The specimens were
physically characterized in the unirradiated
condition. After irradiation, their dimensional
changes were determined as functions of the flu-
ence. Other postirradiation data such as BET sur-
face area, helium densities, open-pore porosities,
and layer heights were measured for some grades of
graphite.

These studies were discontinued at the interrup-
tion of the MSR Program and do not have suffi-
cient priority to be reinstituted at our present fund-
ing level. However, we have been able to extend
this experimental work through funding from the
Air Force Materials Laboratory.' The additional
data acquired on the MSR Program specimens as
functions of accumulated fluence were thermal ex-
pansivity from room temperature to 625°C and
mechanical properties measured at room tempera-
turc. The mechanical properties included brittle
ring strengths, fracture strains, Young’s moduli,
shear moduli, and calculated thermal shock resistiv-
itics. The purpose of the studies for the Air Force
Materials Laboratory was to determine whether
ncutron-damaged graphite had properties beneficial
for some aerospace applications.

12.1 PROPERTY CHANGES IN
NEAR-ISOTROPIC GRADES
OF GRAPHITE IRRADIATED AT
715°C TO FLUENCES AS HIGH
AS 4.2 X 10> NEUTRONS/cm?

W. H. Cook C. R. Kennedy W. P. Eatherly

12.1.1 Materials

Graphite grade AXF was chosen for this study
because of its near isotropy and high strength. We

W. P. Eatherly

included in this study other near-isotropic grades;
principally, grades AXF-5QBG-3, H-395 (ref. 2),
and P-03, to acquire additional information associ-
ated with the aerospace applications and nuclear
rcactors such as the Molten-Salt Breeder Reactors
(MSBRs) and High-Temperature Gas-Cooled Re-
actors (HTGRs). Brief introductory descriptions
and the sources of the four grades AXF, AXF-
5QBG-3, H-395, and P-03 plus a few other grades
uscd for reference purposes are given (Table 12.1).

The typical microstructures of graphite grade
AXF and the other three grades, AXF-5QBG-3,
H-395, and P-03, studied with it are shown in their
unirradiated states (Fig. 12.1). Although all of
these are binderless’ grades of graphite, they tend
to have relatively uniform pore distribution except
for grade P-03. The porosity for grade P-03 ap-
pears to be larger and concentrated between large
particles. Additional relative microscopical charac-
teristics are distinguishable with the use of a rotat-
able sensitive tint plate. The monolithic nature of
these binderless grades of graphite tends to limit
somc measurements to judgments.

Both grades AXF and AXF-5QBG-3 have equi-
axed optical domains approximately 5 um in aver-
age diameter. This means that there are about five
optical domains in the largest particle (25 pm)
uscd to fabricate these materials.

Grade H-395 appears to have been fabricated
with filler particles similar in size to those used in
grades AXF and AXF-5QBG-3. However, the aver-
agc optical domains are slightly larger; there is a
scattering of a few particles that had optical do-
mains two to three times larger than the average;
and the optical domains were not as equiaxed as
those in grades AXF and AXF-5QBG-3. The stock
of grade H-395 is atypical for this grade, because it
1s slightly marred structurally with balls of agglom-
erated filler particles ranging from 250 to 500 um
in diameter (Fig. 12.1¢).

1. Air Force Materials Laboratory/ MXS, Wright-Patterson Air
Force Base, Interagency Agreement No. 40-433-73, Dayton, Ohio.
. 2. Thestock is from an early production of grade H-395 in which
the microstructure is not as uniform as that found in the standard
production of grade H-395.

3. A term applied to grades where there is no microscopically
visible binder; the structure is monolithic.

93

mNmud O

20°0 SIHONI
L—xo00}

O._D.O mo_o,o

2

mmmvd 020'0 S3HONI

o-fld mo.o.o

00. 009

SNO¥OIW

T T
002 00}

T T T——X00}
OONOO@ mzoxu_i

T T
002 00}

. (c) H-395, and (d) P-

F-5QBG-3,

(b) AXI

Fig. 12.1. Photomicrographs of the microstructures of unirradiated graphite grades (a) AXF,

03. As polished. 100,

94

Table 12.1. A summary of the grades of graphite studied and used for reference purposes

Graphite grade

G/ Gl reference
_plots

AFML/MXS
study

Density

(g/ cm’) Form

Maximum particle

size (in.) Comment

AXF? 1.82

AXF-
5QBG-3°

H-395"

P-03¢
AXM®

AXF-UFG*

H-337°

H-364° 1.94

Near-isotropic

Near-isotropic

Near-isotropic

Near-isotropic

Near-isotropic

0.001 No microscopically visible binder,
uniform pore structure with

nominal pore diameter of | um

This is AXF-5QBG

that has been ORNL-
fired at 3000°C for | hr,
grade AXF-5QBG is
manufactured from AXF
and is carbon-impregnated
and regraphitized

Probably a raw coke-based
fabrication process, rated
by manufacturer as
improved H-337 and H-364

0.006
0.001

Molded, medium-grained

Less dense but made
similarly to AXF

<0.001 0.001 usual maximum particle

size in AX series

Made using raw coke,
heavily impregnated with
carbon and regraphitized

Pilot plant product,
precursor to H-337

“ Manufactured by POCO Graphite, Inc., Decatur, Texas.

* Manufactured by Great Lakes Carbon Corp., New York, New York.
¢ Manufactured by Pure Carbon Company, St. Mary's, Pennsylvania.

¢ Change in shear modulus.

Grade P-03 has large and small filler particles.
The large particles range from 100 to 150 um and
the smaller particles appear to be 5 to 25 um in
size. The optical domains are not equiaxed and
their dimensions range from 20 to 40 um with
smaller ones intermingled that are approximately 6
um in diameter.

12.1.2 Testing

We used the MSR specimens that were nomin-
ally 0.400 in. OD by 0.126 in. ID by 0.500 in. long
which had been irradiated in the High Flux Iso-
tope Reactor (HFIR) at 715°C to various fluences
up to 4.2 X 10%* neutrons/cm’ (E > 50 keV). The
initial Young’s and shear moduli were determined

on the existing specimens by ultrasonic techniques.*
The thermal expansion of the specimens in the di-
rections of their axes of rotation was determined
from room temperature to 625°C in a quartz dila-
tometer.” These measurements concluded the neces-~
sary nondestructive testing of the specimens. The
flexural strength tests were destructive and were
obtained using the brittle-ring technique.

4. This work was done by the Nondestructive Testing Group of
the Metals and Ceramics Division of the Oak Ridge National
Laboratory, Oak Ridge, Tennessee.

5. This work was done by the Mechanical and Physical Proper-
ties Laboratory of the Produets Certification Division of the Y-12
Plant, Oak Ridge, Tennessee.
12.1.3 Property Changes as Functions of Fluence

The volume changes of these grades of graphite
were reported previously,”’ but are shown again for
reference purposes (Figs. 12.2 and 12.3). The dam-
agce model used to explain these data was reviewed
in some detail by Engle and Eatherly.®

The physical properties of the unirradiated and
irradiated grades are summarized in Table 12.2.
These effects of the fluence are respectively shown
graphically for the coefficients of thermal expan-
sion, brittle-ring stresses (strengths), changes in
Young’s moduli of elasticity, changes in shear mod-
ull, and fracture strains (Figs. 12.4, 12.5, 12.6, 12.7,
12.8 and 12.9).

6. C. R. Kennedy, MSR Program Semiannu. Progr. Rep. Aug.
31, 1969, ORNL-4449, pp. 175-77.

ORNL-CWGC 74-47533

25
20
~ s /
3o /xxm AXF- 508G
N
= 10 //AXF-UFG
[}
L]
3 | //
/7 //Axp
0 ¥ A/
-____../
-5
0 5 10 5 20 25 30 35 (x102h)

FLUENCE [neutrons /cm€ (£ > 50 kev]

Fig. 12.2. A comparison of volume changes of various POCO
grades of graphite vs fluence accumulated at 715°C.

ORNL-DWG 74-4757

20
// HL-18

Vo)
-
NS
\

3Li0
q.t / /
o
o / H364
g ° / L~
j H395 A_-
’//
° T —
———5550 |
5 5 1 15 20 25 30 35 {x10%

FLUENCE [neutrons/cm? (£ > 50kev | ]

Fig. 12.3. A comparison of volume changes of various
binderless grades of graphite vs fluence accumulated at 715°C.

95

CRNL-DWG 74-7674R

(x107%
P
o \

W o

=] 8 8

= W0

Go . A \ N,

=} / \

Eg 1 l . \

Oz 6 -

© O

VAN

2 s N

B “/ g \\N P~03
X -

24 s
I s “~AXF-5QBG-3
g, — s AXF
=

2
o 5 1 15 20 25 30 3% 40 (x10%Y

FLUENCE (neutrons - em 2, £ 50 keV)

Fig. 12.4. The average coefficients of thermal expansion from
20 to 600°C vs fluence accumulated at 715°C for graphite
grades AXF, AXF-5QBG-3, H-395, and P-03.

ORNL-DWG 74-7677R

puod)

"51/ Ri

\.l \ °

= / />q§<\\
v L L~ ™~ \N °
w 40 7 ‘\AXF
lfi:J d \ \
[ LFS \ .
& AXF-50BG-3
g . o
& a
w 5 -
= a
o
=
@
m

0 21

0 5 10 15 20 25 30 35 a0 (wo?h

FLUENCE {neutrons®cm~=2, £>50 kev)

Fig. 12.5. Brittle-ring stresses vs fluence accumulated at 715°C
for graphite grades AXF, AXF-5QBG-3, H-395, and P-03.

12.1.4 Conclusions

The MSBR design life for the graphite moderator
has been set as 3 X 0% neutrons/cm” (£ > 50 keV)
(rcf. 9) based on the dimensional changes. Fluences
of 3 X 10** neutrons/cm’ and beyond ultimately re-
duced the values of ail parameters involved in these
most recent studies with the exceptions of the coef-

7. C. R. Kennedy, Metals and Ceramics Division, Ouk Ridge
National Laboratory. Oak Ridge. Tennessce.

8. G. B. Engle and W. P, Eatherly, High Temp. High Pressures
4, 144 (1972).

9. O. L. Smith, W. R. Cobb, and H. T. Kerr, MSR Program
Senviannu. Progr. Rep. Aug. 31, 1968, ORNL-4344 p. 68,
ORNL-DWG 74-7681R

| & % \Km

)’/)’0 CHANGE N
YOUNG'S MODULUS OF ELASTICITY
\\\

\ \ \

o 5 0 15 20 25 30 35 40  (x0?
FLUENCE (neutrons - cm 2, £3> 50 keV)

Fig. 12.6. Changes in Young’s moduli of elasticity vs fluence ac-
cumulated at 715°C for graphite grades AXF, AXF-5QBG-3, H-
395, and P-03.

ORNL-DWG 74-7678

] T T

o AXF AND AXF-5QBG ¢
‘:j: 5 | o AXF-UFG e S
3 o AXM-5Q \
2 < °
= A N ° Ne
< L~ ~N \
4 /"‘\
[¥2] . \ ?
: /’ N
w
Q. o ©
g d \
[&] fi \
&,
o \

0 5 10 5 20 25 30 35 40 (x10%h
FLUENCE (neutrons - cm 2, £ >50 keV)

Fig. 12.7. Changes in shear moduli vs fluence accumulated at
715°C for graphite grades AXF, AXF-5QBG-3, AXF-UFG, and
AXM. ’

ficient of thermal expansion (CTE) of grade P-03
and the strength of grade H-395. In general, the
results of these studies do not reveal any changes in
the specific parameters studied for these grades of
graphite that would seriously affect their perfor-
mance in an MSBR to a fluence of 3 X 10> neu-
trons/cm®. The data acquired make possible a more
rigorous analysis of the preceding general state-
ment, and suggest a satisfying degree of potential
ranges of control for attaining desired properties
through the proper fabrication processes.

The CTE values were not measured beyond
600° C to prevent annealing out of property changes
acquired through the neutron irradiations at 715°C
that were to be subsequently measured on the same

ORNL-DWG 74-7679

-
=7 |
(é) //.— ™~
:8) / %o N
= ) / % N \
E / \ N
z L7 o\ N\
g AN
1 \
U. v H385 \ \
& o H364
G a H337
o P-03
0
© 5 w© 15 20 25 30 35 40 (x0%

FLUENCE (neutrons - cm™2, £ >50 keV)

Fig. 12.8. Changes in shear moduli vs fluence accumulated at
715°C for graphite grades H-395, P-03, H-364, and H-337.

ORNL-DWG 74-8896

- N\,
b= L AXF
o \
a \H—395\
¥ 05 PO N
= \'\ \
S
Q P-03 ~
& \\'—" }"‘fi._;
(e — —
0 21
0 5 10 15 20 25 30 35 40 (x10%)

FLUENCE (neutrons - cm 2, £>50 keV)

Fig. 12.9. Fracture strains vs fluence accumulated at 715° Cfor
graphite grades AXF, AXF-50BG, H-395, and P-03.

specimens after the CTE determinations. All CTE
values for graphite grades AXF, AXF-5QBG-3, H-
395, and P-03 behaved ina similar manner (Fig. 12.4).
With increasing fluence, they all initially increased
by 10 to 40% and then decreased to 40 to 60% de-
pending on the grade involved, with the exception
of grade P-03 which returned essentially to its orig-
inal value. Fortunately, the CTE values are small
enough that these should not create dimensional de-
sign problems. For example, in a 1000-MWe power
MSBR with a diameter of approximately 20 ft,'" the
diameter change from room temperature to the op-
erating temperature would be 0.2 to 1.2 in. in the
absence of irradiation. Therefore, a 740 to 60%

10. M. L. Lundin, MSR Program Semiannu. Progr. Rep. Feb.
29, 1972, ORNL-4782, p. 10.

97

change in these values does not appear to create HFIR-type sample by cutting the samples trans-
serious design problems. versely to their axes of rotation. The brittle-ring
Brittle-ring strength measurements have been stress values are plotted vs fluence using averages of

shown by C. R. Kennedy of this Laboratory and threc values (Fig. 12.5). These values are high be-
others'' to be a simple, accurate method for deter-

mining the strengths of various grades of graphite. 11, S. A, Bortz and H. H. Lund, “Evaluation of a Tension Test
This method is well suited for the HFIR-type sam- for Brittle Materials,” p. 531 in the Proceedings of the Fourth
ples. We obtained three specimens from each Conference on Carbon, New York, 1960.

Table 12.2. Some physical properties of graphite grades AXF, AXF-5QBG-3, H-395, and P-03 as functions of fluences from
0 to 42.1 X 10°' neutrons /cm? 50 keV) accumulated at 715°C*

Pore * * *

entrance Permeability Electrical Thermal
Speci- diameter to helium resistivity conductivity
men - Fluence®  Density, g/cm3 Porosity, % range for  BET at 1 atm (um-cm) (cal{sec-cm® C)
number (107 90%ofopen surface at room
(orien- neutrons/ Bulk  Helium Open Closed” Total porosity area  tempera- With Across  With Across
tation)® cm?) (um) (mz/g) ture grain grain grain  grain

(107%

Grade AXF
122 0 1.82 2.14° 15.4° 4.1 19.5 1.0 0.34° 40 1475 0.154'
120 54 2.08 130 8.0 21.0 09
118 8.9
119 4.2 2.07 150 84 234 1.9

Grade AXF-5QBG-3*

174 0 1.87(39) 17.3 0.5-2.0 ~ 0.62 1020(6) 1095(6)
i70 7.6 - 2.11° 10.4° 6.7 17.1 0.7°
173 10.2 2.14 121 53 17.4 0.5
[71 38.3 2.11 142 6.7 209 4.1
177 42.3 2.11 204 6.7 27.1 3.2

Grade H-395
8351y 0O 1.79(30) 20.8 . 1260(6) 1300(6)
832(jly 4.8
833(iy 20.5
834(Ily 23.0
840(L)y O
839(1L) 2.7

Grade P-03
759(11) O 1.82(30) 19.5 1600(6) 1655(6) 0.145 0.135
756(]|)y 5.7
757()1y 37.3
758(]|) 42.1
769(L) O

“All data were taken directly on the actual control or irradiated specimens except in those columns headed with an asterisk (*). In the
columns headed with an asterisk (*), the data were taken on other special specimens taken from the same parent stock or they are
representative data. ) .

"T'he axis of rotation of the specimen is parallel with the axis of rotation of the stock for the symbol {|; it is perpendicular to the axis of
rotation of the stock for the symbol 1. No symbol means the material is near isotropic.

E > 50 keV.

“Porosity that is closed to helium.

“Data in this column were reported by C. R. Kennedy, MSR Program Semiannu. Progr. Rep. Aug. 31, 1970, ORNL.-4622.p. 147.

/At 600° C, ORNL Intra-Laboratory Correspondence of July 7, 1970, from D. L. McElroy to A. F. Zulliger.

¥This grade AXF-5QBG was fired for 1 hr at 3000°C.

"Suffix numerals enclosed in parentheses indicate the number of values averaged.

A
causc the stresses were concentrated in a small vol-
umec; however, direct experimental data show that
the valucs may be readily and accurately converted
to truc flexural stress values by multiplying by 0.75.
As shown in Fig. 12.5, irradiation is ultimately
deleterious to the strengths of grades AXF, AXF-
5QBG-3; and P-03 while the strength of grade H-
395 is still increasing at a fluence level of 2.3 X 10*
neutrons/cm’ (E > 50 keV). No evidence exists of
any saturation effects for the first three grades; how-
cver, their strengths at the design life of 3 X 10
ncutrons/cm’ are several factors greater than the
_strength required for the MSBR graphite. There is
no obvious reason for the different strength behav-
ior of grade H-395. This strength increase 1s also
present in samples taken perpendicular to each
other from the parent stock (Fig. 12.5). Grade H-
395 and AXF tend to have similar characteristics
in most properties except strength. Unfortunately,
thc maximum fluence accumulated on a sample of
grade H-395 was only 2.3 X 10%* neutrons/cm”.
The changes in Young’s and shear moduli natu-
rally reflect the changes in strength and fracture
strain values. The latter is the parameter in which
all four grades of graphite, AXF, AXF-5QBG, H-

08

395, and P-03, behave in the most strikingly similar
manner with increasing fluence values. The frac-
turc strains for all four grades (Fig. 12.9) have de-
crcased as much as 409 for an accumulated fluence
of only 1 X 10% neutrons/cm’ (E > 50 keV), and
they all appear to be saturating between 60 to 85%
of their original values at fluences greater than 2.5
X 10?* neutrons/cm’. Apparently, all grades are ap-
proaching the same critical flaw size. The signifi-
cance of these mechanical changes will have to be
dctermined by more detailed analyses.

Further studies should include measurements of
pore cntrance diameter distributions, permeabilities
to helium gas, and thermal conductivities. The sur-
prisingly different strength behavior of grade H-395
vs fluence also warrants additional study. Graphite
rescarch programs have reached a significant pla-
tcau in qualitatively relating fabrication processes
and irradiation behavior. An irradiation program
within relatively narrow limits of graphite types,
but with heavy commitments for characterization of
raw materials and of the unirradiated and irradi-
atcd properties of the finished graphite bodies,
would have appreciable benefits.
Part 4. Fuel Processing for Molten-Salt Reactors

J. R. Hightower, Jr.

Part 4 deals with the development of processes for
the isolation of protactinium and for the removal of
fission products from molten-salt breeder reactors.
During the period covered by this report, we resumed
work on the chemistry of fluorination and
reconstitution. An experimental batch reactor facil-
ity has been reconstructed for delivery and metering
of fluorine, hydrogen, HF, UFs, and argon to a gold-
lined nickel reaction vessel.

Studics of equilibria in fused salt-liquid alloy
systcms were completed and the results have been
published.

The metal transfer experiment MTE-3 was com-
pleted and the equipment used in that experiment
was examined. The examination showed that no
serious corrosion had occurred on the internal sur-
faces of the vessels, but that serious air oxidation
occurred on the external surfaces of the vessels.
Analyses of the bismuth phases indicated that the
surfaces in contact with the salts were enriched in
thorium and iron. We plan to repeat experiment
MTE-3 to determine the cause of low mass transfer
ratcs scen in that experiment. The carbon steel ves-
sels are being replaced with new vessels; and puri-
fied fuel carrier salt, bismuth, and lithium chloride
will be used.

Mass transfer coefficients in the mechanically
agitated nondispersing contactors were measured in
the Salt/Bismuth Flow-through Facility. . The
measured mass transfer coefficients are about 30 to
409 of those predicted by the preferred literature cor-

99

relation, but were not as low as those seen in some of
the runs in MTE-3. Additional studies using water-
mercury systems to simulate molten salt-bismuth
systems have indicated that the model used to inter-
pret results from previous measurements in the
water-mercury system has significant deficiencies.
We must analyze the results differently to extract in-
formation useful for designing contactors in molten
salt-bismuth systems.

Autoresistance heating studies were continued to
develop a means of internal heat generation for
frozen-wall fluorinators. Equipment was built to test
a design of a side arm for the heating electrode.
Results of experiments with this equipment indicate
that for proper operation the wall temperature must
be held much lower than that for which the equip-
ment was designed. Studies with an electrical analog
of the cquipment indicate that no regions of ab-
normally high current density exist in the side arm.

We are beginning engineering studies of fuel re-
constitution, Equipment is described in this report
for carrying out the reaction of gaseous UFs with UF,
dissolved in molten salt and the subsequent re-
duction with hydrogen of the resultant UFs.

Future development of the fuel processing opera-
tions will require a large facility for carrying out
enginecering experiments. Work was initiated for a
conceptual design of a processing engineering labo-
ratory where these engineering experiments can be
carricd out.
100

13. Processing Chemistry
A. D. Kelmers

Studies of certain aspects of the chemistry of the
metal transfer process and of fuel reconstitution were
conducted between September 1972 and the tem-
porary termination of the MSR Program in March
1973. Investigation of the reaction of gaseous UFs
with UF4dissolved in LiF-BeF,-ThF, (72-16-12 mole
%) to form dissolved UFs was partially completed,
and the results were published. Studies of the
distribution of Li;Bi between molten LiCl and liquid
Li-Bi alloys were completed and the results were
published. After reactivation of the MSR Programin
February 1974, apparatus was assembled for
conducting studies on the chemistry of fluorination
and reconstitution of MSBR fuel salt. These studies
are Now in progress.

13.1 CHEMISTRY OF FLUORINATION
AND FUEL RECONSTITUTION

M. R. Bennett L. M. Ferris A. D. Kelmers

The reaction of gaseous UF¢ with UF,dissolved in

molten LiF-BeF-ThF. (72-16-22 mole %) had been

under investigation for some time.' The results

obtained prior to March 1973 have been published;’
an abstract follows:

The reaction of gaseous UFs with UF, dissolved
in molten LiF-BeF,-ThF, (72-16-12 mole %) was
studied over the temperature range 550-650°C.
Chemical and spectrophotometric evidence
showed that U(V) was formed in the salt according
to the reaction UF¢(g) + UF4«d) = 2UFs(d), in
which (g) and (d) denote gas and dissolved species,
respectively. The studies were conducted using
gold apparatus, since gold was the only material
found to be inert to both gaseous UF, and dis-
solved UFs. When U5 was present in low concen-
trations, it disproportionated slowly according to
the reverse of the above reaction; the rate was
second-order with respect to the UFs concentra-
tion. The results indicated that, when UFs was pres-
ent in high concentrations, its vaporization from
the salt and subsequent disproportionation in the

I. M. R.Bennettand L. M. Ferris, MSR Prograin Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 164.

2. M. R. Bennett and L. M. Ferris, J. Inorg. Nucl. Chem. 36,
1285 (1974). '

vapor phase were significant. Under all conditions
studied, however, the pressure of uranium-
containing species in the vapor phase was less than
0.004 atm.

An experimental batch reactor facility has been
constructed so that studies of fluorination and fuel
reconstitution can be continued. The facility in-
cludes systems for the delivery and metering of
fluorine, hydrogen, HF, UF¢, and argon. The initial
experiments will be conducted in a gold-lined nickel
reaction vessel to investigate reactions involved in
continuous fluorination and the subsequent fuel

reconstitution. The overall reactions may be
represented as follows:
for fluorination,
" UF4(d) + Fx(g) — UF«(g) (1)
and for reconstitution,
UFs(g) + Ha(g) — UF4(d) + 2HF(g). (2)

The above reactions are useful for calculating stoi-
chiometric chemical reagent use, but they fail to re-
veal the complexity of the actual operating
conditions.

The continuous fluorinator is envisioned as a
countercurrent device operating at about 510° C with
fuel salt entering one end and fluorine gas being
introduced at the other end. Thus, UFs gas and
unreacted F» gas would exit from the reactor coun-
tercurrent to the fuel salt, and salt depleted of ura-

~ nium would exit countercurrent to the fluorine. The

following reactions can occur in the fluorinator

UF4(d) + UFs(g) — 2 UFs(d), 3)
UFs(d) + 1/2Fxg) — UF«(g), @)
UFE4(d) + 1/2Fx(g) — UFs(d). (5)

In addition, UFs can distribute to the gas phase

UFs(d) = UFs(g) (6)
where it would be oxidized to UF¢ by F»:
UFs(g) + 1/2Fa(g) — UFs(g). (7)

Preliminary experiments’ have investigated some
of these reactions. The results indicate that the oxi-
dation reactions (3) and (4) are rapid and that the
equilibrium constants must be quite large. The ex-
istence of dissolved UFs was confirmed, but a
measure of the partial pressure of UFs(g) was not
obtained.
In the fuel reconstitution step the gas entering from
the fluorinator will be a mixture of UFs with the
unreacted F.. These gases will be reduced with an
excess of UF4 contained in a circulating salt stream:

UFe(g) + UF4(d) — 2 UFs(d) 3)

1/2 Fag) + UF4(d) — UFs(d). (5)
Thus, for this processing step it is also important to
have an estimate of the equilibrium constant for re-
action (3).

Ultimately, the dissolved UFs will be reduced in a

separate chamber with hydrogen gas

UFs(d) + 1/2 Ha(g) — UFu«(d) + HF(g). (8)
It is necessary to conduct the two fuel reconstitution
reduction reactions separately to avoid mixing H
and F, gases. The situation in the hydrogen re-
duction vessel is further complicated by the simul-
tancous gas-phase disproportionation of UF;

2UFs(g) — UFa(c) + UFs(g) )

where (c) represents the condensed or solid phase.
This disproportionation could lead to isolation of
some of the uranium as solid UF, in regions of the
vessel not 1n contact with salt. This reaction should
not be a complication where F; is present in the gas.

Preliminary results® indicate that reaction (8) is
slower than reaction (3) and, when carried out in a
batch contactor, many times the stoichiometric
quantity of H: was required to reduce all the UFs.
This may be due to the expected low solubility of H
in fuel salt melts. Additional experiments are planned
to investigate the fuel reconstitution reactions.

13.2 EQUILIBRIA IN FUSED
SALT-LIQUID ALLOY SYSTEMS

L. M. Ferris M. A. Bredig’

J. F. Land

F. J. Smith

Preliminary results from a study of the distribu-
tion of lithium and bismuth between molten LiCl and

liquid Li-Bi alloys were reported previously.® This
study was completed and the results were pub-
lished;’ an abstract follows:

The distribution of lithium and bismuth be-
tween Li-Bi alloys and molten LiCl was measured
at several temperatures between 650 and 800°. The
extent of their distribution to the LiCl increased
dramatically at each temperature with a moderate
increase in the lithium concentration of the alloy;
for example, at 650°, the bismuth concentration in
the LiCl increased from about 5 to 4800 ppm as the
lithium concentration in the alloy increased from
10 to 50 at. %. The ratio of “excess” lithium to
bismuth in the LiCl generally was about 3,
suggesting that salt-like LisBi was selectively
dissolved from the alloys. The measured
cquilibrium bismuth concentrations in the LiCl in
the temperature range 650-800° can be expressed
as

In Nai(d) =4 In NLi(m) + 2-579NLi(m) + 2.211
— 9465/ T(°K) + 0.0832 exp[3090/ T(° K)],

with an estimated uncertainty in In N of +0.2,
In this expression, N, (d), and (m) denote mole
fraction, dissolved in salt phase, and dissolved in
alloy phase, respectively. Some data were obtained
using molten LiBr as the salt phase at 650°.
Bismuth concentrations (mole fractions) in the
LiBr were about twice as high as those obtained
with LiCl at the same alloy concentrations.

3. Consultant to the Chemistry Division, ORNL.

4. L. M. Ferris and J. F. Land, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL-4832, p. 160.

5. L. M. Ferris, M. A. Bredig, and F. J. Smith, “Equilibrium
Distribution of Lithium and Bismuth Between Liquid Lithium-
Bismuth Alloys and Molten Lithium Chloride at 650-800°,” J.
Phys. Chem. 77, 2351 (1973).
102

l4. Engineering Development of Processing Operations

J. R. Hightower, Jr.

Studies related to the development of a number of
processing operations were resumed in January 1974,
after the interruption of the MSR Program. During
this report period, the equipment in which metal
transfecr experiment MTE-3 was carried out was
examined, indicating that no serious corrosion had
occurred on the interior of the vessels, although
serious air oxidation had occurred on the external
surfaccs. Analyses of the bismuth phases indicated
that those portions of the bismuth which were
adjacent to a salt-bismuth interface were enriched in
iron and thorium. Evidence of oxide was found only
in the stripper vessel. We plan to repeat experiment
MTE-3 to determine the reason for the low mass
transfer rates seen in that experiment. The carbon
stcel vessels are being replaced with new vessels and
purified fuel carrier salt, bismuth, and lithium
chloride will be used. Results of tests with a new
oxidation resistant coating indicate its superiority
over the nickel aluminide used on the previous test
vessels.

A flow-through mechanically agitated nondis-
persing contactor was installed in place of the packed
column in the Salt/ Bismuth Flow-through Facility.
Mass transfer rates measured in this contactor are
lower than those predicted by literature correlations.
Studies in the same type contactor using a water-
mercury system indicate that the model previously
used for describing mass transfer in this system had
serious deficiencies and that a more accurate model
must be found.

Studies were continued of autoresistance heating
as a means of internal heat generation for frozen-wall
fluorinators. Equipment is described for testing one
design of a side arm for the heating electrode. Results
of experiments with this equipment indicate that for
proper operation the wall temperature must be held
much lower than that for which the equipment is
designed. Studies with an electrical analog of the
equipment indicate that no regions of abnormally
high current density exist in the side arm.

Equipment is described for studying fuel recon-
stitution via the reaction of UFs with UF:dissolved in
molten salt and the subsequent reduction with H; of
the resultant UFs.

Work was initiated on the conceptual design of a
laboratory for carrying out large engineering exper-
iments on fuel processing.

14.1 EXAMINATION OF EQUIPMENT
FROM METAL TRANSFER
EXPERIMENT MTE-3

H. C. Savage

The reference processing flowsheet' shows the
metal transfer process for removing rare earths from
molten-salt breeder reactor fuel salt. In this process,
fucl salt, which is free of uranium and protactinium
but contains the rare earths, is countercurrently
contacted with bismuth that contains reductant to
extract the rare earths into the bismuth. The bismuth
stream, which contains the rare earths and thorium, 1S
then countercurrently contacted with lithium
chloride. Because of favorable distribution
coefficients, significant fractions of the rare earths
transfer to the lithium chloride with a negligible
amount of thorium. The final steps of the process
consist of extracting the rare earths from the lithium
chloride by contact with bismuth having lithium
concentrations of 5 and 50 at. %.

Four engineering scale experiments (MTE 1, -2,
2B, and -3) have been carried out to study the steps n
the metal transfer process and to obtain information,
such as mass transfer rates of the rare earths between
the salt and bismuth phases.that are necessary to
determine the size and type of equipment needed for
the process. Results from these experiments have
been reportf:d.z_6 Experiments MTE-1, -2, and -2B
demonstrated the selective removal of rare earths
(lanthanum and '“7Nd) from MSBR fuel carrier salt
(72-16-12 mole % LiF-BeF:-ThF4) and transfer to Li-

. D. E. Ferguson, Chem. Technol. Div. Annu.
Mar. 31, 1972, ORNL-4794, p. 1.

2. L. E. McNeese, Engineering Development
Molten-Salt Breeder Reactor Processing No. 7,
3257, pp. 29-46 (Feb. 1972).

3. L. E. McNeese, Engineering Development
Molien-Salt Breeder Reactor Processing No. 9,
3259, pp. 167-94 (Dec. 1972).

4. L. E. McNeese, Engineering Development
Molten-Salt Breeder Reactor Processing No. 11,
3774 (in preparation). -

5. L. E. McNeese, Engineering Development
Molten-Salt Breeder Reactor Processing No. 17,
4178 (in preparation),

6. Chem. Technol. Div. Annu. Progr. Rep. Mar. 31, 1973,
ORNL-4883, p. 25.

Progr. Rep.

Studies for
ORNL-TM-

Studies for
ORNL-TM-

Studies for
ORNL-TM-

Studies for
ORNL-TM-
103

Bi acceptor alloy. Experiment MTE-3 was a larger
experiment that included features such as mechanical
agitation of the saltand metal phases to improve mass
transfer rates of the rare earths across the three salt-
metal interfaces. This equipment was designed to
measure the mass transfer rates across the three inter-
faccs as a function of agitator speed for comparison
with calculated values using a mass transfer corre-
lation developed by J. B. Lewis.’

Metal Transfer Experiment MTE-3 (Fig. 14.1)
consisted of three interconnected vessels: a 14-in.-
~diam fuel salt reservoir, a 10-in.-diam salt-metal
contactor, and a 6-in.-diam rare earth stripper. The
salt-metal contactor was divided into two compart-
ments that were interconnected through an opening
in the bottom of the divider by a pool of bismuth
containing thorium and lithium. The stripper con-
tained a lithium-bismuth solution. Mechanical agi-
tators were used in both compartments of the con-
tactor and in the lithium-bismuth stripper to pro-
motc mass transfer across the three salt-metal inter-
faces. All vesscls were carbon steel with an oxida-
tion-resistant coating of nickel aluminide on the
outside surfaces.

Thce purpose of MTE-3 was to evaluate this type of
contactor and stipper for use in removing rare earths
from molten-salt reactor fuel. Fluoride salt and LiCl

. 1. J. B. Lewis, Chem. Eng. Sci. 3, 248-59 (1954).

flow rates in the MTE-3 system were about 1% of
those required for removing rare earths from a 1000-
MWe MSBR. The operating temperature was about
650° C; phases were LiF-BeF:-ThFs (72-16-12 mole
%) fuel carrier salt, 0.13 at. % thorium-bismuth
solution containing about 0.3 at. 9 lithium as
reductant, and a 5 at. 9 lithium-bismuth solution in
the stripper. In this process the rare earths are
extractcd from the carrier salt into the thorium-
bismuth solution. Next, the rare earths are extracted
from the thorium-bismuth with molten lithium
chloride and finally, they are stripped from the LiCl
into bismuth-lithium alloy.

Mass transfer coefficients for radium, europium,
lanthanum, and neodymium were measured during
13 runs using agitator speeds ranging from 100 to 400
rpm for comparlson with values predicted by a
literature correlation’ based on data obtained with
aqueous-organic systems. Many of the mass transfer
coefficients were lower than those predicted by
factors of 6 to 25.° Termination of the MSR program
precluded further investigation of the mass transfer
rates and postoperative examination of the MTE-3
equipment to determine the reason for the lower than
cxpected transfer rates. The system was shut down
and was placed in standby condition in February
1973 with all salt and metal phases frozen.

During operation, metal transfer experiment
MTE-3 was maintained at approximately 650° C for
about | year (about 8700 hr). During the final run

ORNL -DWG-71-147-R1I

AGITATORS
VENT LEVEL '
ELECTRODES LEVEL
ELECTRODES
FLUORIDE
" SALT PUMP ( VENT
=] =]
ARGON
SUPPLY
ARGON
SUPPLY -
‘72-16-12 mole % Bi-Th Li-8Bi
LiF'BGFz'ThFQ
FLUORIDE SALT- METAL RARE EARTH
CONTACTOR STRIPPER

SALT
RESERVOIR

Fig. 14.1. Flow diagram of metal transfer experiment MTE-3.
(EU-9), with agitator speeds of 400 rpm, fluoride salt
was entrained into the LiCl in the contactor’ (about
10 wt % fluoride into the LiCl). This also resulted in
the transfer of thorium into the LiCl which was
extracted into the lithium-bismuth solution in the
stripper.

After the MSR Program was reinstated in January
1974, the vessels used in MTE-3 were opened for
inspection. To remove the three agitator assemblies
for examination of the molybdenum shafts and
agitator blades, it was necessary to melt the salt and
mectal phases in the contactor and stripper vessels.
After a check of all heater circuits and recalibration of
instrumentation, the contactor and stripper vessels
were heated to about 650°C and the agitator
asscmblics were removed without difficulty. The
agitator ports were immediately sealed with blank
flange cover plates and the vessels were cooled to
ambient temperature. During the removal of the
agitators, an argon atmosphere was maintained in the
vessels. '

From a cursory visual examination, the molybde-
num shafts and blades appeared to be in good con-
dition. No areas of heavy corrosion attacks were
noted. However, some areas of the shaft and blades
had deposits of salt and metal. Figure 14.2 is a
photograph of the molybdenum agitator shafts taken
immediately after they were removed.

We removed the thermal insulation from MTE-3
and some of the stainless steel shimstock that sur-
rounded the electric heaters. The outside surfaces of
the carbon steel vessels were found to be badly oxi-
dized due to failure of the nickel aluminide oxida-
tion resistant coating. The vessels had been main-
tained at about 650°C for approximately one year
(8700 hr). The depth of oxidation is estimated to be
1/16in. or less and thus did not significantly affect the
mechanical integrity of the vessels (initial wall
thickness = 3/8 in.) at the operating conditions of
MTE-3. The oxidation resistant protective coating
consisted of a 0.015 to 0.020-in.-thick nickel alumi-
nide coating applied by flame spraying with wire
flame spray equipment. The nickel aluminide mate-
rial is identified as METCO No. M405-10 (METCO,
Inc., Westburg Long Island, New York). A
representative of METCO recommended thata much
better oxidation-resistant coating would be obtained
by using a nickel chromium alloy with 6%aluminum
composite in powder form (METCO No. P443-10)
applied with plasma spray equipment. Test sections
of carbon steel pipe with the two coatings described
above were prepared and tested for oxidation

104

resistance in air at temperatures of 700 and 815°C.
Rcsults arc discussed in Sect. 14.2.

Wc obtained samples of the salts (fuel carrier salt
and LiCl) and bismuth from the MTE-3 vessels at
various locations by removing 2-in.-diam plugs using
a holec saw. No foreign material was seen at the
interfaces between the fluoride salt—Bi-Th and the
LiCI-Bi-Th phases in the contactor whereas there
was a layer of material of different structure (about
1/32-in. thick) between the LiCl-Li-Bi phase in the
stripper. The surface of the Bi-Th phase in contact
with the LiCl in the contactor had an etched-
appearance whereas the surface in contact with the
fluoride salt did not. Corrosion of the inside surfaces
of the 2-in.-diam vessel wall sections appeared
minimal on visual examination of the 2-in.-diam
plugs.

A small (about 1/8-in. by about 1/4-in.) sample of
bismuth was taken at each of the three metal-salt
interfaces for metallographic examination. Each
section included the bismuth surface in contact with
the salt phase. Photomicrographs of these samples
arc shown (Figs. 14.3, 14.4 and 14.5). As seen in the
photomicrograph some foreign material exists near
cach surface of the bismuth that was in contact with
the salt phase (fuel carrier salt and LiCl). Whether
this material was present at the interface during
operation of metal transfer experiment MTE-3 or
segregated at the surface when the phases were
cooled without agitation is unknown.

Scanning electron microprobe analyses identified
the material seen at the interfacial surfaces of the
bismuth (Figs. 14.6, 14.7 and 14.8). The bismuth in
contact with the fluoride carrier salt contained a
surface layer enriched with thorium and iron along
with some manganese and lanthanum (lanthanum
was one of the rare earths used to measure mass
transfer rates in MTE-3). Thorium and iron were
identified at the surface of the bismuth in contact
with the lithium chloride in the contactor. Iron and
thorium were also identified at the surface of the
bismuth in contact with the lithium chloride 1n the
stripper. The thorium in the stripper was most
probably transferred over when fuel salt was en-
trained into the LiCl at the end of experiment
MTE-3.

Samples of material were removed from the vi-
cinity (within 1/32 in.) of each of the three salt-metal
interfaces in the MTE-3 equipment and analyzed by
x-ray diffraction (Table 14.1). The only oxides found
in the system were found at the interface betwen LiCl
and the Li-Bi stripper alloy in the stripper vessel. A
105

PHOTO 1070-74

Fig. 14.2. Photograph of agitator shafts removed from MTE-3 equipment

106

ORNL DWG. 74-10113

SURFACE IN CONTACT WITH SALT ~F

375

MICRONS

; Fo
olo
S
AT “fo
= 7 L
= / &
/

Fig. 14.3. Photomicrograph of the Bi-Th phase from the contactor at the fluoride salt interface, MTE-3.

ORNL DWG. 74-10114

MICRONS

- = olO
[ARS
‘ BISMUTH 13

Fig. 14.4. Photomicrograph of the Bi-Th phase from the contactor at the LiCl interface, MTE-3

high concentration of ThO; was found in the LiClat
this interface. Although no thorium should have been
present during most of the operating time, the
thorium could be expected to combine with any
oxides present when the fluoride salt (containing Th)
was entrained into the chloride salt.

No oxides were detected at the LiCl-Bi-Th inter-
face in the contactor. The analysis of the mass
transfer rates observed during this experiment sug-

107

gested that some hindrance of mass transfer was
occurring at this interface. Oxide films were sus-
pected. However, analyses of the material near this
interface do not support the theory that oxide films at
the LiCI-Bi-Th interface slowed down the mass
transfer.

The samples taken for x-ray diffraction analysis
were extremely small (of the order of 2-3 mm’ mate-
rial was removed from the interface for each sample)

ORNL DWG. 74-10115

SURFACE IN CONTACT WITH SALT

BISMUTH

-~
100 200 MICRONS 600 700
L L1 00X—L 1 It
T T
0.005 0.010 INCHES 0.020 0.025

Fig. 14.5. Photomicrograph of the Li-Bi phase from the stripper at the LiCl interface, MTE-3.

108

Y-124404

+————500 m ———=

ThMa X-Rays FeKa X-Rays

Fig. 14.6. Electron beam scanning images of elements in the Bi-Th phase, MTE-3 contactor.

Y-124405

Fe Ko X-Rays Mn Ka X-Rays Lala X-Rays

~——— 500 pm ——

Fig. 14.7. Electron beam scanning images of elements in the Bi-Th phase at the fluoride salt interface, MTE-3 contactor.

ThMa X-Reys

109

Y-124403

"~ Fe Ko X-Rays

Fig. 14.8. Electron beam scanning images of elements in the Li-Bi phase at the LiCl interface, MTE-3 stripper.

and cffects of segregation during freezing will make
interpretation of results of these analyses very dif-
ficult. For instance, in the sample of LiCl from the
interface in the stripper vessel, no LiCl was detected.

Bismuth was detected in each of the salt samples
analyzed by x-ray diffraction. It is likely that during
freczing bismuth was forced into the already frozen
saltand was not present in the salt when both phases
were molten.

Petrographic examination was made of a sample of
the fluoride salt from the contactorata point near the
Bi-Th phasc and the tank wall. This examination
indicated that the sample was predominately crystals
of 3 LiF-ThFs. From the phase diagram of the 72-16-
12 mole % LiF-BeFa-ThFs, the 3 LiF ThFs would be
expected to crystalize from solution on cooling.

Samples of the Bi-Th, Li-Bi,and LiCl phases taken
at or near the salt-metal interfaces contained signifi-
cant amounts of iron. Significantly lower con-

centrations of iron were found in samples taken some
distance (2 to 8 in.) from the interface (Table 14.2).

Some black material was seen when sampling the
LiClin the stripper about | in. above the LiCl- Li-Bi

interface. We were able to separate some of this black
material (about 1/2 g) from the salt for analy
Chemical analysis of the material gave the following
composition (wt 04): bismuth, 52.7; thorium, 18.7;
chlorine, 4.1; lithium, 2.8 fluorine, 2.4; iron, 0.26;
and beryllium, 1 ppm

These constituents total 819 and leave 19% un-
accounted for. No other elements, excepting the
possibility of oxygen or water contamination, are
known to be present in the system. If one assumes
that the remainder is predominantly oxygen, this
amount would be sufficient to form oxides and/or
hydroxides of all of the cationic species (Bi, Th, Li,
Fe). The gram equivalent of these species is about
0.015 while the anion gram equivalent of iron and
chlorine is only 0.0025. For 19 wt ¥ oxygen in the
en gas equivalent is 0.024 or 0.011

is.

matcrial, the oxy
of hydroxide.

A scction of the Li-Bi~LiCl across the interfacial
arca in the stripper (which included the layer of dif-
ferent structure between the Li-Bi and LiCl) was
cxamined by scanning electron microscopy. Scan-
ning clectron photomicrographs at magnifications of

110

Table 14.1. Results of x-ray diffraction analyses of material removed from
vicinity of salt-metal interfaces in experiment MTE-3

Identified Approximate
Phase material composition(wt%)”
LiCl in stripper Bi 25-50
LisThE; 10-30
ThQ; 40-80
unidentified
solid solution
Li-Bi alloy in stripper Bi 7595
BeF; 2-10
Bi-Th in contact with LiCl Bi 90%
in contactor
LiCl in contactor Bi 30-70
LiCl 30-70
LisThF; 10-30
Fluoride salt in contactor Bi 2040
LisThF; 40-80
LisTheFa 5-15
Bi-Th in contact with fluoride Bi 60—90
salt in contactor Th (unidentified
compound)
Fe (no x-ray data trace
available)

*X-ray diffraction can only detect >5%.

Table 14.2. Iron in samples of metal and salt phases
from metal transfer experiment MTE-3 at various
distances from the metal-salt interfaces

Material Location Iron (ppm)
Contactor, LiCl side
Bi-Th ~1/8 in. from interface 2500
~2in, {from interface 18
~4-1/2 in. from interface 17
Contactor, fluoride salt side
Bi-Th ~1/4 in. from interface 1100
~2 in. from interface 37
~4 in. from interface 9
Stripper
Li-Bi <1/8 in. from interface 1400
~2 in. from interface 108
~5 in. from interface 7
Stripper
LiCl ~1/2 in. from interface - 650
~1 in. from interface 200
~6-3/4 in. from interface 115
~8-3/4 in. from interface 35
Contactor
LiF-BeF,-ThF. ~1/8 in. from interface 320, 88"

(72-16-12 mole %)

“Results of two different samples. All other results are for one sample.
20. 100, 500, and 2000X were taken of the interface
(Fig. 14.9). These photographs were taken of the
rough unpolished surface. No identifiable film or
forcign material can be seen at the interface in these
photomicrographs.
X-ray fluorescence analyses were made at five lo-
cations across the LiCl-Li-Bi interface and are
identified as 1, 2, 3, 4, and 5 on the 20X photo-
micrograph of Fig. 14.9. Results are given below:
1. In the LiCl some distance from Bi-LiCl bound-
ary, some Th and a small amount of Bi and La
were detected.

. In the LiCl near the Bi-LiCl boundary, very small
amounts of Th, Fe, and Bi were detected.

=

111

3. On the Bi-LiCl boundary Th and Fe as well as Bi
and CI were detected.

. In the Bi area near the Bi-LiCl boundary, large
amounts of Th, Fe, and Clas well as the Bi were
detected.

5. In the Bi area some distance from the Bi-LiCl

boundary, only Bi was detected.

The presence of iron (or iron oxides) at or near the
interface is consistent with the relative densities of
bismuth (9.6 g/cm’) and iron (7.6 g/cm’) or possibly
iron oxides (about 5.5 g/cm’) that would be expected
to come out of solution during cooling of the bismuth
phase.

ORNL DWG. 74-10116

MAG. 500 x

= |
- J,0.04 mm
S

Fig. 14.9. Scanning electron photomicrographs of the interface between LiCl and Li-Bi in the stripper from experiment MTE-3.
Unpolished

- Firm conclusions cannot be drawn about the re-
lationship of these observations to the low mass
transfer rates seen in experiment MTE-3. The
transfer of fluoride salt into the chloride salt just prior
to shutdown and the length of time between shut-
down and inspection (from February 1973 to
February 1974) introduce a great deal of uncertainty
into the interpretation of the analyses.

14.2 INSTALLATION OF METAL TRANSFER
EXPERIMENT MTE-3B

H. C. Savage

Wec plan to continue studies of the mass transfer
ratcs of rare ecarths between fuel carrier salt, bis-
muth, and lithium chloride in a new experiment
dcsignated MTE-3B. New vessels, salts, and bis-
muth will be used.

During this report period fabrication and as-
sembly of new carbon steel vessels was completed.
An oxidation resistant protective coating, METCO
No. P443-10, about 0.015 in. thick was applied to the
outsidc surfaces of the carbon steel vessels and
interconncecting lines to prevent air oxidation at the
opcrating temperature of 650°C. A new pump for
transfcrring fuel carrier salt between the salt reser-
voir and contactor was fabricated and installed. New
molybdcnum agitator shafts and blades have also
bcen obtained.

The vessels and their contents (fuel carrier salt,
-lithium chloride, and bismuth) previously used in
experiment MTE-3 were removed from Bldg. 3541
and scnt to the burial ground for disposal. The new
vesscls have been installed and we are renovating the
cxperimental area where the metal transfer ex-
periment is located. The renovation includes rerout-
ing some of the service lines to improve access to
sampling stations; relocation of some pressure gauges,
flowmeters, and valves for better visibility and
access; calibration and replacement (where re-
quired) of pressure gauges, flowmeters and solenoid
valves; and recalibration of temperature controller
and recorders.

Wc have completed tests on two different oxida-
tion-resistant protective coatings using a different
mcthod of application for each type coating. The
coatings were applied to test sections made from
longitudinal half-sections of 6-in. sched 40 mild steel
pipc. The test sections were coated on all exposed
surfaces. One test piece was coated with the nickel
aluminide material (METCO No. M405-10 wire) by

112

flame spraying with a wire gun to obtain a coating
thickness of ~0.010 in. duplicating the coating used
on the MTE-3 vessels. The second test piece was
coated with a nickel chromium alloy containing 6%
aluminum (METCO No. P443-10 powder) applied
with a plasma spray gun to a thickness of about0.010
in. as reccommended by the manufacturer. The two
picces were placed on fire bricks inside a furnace and
heated to the test temperature. During the test, the
picces were thermally cycled several times by cooling
the furnace to about 100°C, then returning to the test
tempcrature.

The initial test temperature was 700°C. After
about 400 hr at 700° C with eight thermal cycles the
picces were examined, weighed, and photographed
(Fig. 14.10a). The M405-10 coating was covered ex-
tensively with rust-colored oxidation product, but
there was no sign of spalling of the coating. The piece
coatcd with P443-10 had a much better appearance
with onc cdge showinga rust color. Weight gain of the
P443-10 test coated test piece was about 7.3 mg/cm”
whilc the¢ M405-10 coating gained about 10.3
mg/cm’.

The test temperature was then increased to 815°C
for cvaluation at a higher temperature. The test was
continucd for 520 hr at 815°C with eight thermal
cycles to 100°C. The test pieces were again
cxaminced, weighed, and photographed (Fig. 14.104).
The M405-10 coating had deteriorated to the point
where spalling had occurred, while the P443-10
coating had no spalling except along one edge where
thcrc was some rust-colored oxidation product and
somc spalling. Total weight gain of the M405-10
coated test piece was 69.2 mg/cm’ and 20.5 mg/cm’
for the P443-10 test piece. Metallographic
cxaminations were made of the best and worst
appcaring areas of the two specimens. In the best
appearing areas, both coatings were adherent, but a
thin layer of oxide formed at the metal-coating
intcrfacc of the specimen coated with METCO
M405-10. This indicates that oxygen diffused
through the coating. The metal-coating interface of
the specimen coated with METCO P443-10 was free
of visible oxide, indicating that it isa good barrier to
oxygen diffusion. For the worst areas of both
coatings, oxide at the metal-coating interfaces did
cause some spalling where the coating may not have
been thick enough to prevent oxygen diffusion
through the coatings. The oxide that formed
underneath the P443-10 coating (in a spalled area
occurring only along one edge) appeared to be more
dense and protective than that formed on the M405-
10 coated specimen. This indicates that some of the

113

ORNL DWG. 74-10117

NICKEL ALUMINIDE m MICKEL CHEOME ¢ 6% ALUMINUM

METCO AM03-10 METCO Pax3-10

e : m
(b)

Fig. 14.10. Photographs of 6-in. sched 40 mild steel pipe with oxidation resistant protective coatings: (a) after 400 hrat 700° C in air, (b)
after an additional 500 hr at 815°C in air.

elements in the P443-10 coating may have diffused
into the base metal and imparted a measure of
protection even though the coating had separated in
this area. : ,

This limited test clearly demonstrated the superi-
ority of the plasma spray P443-10coating over M405-
10 wirc gun sprayed material previously used for the
MTE-3 vessel, and the plasma spray coating has been
applicd to the external surfaces of the MTE-3B
vesscls that will operate at elevated temperatures.

14. 3 DESIGN OF THE METAL TRANSFER
PROCESS FACILITY

-.W._ L. Carter - H. C. Savage

Design of the metal transfer process facility in-
which the fourth metal transfer experiment (M TE-4)

will be carried out was under way when the MSR
program was terminated in 1973.° Briefly, MTE-4 is
ancnginecring experiment that will use salt flow rates
thatarc 5 to 109% of the rates required for processinga
1000-MWe MSBR. Conceptual designs of the three-
stage salt-metal contactor, made of graphite, and its
containment vessel were completed.” The primary
purposcs of MTE-4 are: '

I. Demonstration of the removal of rare-carth fis-
sion products from MSBR fuel carrier salt and
accumulation of these materials in a Li-Bi solu-
tion in equipment of a significant size;

. Dctermination of mass transfer coefficients be-

_ tween mechanically agltated salt and bismuth
phases;

earths from the fluorlde
. equipment; , : :
. Evaluation of potential materlals of construction,
\"graphlte in particular;- :

. Testing of mechanical devices, such as pumps and

salt n mul_tlstage

agitators, that will be required in a processing .

plant; and

. Development of instrumentation for measure-
ment and control of process variables such as salt-
metal interface location, salt flow rate, and salt
or bismuth liquid level.

We are currently reviewing the design of metal
transfer experiment MTE-4. A mathematical model

8. Part-time.
9. W, L. Carter et al., MSR Program Semiannu. Progr. Rep.
Feb. 29, 1972, ORNL-4782, pp. 224-25.

. Determination of the rate of removal of rare -

114

of thc system has been devised, and a computer
program-has been written to simulate transient
opcration of the experiment. Computations can be
madc to determine the concentration of each nuclide
being transferred at each stage and at the feed and
receiving vessels as a function of operating time. The
program allows the experimenter to make parametric
studics for such design features as interfacial area,
number of stages, flow rates, agitator speed, and
inventorics of materials. The program (METTRAN)
is being used to analyze the MTE-4 experiment to
asccrtain the significance of various design features
on mctal transfer rates in order to fix optimal design

: .condltlons

‘Duc. to space limitations in Bldg. 4505 we are
planmng to locate MTE-4, along with several of the
engineering experiments on molten-salt processing
in thc MSRE Building (7503). General cleanup and

" checkout. of existing building services (electrical

circuits, ventilation, and air filtration systems) is
under way. A 480-V, 3-phase, 60 Hz, 300-kW diesel
generator will be installed in the existing generator
building at the MSRE site to provide emergency
power for maintaining portions of engineering ex-
pcriments  which contain salt or bismuth at
tempcratures above the liquidus temperatures. A
purchase order for the generator set has been issued
and thc system design has been completed.

14.4 SALT-METAL CONTACTOR
DEVELOPMENT: EXPERIMENTS
WITH A MECHANICALLY AGITATED
NONDISPERSING CONTACTOR IN THE

SALT-BISMUTH FLOW-THROUGH FACILITY

J. A Klein C.H. Brown,Jr. J.R. Hightower, Jr.

Mechanically agitated nondispersing contactors

are being considered for extracting protactiniumand

the rare-earth fission products from the fuel—carrler

“salt of a molten-salt breeder reactor into molten

bismuth that contains dissolved reductant.
Mass transfer rates were measured in a mild steel

' contactor that was fabricated and installed in the

Salt-Bismuth Flow-through Facility. This ex-

perimental system allows (1) periodic purification of

the feed salt and metal, (2) removal of surface
contamination from the salt-metal interface, and (3)
variance of the distribution ratio of the material
interest between the salt and bismuth.

The new stirred interface contactor uses the
existing piping equipment and instrumentation pres-
ent in the Salt-Bismuth Flow-through Facility that
was used to study salt-bismuth flow in packed
columns.'® ™ A simplified flow diagram of the facility
is shown in Fig. 14.11. The facility consists of five
vessels, the associated piping, and the contactor
assembly. The vessels consist of a salt-feed tank, a
salt-collection tank, a bismuth-feed tank, a bismuth-
collection tank, and a vessel (the treatment vessel, T-
5) for sparging both salt and metal with mixtures of
HF and H.. To conserve space, feed and collection
tanks for both salt and metal are concentric tanks.
Feed and collection tanks and piping are made of
low-carbon steel, and the treatment vessel is made of
stainless steel and has a graphite liner. .

A diagram of the contactor is shown in Fig. 14.12.
The contactor consists of a 6-in.-diam low-carbon
steel vessel that contains four l-in.-wide. vertical
baffles. The agitator consists of two 2 7/8-in.-diam
stirrers with four 3/4-in.-wide noncanted blades. A
3/4-in.-diam overflow at the interface allows the
removal of mnterfacial films with the salt and metal
effluent streams. Salt and bismuth are fed in the
contactor below the surface of the respective phase.
The system was operated in essentially the same
manner as that employed with the packed
column.'™ The salt and bismuth phases were
equilibrated prior to an experiment. After transfer of
the salt and metal phases to the feed tanks, *’Zr and
#7U tracers were added to the salt. With this
technique, the rates at which °~'Zr and *’U tracers
transfer from the salt to the bismuth could be
measured in a system that was otherwise at chemical
equilibrium. _

The experimentally determined data from the
system- are sufficient to allow three independent
expressions for the overall mass transfer coefficient
to be derived for the contactor in the Salt-Bismuth
Flow-through Facility from steady-state material
balance relationships.'” These expressions for the

10. B. A. Hannaford, C.- W. Kee, and L. E. M¢Neese, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1971, ORNL-4676,
p. 256,

Il. B. A. Hannaford et al., MSR Program Semiannu. Progr.
Rep. Aug. 31,1971, ORNL-4728, p. 212.

12. B. A. Hannaford, C. W. Kee, and L. E. McNeese,
Engincering Development Studies for Molien-Salt Breeder Reac
tor Processing No. 8, ORNL-TM-3258, p. 64 (May 1972).

13. B. A. Hannaford, C. W. Kee, and L.. E. McNeese, Engi-
neering Development Studies for Molien-Salt Breeder Reactor
Processing No. 9, ORNL-TM-3259, p. 158 (Dec. 1972).

14. B. A. Hannaford, Engineering Development Studies for
Molien-Salt Breeder Reactor Processing No. 10, ORNL-TM-
3352, p. 12 (Dec. 1972). '

115

ORHL OWG 74—6594R1

EtReam
MUT
SAMPLER ./ BiTRent,

SAMPLER

------- l' FV—1 Y
——— Fv-2
...-.—.-{ -— ——— ——
Y F ] i ' Fv-3 _?
! Vo tam e —_——
: LI Pvea R
i ! ;
! i
: 3 1
! i R Vo
H T I )
ool ; P d
] ' 1 }
2! 110 | T417-3; 1 T-5 |
vl ' \lfi H

SALT AND METAL
TREATMENT VESSEL

METAL FEED AND
COLLECTION TANK

SALT FEED AND

CONTACTOR _COLLECTION TANK

Fig. 14.11. Flow diagram of the Salt-Bismuth Flow-through
Facility with the mechanically agitated contactor installed.

GRNL DWG 74-10073

A-Bismuth |nlet
B-Satt Inlet

C-Four |-in. Baffles
D-2-7/8-in. x 3/4-in. Blades
E-Gas-Salt Interface

F -Bismuth - Salt Interface
G-Interface. Removal

H-Sait Effluent

I-Bismuth Efftuent

- Fig. 14. 12. Diagrarh of the mechanically agitated nbndispersing
contactor installed in the Salt-Bismuth Flow-through Facility.

overall mass transfer coefficient are given below in
terms of the measured quantities Cy, Cs, Cm, F1, F2, D,
and A: :

K = H[—(Cj
SA(C C) A A DNC] COF Y F)y— (A DY F)

)

— ) F?. (Cm/C])
" A= ACu] CYF2f F)— (Cu/C XA/ D)’

K (2)

150 0 AL Klein, Engineering Developnient Studies for Molien-
Salt Brecder Reactor Processing No. 18, ORNL-I'M-698 (in
prepatition).
and
K. = F> (Cn/ Cy) - 3)
U A (Cu/C)AI D)
where »
F1 = flow rate of salt, cm’/sec,
F, = flow rate of metal, cm’/sec,
C, = tracer concentration in salt inflow,
units/cm’,
Cs = tracer concentration in salt outflow,
units /cm’,
C., = tracer concentration In metal outflow,

. 3
units/cm”,

. . 2
A = interfacial area, cm”,

D = distribution coefficient = ratio of concen-
tration in metal phase to concentration
in salt phase at equilibrium, (moles/cm’)/
(moles/cm’).

The subscripts 1, 2, and 3 on K| indicate the equation
used to evaluate the overall mass transfer coefficient
K. The overall mass transfer coefficient is related to
the mass transfer coefficients in the individual phases
through the relation

1/Ks=1/k, + 1/(Dk,), (4)

where

K = overall mass transfer coefficient based on salt-
phase concentrations, ¢m/ sec,

k, = individual mass transfer coefficient in salt

phase, cm/sec,

km = individual mass transfer coefficient in metal
phase, cm/sec.

Within experimental error, the distribution coef-
ficient D can be set as desired. To minimize effects of
uncertainties in the value of D on the calculated value
of the overall mass transfer coefficient, D should be
made fairly large. For the values of concentrations
and flow rates used in these experiments, the terms
which contain D in Egs. (1), (2), and (3) are less than
5% of the values of the other terms for values of D
greater than 20 and can be neglected with little error.
By assuming that the terms that contain D can be
neglected, Egs. (1), (2), and (3) reduce to

]

F

A

1—(Cs/C))

K, =
(Cs/Ch)

)

116

. _fg (Cm/cl)

KS'_A ]_(FE/FJ)(Cm/Ci)} | ©)
and

K, = (F1/A) (C/Cy) . (7

Uncertainties in the distribution coefficient do not

affect the accuracy of the overall mass transfer
coefficient. However, as shown by Eq. (4), when D is
very large, the overall mass transfer coefficient is
essentially the individual salt-phase coefficient, since
resistance to mass transfer in the metal phase is
negligible in comparison. .

Six runs have been completed to date, and all runs
were performed by using the same procedure. While
the fluoride salt and bismuth are in contact in T35, the
treatment vessel, sufficient beryllium is added elec-
trolytically to the salt to produce the desired distribu-
tion coefficient, D. Since complete exclusion of
oxidants is impossible, berylliuimn must be added
periodically to maintain a relatively high distribution
coefficient. Salt and bismuth are transferred
periodically throughout the system to keep the
system in equilibrium.

Before a run, the salt and bismuth phases are
separated by pressurizing the salt-bismuth purifica-
tion vessel and transferring salt and bismuth to their
respective feed tanks. Approximately 7 mCi of *'Zr-
*’Nb and 50 to 100 mCi of *’U;0s are allowed to
dissolve in the salt phase about 2 hr prior to an
experiment.

Salt and bismuth streams are passed through the
contactor vessel at the desired flow rates by con-
trolled pressurization of the salt and bismuth feed
tanks. The contactor is maintained at about 590°C
for all runs. Both phases exit through a common
overflow line, separate, and return to the salt and
bismuth catch vessels. Both exit streams are sampled
periodically by means of flowing-stream samplers
installed in the exit lines from the contactor.

Samples were analyzed by first counting the
sample capsules for the radioactivity of 27U (207.95
keV B) and the radioactivity of 7Zr-*’Nb (743.37
keV and 658.18 keV B, respectively) after secular
equilibrium was reached between “’Zr and its
daughter, *’Nb. The material in the sample capsules
was then dissolved and the radioactivity of **’U was
counted again after the *’Zr-"’Nb radioactivity had
decayed to a very low level. This procedure was
followed to correct for self-adsorption in the solid
samples. Mass transfer rates were then calculated
from the ratios of tracer concentrations, as discussed
previously.
Results of measurements of overall mass transfer
coefficients are shown in Table 14.3. Three different
equations for calculating mass transfer coefficients
were used to determine the results for any one run,
and the results are presented as an average of three
calculated values with the standard deviation. Values
are given both for results based on the uranium
counting data and for results based on the zirconium
counting data. _

Run TSMC-1 was mainly a preliminary run
designed to test the procedure. Salt and bismuth
flows were about 200 cm’/min, and the stirrer rate
was 123 rpm. Unfortunately, the distribution coef-
ficient was too low to effect any significant mass
transfer; thus, mass transfer rates could not be
determined accurately; thus, Table 14.3 does not
include results for this run.

During run TSMC-2 the equipment operated very
smoothly. The salt and bismuth flow rates were 228
and 197 cm’/min, respectively. The distribution
coefficients were higher than for the previous run but
were lower than desired with a concomitant large
degree of uncertainty. Several determinations of Dy
were made ina range of 0.94 to 34. One determination
of Dz gave the value 0.96. Consequently, only a range
of possible values for the overall mass transfer
coefficient could be stated for the results based on
uranium. A value for overall mass transfer coefficient
based on zirconium is given, but, since the value for
Dz must be considered to be uncertain, more
uncertainty exists in the mass transfer coefficient
than is indicated by the reported standard deviation
given in Table 14.2.

If one assumes that the ratio of salt-side mass
transfer coefficient to bismuth-side mass transfer
coefficient can be determined from the Lewis correla-
tion, then for the results based on zirconium, the salt-
side mass transfer coefficients for run TSMC-2 are
higher than the overall mass transfer coefficient by a
factor of 1.55.

117

A bismuth-line leak occurred immediately pre-
ceding run TSMC-3. During the resulting delay
for repairs, the *’Zr decayed and only the **’U tracer
could be used. The remainder of the run went
smoothly. Salt and bismuth flow rates were 166 and
173 cm®/min, and the stirrer rate was 162 rpm. Ahigh
value for Dy (greater than 34) was maintained for this
run.

In run TSMC-4, flow rates of 170 and 144 cm’/ min
were set for the salt and bismuth flows, and a stirrer
rate of 205 rpm was maintained. The distribution
coefficients as determined from samples taken
before, after, and during the run were greater than
172 for Dy and greater than 24 for Dz,. This value is
large enough to make Egs. (5), (6), and (7) valid.
Large distribution coefficients cannot be determined
precisely due to the inability to determine very small
amounts of uranium in the salt phase. No problems
arose during this run.

Runs TSMC-5 and TSMC-6 were performed
without incident. The distribution coefficients were
maintained at high levels for both runs. Run TSMC-5
had a stirrer rate of 124 rpm and salt and bismuth
flows of 219 and 175 ¢cm’/min. Salt and bismuth
flows for run TSMC-6 were 206 and 185 cm’/ min,
respectively, with a stirrer rate of 180 rpm.

As mentioned previously, when the distribution
coefficient is large, the overall mass transfer coef-
ficient is essentially the individual salt-phase mass
transfer coefficient. Thus, results for runs TSMC-3
through TSMC-6 can be compared directly with the
Lewis correlation. The Lewis correlation’ for mass
transfer in the nondispersing contactor is

60k1/vi = (6.76 X 107°) [Re, + Rez (n2/ m)]"* + 1
(®)

where

k = individual mass transfer coefficient, cm/sec,
v = kinematic viscosity (/p), cm?/sec,

Table 14.3. Experimental results from the salt-metal contactor

Stirrer Fraction of | K, (cm/sec)
Salt flow Bismuth flow rate tracer
Run” (cm”/ min) {em’/ min) (rpm) Dy Dz, transferred” Based on U Based on Zr
TSMC-2 228 197 121 0.94 10 34 0.96 0.17 0.0059 to 0.0092 0.0083£0.0055
TSMC-3 166 173 162 >34 0.50 0.012 £ 0.003
'SMC4 170 144 205 >172 >24 0.78 0.054 £ 0.02 0.035 £ 0.02
TSMC-5 219 175 124 >43 >24 0.35 0.0095 = 0.0013 0.0163£0.0159
=24 0.64 0.049 = 0.02 0.020 = 0.01

TSMC-6 206 185 180 >172

“T'urbulent salt-metal contactor.
hy- . . . N .
Iraction of tracer transferred = 1 - ¢, ¢,
n= v1sc051ty, p01ses

p = densnty, g/cm

Re = nd’ /v, dimensionless,

d = agitator diameter, cm,

‘n = agitator speed, rps.

The subscripts 1 and 2 refer to salt and bismuth,
respectively. Fig. 14.13 shows a comparison of the
measured values of mass transfer coefficient with the
Lewis correlation. The. effects of mass transfer
resistance in the bismuth phase cannot be accounted
for accurately in the results of run TSMC-2; thus,
these results are not included in Fig. 14.13.

Figure 14.13 shows that two runs produced values
of mass transfer coefficients that included 30 to
40% of the values predicted by the Lewis corre-
lation, whereas two runs produced values slightly
greater than those predicted by the Lewis correlation.
The mass transfer coefficients based on zirconium are
consistently lower (though only slightly) than the
values based-on uranium. The mass transfer coef-
ficients based on zirconium are considered to be less
reliable than those based on uranium, since in all runs
only about 609%, of the "7Zr tracer could be accounted
for, whereas more than 809 of the.?>’U tracer could

ORNL DWG T4-6833IR)}

10000 —T— T T T T T T TR
= LEWIS -
- CORRELATION?” ]
[ S
1000~ ’/ —
: / E
Tt ,/ .
ols 100k / -
© S .8’ EXPERIMENTAL VALUES 3
- Zg FROM MERCURY-WATER 7
o ‘ SYSTEM ]
i & EXPERIMENTAL VALUES ]
FROM ORGANIC -WATER

SYSTEMS =
10 - —
® BASED OCN U -
= A BASED ON Zr -
Io L 1 l'lIIIII i 1 Illllll L J L 1L 1li)

10 IOOO"7 000

[6.76 x107%] [Re; + Rez 22 ]'-6"’
Fig. 14.13. Experimental results from the salt-bismuth contac-
“tor. '

118

be accounted for. This discrepancy is probably
related to self-absorption of the 743.37 keV 8 from

"Zr in the bismuth samples.

We believe that at a stirrer speed of 160 to 180 rpm,
entrainment of salt into the bismuth begins to occur,
and the apparent increase in mass transfer coefficient
is a manifestation of an increase in the surface area
for mass transfer due to surface motion. Experiments
with water-mercury and water-methylene bromlde
systems support this belief.

If, as Fig. 14.13 indicatés, the mass transfer
coefficients in the salt-bismuth system are lower than
those predicted from the Lewis correlation by a factor
of about 0.35, this corrected correlation could be used
to estimate the area and the bismuth flow rate
required in the rare-earth-removal contactor in the
present flowsheet. The Lewis correlation predicts
that mass transfer coefficients are approximately
proportional to the agitator diameter, 4, raised to the
3.30 power. Since studies show that the allowable
agitator speed below which no phase dispersal occurs
is inversely proportional to d"* (ref. 16), at speeds
slightly below the limiting agitator speed, the mass
transfer coefficient appears to be proportional to
d"**. These results were used to estimate the area
requ1red for a single-stage contactor and for the
bismuth flow rate to remove cerium, which has the
shortest removal time (16.6 days) of the rare earths in
the reference flowsheet. For a salt flow rate of 0.88
gpm (which corresponds to the 10-day cycle time used
in the reference processing flowsheet) and for a
distribution coefficient of 0.062 (the distribution
coefficient assumed in the reference processing
flowsheet), a mass transfer area of 10 ft? will allow
removal of cerium on a 16.6-day cycle if the bismuth
flow rate through the contactor is 30 gpm and if a 3-ft-
diam agitator is used. This bismuth flow rate is only
about 2 1/2 times the flow rate specified in the -
processing plant flowsheet on the basis of equilibrium
calculations. For example, use of a higher distribu-
tion coefficient and of multiple stages can result in a

“reduction of-either the bismuth flow rate or the mass

transfer area. However, these changes will result in
changes in other areas of the flowsheet, and more
extensive calculations are required to evaluate these
effects.

16. H. O. Wceren and L. E. McNeese, Enginéert’ng
Development Studies for Molten-Salt Breeder Reactor Processing
No. 10, ORNL-TM-3352 (December 1972), p. 53.
14.5 SALT-METAL CONTACTOR
DEVELOPMENT: EXPERIMENTS
WITH A MECHANICALLY AGITATED
NONDISPERSING CONTACTOR USING
WATER AND MERCURY

C. H. Bfown, Jr.

A critical part of the proposed MSBR processing
plant is the extraction of rare earths from the
fluoride-fuel-carrier salt to an intermediate bismuth
stream. One device being considered for this extrac-
tion is a mechanically agitated nondispersing contac-
tor in which bismuth and fluoride salt phases are
agitated to enhance the mass transfer rate of rare
earths across the salt-bismuth interface. Previous
reportsls’lj’18 have shown that the following reaction
in the water-mercury system is suitable for simulating
and for studying mass transfer rates in.systems with
hlgh density differences:

Pb>*(H.0) + Zn(Hg) — Zn*'(H0) + Pb(Hg) . (9)
A large amount of data has been reported'® for the
water-mercury system under the assumption that the
limiting resistance to mass transfer exists entirely in
the mercury phase, as suggested by literature cor-
relations. During this report period, a series of
experiments was performed in the water-mercury
contactor to determine which phase actually controls
the rate of mass transfer and also at what concentra-
tion of Pb*" the control of mass transfer changes from
one phase to the other.

14.5.1 Theory

The reaction under consideration, Eq. (9), is a
liquid-phase ionic reaction that occurs entirely at the
mercury-water interface, since zinc metal and lead
metal are insoluble in water and no ionic lead or zinc
can be in the mercury. Because this is a typical ionic
reaction, it is assumed to be essentially instantaneous
and irreversible. The equilibrium constant for the
reaction is given by the equation

K =Cpy Czp2+/Cpp2+ C (10)

o JoACKlein, Engineering Development Studies for Molien-
Salt Breeder Reactor Processing No. 14, ORNL-TM-4018 (in
preparation). '

18. J. A Klein, Engineering Development Studies for Molien-
Salt Breeder Reactor Processing No, 15, ORNL-TM=-4019 (in
preparation),

119

where
K = equilibrium constant,

Ce» = concentration of Pb metal in mercury,
gmole/liter,

Crp?* = concentration of Pb*"

Czn concentration of Zn metal in mercury,
gmole/liter,

in water, gmole/liter,

Cz?* = concentration of Zn*" in water gmole/llter

The equilibrium constant is very large, which implies
that at equilibrium the ionic lead and metallic zinc
cannot coexist at appreciable concentrations at the
interface. Since the equilibrium reaction is assumed
to be an extremely fast reaction, it must be satisfied at
all times near the interface.

For the instantaneous irreversible reaction dis-
cussed previously, two situations could occur near
the liquid-liquid interface, depending on the relative
magnitudes of mass transfer coefficients of the
individual phases and on the bulk phase concen-
trations of the transferring species in each phase.'
Figure 14.14 illustrates these two conditions.

In Fig. 14.14a, the limiting resistance to mass
transfer is assumed to occur in the mercury phase.

Equating the tranfer rate of equivalents of zinc to the
interface to the transfer rate of equivalents of Pb*" to

19. 0. Levenspiel, Chemical Reaction Engincering, 2nd ed., p.
387, John Wiley & Sons, Inc., New York, 1962,

ORNL DWG. 74-1010

e
Lt
EE
™
v @
ey
a2
5 .
i CF'b",B
I
CZn'B i (a)
]
MERCURY Cp"i | WATER
i 1
2 1
L—W_Lfl__._._}
INTERFACIAL FILMS
POSTULATED BY LEWIS AND WHITMAN
CZn.B : }I
! ! Cee'ie  (b)
MERCURY : : WATER
| |
! ~ CF'b"l:

Fig. 14.14. Interfacial behavior for an instantaneous irreversi-
ble reaction occurring between two liquid phases. () Mercury
phase controlling mass transfer. () Water phase controlling mass
transfer.
the interface can show that the product of the bulk-
phase concentration of reactant and the individual-
phase mass transfer coefficient in the phase where the
limitation occurs must be less than the product of the
bulk-phase concentration of the other reactant and
the individual-phase mass transfer coefficient in the
other phase. The concentration of zinc in mercury
near the interface decreases from the bulk concentra-
tion to near zero at the mercury-water interface. The
concentration at the interface is very small because of
the instantaneous irreversible reaction that occurs at
the interface. At the interface in the water phase, the
concentration of Pb’" has a finite value and increases
through the interfacial film to the bulk-phase value.

Figure 14.14b illustrates the condition under which
the limiting resistance to mass transfer is assumed to
occur in the water phase. The explanation of the
concentration gradients in the interfacial films is
entirely analagous to the case explained above. The
concentration profiles of lead in the mercury, Zn* in
the water, and NO; (the anion) in the water are not
shown in Fig. 14.14.

Several correlations have been developed and
presented in the literature for predicting individual-
phase mass transfer coefficients in nondispersing,
stirred interface contactors. For the mercury-water
system, all these correlations predict that the
mercury-phase mass transfer coefficient would be
smaller than the water-phase coefficient.

In all previous work done with the mercury-water
system,m‘” the concentrations of the reactants were
equal. This condition, coupled with the fact that the
mercury-phase mass transfer coefficient was
predicted to be significantly smaller than the water-
phase coefficient, indicated that the limiting
resistance to mass transfer should occur in the
mercury phase.

To test the assumption that mass transfer is
controlled by the mercury phase, we can write the
following relations for transfer of the reactants from
the bulk phase to the interface where they react, based
on the two-film representation shown in Fig. 14.14:

NZn = ngA(CZn,B - CZn,i) s (ll)
Npp2+ = kH2 0A(Cpy2+ g — Cpp2+ ) (12)
where
k = individual-phase mass transfer coefficient,
cm/ sec,

N = rate of mass transfer to the interface, g/sec,
A = interfacial area, cm’,

120

C = concentration (B denotes bulk-phase concen-
tration, 1 denotes interfacial concentration),
g/ cm’

The subscripts Hg and H:O indicate the phase being
considered. As stated above, we assumed that the rate
at which reaction (11) proceeds is controlled by the
rate of transfer of zinc through the mercury phase to
the interface. The conditions necessary to the validity
of this assumption are:

1. The equilibrium constant for the reaction must be
large (i.e., the reaction should be irreversible).
The product of the mass transfer coefficient times
the bulk concentration of reactant in the phase
where the rate of mass transfer is limiting must be

less than that product in the other phase.

2.

Since 1 mole of Zn is equivalent to 1 mole of Pb*
according to Eq. (1), Nev2*= Nz.. By substituting Egs.
(11) and (12) into this expression and by assuming
that the controlling resistance is in the mercury phase,
the following expression is obtained for the apparent
mercury-phase mass transfer coefficient:

Kyg.a = [kHQO(CPb%,B - CPb2+,i)/CZn,B] (13)
where
kug 4 =apparent individual-phase mass transfer

coefficient for the mercury phase, cm/sec,
ku,o=true individual-phase mass transfer coef-
ficient for the water phase.

In Eq. (13), the actual mass transfer coefficient differs
from the apparent mass transfer coefficient only
when the mercury phase does not control the rate of
mass transfer. The transient method used to deter-
mine the mass transfer coefficient has been described
pre,vious]y.16

From an argument similar to the one that leads to
the development of Eq. (13), an equation for the
apparent aqueous-phase mass transfer coefficient can
be written for the case where the limiting resistance 1s
in the water.

The concentration of Pb?* in the water, below which
the limiting resistance to mass transfer is in the water
(at a fixed concentration of zinc in the mercury), can
be determined as follows. If the Pb®* concentration is
sufficiently high, the limiting resistance to mass transfer
will be in the mercury, and the interfacial concentration
of Pb?* and Cp2+; will have a finite value. If Cp 2+
is lowered by an amount A (not large enough to cause
the limiting resistance to change phases), Eq. (13)
indicates that Cpy,2+; must also decrease by the same
amount A, which would keep the difference Cpy 2+ g —

Cpp2+; constant. The water-phase mass transfer coeffi-
cient, although the value for which is not known, is
assumed to remain constant. If Cpyo+ p is reduced
sufficiently, Cpp2+; will decrease to zero, and if
CPb2+,B ‘is further reduced, Cp,2+; will remain zero,
and the limiting resistance will change from the
“mercury phase to the water phase. At high values of
Cpp2+ g, the apparent mercury-phase mass transfer
coefficient will remain constant-as Cp 2+ p is lowered
to the point where the limiting resistance moves into
the water phase. As Cpy 2+ p is further reduced, Cpy2+
will be zero or very nearly zero, and Eq. (13) shows
that the apparent mercury-phase mass transfer coeffi-
cient will vary directly with Cpp2+ 5, since Cy, 5 is
fixed, and the true water-phase mass transfer coefficient
is assumed to be constant. Figure 14.15 shows this
dependence of the apparent mercury-phase mass trans-
fer coefficient on the concentration of Pb®" in the
water. The dependence of the analogously defined
apparent water-phase mass transfer coefficient is also
shown. Thus, by calculating the apparent mercury-
phase mass transfer coefficient as a function of the
initial aqueous-phase lead concentration for a single
agitator speed, the transition from mercury-phase
controlling to aqueous-phase controlling should be

identifiable by a change in slope of the line that is -

determined by plotting the apparent mercury-phase
mass transfer coefficient vs the initial aqueous-phase
lead concentration. '

ORNL DWG. 74-1011

MERCURY PHASE
CONTROLLING

/iuq,n

WATER PHASE
CONTROLLING =-—{—»

Hy0, A

TRANSFER COEFFICIENT

APPARENT INDIVIDUAL PHASE MASS

INITIAL AQUEOUS PHASE LEAD CONCENTRATION

Fig. 14.15. Theoretical relationship between the apparent mass
transfer coefficients and the initial aqueous-phase lead concentra-
tion.

121

14.5.2 Experimental Apparatus

A series of mass transfer experiments was per-
formed in a mercury-water contactor to determine
the point at which control of mass transfer changes
from the mercury phase to the aqueous phase. The
experimental apparatus used in the study is shown
schematically in Fig. 14.16. The equipment consists
of a 5- by 7- by 10-in. (height) Plexiglas vessel that
contains two phases, each 3 in. deep. An agitator
shaft to which two 4-vaned, flat paddles are attached
is suspended from a variable-speed motor. The
paddles are positioned at the midpoint of each phase.

In this study, the two phases were 1.8 liters of clean
mercury that contained 0.1 M Zn and 1.8 liters of
distilled water that contained 0.02 to 0.10 M
Pb(NO3)..

14.5.3 Results and Conclusions

Five experiments were run to measure the apparent
mercury-phase mass transfer coefficient asa function
of the initial aqueous-phase lead concentration. The .
initial mercury-phase zinc concentration was held
constant at 0.1 M. Phase volumes and agitator speed
were also held constant at 1.8 liters and about 150
rpm, respectively.

The experimental results are presented graphically
in Fig. 14.17. The apparent mercury-phase mass
transfer coefficient decreased for all initial lead
concentrations less than those used in previous
experiments. This decrease indicates that under these
conditions, the limiting resistance to mass transfer
does notappear to be in the mercury phase. However,

ORNL DWG. 74-10112

] VARIABLE SPEED DRIVE
MOTOR

ROTATING AGITATOR SHAFT

L——— H,0

AGITATOR PADDLES <]

{CENTERED IN .EACH
PHASE)

5in. x7in. PLEXIGLAS
CONTACTOR

Fig. 14.16. Schematic diagram of the Plexiglas contactor used
for the mercury-water system,
ORNL DWG. 74-10109

T I 1 1
V- APPARENT MERCURY
. 0010} PHASE MASS .
g% TRANSFER COEFFIC-
4 IENT
(o) O- APPARENT AQUEOUS
- PHASE WASS TRANS-
z FER COEFFICIENT
o i
S 0.008f
I
W
w
o
(&)
x 0.006} J
w
[l
[72]
z
<X
[1 4
" o004l
[/)]
[73]
L4
=
[m]
w 0002} —~
14
@ v
=
w
=
0.0 - H 1 L
0.0 002 004 06 ocs 040

moles
liter

INITIAL AQUEOUS PHASE LEAD CONCENTRATION,

Fig. 14.17. Experimental results showing the apparent
mercury-phase transfer coefficients as a function of the initial
aqueous lead concentration at constant agitator speed.

based on the one point at the lowest Pb”* concentra-
tion, the apparent aqueous-phase mass transfer
coefficient does not seem to remain a constant, as
would be predicted by the model described above.
This point must be clarified by additional data.

From these results, the conclusion can be drawn
that, for the conditions under which these and all
other runs were performed, the resistance to mass
transfer was not in the mercury phase, as previously
believed. Further studies are needed to test other
assumptions used in analyzing results from the
transient mass transfer equipment,

14.6 CONTINUOUS FLUORINATOR
DEVELOPMENT: AUTORESISTANCE
HEATING TEST AHT-3

J. R. Hightower, Jr. R. M. Counce

A nonradioactive demonstration of frozen salt
corrosion protection in a continuous fluorinator re-
quires a heat source in the molten salt which is im-
mune to attack by the gaseous fluorine. We have
previously demonstrated that autoresistance heating
of the salt is feasible,” and we have designed and
built a fluorinator mockup to test a design for an

electrode side arm in which the electrode does not
contact the fluorine stream.”'

Five experiments (experiments AHT3-1 through
AHT3-5) were run using this equipment before the
Molten-Salt Reactor Program was interrupted be-
tween January 1973 and January 1974. This equip-
ment remained idle but intact during the interrup-
tion of the program. The equipment has been put
back into operating condition, and tests of autore-
sistance heating have resumed.

14.6.1 Experimental Equipment and Procedure

A simplified flow diagram of the equipment used
in the autoresistance heating tests is shown in Fig,
14.18. The equipment, which is located in cell 3 of
Bldg. 4505, consists of a feed tank, a test vessel, a
vessel into which salt remaining in the sidearm can
be drained, and a 60-Hz, 100-V, 100-A power sup-
ply to provide the necessary internal heat gen-
eration.

The test vessel (Fig. 14.19) is made from sched 40
nickel pipe. This vessel is essentially a shortened
version of the fluorinator which would be used in a
proposed continuous fluorinator experimental facil-
ity. The vertical portion of the test section (corre-
sponding to the fluorination zone) is 3 ft long. An
angled gas inlet, which can be protected from cor-
rosion by a frozen salt layer, is provided at the bot-
tom of the section. The side arm is also made from

20. J. R. Hightower, Jr.. Engineering Design Studies for
Molien-Salt Breeder Reactor Processing No. 16, ORNL-TM-4020
(in preparation).

21, J. R. Hightower, Jr.. Engineering Design Stucdies for
Molten-Salt Breeder Reactor Processing No. 17, ORNL-TM-4178
{in preparation).

ORKL DNG. 73-2447

OFF-
ARGON — et td— aVACUUM GAS  ARGON—ep3——t—e VACUUM
AUTORESISTANCE
ARGON HEATIN SECTION OF
POWER SUPPLY PLASTIC_TUBING
FOR ELECTRICAL

ISOLATION

~|~ELECTRICALLY
INSULATING
GASKET

ot
DRAIN
TANK
ELECTRICALLY
INSULATING
BASE

Fig. 14.18. Flow diagram of equipment for autoresistance
heating tests.

FEED "
TANK

PHOTO 3431-72

Fig. 14.19. Photograph of the test vessel used in autoresistance
heating test AHT-3.

6-in. sched 40 nickel pipe so that the autoresistance
heating current density in the side arm will equal
that in the fluorination section.

A diagram of the upper electrode is shown (Fig.
14.20). The electrode is made from 1/2-in. Inconel
tubing and also serves as the dip tube for draining
the side arm. The electrode is inserted into the ver-
tical portion of the side arm and is insulated from
the test vessel by means of a Teflon gasket. Car-
tridge heaters are provided in the electrode to keep
the salt molten in the vertical portion of the side
arm.

At the beginning of an experiment, molten salt is
transferred from the feed tank and from the side
arm drain tank into the test vessel. The test vessel
is maintained initially at a temperature above the

liquidus temperature (505°C) of the salt mixture
[LiF-BeFz-ThFs (72-16-12 mole %)]. The insulation
is then removed from the test section. and the wall
of the test section is allowed to cool by radiation
and convection to the cell environment. Frozen salt
forms on the vessel wall during this period. When
the wall reaches the desired temperature, the auto-
resistance heating current is turned on, and the
heating rate is adjusted until steady-state conditions
are reached. The attainment of steady state is deter-
mined by the condition that the wall temperatures
in the test section must remain constant at some
value below the salt liquidus temperature with a
molten zone throughout the test section. After the
equipment is maintained at the desired conditions
for the desired time period, the feed tank and the
side arm drain tank are evacuated and the molten
salt is removed from the test vessel. leaving the
frozen material on the wall. The equipment is then
cooled to room temperature and opened, and the
frozen salt layer is examined.

ORNL DHG. 73-2448

~= TO SIDE ARM
(7 DRAIN TANK

NOZZLE FOR SIDE

ARM OFF-GAS:
LUG FOR
ELECTRICAL
FEANGE F(counecnon
2-in NICKEL
PIPE———_|

-COPPER INSERT
CONTANING CARTRIDGE
HEATERS

=T

V2-in. INCONEL
" Tuse

Fig. 14.20. Diagram of upper electrode used in autoresistance
heating test AHT-3.

14.6.2 Experimental Results

During this report period experiments AHT3-
6A, -6B, -7, -8, and -9 were run. In runs AHT3-6A,
-6B, -8, and -9, the autoresistance heating was be-
gun when the wall temperature. of the test section
was cooled to 350°C, the lowest temperature at
which liquid can exist in the LiF-BeF:-ThFs sys-
tem. In run AHT3-7, no internal heating was at-
tempted.

The circuit resistance during cooldown in runs
AHT3-6A and AHT3-6B remained at about 0.01 to
0.03 ) until the wall temperature dropped to ap-
proximately 500°C, the temperature at which the
fuel salt begins to freeze. As the wall temperature
dropped from 500 to 350°C, the resistivity in-
creased; at about 350°C, the resistance increased
sharply. Presumably, this event occurred because
the salt layer becomes nonconducting; shorting to
the wall is prevented only when the temperature is
low enough that no liquid can exist in the frozen
salt layer. For the LiF-BeF:-ThF4system, no liquid
can exist below 350°C.

When the wall temperature reached 350°C, auto-
resistance heating was started. We had planned to
increase the current through the salt until the wall
‘temperatures indicated steady state. Before this con-
dition was reached, the resistance through the con-
ducting path dropped suddenly from 0.5 {} in run
6A and 0.4 Q in run 6B to zero, indicating a loss of
film integrity and subsequent shorting of the heat-
ing current. Increases in the wall temperatures at
the time of shorting indicated that the shorting oc-
curred near the bend in the electrode side arm
nearest to the electrode. The cause of this shorting
is not known,

In run AHT3-7 the wall temperature of the test
vessel was lowered to 425°C, and the remaining
liquid salt was transferred to the feed tank. Internal
heat generation was not attempted. The frozen -salt
wall was inspected after the equipment was cooled
to room temperature. The salt was almost com-
pletely frozen across the vessel just above the lower
side arm inlet, although a much thinner salt layer
(about 3/4 in.) was present both below the inlet
and higher in the test section. This freezing was
caused by the unheated argon contacting the liquid
salt. Heaters have been added on the argon lines to
the test vessel to reduce this problem.

The maximum resistance during the autoresis-
tance heating in runs AHT3-6A and AHT3-6B was
0.5 and 0.4 Q, respectively. In run AHT3-8, which
was made after argon heaters were added, the maxi-

124

mum resistance during autoresistance heating was
2.4 ; the autoresistance heating period lasted for
25 min. During this time, the current was increased
from 5 A to 24 A. Although the wall temperatures
continued to decrease during this time, they ap-
peared to be approaching a constant value above
room temperature, After 25 min of autoresistance
heating, the resistance suddenly decreased at the
same time that the temperature rose rapidly in the
bend of the electrode side arm near the electrode,
indicating that this was the area where the salt film
had melted.

Run AHT3-9 was performed using the same pro-
cedure as before, except that the current was lim-
ited to a maximum of 17 A. The electrode in these
tests consisted of a 1/2-in. Inconel tube, as de-
scribed previously, that extended to the intersection
of the centerlines of the vertical section and the
sloping section of the electrode side arm. In the
first nine runs, the salt depth in the electrode side
arm above the end of the tube was measured by
bubbling argon through the tube. Before run
AHT3-9 was begun, this level measurement indi-
cated that the salt level in the electrode side arm
was lower than the salt level measured in the main
compartment (gas-salt contacting zone) of the test
vessel. Also, abnormally high resistances were ob-
served during this run. The highest resistance dur-
ing autoresistance heating in run AHT3-9 was 4.3
(). At the end of the experiment the salt was trans-
ferred to the feed tank from the test vessel. The
captive volume of salt in the electrode side arm
could not be removed. After allowing the equip-
ment to cool, the electrode was removed and ex-
amined. The lower 6 in. of the 1/2-in. Inconel tube
of which the electrode was made had been melted

“off during an attempt to heat the transfer line to the

side arm drain tank with no frozen salt present in
the side arm. With no frozen salt present, the cur-
rent for heating the transfer line took the path of
low resistance through the molten salt. The low re-
sistance led to much higher currents than would
otherwise have occurred, and the tube was melted.

14.6.3 Distribution of Current Densities
in the Autoresistance Heating Equipment

We have constructed an electrical analog of the
electrode side arm to determine if there are regions _
of localized high current density that might be
causing the frozen salt film near the electrode to
melt. The electrical analog was made from eclectri-
cally conducting photographic paper cut in the
shape of the side arm and the contained electrode.
With the use of this electrically conducting paper, a
two-dimensional model may be constructed for the
study of steady-state current flow in simple or com-
plex geometry.”

The geometric model used isopotential noncon-
ducting wall conditions that were represented in the
model by lines of conductive paint. By giving the
opposing walls separate and distinct potentials, an
electric field was created which could be mapped
by using a high impedance voltmeter with an elec-
tric probe. The isopotential lines generated between

22, C. K. Kayan. “An Electric Geometrical Analogue for
Complex Heat Flow.™ Trans. ASME 67, T13-16 (1945).

125

the walls in this manner represented current flow
lines. '

The paper used in the model was black photo-
graphic paper of grade 506-L by Knowlton Bros.,
Watertown, New York. The paper has a resistance
sufficiently high for this purpose. Silver paint made
by Dupont Electrochemicals of Wilmington, Dela-
ware was used. The Bivar potential source and
probe used to measure the electric field was made
by Electronic Associates, Inc., of Long Beach, New
Jersey. An accompanying Variplotter, also by Elec-
tronic Associates, Inc., provided the connections to
the voltage source and an insulated surface to hold
the model. A high impedance voltmeter was re-
quired to measure the field potentials.

The lines in Figs. 14.21, 14.22 and 14.23 represent

~current stream lines. The distance between two

ORNL DWG 74-7326

100 0

100 .

R S A T TR TR R T S e e
AT ATATETHGGI|TTITE A GG GGIGAG|TETEITGG;G|G|T|TITTTEIEITIHE T Tttt tittttttL Ll RS

N
3
N

Fig. 14.21. Calculated current distribution in the electrode side arm of autoresistance heating test AHT-3. The percentage of the total
current flowing between two stream lines is equal to the difference between the number on the stream lines.
126

ORNL DWG 74-6372

100

100 __—]

100
90
80
70
60 ———
50
40
30
20
10

&in

L7 L7777 P77 77

Fig. 14.22. Calculated current distribution at the junction of the electrode side arm and the fluorination section in autoresistance
heating test AHT-3. The percentage of total current flowing between two stream lines is equal to the difference between the numbers on the

stream lines.
127

ORNL OWG 74-7328
100 0

IO ITTIT IV 77777777 T ITF7 77 7T T 777 T T ITT 77 I 77T TI 7T T TTIT7TT I 7T TTIT T T T T T T IS,
i dr P TIPS IITIIBTIIIIIIIIIVIIIIII I I ISV NI I I IV I TPV Y Iy LLLLLL L LLLLEEE

Fig. 14.23. Calculated current distribution in the electrode side arm using a new electrode design. The percentage of the total current
flowing between two stream lines is equal to the difference between the numbers on the stream lines.

stream lines in Fig. 14.2] represents the passage of
12.5% of the total current. In Figs. 14.22 and 14.23,
the distance between stream lines represents 10% of
the total current.

Figures 14.21 and 14.22 show that there are no
regions of abnormally high current densities in the
vicinity of the two bends in the electrode side arm.
The present side arm design seems adequate with
regard to localized heat generation rates.

Figure 14.23 shows an alternate electrode design
that would further reduce localized heating at the
side arm bend near the electrode. This design has
the additional advantage that it could also be used
as the device for introducing the feed to the
fluorinator.

147 FUEL RECONSTITUTION
DEVELOPMENT: DESIGN OF A FUEL
RECONSTITUTION ENGINEERING
EXPERIMENT

R. M. Counce

The reference flowsheet for processing the fuel
salt from a molten-salt breeder reactor is based
upon removal of uranium by fluorination to UFs as
the first processing step.' The uranium removed in
this step must subsequently be returned to the fuel
salt stream before it returns to the reactor. The
method for recombining the uranium with the fuel
carrier salt (reconstituting the fuel salt) is to absorb
128

gascous UFs into a recycled fuel salt stream con-
taining dissolved UF: as described by the reaction

UFs(g) + UFas(d) = 2UFs(d) . (14)

The resultant UFs ‘would be reduced to UFs with
hydrogen in a separate vessel according to the
reaction

UFs(d) + 1/2 Ha(g) = UF«(d) + HF(g) . (15)

We are beginning engineering studies of the fuel
reconstitution step to provide the technology neces-
sary for the design of larger equipment for recom-
bining UFs generated by fluorinators in the process-
ing plant with the processed fuel salt returning to
the reactor. During this report period, equipment
for studying the fuel reconstitution process has been
designed. The equipment is described in this
section. _ ;

A flow diagram of the equipment to be used for"
the fuel reconstitution engineering experiment
(FREE) is shown in Fig. 14.24. The equipment for

this experiment consists of a 36-liter feed tank; a .

UFs absorption vessel; a H: reduction column; an -
effluent stream sampler; a 36-liter receiver; NaF
traps for collecting excess UFs and for disposing of
HF; gas supplies for argon, hydrogen, and UFs; and
means for analyzing the gas streams from the reac-
tion vessels.

The experiment is operated by pressurizing the
feed tank with argon to displace salt from the feed
tank to the UFs absorption tank at rates from 50 to
300 cm’/ min. From the UFs absorption tank, the
salt is siphoned into the H: reduction column.
From the H: reduction column, the salt flows by
gravity through the effluent stream sampler to the
receiver. The feed salt will be LiF-BeF:-ThF.: (72-
16-12 mole %) MSBR fuel carrier salt contain-
ing up to 0.3 mole % UFi. Absorption of UFs by
reaction with dissolved UFa will occur in the UFs
absorption vessel, and the resultant UFs will be re-
duced with hydrogen in the H: reduction column.
The UFs flow rates from 60 to 360 std em’/ min and
the H; flow rates from 60 to 720 std cm®/ min will
be used. The effluent salt will be collected in the
receiver for return to the feed tank at the end of the
run. The off-gas from the absorption vessel and
the reduction column will be analyzed for UFs and
for HF, respectively. The salt in the feed tank will
be analyzed, and salt samples from the efffluent
salt from the column will be analyzed for uranium.
The performance of the column will be determined
from these analyses. The effluent HF and any UFs
passing through the column will be collected on the

NaF traps before the gas is exhausted to the off-gas
system.

The fuel reconstitution engineering experiment
will be installed in the high bay area at the MSRE
site in Bldg. 7503. Scaffolding erected for a previous
experiment will serve the FREE equipment ade-
quately. The system, except for the upper section of
the column, will be enclosed by a splash shield.

The UFs absorption vessel will have an inside di-
ameter of 4 in. and an inside height of 11 in. The
vessel will be constructed from 4 in. sched 40 nickel
pipe mounted vertically with a 150-1b standard .
Monel flange at the top and a 0.25-in. nickel plate
at the bottom. Two 1/2-in. nozzles provide for the
salt to flow in and out of the vessel. Two 3/8-in.
nozzles provide for the UFs(g) to flow in and off-
gas to flow out of the tank.

The H: reduction column will have an overall
height of 115 in. It will be constructed of 1 1/2-in.
sched 40 nickel pipe mounted vertically with a 150-
Ib Monel flange on the upper end and a sched 40

- nickel welded cap at the lower end. Gas enters

thrcugh a 3/8-in. sched 40 nickel side arm at the

ORNL DWG 74-11666

L

o
—

—
UFg HF UFg
ANALYSIS | [anacysis HF TRAP
TRAP —
=
) | | p BUILDING
, OFF-GAS
UFg *—fi : SYSTEM
} EFFLUENT
i | STREAM

e

UFg ABSORPTION
VESSEL

FEED TANK RECEIVER

Ha

A Y

L
Hp REDUCTION
COLUMN

Fig. 14.24. Flow diagram for equipment used in the fuel
reconstitution engineering experiment.
bottom, while 1/2-in. and 3/8-in. nozzles provide
for salt entrance and an off-gas exit.

Because of the highly corrosive nature of dis-
solved UFs, all equipment exposed to significant
quantities of UFs will be gold or gold-plated after a
startup and demonstration period using the equip-
ment described in this report.

The 36-liter feed tank will have an inside diam-
eter of 10 in. and an inside height of 33 in. It will
- be composed of nickel with 0.25-in.-thick walls,
equipped with two 1/2-in. nozzles for salt exit and
entrance lines, two 3/8-in. nozzles for argon sparg-
ing and off-gas exit, and a 3/4-in. sampling port.
The receiver vessel will be essentially identical to
the feed tank. ' -

All piping exposed to the liquid phase (MSBR
carrier salt) or temperatures in excess of 100° C will
be nickel pipe or tubing. Off-gas lines carrying UFs
or HF will consist of nickel tubing, and all other
gas lines will consist of copper tubing. Off-gas lines
and liquid phase piping will be 1/2-in. tubing.

Standard brass valves can be used throughout
the gas system except on gas lines carrying UFs or
HF, where nickel bellows-seal valves are required.
Stainless steel, full-bore ball valves will introduce
the sample ladles into the flowing stream salt
sampler.

Two NaF traps are required for UFs absorption
and for HF absorption. These traps are identical ex-
cept for their lengths. The UFs removal trap is 55
5/8 in. long, and the HF trap is 31 5/8 in. long. The
traps are constructed of 4-in. sched 40 Monel pipe
mounted vertically with 150-1b standard flanges at
both ends for charging and removing NaF pellets.
The gas enters from 1/2-in. nickel tubing through
the upper flanged end and exits through the lower
flanged end. A 6-in. section in the lower end of the
trap contains 4-in.-diam Monel York mesh to keep
pellets from plugging the exit.

The experiment will be monitored by analyzing
the off-gas from each leg of the column. GOW-
MAC gas density detectors are being considered for
this analysis. The detector elements of this analysis
system are not exposed to the measured gas stream
which, in this case, i1s corrosive.

The liquid phase will be sampled periodically by
withdrawing salt samples from the effluent stream
from the hydrogen reduction column. The sampler
is a vertically mounted 3/4-in. sched 40 pipe 20 3/4
in. long with a stainless steel full-bore valve located
12 1/4 in. from the lower end to allow access to the
flowing salt stream. The sampler is vented to the
off-gas system above and below the ball valve. The

129

salt flows into the lower end of the sampler from
1/2-in. tubing and exits 1 1/4in. from the end also
by 1/2-in. tubing. This flow assures a minimum of
1/4 in. of salt for insertion of a sampler. The sam-
pler is a stainless steel 0.250-in.-diam by 1-in.-long
tube with a 0.050-in. fritted metal filter on the
lower end. It is connected to a 1/16-in. stainless
steel tube, which in turn is attached to a vacuum
pump.

Engineering sketches for all the equipment, ex-
cept the UFs supply, have been completed. The
scaffolding in Bldg. 7503 is being cleared to make
room for this experiment. '

o~
(=

14.8 PERFORMANCE OF OPEN
BUBBLE COLUMNS

Conceptual designs for the primary fluorinator
and UF;s reduction columns in a MSBR fuel repro-
cessing plant call for open bubble columns. These
columns are simple, unpacked gas-liquid contactors
with a continuous liquid phase. The gas phase is
introduced into the bottom of the column and bub-
bles travel upward through the liquid. In open col-
umns, the wall can be protected from corrosive
fluorine or UFs in the salt by freezing a layer of salt
on the vessel walls. Although open bubble columns
look attractive for these MSBR applications, there
are few data available for designing such systems.
We have developed an experimental program to
produce and correlate the necessary data. These ef-
forts have included measurements of axial disper-
sion and mass transfer coefficients in open bubble
columns. Dispersion is particularly severe in these
systems and greatly affects the performance of the
columns.

During this report period, studies of open bubble
columns continued as part of the Chemical Engi-
ncering Research Program (sponsored by the AEC
Division of Physical Research). The results of these
studies have been reported in other publica-
tions;”_26 therefore, all the results will not be re-
ported here. Only the most recent results on axial

23. C. D.Scottetal. (eds.), Experinienial Engineering Section
Semiannual  Progress Report (Excluding Reactor Programs),
ORNL-TM-4602 (September 1, 1973 to February 28, 1974).

24. D.J. Conklin, L. A. Lamadrid. and J. G. Arquette, A .xial
Mixing in An Open Bubble Column, Part V: Mass Transfer
Lifects, ORNL-MIT-174 (1973).

25. A, 5. Y. Ho. A. L. El-Twaty, K. J. Kaplan, and T. L.
Montgomery., Axial Mixing in Open Bubble Coltunns, Part VI
Muass Transfer Effects, ORNL-MIT-183 (1974).

26, . ). Toman et al., Avxial Mixing in An Open Bubble
Colunin, Part VI: Mass Transfer Effects, ORNL-MIT-175(1973).
dispersion and mass transfer rates are given below.
Some of the earlier data are only summarized here,
but the complete data are available from the cited
references.

14.8.1 Axial Dispersion
in Open Bubble Columns
J. S. Watson  H. D. Cochran, Jr.

Measurements of gas holdup and axial dispersion
in open bubble columns with diameters of 1, 1.5, 2,
3, and 6 in. have been made over a wide range of
operating conditions and liquid properties.”” Dur-
ing this report period these data, along with a large
collection of the best data in the literature, have
been reduced to generalized correlations based on
dimensionless groups.

The hydrodynamic performance of open bubble
columns is characterized by two flow regimes with
an intermediate transition region. At low super-
ficial gas velocities, the gas rises in the form of dis-
crete single bubbles. This flow regime is called the
bubbly regime. At very high superficial gas veloci-
ties, the gas surges through the column in large
cylindrical slugs that occupy nearly the entire cross
section of the column. This flow regime is called
the slugging regime. In the intermediate transition
region, discrete bubbles coalesce as they rise
through the column to form slugs near the top. We
have found it useful to distinguish between the two
regimes of flow 1n correlating hydrodynamic
results. :

In the bubble flow regime the axial dispersion
coefficient has been correlated by using dimension-

less groups as follows:
45

Nee= 11,12 NR¥ NS, (16)
where '
Ne. = Peclet number = DU/E,
Ngre. = Reynolds number = DU/,

N4, = Archimedes number = D” g/+?,

D = column diameter, cm,

U = superficial gas velocity, cm/sec,

E = axial dispersion coefficient, cm?/sec,
v = liquid kinematic viscosity, g/cm-sec,
g = acceleration of gravity, cm/sec’.

27, J. 5. Watson and L. E. McNeese, Engineering Devélopment
Sfor Molten-Salt Breeder Reactor Processing No. 9, ORNL-TM-
© 3259 (December 1972), pp. 52-148.

130

In the slugging regime the axial dispersion coef-
ficient has been correlated using dimensionless
groups as follows:

Np. = 0.387 N> N,

where

(17)

~ Ns, = Suratman number = Dpa/u? |
o = interfacial tension, dyne/cm,
p = hquid density, g/cm3,
- u = liquid viscosity, g/cm-sec.

The transition between bubbly flow and slugging
flow may be reliably predicted from the intersection
of Egs. (16) and (17). The observed Peclet number
tends to follow the lower (greater axial mixing)
value of the Peclet number across the transition.

This method predicts the transition 1h flow re-
gime, and correlates the gas phase holdup of each
regime; in the bubbly regime,

6 = 0.322(pU"/g0)" ™" ; (18)
in the slugging regime, :
6 = 0.848(U° /g D)%% ; (19)

where 6 = gas holdup fraction.

The correlations for axial mixing and gas holdup
comprise a complete description of the hydrody-
namic performance of open bubble columns.

14.8.2 Mass Transfer Rates
in Open Bubble Columns -

J. M. Begovich  H D. Cochran, Jr.
J. S. Watson™

Studies of mass transfer rates in open bubble col-
umns have been made by several groups of MIT
Practice School students and by ORNL personnel
during the period covered by this report. Because
most of these results have been reported elsewhere,
only the most recent data will be presented here. All
studies have measured CO: sorption rates in water
or aqueous solutions. The most recent group of
MIT students obtained data that confirmed results
from the earlier studies with COz-water ina 1 1/2-
in. column, and the group extended these results to
a larger 3-in.-diam column. In addition, subse-
quent ORNL studies have produced data for high
viscosity solutions of glycerol (approximately 10
cP). .

28. Part of this work wasdone by students in the MIT Schoolof
Engineering Practice.
Experimental. The experimental technique has
been used in all bubble column mass transfer stud-
ies and has been described previously.” > At the
conclusion of the last MIT study, two modifications
were made in the experiment. An automatic titrator
(Radiometer Type TTTla) significantly reduced the
time involved per titration and appeared to improve
the precision of the measurements, This instrument
titrates the samples to a preset pH, and the titration
time can be adjusted by controlling the maximum
solution flow rate. We are currently allowing ap-
proximately 20 to 30 sec per titration. This rate is
faster than most operators can titrate accurately, so
there is less time for CO: sorption or desorption
from the atmosphere during titration.

A basic modification in the equipment piping was
required to use the glycerol solutions. The appara-
tus currently in use is shown (Fig. 14.25). Two 55-
gal drums have been installed as feed or receiving
tanks; glycerol and water can be mixed to the de-
sired composition in either tank. The feed pump
can be used to mix the solutions. There are two
ways to operate the system. The glycerol solution
can be pumped into the column from one tank and
then returned to the other tank. This mode of oper-
ation is convenient for small column diameters (and
heights) and for low liquid rates. Once the feed
tank is empty, the CO; must be stripped from the
receiving tank with an air sparge before starting
again. Since the receiving tank can be sparged as it
1s filled, this does not necessarily delay the experi-

131

ments greatly, but care must be taken to ensure that
the COn content of the feed is known and is prefer-
ably low. With higher liquid rates (such as those
required for the 3-in. column), the first mode of op-
eration does not allow time to take enough samples
after steady state is reached. Under these condi-
tions, two columns can be operated, one as a sorber
and one as a stripper. This is a steady-state opera-
tion, and to reach steady state rapidly, it is desir-
able to maintain the inventory of liquid in the two
drums as low as practical. Both techniques have
been used successfully. .

The glycerol was obtained from Fisher Scientific
Company, and the viscosity of the solution used
was checked regularly with a Brookfield viscom-
eter. The applicability of Henry's law to glycerol
solutions was verified.

Effects of column diameter. The MIT students
collected data with a 3-in.-diam column and com-
pared their results with the results from previous
groups that used a [.5-in. column. For the same
superficial gas velocity, the. larger column gave
larger mass transfer coefficients. The mass transfer
coefficient could be correlated by the equation

Kia = 0.00288 (UD)"72013 (20)

where U is the superficial gas velocity (cm/sec), D
is the column diameter (cm), and Kia 1s the volu-
metric mass transfer coefficient (sec ).

Studies on the effect of liquid viscosity on the
mass transfer coefficient were initiated during this

ORNL DWG 74-106888

VENT
f

4 VENT

z =
2 s JACK
2 35 LEG
(3 [=]

L
[ I—
w JACK
a LEG
7 8
w &
< a
2 2
) <
" a

o

 SAMPLING
PORT

LiIQuID
ROTAMETER

55 GALLON
STAINLESS
STEEL DRUM

GATE
VALVES

ALR

EASTERN 2
CENTFUGAL PUMPS

Fig. 14.25. Flow diagram of equipment used to measure mass transfer coefficients in bubble columns.
report period. At this time, mass transfer coeffi-
cients have been measured only for the 46 volume
% glycerol solution having a viscosity of about 10
cP and a single gas rate of 6.7 cm/sec in a 1.5-in.
column. The liquid flow rate was 4.10 cm/sec, and
several column heights were used. The mass trans-
fer coefficient, K,a, measured under these condi-
tions was 0.0154 sec", and the standard error in
the coefficient was £23%,.

This initial information can be compared with
the MIT Practice School results to show the effects
of viscosity. For the same flow rates, Eq. (20) de-
scribing the data with water (viscosity of about 1
cP) predicts a coefficient of 0.030 sec”', which is
about twice the value mentioned above. Thus, the
higher viscosity reduces the mass transfer rate sig-
nificantly. Since the MIT results suggest that the
mass transfer coefficient is proportional to approxi-
mately equal powers of U and D, the dependence
on viscosity may be the inverse of this power; that
is, the three terms may be grouped into a Reynolds
number. Although the present data are limited, vis-
cosity appears to have less effect upon mass trans-
fer rates than gas velocity or column diameter. The
exponent on viscosity currently appears to be ap-
proximately 0.3. Viscosity probably enters into
other terms besides the Reynolds number. Further
data will elucidate the effects of these parameters.

149 CONCEPTUAL DESIGN OF AN MSBR
PROCESSING ENGINEERING LABORATORY

J. R. Hightower, Jr.

A conceptual design is being prepared for a $12
million MSBR Processing Engineering Laboratory

132

for which FY 1977 authorization is proposed. The
building will provide space for the engineering de-
velopment of the fuel processing technology devel-
oped to date only on a scale of about 1% of that
required for a 1000-MWe MSBR. With the pro-
posed building, operation of the entire processing
system should be possible on a scale that is 25 to
50% of that required for a 1000-MWe MSBR.
Near-full-scale testing of the complete processing
system in the absence of high radiation levels will
be adequate to demonstrate most aspects of com-
ponent performance and to provide the technology
necessary for design of processing equipment for
further evaluation in reactor systems. No facilities
exist at ORNL that meet the specific requirements
for engineering development of the MSBR process-
ing system.

The building will have offices for technical staff,
laboratories for chemical and small-scale engineer-
ing experiments, shop areas equipped to handle
beryllium-contaminated items, a three-story high-
bay area for large-scale engineering experiments,
and, if space and money permit, facilities for salt
purification and preparation.

Candidate floor plans are being considered. Cur-
rent plans call for a three-story concrete and ma-
sonry building approximately 125 ft x 130 ft with a
three-story high-bay area 125 ft x 80 ft. The build-
ing will be suitable for housing beryllium-contain-
Ing experiments,

Part 5. Salt Production

R. W. Horton

The salt production facility at Y-12 is operated to
supply purified mixtures of fluoride salts for MSR
Program experiments and for the Metals and
Ceramics, Chemistry, and Chemical Technology
Divisions. Supplies of lithium fluoride, beryllium
fluoride, thorium fluoride, and uranium fluoride

133

which are maintained can be mixed in any of the
compositions needed, and melted and processed in,
specially designed equipment. Processing involves
treatment with hydrofluoric acid and hydrogen, and
also filtration to remove solids.
134

15. Production of Fluoride-Salt Mixtures for Research
and Development Program

R. W. Horton F. L. Daley

Reactivation of the MSR Program has required
that the source of salt for experimental purposes be
reestablished. This has been done, and production
on a limited scale has been started. Considerable
time was spent modifying the facility and locating
materials and supplies used previously to meet
safety and operational requirements.

The current objectives of the program are (1) to
maintain salt production facilities already available
to satisfy the more immediate demands, and (2) to
start production in the larger-scale equipment as
soon as possible to accumulate an inventory of salt
for future larger-scale test programs. Qur best esti-
mate is that 3100 kg (350 liters) of salt will be
needed by July 1, 1976.

The preoperational work at the salt production
facility included an inventory period during which
the new operating group became familiar with the
equipment, operating areas, and related items. Spe-
cific units were chosen for immediate operation;
these units were completely tested and rebuilt as
needed. During the same time period, two large re-
actor vessels were designed to code standards and
built in local shops. The accumulated inventory of
old vessels, receivers, furnaces, instruments, process
chemicals, and other items was examined and
sorted for disposition. The older vessels containing
varying amounts of salt for which no record existed
(or the information available indicated that recovery

would be unwise) were sent to the burial ground,
source and special nuclear materials included were
written off. The chemical supplies, in metal or
cardboard drums, that were sufficiently identified
were stored in a Y-12 warehouse for recall when

needed.

As these preliminaries were ending, the educa-
tional period for the operating group progressed to
the actual practice of melting, treating, sampling,
and transferring salt. Methods for controlling flow
of process gases were developed and checked.
Safety equipment was acquired and some gas sys-
tem modifications were made to bring hydrogen gas
handling practice up to the level suggested by Y-12
fire protection and safety groups. We also worked
with health physics and industrial hygiene person-
nel to ensure that our methods for handling toxic
chemicals were acceptable.

Processing started at the end of June 1974 and
proceeded concurrently with some of the activities
mentioned. By the end of August 1974, we had
processed 72 kg of 72-16-12 fuel salt and 2 kg of
L:B; a 35 kg batch of 72-16-11.7-0.3 (uranium-
bearing fuel salt) was in process. Operation was
generally satisfactory during this initial period, and
although complete analytical data were not avail-
able, we believe that we are capable of producing
satisfactory material for the MSR Program.
CT
CT
CT
CT
CT

135 . .
r’; ‘\\/ ;
)
L. E. McNEESE, PROGRAM DIRECTOR
b
i
i
|
MSBR DESIGN, DEVELOPMENT, MATERIALS MSBR PROCESSING DEVELOPMENT MSBR CHEMISTRY ! SALT PRODUCTION
AND SAFETY ° . |
H. E. McCOY™* M&C J. R, HIGHTOWER, JR.** CT L. M. FERRIS , CT R. W. HORTON
-J. R. ENGEL** R A.D. KELMERS | cT F. L. DALEY
L. MAYA | CT W. J. BRYAN
HASTELLOY N STUDIES HEMICAL DEVELOPMENT ! .
DESIGN AND SYSTEMS ANALYSIS . ¢ VELOP B. F. HITCH" cT P. P. HAYDON"
E J ALLEN® " H. E. McCOY M&C A. D. KELMERS cT S. CANTOR® cT A L. JOHNSON
e . O.N.BRASKI M&C M, R. BENNETT cT C. E. BAMBERGER* cT
. T. KERR R J. BRYNESTAD M&C
G. T. MAYS . D. E. HEATHERLY cT
o R g- S 5;8323‘-6 ng ENGINEERING DEVELOPMENT D. Y. VALENTINE cT
CONSULTANTS -9 .
s J. R. DISTEFANO** M&C J. R.HIGHTOWER, JR. cT
D.W. CARDWELL" 'R, KEISER MG W. L. CARTER® or ANALYTICAL CHEMISTRY
SYSTEMS AND COMPONENTS B.C. LESLIE" M&C R. M. COUNCE cT RESEARCH AND DEVELOPMENT
B. McNABB M&C C.W. KEE" cT . '
DEVELOPMENT T. K. ROCHE M&C J. A. KLEIN CT A. i “:EZEF?LE AC
’ . F. . . AC
R. H. GUYMAN R A. C. SCHAFFHAUSER* M&C H.C. SAVAGE cT B. R CLARK* ' AC
» J. W. WOoODSs* M&C J. BEAMS CcT - A.
J. L. CROWLEY R J. M. DALE* LA
: P. P. HAYDON* cT .M. c
W. R. HUNTLEY R o L MANNING
R.O. PAYNE CT - L AC
A.N. SMITH R GRAPHITE STUDIES 1 P. YOUNG® AC
E. L. BIDDLE R o E. MeCOY " e
H. E. ROBERTSON R .E.Mc .
W. H. COOK* M&C ANALYSES '
REACTOR SAFETY W. P. EATHERLY M&C L T. CORBIN* AC
J. H. COOPER* AC
J. R.ENGEL R CHEMICAL PROCESSING MATERIALS W. R. LAING* AC
CONSULTANT J. R. DISTEFANO" " M&C CONSULTANT
* B »*
E.S.BETTIS* 0. B. CAVIN M&C G. MAMANTOV* uT
A. J. MOORHEAD* M&C
TECHNICAL SUPPORT
INORGANIC AND PHYSICAL CHEMISTRY
K. W. BOLING* M&C .
C.E. DUNN* M&C L. M. TOTH cT
W. H. FARMER* M&C L. 0. GILPATRICK" C
J. C. FELTNER M&C R. M. WAI:LER*** ' c
J. L. GRIFFITH* M&C F. A. POSEY . ! c
L. HEATHERLY* M&C H. R. BRONSTEIN c
. J. W. HENDRICKS"* M&C C. F. BAES, JR.* c
"PART TIME ON MSRP T HENSON* M&C W. C. WAGGENER"* c
DUAL CAPACITY .
“** DECEASED J. D. HUDSON M&C
E.J. LAWRENCE M&C
AC  ANALYTICAL CHEMISTRY DIVISION G. A. POTTER . M&C ’
C  CHEMISTRY DIVISION L. G. RARDON M&C
CT CHEMICAL TECHNOLOGY DIVISION R. H. SHANNOT M&C
M&C METALS AND CERAMICS DIVISION L.J. SHERSKY_ R
R REACTOR DIVISION C. K. THOMAS . M&C
UT  UNIVERSITY OF TENNESSEE L. R. TROTTER M&C
C. A. WALLACE"* R
J. J. WOODHOUSE* M&C