ORNL-5078.txt

OCKHEED MARTIN ENERGY RESEARCH UBRARIES ORNL-5078

AT et

3 445k D445252 |

MOLTEN-SALT REACTOR
PROGRAM

Semiannual Progress Repont
Peniod Ending August 31,197

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

Name/Title: Leesa Laymance, ORNL TIO
Date: 7/27/2017

OAK RIDGE NATIONAL LABORATORY
CENTRAL RESEARCH LIBRARY
DOCUMENT COLLECTION

LIBRARY LOAN COPY
DO NOT TRANSFER TO ANOTHER PERSON

If you wish someone else to see this
document, send in name with document
and the library will arrange a loan

OAK RIDGE NATIONAL LABORATORY

OPERATED BY UNION CARBIDE CORPORATION FOR THE ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION

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 $8.50; Microfiche $2.25

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-5078
UC-76 — Molten-Salt Reactor Technology

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM
SEMIANNUAL PROGRESS REPORT
FOR PERIOD ENDING AUGUST 31, 1975

L. E. McNeese
Program Director

FEBRUARY 1976

OAK RIDGE NATIONAL LABORATORY
Oak Ridge, Tennessee 37830
operated by
UNION CARBIDE CORPORATION
, for the
ENERGY RESEARCH AND DEVELOPMENT ADMINISTRATION

i

3 445k O4y5R52 b

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.

ORNL-2474
ORNL-2626
ORNL-2684
ORNL-2723
ORNL-2799
ORNL-2890
ORNL-2973
ORNL-3014
ORNL-3122
ORNL-3215

ORNL-3282

ORNL-3369
ORNL-3419
ORNL-3529
ORNL-3626
ORNL-3708
ORNL-3812

ORNL-3872 .

ORNL-3936
ORNL4037
ORNL4119
ORNL4191
ORNL4254
ORNL-4344
ORNL-4396
ORNL4449
ORNL4548
ORNL-4622
ORNL4676
ORNL-4728
ORNL4782
ORNL-4832
ORNL-5011
ORNL-5047

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
Period Ending August 31,1974
Period Ending February 28, 1975
Contents
SUMM A RY . . e e
PART 1 — MSBR DESIGN AND DEVELOPMENT
1. SYSTEMS AND ANALY SIS . ..o e e e e e 2
- 1.1 Tritium Behavior in Molten Salt-Systems . . ... . i e .. 2
1.1.1 MSBR Calculations . . ... ... e e 2
'1.1.2 Coolant Salt Technology Facility .. ....................... .. .. oo i 3
1.2 Xenon Behavior in MSBR . . ... . e 8
1.3 Neutronic ANalysis . ... ..o et e it e e e e et e e e e e e e e e 9
1.3.1 MSBR Studies .......... ..t e 9
1.32 TeGen Capsules .. ... ... e 12
1.4 High-Temperature Design Methods . .. ... ... .. . . 12
2. SYSTEMS AND COMPONENTS DEVELOPMENT ... ... ... . i 16
2.1 Gas-Systems Technology Facility . ........ .. i e 16
2.1.1 Cavitation and Salt-Pump Shaft Oscillations . . . . ... ... ... .. . 16
2.1.2 Salt-Pump Performance Data and Calibration of Variable-Flow Restrictors , ............... 18
2.1.3 Salt-Pump Fountain Flow .. ... . ... ... . ... . . . . . . . e 19
2.1.4 Densitometer Studies .............. ... ... .. ... .. P 22
2.2 Coolant-Salt Technology Facility (CSTE)} ... tvnn e et e et 22
2.2.1 LoopOperation ...................... e e e e 22
2.2.2 Salt Mist Test . ..ot e e 23
2.2.3 Tritium ExXperiments . .. .. ... e e 24
2.3 Forced Convection LOOPS ... ...t i e 26
2.3.1 Operation of MSR-FCL-2b ............ SR 26
2.3.2 Design and Construction of FCL-3 and FCL4 . ...... ... ... ... ... . . . .. ... .. . ..., 27
PART 2. CHEMISTRY

3. FUEL SALT CHEMISTRY ... ... e i [ ... 29
3.1 Compounds in the Lithium-Tellurium System .............. [ 29
3.2 Spectroscopy of Tellurium Species in Molten Salts . .......... ... .. ... ... . .. .. 30
3.3 The Uranium Tetrafluoride-Hydrogen Equilibrium in Molten Fluoride Solutions . ............ ... 31
3.4 Porous Electrode Studies in Molten Salts . ......... ... .. . . . 32
3.5 Fuel Salt-Coolant Salt Interaction Studies ...................... R L
3.6 Lattice and Formation Enthalpies of First-Row Transition Metal Fluorides . .................. 37

iii
iv

4. COOLANT SALT CHEMISTRY . ... . e e e R 41
4.1 Chemistry of Sodium Fluoroborate ... ....... ... ... i i e 41
4.2 Corrosion of Structural Alloys by Fluoroborates .. ......... . .. .. . e 42
5. DEVELOPMENT AND EVALUATION OF ANALYTICALMETHODS . .......... .. ... .. ... .. 44
5.1 In-ine Analysis of Molten MSBR FUEL . . . .« .+ oo e oo e e e 44
5.2' Tritium Addition Experirfients in the Coolant-Salt Technology Facility ........................ 45

5.3 Electroanalytical Studies of Iron(II) in Molten LiF-BeF,-ThF,
(72-16-12 MO0LE TB) . . o v oo e ettt e e e e e e e 47

5.4 Voltammetric Studies of Tellurium in Molten LiF-BeF, -ThF,
(72-16-12 MO0LE ZB) v oo oottt e e e e e e e 48

PART 3. MATERIALS DEVELOPMENT

6. DEVELOPMENT OF MODIFIED HASTELLOY N .......... .. ... ... ... . ... e 52
6.1 Development of a Molten-Salt Test Facility ........... . .. . . . i 52
6.2 Procurement and Fabrication of Experimental Alloys .................. P e 65

6.2.1 Production Heats of 2% Ti-Modified Hastelloy N . . .. ... ... . . i e 65
6.2.2 Semiproduction Heats of 2% Ti-Modified Hastelloy N
Containing Niobium .. ... .. ... ... . . e 69
6.3 Weldability of Commercial Alloys of Modified Hastelloy N . .................. e 69
6.4 Stability of Various Modified Hastelloy N Alloys in the
Unirradiated Condition ............. ... .. . ... e h e aaee e iaaaeiaaan 74
6.5 Mechanical Properties of Titanium-Modified Hastelloy N Alloys
in the Unirradiated Condition .. ........ ... ... ... . ... ..... e e 78
6.6 Postirradiation Creep Properties of Modified Hastelloy N .. ... ... . ... . .. . .. . . .. ... 82
6.7 Microstructural Analysis of Titanium-Modified Hastelloy N. . .. .. e T 84
6.7.1 Microstructural Analysis of Alloys 503 and 114 .. ... ... . ... ... ... .. P 85
6.7.2 Homogeneous Hastelloy N Alloys . ........ ... ... ... ... ... ........ e 88
6.8 Salt Corrosion Studies . .. . . P P P e R 91
6.8.1 Fuel Salt Thermal Convection Loops . .................... e 93
6.8.2 Fuel Salt Forced Circulation Loop ....... ... ... ........... IR T 94
6.8.3 Coolant Salt Thermal Convection Loops . ... ... ... .. . .. 94
6.9 Corrosion of Hastelloy N and Other AlloysinSteam .......... ... ... ... ... ... .. ... ... ..... 97
6.10 Observations of Reactions in Metal-Tellurium-Salt Systems . . . ... ..... e 100
6.11 Operation of Metal-Tellurium-Salt Systems ......... e T 101
6.11.1 Tellurium Experimental Pot Number 1 .. ... .. ... . . . . . 101
6.11.2 Chromium Telluride Solubility Experiment . ............ e 102
6.11.3 Tellurium Experimental Pot Number 2 .. ... ... .. .. . . . . . .. . . . . e 103
6.12 Grain Boundary Embrittlement of Hastelloy N by Tellurium .. .. ....... . ........ B 103
6.13 X-Ray Identification of Reaction Products of Hastelloy N o
' Exposed to Tellurium-Containing Environments . .................. P AP 107
6.14 Metallographic Examination of Samples Exposed to
Tellurium-Containing Environments . . . ... .. ... e e e 108
6.15 Examination of TeGen-1 ... .. . . . . e e 119
6.15.1 Metallographic Observations ... . ... . ...t 123

6.15.2 Chemical Analyses for Tellurium . ... ... ... .. . . e 124
6.16 Salt Preparation and Fuel Pin Filling for TEGen-2and -3 .. ......... .. ... . ... .. oot 131
7. FUEL PROCESSING MATERIALS DEVELOPMENT . ... .. . . 132
7.1 Static Capsule Tests of Graphite with Blsrnuth and
Bismuth-Lithium Solutions ...................... e e e e 132
7.2 Thermal Gradient Mass Transfer Test of Graphite
inaMolybdenum Loop .. ... i e e 133
7.2.1 WeightChanges ........... ...t .. e e 133
7.2.2 Compositional Changes ......... ... ... ... ..l e 133
7.2.3 Microstructural Changes . . . ... ... . e e 137
7.2.4 Discussion of Results . . ... .. .. e 137

PART 4. FUEL PROCESSING FOR MOLTEN-SALT REACTORS

8. ENGINEERING DEVELOPMENT OF PROCESSING OPERATIONS . ......... ... . ... ... . ... 142
8.1 Metal Transfer Process Development ........................... e 142
8.1.1 Addition of Salt and Bismuth Phases to Metal Transfer '
Experiment MTE-3B ........... e 143
812 RunNd-1 ....... ... . ... .. . ... ... ... ... e e e 144
8.1.3 Run Nd-Z . . e e 145
8.1.4 Discussion of ReSUILS . .. ..t e e e e e e 145
8.2 Salt-Bismuth Contactor Development .......... ... ... . ... . i 147
8.2.1 Experiments with a Mechanically Agitated Nondlspersmg Contactor
in the Salt-Bismuth Flowthrough Facility ........... ... .. ... . . .. . 148
8.2.2 Experiments with a Mechanically Agitated Nondispersing
Contactor Using Water and Mercury ... ... ... i e ... 149
8.3 Continuous Fluorinator Development .. ... ... ... .. . ... . . . i 152
8.3.1 Installation and Initial Operation of Autoresistance Heating _
Test AHT -4 . .. e 152
8.3.2 Design of a Continuous Fluorinator Experiment Facility (CFEF) ....................... 155
8.3.3 Fluorine Disposal System for Bldg. 7503 . ... ..o\t tirtitit ettt e e 156
- 8.3.4 Frozen Wall Corrosion Protection Demonstration ........... e 156
8.4 Fuel Reconstitution Engineering Development .. ....... ... .. .. .. . . i 157
8.4.1 Instrumentation for Analyzing Reaction Vessel Off-Gases ... .................... ... ... 158
8.4.2 Design of the Second Fuel Reconstitution Engineering '
Experiment ........ .. ... e 160

8.5 Conceptual Design of a Molten-Salt Breeder Reactor Fuel
Processing Engineering Center .. ...... ... o 161

PART 5. SALT PRODUCTION
9. PRODUCTION OF FLUORIDE SALT MIXTURES FOR MSR PROGRAM

RESEARCH AND DEVELOPMENT e 163
9.1 Quantities of Salt Produced .. .. ... .. 163
9.2 Operating Experience in 12-in.-diam Reactor ......... .. .. ... . .. . il 163

9.2.1 Charging and Melting of Raw Materials . ......... .. ... .. ... ... . i, 164
9.2.2 Hydrofluorination and Hydrogen Reduction ......... ... ... ... ... .. ... 166
0.3 SUMMATY . oottt ettt e e e e e it ieee e e e 166

ORGANIZATION CHART . .ot e e e e e e e e e et 167
Summary

PART 1. MSBR DESIGN AND DEVELOPMENT
J. R. Engel

1. Systems and Analysis

Calculations of the expected tritium behavior in the
reference-design MSBR were continued with studies of

An updated neutronics model of the 1000-MW(e)
reference-design MSBR is being developed. Multi-
dimensional, multigroup calculations will use the VEN-
TURE code, with neutron cross-section data derived
entirely from the ENDF-IV libraries. Processing of the
cross-section data was completed for 38 of 39 nuclides

~at four temperatures of interest for the planned calcula-

the  possible effects of oxide films on heat exchange -

surfaces in the steam system and on surfaces exposed to
the containment atmosphere. The presence of oxide
films with very low permeability on the heat transfer
surfaces would significantly reduce the rate of tritium
migration to the steam system because of the increasing
importance of the oxide-film resistance at very low par-
tial pressures of hydrogen and tritium. However, the
reduction from this effect alone would be insufficient
to limit the rate of tritium migration to the steam
system to desired values. At high rates of tritium trans-
port to the steam system, the presence of oxide-film
resistances on loop walls tends to increase the rate of
tritium flow into the steam. However, this effect is
insignificant at the low migration rates required.

Potential distributions of tritium in the Coolant-Salt
Technology Facility ‘were estimated for the conditions
of planned experiments. In the absence of tritium inter-
action with the salt, other than simple dissolution, as
much as 99% of the added tritium could be expected to
escape through the loop walls. Removal of significant
fractions in the loop off-gas could be expected only if
the effective permeability of the loop walls were 10 to
100 times less than that of bare metal.

Substantial chemical interaction of tritium with
NaBF,-NaF was observed in the two tritium addition
tests performed. Ratios of combined-to-elemental trit-

tions. Cross-section data are also being examined for the

two-step thermal reaction *®*Ni(n,y)*°Ni(n,a)®°Fe,
. which is expected to be the principal source of helium

ium in the salt, inferred from elemental concentrations

in the off-gas and combined concentrations in the salt,
~ were 50 and 530 for the two tests. Approximately ' to
Y. of the added tritium was removed in the off-gas
stream, principally in a chemically combined, water-
soluble form.

vii

in MSBR structural metals. _

A review of the data and calculations used to estimate
tellurium inventories in the TeGen-1 experiment indi-
cates an uncertainty of *20%.

Work is continuing on the study of thermal ratch-
etting and creep fatigue in reactor structural materials.
Analytical methods are being developed which will be
applied to the reference-design MSBR to evaluate the
significance of these processes in Hastelloy N.

2. Systems and Components Development -

The Gas-Systems Technology Facility was operated
with water throughout the report period. Efforts to
reduce the amplitude of the salt-pump shaft oscillations
have been unsuccessful. The amplitude of these oscilla-
tions is largely dependent upon shaft speed, so a larger-
diameter impeller, which will give the design flow and
head at lower speeds, is being fabricated. A method was
developed for estimating the pump fountain flow. Since
this flow was highier than desirable, back vanes will be
used on the new impeller to limit the flow. Tests made
at the loop indicate that the densitometer can be used
to determine bubble-separator efficiencies if short-term
tests are used.

Routine operation of the Coolant-Salt Technology
Facility was established with more than 2500 hr of salt
circulation without plugging in the loop off-gas line.
Measurements of the amount of salt mist in the off-gas
stream showed 100 to 500 ng/cm?® (STP), depending on
the salt temperature and the BF; flow rate into the
loop gas space. The mist trap installed in the salt cold
trap was effective in preventing the plugging that had
been experienced earlier. Two tritium injection tests
were conducted, in which 85 and 97 mCi, respectively,
of tritiated hydrogen were added to the loop during two
10-hr periods. Frequent salt and off-gas samples were
taken to monitor the tritium behavior in the loop.

The forced-convection loop, MSR-FCL-2b, has accu-
mulated 3000 hr of operation with MSBR reference fuel
salt at design AT conditions with the expected low cor-
rosion rates. Data obtained on the heat transfer charac-
teristics of this salt are being analyzed. The design is
essentially complete for forced-convection loops FCL-3
and FCL-4. Components are being fabricated and elec-
trical installation is proceeding.

PART 2. CHEMISTRY

3. Fuel-Salt Chemistry

Relatively pure Li, Te (about 99% on a mole basis)
was prepared by the controlled addition of tellurium to
liquid lithium. The reaction was begun at 250°C, but
ultimately temperatures greater than 500°C were re-
quired to complete the reaction. LiTe; was prepared by
reacting the stoichiometric amounts of Li, Te and tellu-
rium for 2 hr at 550°C.

Apparatus for the spectroscopic study of tellurium
species in MSBR fuel salt has been assembled. Prelimi-
nary work with lithium tellurides in chloride melts has
shown that at least two light-absorbing species are
present with compositions in the range Li, Te to LiTe,.
Furthermore, studies with Te, in LiCl-KCl eutectic have
shown that, in addition to Te,, a second species is
present at high temperatures and/or high halide ion
activity. : - '

Apparatus for the spectrophotometric study of the
equilibrium UF4(d) + %4H,(g) = UF;3(d) + HF(g) has
been assembled, and measurements using Li,BeF, as
the solvent have begun. A preliminary value of about
107% was obtained for the equilibrium quotient at
650°C. This value is in good agreement with the value
obtained previously by other workers.

Development proceeded on porous and packed-bed
electrode systems as continuous, on-line monitors of
concentrations of electroactive species in molten salt
solutions. The packed-bed electrode of glassy carbon
spheres was calibrated using Cd** ions in LiCl-KCl
eutectic before experiments were conducted with Bi**
ions in solution. The results of the experiments demon-
strated the capability of the electrode for monitoring
these and other ions.

viii

Preliminary experiments were conducted to evaluate-
some questions relating to the mixing of NaBF,-NaF
coolant salt with MSBR fuel salt, LiF-BeF,-ThF, -UF,
(72-16-11.7-0.3 mole %). The results showed that the
rate of evolution of BF; gas on mixing was low. Mixing
of small amounts of coolant salt with fuel salt did not
result in the precipitation of uranium- or thorium-
containing compounds. No data were obtained on the
mixing of small amounts of fuel salt with coolant salt.
Results of experiments in which a small amount of cool-
ant salt containing oxide was mixed with fuel salt sug-
gested that oxide species more stable than UQ,. were
present, since no precipitation of UQ, was observed.

A study of lattice enthalpies of first-row transition-
metal fluorides was undertaken to provide a theoretical
basis for evaluating thermochemical data for structural
metal fluorides -being obtained from solid-electrolyte
galvanic cells. Ligand-field corrections to plots of lattice
enthalpy vs atomic number for the two series CaF, to
ZnF, and ScF; to GaF; indicated that the standard
enthalpies of formation (AH7 )of NiF, and VF; were
satisfactory, but that more accurate experimental values
of AHf for TiF;, VF,, CrF,, CrF;, FeF,, and FeF,
would be desirable.

4. Coolant-Salt Chemistry

Af;alyses of samples of condensate collected during
operation of the Coolant-Salt. Technology Facility
indicate that the vapor above the salt is not a single
molecular compound but rather a mixture of simple
gaseous species such as H, O, HF, and BF,. The conden-
sate showed a tritium concentration ratio of about 10°
relative to the salt. This result suggests a possible
method for concentrating and collecting tritium in an
MSBR. Related work showed that NaBF; OH dissolved
in coolant salt undergoes a reaction that reduces the
OH™ concentration in the salt, producing a volatile frac-
tion. Physical and chemical observations were made on
the system NaF-NaBF,-B,0; at 400 to 600°C. Work
with- compositions typical of the usual coolant salt
(oxide concentrations up to 1000 ppm) showed that at
least two oxygen-containing species are present. One
species is Nay;B3F¢O;; the other has not yet been
identified. .

Studies were continued to determine the extent to

- which borides were formed in Hastelloy N and Inconel

600 by reaction with NaBF,-NaF at 640°C. Data ob-
tained thus far indicate some formation of chromium
and nickel borides; however, after four months of ex-
posure of the alloy samples to salt the boride concentra-
tion on the metal surfaces did not exceed 500 ppm in
Hastelloy N and 1000 ppm in Inconel 600. The results
also showed that chromium in these alloys was selec-
tively oxidized by the salt.

5. Development and Evaluation of
Analytical Methods

During this period U*/U* ratios were monitored by
voltammetric techniques in two thermal-convection
loops and one forced-circulation loop. Stable redox con-
ditions continue to exist in thermal-convection loops
21A and 23;the U*/U? ratio is approximately 7.6 X
10° and 5 respectively. In forced-convection l'oop
FCL-2b, the U*/U* ratio is about 80. No attempts
have yet been made to reoxidize the U*" in the melt by
the addition of nickel fluoride or some other oxidant.

The results from the first series of tritium addition
experiments at the Coolant-Salt Technology Facility
show that very little tritium exists in the off-gas in the
elemental state; the bulk of the tritium occursin a com-
bined or water-soluble form. It appears that about 50%
of the injected tritium experienced significant holdup in
the salt and was eventually removed in the system off-
gas stream.

It was observed that the Fe®* — Fe® electrode reac-
tion in molten LiF-BeF,-ThF, (72-16-12 mole %)
closely approximates the soluble product case at a gold
electrode, the insoluble product case at pyrolytic graph-
ite, and, depending on the temperature, both soluble
and insoluble product cases at an iridium electrode.

Voltammetric measurements were made in molten
LiF-BeF,-ThF, following additions of Li,Te in an

effort to identify soluble electroactive tellurium species

in the melt. No voltammetric evidence of such com-
pounds was obtained. These observations were in
general agreement with chemical analysis that indicated
<5 ppm Te in the salt.

PART 3. MATERIALS DEVELOPMENT

6. Development of Modified Hastelioy N

Work ‘is partially complete on the molten-salt test
facility to be used mostly for mechanical property test-
ing. Much of the test equipment is operational.

All products except the seamless tubing of the 2%
Ti—modified Hastelloy N were received. The first heat
weighed 10,000 1b and had a fairly narrow working tem-
perature range. The second heat weighed 8000 lb and
had a wider working temperature range. Seamless tubing
is being fabricated by two vendors. Weldability studies

ix

on these two heats showed that their welding charac-
teristics were equivalent to those of standard Hastelloy
N and that existing welding procedures for standard
Hastelloy N could be used for the 2% Ti—modified
alloy.

The mechanical properties of Hastelloy N modified
with titanium, niobium, and aluminum were evaluated
in the irradiated and unirradiated conditions. These
properties were used to estimate the individual and
combined concentrations of titanium, niobium, and
aluminum- required to produce brittle intermetallic
phases. The .formation of brittle phases in the alloys
containing niobium was enhanced by an applied stress.

Specimens of modified Hastelloy N were exposed to
tellurium from several different sources. The partial
pressure of tellurium above Cr;Te, at 700°C seems
reasonably close to that anticipated for MSBRs. Metal-
lographic examination of the exposed specimens after
straining revealed that alloys containing 0.5 to 1% Nb
were resistant to intergranular cracking by tellurium.

Further analysis of the data from TeGen-1 showed
that most of the tellurium in each fuel pin was concen-
trated on the tube wall. The concentration in the salt
was 1 ppm or less. The salt has been prepared for filling
the fuel pins in TeGen-2 and-3, and the pins for

TeGen-2 have been assembled for filling.

7. Fuel Processing Materials Development

Experiments were continued to evaluate graphite as a
material for fuel processing applications. The penetra-
tion of graphite by bismuth-ithium solutions was found
to increase with increasing lithium concentration of the
solution and pore diameter of the graphite. Decreasing
the pore diameter of the graphite by pitch impregnation
decreased the average depth of penetration. However,
because the structure of the graphite was variable,
greater-than-average penetration occurred in regions of
low density.

A thermal-convection loop constructed of molyb-
denum contained ATJ graphite specimens in hot- and
cold-leg regions and circulated Bi—2.4 wt % (42 at. %)
Li for 3000 hr at 700°C maximum temperature, with a
temperature differential of 100°C. Very large weight
increases (30 to 67%) occurred in all of the graphite:
samples, primarily as a result of bismuth intrusion into
the open porosity of the graphite. Dissimilar-metal mass
transfer between molybdenum and graphite was also
noted. These results and previous capsule test results
suggest that the presence of molybdenum enhances
intrusion of bismuth-lithium solutions into graphite.
Thin carbon layers were noted on the molybdenum.
PART 4. FUEL PROCESSING FOR
MOLTEN-SALT REACTORS

8. Engineering Development of
Processing Operations

Addition of the salt and bismuth solutions to the
process vessels in metal transfer experiment MTE-3B
was completed. Two experiments were performed to
measure the removal rate and overall mass transfer
coefficients of neodymium. In the first run about 13%
of the neodymium originally added to the fuel salt
(72-16-12 mole % LiF-BeF,-ThF,) in the fuel-salt reser-
voir was removed during the 100 hr of continuous
operation. Overall mass transfer coefficients for neo-
dymium across the three salt-bismuth interfaces were
lower than predicted by literature correlations, but were
comparable to results seen in experiment MTE-3.

For the first 60 hr of the second experiment, which
was a repeat of the first experiment, the rate of removal
of neodymium was similar. The second run was termi-
nated because of unexpected entrainment of the fuel
salt into the lithium chloride in the contactor, which
resulted in depletion of the lithium from the Bi-Li solu-
tion in the stripper and stopped further neodymium
transfer.

Future experiments in MTE-3B will depend on deter-
mining the reason for the unexpected entrainment of
fluoride salt into the lithium chloride, and it will be
necessary to remove and replace the lithium chloride
that is presently contaminated with fluoride salt.

A hydrodynamic run intended to determine the effect
of increased agitator speed on the extent of entrainment

of one phase into the other in the salt-bismuth con-.

tactor was performed. No visual evidence of gross en-
trainment was found. Analytical results indicate that
the bismuth concentration in the fluoride salt phase
decreased with increasing agitator speed. This un-
expected result is probably due to sample contamina-
tion.

Development work continued on an electrochemical
technique for measuring electrolyte film mass transfer
coefficients in a nondispersing mechanically agitated
contactor, using an aqueous electrolyte solution and
mercury to simulate molten salt and bismuth. During
this report period experiments with Fe®-Fe®* were
made with improved experimental apparatus. A stan-
dard calomel electrode which enables measurement of
the mercury surface potential was obtained. Electronic
filters were attached to the inputs on the x,y plotter to
damp out noise in the signal to the plotter. Near the end
of the report period, a potentiostat was obtained which
will automate the scan procedure now performed with

the dc power supply. Copper, iron, and gold anodes
have been tested. The gold anode is the most satisfac-
tory choice, since it does not react with the electrolyte
solution. By noting that the active anode area in the cell
could be decreased with no resulting change in the dif-
fusion current, it was determined that the mercury
cathode rather than the gold anode is polarized. Results
indicate that the ferric iron is being reduced by some
contaminant in the system. Further tests with purified
mercury and electrolytes in the absence of oxygen indi-
cate that the contaminant was present in the mercury.
Analytical results for Fe®" and Fe®* concentrations in
the electrolyte phase are inconsistent with expected
results. Qualitative results indicate that a buffered
quinone-hydroquinone system may be useful as an alter-
nate to the Fe**-Fe?* system.

Installation of autoresistance heating test AHT-4, in
which molten salt will be circulated through an
autoresistance-heated test vessel in the presence of a
frozen-salt film, was completed and operation was
begun. A conceptual design was made of a continuous
fluorinator experimental facility for the demonstration
of fluorination in a vessel protected by a frozen-salt
film. Design was completed and installation was begun
on a fluorine disposal system in Building 7503, using a
vertical scrubber with a circulating KOH solution. In-
stallation was completed of equipment to demonstrate
the effectiveness of a frozen-salt film as protection
against fluorine corrosion in a molten-salt system.

Off-gas streams from the reaction vessels in the fuel
reconstitution engineering experiments will be con-
tinuously analyzed with Gow-Mac gas density detectors.
To determine whether hydrogen back-diffusion in the
cell body will be a problem during the analysis of the
HF-H, mixture from the hydrogenation column, the
cell was calibrated with N,-H, mixtures. It was found
that when the reference gas flow rate to the cell is suffi-
ciently high, the effect of hydrogen back-diffusion is
not seen. The second engineering experiment will be
conducted in equipment which is either gold plated or
gold lined to eliminate or minimize effects resulting
from equipment corrosion. Several alternatives for gold
lining or gold plating are discussed. The factors which
must be considered in deciding between lining or plating
are listed.

A design is being prepared to define the scope, esti-
mated design and construction costs, method of accom-
plishment, and schedules for a proposed Molten-Salt
Breeder Reactor Fuel Processing Engineering Center.
The proposed building will provide space for prepara-
tion and purification of salt mixtures, for engineering
experiments up to the scale required for a 1000-MW(e)
MSBR, and for laboratories, maintenance areas, and
offices. The estimated cost of the facility is
$15,000,000; authorization will be proposed for FY
1978.

PART 5. SALT PRODUCTION

9. Production of Fluoride Salt Mixtures
for Research and Development

Activities during the report period fall in three
categories: (1) salt production, (2) facility and equip-

xi

ment maintenance and modification, and (3) peripheral
areas that include preparation of transfer vessels and
assistance to others in equipment cleanup.

Salt produced in this period, totaling about 600 kg,
was delivered in more than 30 different containers.
About one-half of the salt was produced in an. 8-in.-
diam purification vessel and had acceptable purity
levels. The remaining salt was produced in the 12-in.-
diam purification vessel during five runs, each of which
involved about 150 kg of salt.
Part 1. MSBR Design and Development

J. R. Engel

The overall objective of MSBR design and develop-
ment activities is to evolve a conceptual design for an
MSBR with adequately demonstrated performance,
safety, and economic characteristics that will make it
attractive for commercial power generation and to de-
velop the associated reactor and safety technology re-
quired for the detailed design, construction, and opera-
tion of such a system. Since it is likely that commercial
systems will be preceded by one or more intermediate-
scale test and demonstration reactors, these activities
include the conceptual design and technology develop-
ment associated with the intermediate systems.

Although no system design work is in progress, the
ORNL reference conceptual design' is being used as a
basis to further evaluate the technical -characteristics
and performance of large molten-salt systems. Calcula-
tions are being made to characterize the behavior and
distribution of tritium in a large system and to identify
potential methods for limiting tritium release to the
environment. These analytic studies are closely corre-
lated with the experimental work in engineering-scale
facilities. Studies were started, in this reporting period,
to reexamine the expected behavior of xenon in an

MSBR. This work will ultimately use information from

experiments in the Gas-Systems Technology Facilitiy
(GSTF) to further refine !'®° Xe-poisoning projections
and to help define the requirements for MSBR core
graphite. '

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

Additional core neutronics calculations are being
made for the reference MSBR, using widely accepted,
evaluated nuclear data and a two-dimensional computa-
tional model. These calculations will provide updated
estimates of the nuclear performance, as well as addi-
tional information on core characteristics. Analogous
methods and data are employed to provide support for
in-reactor irradiation work.

The GSTF is an engineering-scale loop to be used in
the development of gas injection and gas stripping tech-
nology for molten-salt systems and for the study of
xenon and tritium behavior and heat transfer in MSBR
fuel salt. The facilitiy is being operated with water to
measure loop and pump characteristics that will be re-
quired for the performance and analysis of develop-
mental tests with fuel salt.

The Coolant-Salt Technology Facility is being oper-
ated routinely to study processes involving the MSBR
reference-design coolant salt, NaBF,-NaF eutectic.
Tests are in progress to evaluate the distribution and
behavior of tritium in this system.

Candidate MSBR structural materials are exposed to
fuel salt at reference-design temperatures and tempera-
ture differences (704°C maximum and 139°C AT) and
representative salt velocities in forced-convection loops
to evaluate corrosion effects under various chemical
conditions. These operations, which are principally in
support of the materials development effort, also pro-
vide experience in the operation of molten-salt systems
and data on the physical and chemical characteristics of
the salt. One loop, MSR-FCL-2b, which is made of
standard Hastelloy N, is in routine operation; two
others, to be made of titanium-modified Hastelloy N,
are under construction.
L Systems and Analysis

J. R. Engel

1.1 TRITIUM BEHAVIOR IN
MOLTEN-SALT SYSTEMS

Studies to elucidate the behavior of tritium in large
molten-salt systems were continued in this reporting
period. Additional calculations were made for the
1000-MW(e) reference-design MSBR to examine the
effects that an oxide film on metal surfaces might have
on the distribution of tritium. Analysis of the informa-
tion being generated by the tritium addition experi-
ments in the Coolant-Salt Technology Facility (CSTF)
was begun. As additional data and results are developed,
they will be incorporated into the MSBR studies.

1.1.1 MSBR Calculations

G.T. Mays

Calculations were performed to examine the potential
effects on tritium transport to the steam system caused
by the formation of oxide films on the steam side of the
tubes in the steam raising equipment of an MSBR. The
rate of diffusion of hydrogen (tritium) through metal
oxides typically is proportional to the first power of the
hydrogen partial pressure in the gas phase, as opposed
to the % power for diffusion through metals (i.e., the
diffusion process is molecular rather than atomic). In
addition, at moderate hydrogen partial pressures, the
permeability coefficients of the oxides may be as low or
lower than those of pure metals. Thus, at the very low
hydrogen partial pressures that would be expected in an
MSBR, oxide films could offer substantial resistance to
hydrogen (tritium) permeation. However, the efficiency
of such films would be limited by the degree of metal
surface coverage that could be established and main-
tained during operation of the system.

The computational model! for studying tritium
behavior at steady state provides for variation of the
metal permeability coefficients of the steam-system
tubes, but assumes that diffusion through the tube walls
varies only with the % power of hydrogen partial pres-
sure. Variations in metal permeability were considered
in previously reported results.> However, the model also
includes the effect of a mass transfer coefficient for
tritium transport through a salt film inside the tubes.
Since transport through the salt film depends upon the
first power of tritium concentration (or partial pres-
sure), this value was used to estimate the effects of
oxide films. Effective mass transfer coefficients were

computed which included the resistances of the oxide
films as well as those of the salt films.?

Tritium distribution calculations were made for a
variety of situations in which it was assumed that the
effective permeabilities of the oxide coatings in the
steam system were 1, 107!, 1072, and 1073 times those
of the bare metal at a hydrogen partial pressure of 1
torr (130 Pa). These results were compared with cases
without oxide coatings in which the permeabilities of
the bare metal were reduced by factors of 1, 10, 102,
and 10*. The comparisons were made at three values of
the U** /U ratio (10%, 10%, and 10%) and, in all cases,
sorption of hydrogen or HF on core graphite was as-
sumed to be negligible.

The results (Table 1.1) indicate that a low-
permeability oxide coating would be more effective
than a low permeability in the metal itself for limiting
tritium transport to the steam system. When an oxide
film resistance equal to that of the metal was added, the
rate of tritium transport to the steam system was ap-
proximately halved, as would be expected. (The total
resistance to tritium transport was not doubled because
of the contribution from the salt film.) The results with
a factor of 10* reduction in a steam-tube permeability
due to oxide formation indicate that tritium transport
to the steam system could be limited to the design
objective of 2 Ci/day. However, it may be unreasonable
to expect to obtain and maintain oxide films of this
quality in an operating system. 2

Additional calculations were performed to investigate
the effect of reduced permeability of the primary and
secondary loop walls through the formation of oxide
coatings. These coatings can be expected to form in a
manner similar to those expected on the steam equip-
ment. For a given steam-tube permeability, reducing the
permeabilities of the loop walls would be expected to
increase the amount of tritium transported to the steam

system. With the reduced loop-wall permeabilities, less

1. R. B. Briggs, A Method for Calculating the Steady-State
Distribution of Tritium in g Molten-Salt Breeder Reactor Plant,
ORNL-TM-4804 (April 1975).

2. G. T. Mays, in MSR Program Semiannu. Progr. Rep. Feb.
28, 1975, ORNL-5047, pp. 31 2.

3. Although this calculational approach assumes that the
oxide film is located inside the tubes rather than outside, it can
be shown that, for given oxide and metal permeabilities, this
arrangement slightly overestimates the rate of hydrogen permea-
tion through the wall.
Table 1.1. Effect of oxide films on tritium transport
to the steam system of an MSBRZ

Rate of * H migration to
steam systemn (Ci/day)

Ratio of oxide or
metal permeability  U**/U3*

to neminal metal ratio

Oxid~ film Reduced metal

permeability‘b inside tubes® permeabilityd

1 10° 811 1425

1 10° 656 1169

1 10¢ 115 203
107! 10° 173 1351
107! 10° 138 1114
107! 10° 23 198
1072 10? 19 662
1072 10° 16 575
10_2 104 3 142
1073 107 2 93
1073 10° 1.5 84
1073 10¢ <1 31

INo sorption of H, or HF on core graphite.
bAt a hydrogen partial pressure of 1 torr.
CWith nominal metal permeability.

9No oxide film.

tritium would permeate through the loop walls into the
primary and secondary system containments, elimina-
ting a potential sink for tritium. A higher tritium con-
centration (or partial pressure) in the secondary system
would result, creating an increased driving force for
tritium transport to the steam system.

The results of the calculations did indicate that, with-
out the presence of a chemical getter in the secondary
coolant, more tritium was transported to the steam
system when the primary- and secondary-loop wall per-
meabilities were reduced than in the same cases with
reference permeabilities for loop walls. However, more
importantly, for those cases where tritium transport to
the steam system had been reduced to the design-limit
objective of 2 Ci/day through chemical additions of H,,
HF, or a chemical getter, results showed essentially no
increase over the 2-Ci/day rate. Thus, it appears that
reduced loop-wall permeability has little effect on trit-
ium transport for cases where a tritium exchange mate-
rial is present in the secondary coolant.

1.1.2 Coolant-Salt Technology Facility .

J.R.Engel G.T.Mays

A 1000-MW(e) MSBR is expected to generate about
2420 Ci of tritium per full-power day. Calculations
showed* that, unless a major fraction of this tritium

were converted to a chemical form less mobile than
elemental HT, the rate of migration of tritium through
the metal walls of heat exchange surfaces to the steam
system could be unacceptably high. The purpose of the
tritium addition experiments in the CSTF is to simulate
the general conditions in the MSBR coolant-salt system
to determine the extent to which tritium can be held up
in the NaBF,-NaF salt. There is evidence that hydrogen-
containing compounds in the salt may retain significant
amounts of tritium.

In the experiments, the first two in a planned series,
tritiated hydrogen was diffused into the circulating salt
through the walls of a hollow Hastelloy N tube. The
tritium could accumulate in the salt, pass into the off-
gas system, or permeate through the metal walls of the
loop to the ventilated loop enclosure. Tritium concen-
trations were monitored in the salt and in the loop off-
gas. In the first two experiments, 85 and 97 mCi of
tritium diffused through the Hastelloy N injection tube.
For a detailed description of the experimental condi-
tions, see Sect. 2.2.

The computer program® for calculating the expected
tritium distribution in a 1000-MW(e) MSBR was modi-
fied to describe the CSTF and was used to calculate
potential tritium distributions for the experiments
under the following assumptions:

1. steady-state conditions,

2. only dissolution of elemental tritium (hydrogen) in
the salt with no chemical reaction with the salt or
any of its components,

3. all transport through metal walls varies as the %
power of hydrogen partial pressure.

Calculations were made for addition rates of tritiated
hydrogen equivalent to those achieved in the CSTF ex-
periments, using assumed loop-wall permeabilities rang-
ing from the value expected for bare metal to 1072 of
that value. The results (Table 1.2) show a significant
effect of loop wall permeability on the fraction of the
added material that could escape through the walls. This
table also shows the calculated steady-state concentra-
tions of elemental tritium in the salt (nCi/g) and in the
off-gas (pCi/cm?®) in the same units that are being used
in reporting experimentally observed concentrations.
Also shown are the inventories of elemental tritium in
the loop walls that would be associated with the calcu-
lated transport rates through the walls. Since these cal-

4. G. T. Mays, in MSR Program Semiannu. Progr. Rep. Feb.
28, 1975, ORNL-5047, pp. 3—-12.

5. R. B. Briggs and C. W. Nestor, A Method for Calculating
the Steady-State Distribution of Tritium n a Molten-Salt
Breeder Reactor Plant, ORNL-TM-4804 (April 1975).
Table 1.2. Calculated steady-state tritium distributions in CSTF for experimental addition rates

Addition rate Loop w.a‘ll Time to reach Fraction of addition Elementz.ll tr;tfl-um Eler.n.ental Tritium
Tritiated permeability 85% of steady- te which removal in off-gas tritium .
_ ) % of steady rate which permeates . inventory
mixture (fraction of state conditions? loop walls Fraction of addition concentration . otal walls
hydrogen  Tritium bare-metal " (hr) %) rate removed Concentration in salt (i)
(em®/hr)  (mCi/hr) value) (%) (pCi/em?) {nCi/g)
3.18 7.9 1 0.3 99.4 0.6 400 1.3 0.1
107! 9.5 77.1 22.9 15000 50 0.8
1072 38 14.9 85.1 55000 200 1.6
1072 42 1.6 98.4 64000 220 1.7
3.3¢ 9.3 1 0.3 99.4 0.6 500 1.7 0.13
: 107! 10.3 75.7 243 19000 64 1.0
1072 37 14.3 85.7 67000 230 1.9
10-3 42 1.5 98.5 77000 260 2.0

2Salt and off-gas concentrations only, longer times required for steady-state permeation through toop walls,
by tritium in hydrogen. '
€0.11% tritium in hydrogen.
culations represent steady-state conditions, and the trit-
ium addition experiments involve transients, it is useful
to consider the time required to reach the steady state.
At high loop-wall permeabilities, the concentrations of
elemental tritium in the salt and off-gas are low and
tend to reach steady values quickly for the assumed
addition rates (see Table 1.2). Somewhat longer times
are required to reach the higher concentrations associ-

ated with lower assumed loop-wall permeabilities. In all
cases substantially longer times are required to reach
steady-state rates of tritium release through the loop
walls. However, this has little effect on the ultimate
steady-state levels in the salt and off-gas.

Figures 1.1 and 1.2 show the results of tritium con-
centration measurements in the salt and off-gas from
the CSTF during the first and second tritium addition

ORNL-DWG 75-42897

104 — 7 l ] I —]
s é o SALT (nCi/g) _
- é ® OFF-GAS, WATER SOLUBLE (pCi/em3) | —
B g A OFF-GAS, ELEMENTAL (pCi/cm3) _
2 — / o . . —_
é
103 ?
— . =
n h— —
§ - TRITIUM ADDITION Y —
c 2 ¢ —
Q
5 7
o 102 é
> [— @ h—
S — A ° —
5 5 : / A 4 ® ° @ L :
> [ ° o
S 4l o
: / NE -
- 2 —. / o —_
> 7 o
3 ? N
g 401 Y o
O b - —
(& - ® o) —
v ‘ —
2 5 —
= — _
& — ™ T A A ]
2 (o) A A —
o o o
o
10° :_‘% 4 =
—4 | . =
ST | % -
2 - | ® —
10-! 4
15 17 19 21 23 25 27 29 34| 4
JULY AUG
DATE , 1975

Fig. 1.1. Observed tritium concentrations in CSTF, test 1.
10°

TRITIUM CONCENTRATION (nCi/g or pCi/cm>)
c-Sf\)

10

ot

ORNL-DWG 75-12896

— |V =
= T T T T T T 712
— o SALT (nCi/g) ]
| ® OFF-GAS, WATER SOLUBLE (pCi/m®) —
B A OFF-GAS, ELEMENTAL (pCi/cm?) —
— | Z® —
= . =
— g . =
A | _
\ / .
o | o

/) ®
— ¢ —
— . =
— L—TRITIUM ADDITION ]
| ®

. —
— . —
e p -
e o h—
A O
A
S— o) —
e (o] —
— A _
— A A —
— o] —
N £ ° ]
A A

I o —

/ 0 0
= | ‘ ° =
— ‘% A O—

/
|7 —

% —
- / —

7

Z

5 7 9 1 13 15 17 19 X

DATE, AUGUST, 1975

Fig. 1.2. Observed tritium concentrations in CSTF, test 2.
tests. The concentrations reported for the salt represent
tritium in a chemically combined form, since any ele-
mental HT trapped in the samples would have been re-
leased in preparing them for scintillation counting. The
tritium in the off-gas was present in two distinctly dif-
ferent chemical forms. Part of the tritium activity was
present in a water-soluble form, implying a chemical
compound, since HT does not interact significantly with
water at room temperature. The other form is presumed
to be elemental HT, since it was trapped in water after
passage of the sample stream through a bed of hot CuO.

In all cases the results are presented in Figs. 1.1 and
1.2 as reported, with no corrections for apparent base-
line concentrations. However, the results of samples
taken before and after each test suggest that nonzero
baseline concentrations were present. The apparent
baseline concentrations for the two experiments were:

_ Ist experiment 2d experiment

In salt 1.7 nCi/g 1 nCilg
In off-gas, elemental 1 pCi/cm3 1 pCi/cm3
In off-gas, water soluble 1 pCi/cm3 50 pCi/cm3

During the tritium addition period for the first experi-
ment the tritium concentration in the salt (Fig. 1.1,
open circles) increased, almost linearly, to a maximum
of about 100 nCi/g, and then decreased approximately
exponentially over the following 3 to 4 days to its pre-
test baseline concentration. In the second experiment
(Fig. 1.2, open circles) the tritium concentration in the
salt reached a maximum of 70 nCi/g and returned to the
baseline concentration about 6 to 7 days later. If base-
line corrections are applied to the salt-sample data, the
apparent half-lives for tritium removal from the salt for
the two experiments are 9.2 and 12 hr respectively. '

If tritium removal from the salt is assumed to be a
pure first-order process or combination of such proc-
esses, and if it is assumed that the processes were also
active during the addition period, the buildup of the
tritium inventory in the salt (for a constant addition
rate) should be described by

N =Ly
A

where

N(t) = tritium inventory in salt at any time ¢ during
. the addition,

- A = tritium addition rate,

X =time constant for the removal process (or proc-
esses).

If several first-order processes were involved in :the trit-
ium removal from the salt, the time constant A would
be the sum of the several individual time constants, but
the individual values would not be identifiable. Substi-
tution of the actual addition rate into this equation
gives -the expected tritium inventory in the salt at any
time, if all .of the tritium were reacting with the salt.
Conversely, substitution of observed inventory values
permits evaluation of the effective addition rate (the
rate at which tritium did react with the salt). In either
case the results may be expressed as tritium trapping
efficiencies with values of 85 and 50%, respectively, for
the two experiments. These trapping efficiencies imply
that significant quantities of the added material were
reacting with and being trapped (at least temporarily)
by the salt. Data from the second experiment suggest
the - presence of -other mechanisms with significantly
longer time constants for removal of tritium from the
salt. Because of the apparent scatter in the data at
longer times, the extraction of these time constants was
not attempted.

The water-soluble tritium in the off-gas during the
first experiment (Fig. 1.1, closed circles) did not in-
crease significantly until after the injection was com-
pleted, and then rose to 1750 pCi/cm?. The level then
dropped rapidly to about 50 pCi/cm?®, rose again.to
about 300 pCi/cm® 6.5 days after the addition, and
then decreased to lower values. In the second experi-
ment the watersoluble tritium in the off-gas rose rap-
idly during the addition period and reached a maximum
value at the end of the addition of 13,100 pCi/cm?3.
Also, the ratio of the concentration of water-soluble
tritium in the off-gas to that of the elemental form was
substantially greater in the second experiment than in
the first. . ‘

Owing to the apparent scatter in the data involving
the water-soluble tritium in the off-gas for the first ex-
periment, no ‘quantitative evaluation was attempted.
However, in the second test, the initial decrease in con-
centration has an apparent half-life of 18 hr, but the
data again suggest the presence of other time constants.
An attempt was made to separate the time constants by
assuming that the decay curve was made up of two
simple, first-order exponentials. This led to apparent
half-lives of 9.3 and 37 hr for the two processes. Numer-
ical integration of the water-soluble tritium data for the
second experiment yieided a total flow of 58 mCi

through the off-gas line during the removal period and

7.5 mCi during the addition period. Thus, a total of 65
mCi or about 65% of the tritium added is accounted for
as combined tritium in the off-gas stream during this
test. Since the concentration of elemental tritium was
always less than 0.01 of the combined tritium concen-
tration, the presence of any elemental tritium does not
significantly affect this observation.

The concentration of elemental tritium in the off-gas
samples rose during the addition phase of each experi-
ment and apparently began to decrease as soon as the
addition was stopped. The maximum concentration in
the first test was about 800 pCifcm?® and only 40
pCi/cm? in the second. In both cases the decrease in
concentration with time after the addition was too
irregular to justify any quantitative evaluation.

Although no measure of elemental tritium concentra-
tion in the salt is available, a value can be inferred from
the concentration in the off-gas by (1) assuming that
the elemental tritium in the off-gas samples represents
release from the salt and only from the salt, and (2)
assigning reasonable values to gas stripping parameters
in the CSTF pump tank. Concentrations of elemental
tritium calculated in this way indicate that the ratios of
combined/elemental tritium in the salt were about 50
and 530 in the first and second experiments respec-
tively. It appears that chemical interactions between the
tritium-containing compound in the off-gas and the new
metal of the sample line may have been responsible for
the high concentrations of elemental tritium in the off-

gas samples from the first test and that the actual ratio

of combined/elemental tritium in the salt may have
been higher than 50.

The inferred maximum concentration of elemental
tritium in the salt during the second experiment is
about 0.13 nCi/g. Extension of the calculated tritium
distribution with nominal metal-wall permeability
(Table 1.2) to lower concentrations indicates that, at
0.13 nCi/g, tritium permeation through the loop walls
could account for no more than about one-third of the
trittum added to the system. Since this is close to the
amount not accounted for in the off-gas samples, it
appears that the effective permeability of the loop walls
is near that of bare metal.

1.2 XENON BEHAVIOR IN THE MSBR

G. T. Mays

The computer program MSRXEP (Molten-Salt Reac-
tor Xenon Poisoning) describing the !?*Xe behavior in
the reference-design MSBR was used to perform calcula-
tions to study the effects of the Knudsen diffusion coef-
ficient for xenon in the bulk graphite and graphite coat-
ing of the reactor core on the !3%Xe poison fraction.
The program has been described previously .®»’

Following the fission of the fuel, the decay of tfie
mass-135 fission fragments is assumed to follow the
decay chain shown below:

g 135, B
(18.2 sec)

135 Sb
(1.7 sec)

1351

1 BSmXe
*(15.29 min)

135 (g & 135,

(3.0 X 10° yr) (stable)

1351

(6.59 hr) 5

o
&9 )

135Xe
(9.17 hr)

This diagram illustrates the half-life of each isotope and
the branching ratio of the 351 decaying to '*3MXe
and '3%Xe assumed for this study. Along with this
decay chain the following input data were used:

Bubble concentration: 44 bubbles per cubic centimeter
of salt

Total heljum dissolved in salt and present in gas bub-
bles: 1.0 X 107 mole/cm?

Bubble separator efficiency: 90%

For these conditions the mass transfer correlation in the
program gives a bubble mass transfer coefficient of
0.0166 cm/sec, which leads to a loop-averaged void frac-
tion of 0.55% with an average bubble diameter of 0.65
mm. The calculated ! *® Xe poison fraction is 0.0046.
The reference Knudsen diffusion coefficients for the
bulk graphite and graphite coating associated with the
0.0046 poison fraction are 2.58 X 107° and 2.58 X
107® cm?/sec respectively.* The bulk graphite values
were varied from'2.58 X 107° to0 2.58 X 107 c¢m?/sec,
assuming no graphite coating was present (Table 1.3,
cases 1, 3, 5, 7), to observe the effect on the poison
fraction. The low-permeability graphite coating — 0.28
mm thick — was assigned bulk-graphite values for the
Knudsen diffusion coefficient and porosity, making the
coating part of the bulk graphite for calculation pur-
poses. Under these conditions the porosity of the bulk
graphite was held constant at a value about 31 times

*The complete units for diffusion coefficient are (cm?® gas)/
(sec + cm graphite). .

6. H. A. McLain et al., in MSR Program Semiannu. Progr.
Rep. Aug. 31, 1972, ORNL-4832, pp. 11, 13.

7. H. A. McLain et al., in MSR Program Semiannu. Progr.
Rep. Feb. 29, 1972, ORNL4782, pp. 13, 16—-17.
Table 1.3. '**Xe poison fraction as a function of the
Knudsen diffusion coefficient for the graphite coating
and bulk graphite of the reactor core

Knudsen diffusion coefficient

(cm® gasfsec +cm graphite) ¢
: poison fraction

Bulk graphite? Graphite coating
1. 258x10°¢b No coating 0.0153
2. 2.58% 107 2.58 X 10°° 0.0152
3. 258x 1077 No coating 0.0140
4, 258X 10°¢ 2.58 X 1077 0.0145
5. 2.58x10°® No coating 0.0113
6. 2.58X 10 258X 10°* 0.0107
7. 258X 10°° No coating 0.0077
8. 258X 10°¢ 2.58 X107% ¢ 0.00469

ZBulk graphite porosity value remains constant in all cases, V31
times greater than the value for the graphite coating.

bReference value for bulk graphite.
CReference value for graphite coating.

dPoison fraction for reference case.

greater than that of the graphite coating.* In addition,

the Knudsen diffusion coefficient for the graphite coat-
ing was varied within the same, aforementioned range
while the diffusion coefficient of the bulk graphite was
held constant at its reference value of 2.58 X 107°
cm? /sec to observe the effects on the poison fraction
(cases 2, 4, 6, 8). The previously stated values involving
bubble characteristics and mass transfer were held con-
stant throughout this series of calculations.

The results (Table 1.3) indicate that Knudsen diffu-
sion coefficients for the bulk graphite and graphite coat-
ing at least as low as the reference values (2.58 X 107°
and 2.58 X 107° cm?/sec, i.e., case 8) would be re-
quired to meet the 0.005 target value for the '’*Xe
poison fraction. A diffusion coefficient of less than 2.58
X 107% c¢m? /sec would be required for the bulk graph-
ite with no coating, because of its higher porosity. If the
permeability of the graphite coating did not yield a dif-
fusion coefficient equal to that of the reference value,
such a coating would have little effect on xenon poison-
ing. The penalty for not coating the graphite is about
0.01 in xenon poison fraction, or 0.01 in breeding ratio,
if no attempt is made to decrease the permeability or
porosity of the base material.

It may be noted in case 3 that a slight reduction in the
Knudsen diffusion coefficient for the bulk graphite is

*In practice, substantial reductions in the Knudsen diffusion
coefficient probably would be accompanied by reduced po-
rosity.

Calculated '?*%Xe

more effective in reducing the '?°Xe poison fraction
than a similar reduction in the Knudsen diffusion coeffi-
cient for the graphite coating in case 4. In cases 6 and 8,
where the permeability of the coating is very low, re-
ducing the Knudsen diffusion coefficient for the graph-
ite coating affects the '?°Xe poison fraction much
more strongly.

1.3 NEUTRONIC ANALYSIS

H.T.Kerr D.L.Reed E.J. Allen

The neutronic analysis work during this reporting
period has involved several tasks aimed at additional
description of the neutronic characteristics of an MSBR
and the provision of neutronics information for the
fueled in-reactor irradiation experiments.

1.3.1 MSBR Studies

Neutronic analysis studies for the reference-design
MSBR are in progress in three areas:

1. development of a two-dimensional neutronic com-
putational model of the MSBR, using the computer
code VENTURE and reestablishing the operability
of a reactor optimization code (ROD),

2. updating the neutron cross-section data base used by
various computer programs,

3. calculation of the rate at which helium will be pro-
duced in the reactor vessel of the MSBR.

Computational models. A neutronic computational
model of the MSBR, using the computer code VEN-
TURE,® is being developed. VENTURE is a multi-
dimensional, multigroup, neutron diffusion computer
code. The MSBR model will have nine neutron energy
groups and the (r-z) geometry shown in Fig. 1.3. The
various zones in the model allow for different core com-
positions and cross-section sets. In addition to providing
a check of the design studies made with the ROD? code
using one-dimensional calculations, this model will per-
mit explicit evaluation of the nuclear reactivity effects
associated with localized core perturbations, such as
limited core voiding. Previously such effects were con-
servatively estimated from calculations for an infinite
medium of salt and graphite.

8. T. B. Fowler, D. R. Vondy, and G. W. Cunningham III,
VENTURE: A Code Block for Solving Multigroup Neutronic
Problems Applying the Finite-Difference Diffusion-Theory
Approximation to Neutron Transport, ORNL-5062 (October
1975).

9. H. F. Baumann et al,, ROD: A Nuclear and Fuel Cycle
Analysis Code for Circulating-Fuel Reactors, ORNL-TM-3359
(September 1971).
10

ORNL-DWG 75-13175

I |
I[ ll _ $ 1
| ' Tl
T T T T ! T
L R
| | | | I |
TR o
| | | | I
| I | | | |
| | | | ; l
-9 | | ' | | | 9
L NN o
t
7 |8] 5 4 P32 v 12| 3 4 56| 7
l I
| I
S | :
Co o
A Lo
E o
I A
o S B
|- N
Lo I
| I N
L | ‘1 1 | L
11— T
1T 1 i 1
| |
PERCENT THICKNESS (cm)
ZONE NO. _ FUEL SALT RADIAL AXIAL
C . CONTROL RODS ' 17.2 219.5
2 CORE 1A - 13.2 ©32.8 . 219.5
3 CORE 1A 13.2 50.0 219.5
4 CORE 1B 132 119.5 219.5
‘5 CORE Il A AND B 37.0 g4 - © 25.4
6 SALT ANNULUS 100.0 ' 5.1 5.4
7 GRAPHITE REFL. 1.0 76.2 © 61.0 (MAX.)
8 SALT ANNULUS 100.0 0.6 0.6
9 REACTOR VESSEL ' 5.1 5.1

Fig. 1.3. Two-dimensional computational model of MSBR.
ROD is a computer program for nuclear and fuel-cycle
analyses of circulating-fuel reactors. It consists essen-
tially of a neutronics subprogram, an equilibrium-
concentration subprogram, and an optimization subpro-
gram. Variables such as breeding ratio, fuel
composition, etc., can be optimized with respect to
cost. o

The operational status of the ROD code has been re-
established by running a test case for the reference-
design 1000-MW(e) MSBR, using the old cross-section
data previously generated for the MSR program. The
test case will be rerun using the ENDF/B-IV cross sec-
tions, and any significant differences will be evaluated
and reported.

Generation of updated cross-section data. The neces-
sary descriptive information for the neutronic model for
use in the computer code VENTURE has been col-
lected, and the most recent ENDF'? cross-section data
are needed. (The neutron cross-section data used for
 MSBR analysis were originally derived from the GAM-II
and ENDEF/B-I libraries with some ORNL modifica-
tions,!! and no recent updates have been made.) The
new cross-section data are being obtained exclusively
from the ENDF/B-IV data files, using the AMPX!?
processing system. This effort will provide evaluated
cross-section data and neutron energy spectra for

11

4. Perform a one-dimensional neutron transport calcu-
lation of the MSBR core to determine 123-group
spectra and collapse the 123-group cross-section set
to nine groups for each of the various zones in the
model. '

. Reorder the nine-group set from nuclide ordering to

group ordering; the cross sections are then ready for
use in VENTURE and ROD.

The initial processing step is capable of treating nuclides
in groups of from 1 to 3, depending upon the amount
of data in the ENDF/B-IV file for each nuclide. This
step is now complete for all the nuclides of interest,
except 232Th, -
Helium production in reactor vessel. The helium pro-
duction in the reactor vessel for the present reference

~ design and possible alternate designs will be estimated in

typical regions of an MSBR and will serve as the data

base for subsequent MSBR nuclear analyses. The steps
involved in this process are:

1. Calculate 123-neutron-energy-group cross sections
from the ENDF/B-IV library. The ENDF point data
for 39 nuclides are weighted over an assumed energy
spectrum to derive multigroup cross sections.

Thermal scattering cross sections are treated at 300,

600, 900, and 1200 K for each nuclide.

. Determine contributions to the multigroup cross sec-

tions from resolved resonances; resonance self-
shielding is treated for the various fuel configura-
tions at 900 K.
Perform fuel-moderator cell calculations for four
geometries to adjust the cross sections for the flux
depressions in regions having a high concentration of
fuel or moderator (cell homogenization calcula-
tions).

10. ENDF/B-IV is the Evaluated Nuclear Data File-Version
IV and is the national reference set of evaluated cross-section

data.
11. O. L. Smith, Preparation of 123-Group Master Cross Sec-

tion Library for MSR Calculation, ORNL-TM4066 (March
1973).

12. N. M. Greene et al.,, AMPX: A Modular Code System for
Generating Coupled Multtgroup Neutron-Gamma Libraries from
ENDF/B, ORNL-TM-3706 (1974).

conjunction with the neutronic modeling of the MSBR.
(Neutron energy spectra and flux magnitudes in the
reactor vessel as obtained from the. neutronic model
provide the basis for calculating helium productlon
rates.) ‘

Helium is produced in nickel- base alloys primarily
from these reactions:

ENi M5 Ni M

"®Ni 5 sSFe +4He,

56Fe + “He,

SONi 7, $7Fe+“He.

The ®Ni(n,«) and the ®°Ni(n,a) reactions are induced
only by high-energy neutrons, whereas the *®Ni(n,y)
and 5°Ni(n,a) reactions are induced primarily by low-
energy neutrons. In highly thermalized neutron energy
spectra, as in the MSBR vessel, the two-step reaction
58Ni(n,y)*?Ni(n,a) is the principal source of helium.
The *?Ni cross sections are not well known, but differ-
ential measurements are being made by ENDF partici-
pants.}® Also, helium analyses are available from several
irradiated nickel specimens, and effective integral cross

“sections will be derived from these data for comparisons

with the measured cross sections. : :

At present, some cross-section information is avallable
for the *?Ni(n,a) reaction. Values for.the 2200-m/sec
(i.e., 0.0253-eV) cross section have been reported as
13.7 barns!® and 18 barns.!® It has also been re-
ported!? that a large resonance occurs at 203.9 eV with
a total width, T, of 13.9 eV. From this information a
preliminary estimate of the shape and magnitude of the

13. F. G. Perey, Report to the U.S. Nuclear Data Commzttee
ORNL-TM-4885 (April 1975).
14. H. M. Eiland et al., Nucl. Sci. Eng., 53, 1 {(January 1974).
cross section can be deduced and 123-group cross sec-
tions generated.
From the Breit-Wigner one-level formula,

o) = (KIE) {1/1E-E,)" + 1714
where

K = constant,

E = neutron energy,

E, = resonance energy (203.9 eV),
' = total width (13.9 eV).

The constant K can be determined from the value of the
cross section at 0.0253 eV, which for this study is
assumed to be either 13.7 or 18 barns. Energy-
dependent cross sections can be generated, and the
helium production can then be estimated with the fol-
lowing equation:
Nye(t) = [03 0,/(02 — 0,)] N°
x {11 —exp (- 0, 9010,
~ [l —exp (0, 01 fos },

where

g, = (n,y) cross section of 3®Ni,
o, = absorption cross section of * ?Ni,

03 =(n,a) cross section of 3 °Ni,

N°

¢ = neutron flux,

initial >®Ni concentration,

t =time,

Ny (1) = helium concentration at time ¢.

1.3.2 Analysis of TeGen Experiments

Fission rates and tellurium production rates for the
fuel pins in the TeGen-1 irradiation experiment were
reported in the preceding MSR' semiannual report.'®
The fission rates were estimated by a flux mapping
experiment, direct flux monitoring of the TeGen-1 cap-
sule, and computational analyses. The tellurium concen-
trations in the fuel pins were calculated from these fis-
sion rates, but no estimates for the accuracy of the
calculated tellurium concentrations were given in the
report.

The accuracy of the ?33U fission product yield
data'® leads to an estimated uncertainty for the yield
of tellurium in the TeGen-1 capsule of about 13.5%.
Assuming that the uncertainty in the estimated fission
rates 1s *15%, the uncertainty in the reported tellurium
concentrations is about 20%.

12

The TeGen-2 experimental capsule is scheduled to be
inserted into the ORR for irradiation in October 1975.
Flux monitors will be loaded into the capsule prior to
the capsule’s insertion into the ORR. After the TeGen-2
capsule is removed from the reactor, the monitors will
be recovered and their induced activities measured to
develop estimates of the tellurium production rates for
TeGen-2.

1.4 HIGH-TEMPERATURE DESIGN METHODS

G.T. Yahr

Thermal ratchetting and creep-fatigue damage are
important considerations in the structural design of
high-temperature reactor systems. Simplified analytical
methods in ASME Code Case 1592 (ref. 17) and RDT
Standard F94T (ref. 18) permit the assessment of
ratchetting and creep-fatigue damage on the basis of
elastic-analysis results, provided a number of restrictive
conditions are met. Otherwise, detailed inelastic analy-
ses, which are usually quite expensive for the conditions
where they are currently necessary, are required to
show that code requirements are met. Analytical investi-
gations to extend the range over which simplified ratch-
etting and creep-fatigue rules may be used to show
compliance with code requirements are being performed
under the ORNL High-Temperature Structural Design
Program, which is supported in part by the MSRP.
Modeling procedures for applying the simplified ratchet-
ting rules to geometries and loadings prototypic of
those encountered in LMFBR component designs are to
be identified. Then the conservative applicability of
these ratchetting rules and procedures and of elastic
creep-fatigue rules will be demonstrated and placed on a
reasonably sound and defensible engineering basis.
Finally, an assessment will be made of the applicability
of the simplified design methods to Hastelloy N under
MSBR design conditions, and the importance of thermal
ratchetting in an MSBR will be determined.

15. H. T. Kerr and E. J. Allen, in MSR Program Semiannu.
Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 14—15.

16. M. E. Meek and B. F. Rider, Compilation of Fission
Product Yields Vallecitos Nuclear Center, 1 974, General Elec-
tric Company, NEDO-12154-1, (January 26, 1974).

17. Code Case 1592, Interpretations of ASME Boiler and
Pressure Vessel Code, American Society of Mechanical Engi-
neers, New York, 1974.

18. RDT Standard F9-4T, Requirements for Construction of
Nuclear System Components at Elevated Temperatures (Supple-
ment to ASME Code Cases 1592, 1593, 1594, 1595, and 1596),
September 1974.
The detailed plans for achieving the stated objectives
were given in a previous progress report.'® The basic
approach is to perform a relatively small number of
carefully planned and coordinated rigorous elastic-
plastic-creep ratchetting-type analyses of the geometries
illustrated in Fig. 1.4. Each geometry is subjected to the
axial, bending, thermal transient, and pressure loadings
described in Table 1.3.1 of ref. 19. Structural problems
1 and 2 are being analyzed at ORNL, using the
PLACRE computer program,*® while problems 3 and 4
are being analyzed by Atomics International and Com-
bustion Engineering, respectively, using the MARC com-
puter program.”! Each inelastic analysis will include a
complete code evaluation for accumulated strains and
creep-fatigue damage. Also associated with each in-

TYPE I: NOTCHED CYLINDRICAL SHELLS

TYPE 2' CYLINDRICAL SHELLS

Q=o—!—- ——]—-Q=0

AT,

{b) UNIFORM WALL WITH
DIFFERENTIAL RATCHETTING

(a) STEPPED WALL THICKNESS

r 'Tz

NNNNARNARNNNY

_1(‘: [

(¢) UNIFORM WALL WITH {d} BUILT-IN CYLINDER

AXaL TEMPERATURE VARIATION

13

elastic analysis are a number of elastic analyses to pro-
vide the input parameters required to apply the various
simplified ratchetting rules and procedures and elastic
creep-fatigue rules. The progress to date on these studies
is discussed below.

Both Al and CE have encountered difficulties in their
three-dimensional inelastic analyses. Although consider-

19, J. M. Corum and G. T. Yahr in MSR Program Semiannu.
Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 15-22.

20. W. K. Sartory, “Finite Element Program Documenta-
tion,” High-Temperature Structural Design Methods for
LMFBR Components Quart. Progr. Rep. Dec. 31, 1971,
ORNL-TM-3736, p. 66.

21. MARC-CDC, developed by MARC Analysis Research
Corporation, Providence, RI.

ORNL-DWG 757768

TYPE 3: NOZZLE-TO-SPHERICAL SHELL

|_—16"0D. X |
0375" WALL

Z e

S\

JUNCTION DETAIL
{TYPES 384!

76"1D. X 11875" WALL

TYPE 4 NOZZLE-TO-CYLINDRICAL SHELL
(IHX INLET NOZZLE)

18"0D. X 0.375" WALL

76" 1D X
11875 WALL

Fig. 1.4. Structural configurations used in the analytical investigation of the applicability of simplified ratchetting and creep-

fatigue rules.
able effort has gone into developing finite-element
models that are of a size that can be accommodated on
present-day computers and into improving the MARC
computer program, the large 3-D inelastic analyses are
proving considerably more expensive to. run than had
been expected.

The experience at Al and CE indicates the importance
of developing simplified methods of analysis. Three-
dimensional inelastic analysis of many realistic com-
ponent geometries is too expensive and time consuming
at present to be used routinely. Although developments
in computers and stress analysis programs may bring the
cost down in the future, it is desirable, meanwhile, to
minimize the number of inelastic analyses that must be
done.

1.4.1 Circular Cylindrical Shells*

Nine cases of circular cylindrical shells have been pro-
posed for the present study.!® Two of the cases involve
notched shells. The other seven cases involve axial varia-
tions in temperature, pressure, and/or wall thickness, or
a built-in wall. All nine cases were to be analyzed using
the ORNL in-house finite-element program PLACRE.

A ten-cycle inelastic analysis and a one-cycle elastic
analysis have now been completed for all nine cases.
Both the inelastic and the elastic results for all nine
cases have been completely postprocessed.

Because of modifications to the creep-fatigue damage
rules presently under study by the ASME Boiler and
Pressure Vessel Code Working Group on Creep/Fatigue,
it may be necessary to modify the ORNL postprocessor
and repeat some of the postprocessing to keep the
present study up to date.

1.4.2 Nozzle—to—Spherical Shell®

After some difficulties, the MARC computer code is
operational on the IBM computer at the Rockwell Inter-
national Western Computing Center, and check cases
have demonstrated that this code will perform satisfac-
torily.

Considerable effort has gone into developing the
finite-element model of the nozzle—to—spherical shell.
An isoparametric three-dimensional 20-node brick ele-
ment will be used to model the entire geometry. Be-
cause of symmetry about the plane of the applied
moment, only half of the nozzle—to—spherical shell has
to be modeled. There are six 30°-wide elements around

*Work at ORNL, by W_K. Sartory.
TWork at Atomics International, by Y. S. Pan.

14

the half-model. There are three elements through the
wall at the root section of the nozzle and only one
element through the wall in both the nozzle and the
sphere away from the intersection region.

A series of elastic analyses must be done, since this is
a thermal stress problem in which temperature varies
with time. Since the moment applied to the nozzle is
the only nonaxisymmetric load, the principle of super-
position will be used to reduce the cost of the elastic
analyses. A series of axisymmetric analyses were done
to determine the stresses due to the internal pressure
and temperature, and one three-dimensional analysis
was done to determine the stresses due to the moment
applied to the nozzle. The stresses from the three-
dimensional analysis will be added to the stresses from
the axisymmetric analyses to obtain the total elastic
stresses.

The axisymmetric model in the elastic analyses was
used to determine what maximum thermal load incre-
ment may be employed without having to do an exces-
sive number of iterations during each increment. On this
basis, the first cycle of the three-dimensional inelastic
analysis was divided into 32 increments. The first three
increments of the three-dimensional inelastic analysis
have been completed. The computer cost for these three
increments was higher than anticipated. Efforts will be
made to find some way to reduce the cost to an accept-
able level.

1.4.3 Nozzle-to-Cylinder Intersection®

The original concept for the inelastic ratchetting-type
analysis of the nozzle-to-cylinder intersection was to
perform two separate analyses: (1) a thin-shell analysis
of the whole structure, and (2) a detailed three-
dimensional solid analysis of :the intersection only. Dis-
placements and forces to be applied at the boundaries
of the three-dimensional solid model of the intersection
were to be determined from the shell model at the end
of each loading increment. The total computer time of
the two analyses would be less than that required for
the solution of the problem, using one model of the
complete nozzle-to-cylinder intersection with suffi-
ciently small elements in the intersection region. How-
ever, the transfer of the forces and deflections from the
shell analysis to the three-dimensional solid analysis was
found to be more difficult than anticipated. Because the
shell element and solid element have different displace-
ment functions, a special constraint must be imposed on
the shell elements at the boundaries of the three-
dimensional solid model to assure compatibility. This

FWork at Combustion Engineering, by R. S. Barsoum.
stiffens the intersection in the shell model. When run-
ning the initial elastic analyses, it was found that small

15

changes in the displacement boundary conditions

applied to the solid model would produce large changes
in the results of the analysis. From a pragmatic view-
point, the biggest difficulty with the two-model method
is assuring that the correct data are transferred from the
shell analysis to the solid analysis at every increment in
loading. ' '

Due to the above considerations, it was decided to do
the "analysis by using only one model, made up of a
combination of a reduced integration shell element and
a 20-node solid element which are fully compatible with
each other. - ‘ ,

It was necessary to restructure a large portion of the
MARC program to perform the inelastic analysis for the

3-D model of.the nozzle-to-cylinder intersection. This
restructuring made a larger core available for the analy-
sis. The restructuring involved stripping unneeded por-
tions of the program, putting common space on low-
cost storage, and eliminating mesh optimization and its
correspondence table.

The inelastic analysis of the nozzle-to-cylinder inter-
section was - started. The full pressure and nozzle-
moment loadings were imposed on the structure, which
resulted in stresses less than 0.936 of the yield stress at
870 K (1100°F). When the first increment of thermal
load was applied, convergence was not obtained because
of an error in the computer program, which is being
corrected.
2. Systems and Components Development

R. H. Guymon

2.1 GAS-SYSTEMS TECHNOLOGY FACILITY

R. H. Guymon G.T. Mays

After a brief shutdown at the beginning of this report-
ing period to modify running clearances in the pump,
water operation of the Gas-Systems Technology Facility
(GSTF) was resumed on March 11, 1975, with the
bypass loop blanked (Fig. 2.1). Considerably larger salt-
pump shaft oscillations were encountered than before
the labyrinth clearances were increased.! After obtain-
ing calibration data for the main-loop variable-flow re-
strictor and for the salt pump at low flows, the loop was
shut down to install the bypass loop variable-flow re-
strictor. Water testing was then resumed on April 14
and continued throughout the period.

Data for calibration on the bypass loop variable-flow
restrictor and for the salt pump were obtained. At
normal pump speed the head-capacity performance of
the installed impeller was ~3% below the nominal loop
design conditions. At the nominal liquid flow rate and
pressure drop in the main loop, the flow rates from the
gas outlets of the bubble separator were satisfactory.
Although loop cavitation (as indicated by noise level)
was reduced by replacing the variable-flow restrictors
with orifices, the amplitude of the salt-pump shaft oscil-
lations was not reduced appreciably. Preliminary infor-

1. R. H. Guymon and W. R. Huntley, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 23-25.

SALT PUMP
PURGE GAS

SALT PUMP
OFF-GAS

BUBBLE
GENERATOR

BULK SALT SEPARATOR

G
|

mation shows that leakage past the salt-pump shaft
labyrinth is higher than desirable, and attempts will be
made to reduce this.

Tests under actual operating conditions with water in
the loop indicated that the densitometer will be satisfac-
tory for salt operation. Preliminary information ob-
tained from saturating the loop water with air and then
stripping the air by injecting helium at the bubble gener-
ator indicated the need for monitoring the oxygen con-
centration in the off-gas from the bulk salt separator, in
the off-gas from the salt pump, in the loop water, and
perhaps in the water in the pump tank. Difficulties were
also encountered with the response time of the oxygen
monitors and with the reproducibility of their readings.

2.1.1 Cavitation and Salt-Pump Shaft Oscillations

Data on the salt-pump shaft deflections and oscilla-
tions obtained during the previous period! indicated
that the running clearances at the labyrinth (fountain
flow area) and at the impeller hub should be increased
to prevent contact of the metal surfaces during opera-
tion with salt (Fig. 2.2). After increasing the clearances,
water operation was restarted with the bypass loop
blanked off. The shaft oscillations were much larger
than they had been previously under similar conditions.
Turbulence or cavitation as indicated by noise was the
apparent cause. For more flexibility in operating condi-
tions, the bypass-loop variable-flow restrictor was in-
stalled. Loop parameters were then recorded at many

ORNL-DWG 75- 12614

BULK SALT SEPARATOR
OFF-GAS

FLOW
REST

......

Fig. 2.1. Gas-Systems Technology Facility.
ORNL -DWG 75-12642

/;’ SHAFT
\ \ 7
. BOWL N
\ \ ?
N N
N ! |~ SLINGER
|  FOUNTAIN FLOW . el L—— FOUNTAIN FLOW
N el ==_"-F UPPER AND LOWER
N / - =- = = LABYRINTH
\ p— —_— —_ T x e— — f .
\:__., - ’___—..____ '
N——— - voLuTe — =
Q———-- - - —— -NIMPELLER"— — T
\‘_ —_— =
N

—— IMPELLER HUB
LABYRINTH

Fig. 2.2. GSTF salt pump.

combinations of salt-pump speed and settings of the
main-loop and bypass-loop variable-flow restrictors.

Log-log plots were made of pressure drops across
various sections of the loop as functions of the flow
rates through the segments. Since the head loss for a
fixed resistance is proportional to a fixed power of the
fluid velocity, the curves should be straight lines unless
the character of the resistance changes due to cavita-
tion. The plots indicated that cavitation was occurring
in the main loop between the inlet to the main-loop
variable-flow restrictor (FE-102A) and the throat of the
bubble generator at flow rates above 320 gpm (1200
liters/min) with the variable-flow restrictor set at 1 in.
(25 mm), above 470 gpm with the variable-flow restric-
tor at 2 in. (1800 liters/min at 51 mm), above 600 gpm
with the variable-flow restrictor at 3 in. (2300 liters/min
at 76 mm), and above 630 gpm with the variable-flow
restrictor at 4 in. (2400 liters/min at 102 mm). The data
were not sufficiently precise to determine whether cavi-
tation was also occurring in the bypass loop; however,
noise indicated that it was.

Since the loop turbulence and/or cavitation as indi-
cated by noise and the salt-pump shaft oscillations were
unacceptable at conditions required by the bubble
separator design, changes were made in the main-loop
and bypass-loop flow restrictions. By replacing each
variable-flow restrictor with two or more orifices in
series, the loop noise level was decreased, but there was

17

little or no decrease in the amplitude of the shaft oscil-
lations.

The amplitude of the shaft oscillations was plotted as
a function of salt-pump speed at various operating con-
ditions (Fig. 2.3). At salt-pump speeds less than about
1600 rpm, the oscillations were reasonably small and at
any given speed appeared to be unaffected by: (1) flow
rates between 450 and 1050 gpm (1700 to 4000 liters/
min), (2) salt-pump overpressures between 5 and 15 psig
(1.3 X 10° to 2.0 X 10° Pa), (3) type of restriction
(variable-flow restrictors, orifices, or a combination of
these), or (4) flow route (through the main loop, bypass
loop, or both). At higher speeds the oscillation ampli-
tude increased rapidly with increases in speed, and there
was more scatter in the data, making it difficult to eval-
uate improvement in cavitation and effects of other
variables. However, at any given speed above about
1700 rpm, increasing the flow rate (between 450 and
1050 gpm) caused larger oscillations.

One possible explanation for the increased amplitude
of the oscillations at higher speeds is that the shaft is
approaching its critical vibration frequency and is there-
fore more sensitive to disturbances, such asloop turbu-
lence or cavitation. The critical speed of this impeller
assemnbly is 2280 rpm in air,' which would indicate a
maximum normal operating speed of 1710 rpm, using
the normal industrial practice of operating pumps at less
than 75% of critical speed.

If the pump shaft oscillations were, in fact, a con-
sequence of operation near the critical speed of the
rotating assembly, two obvious alternatives were avail-
able to reduce the amplitude of the oscillations:

1. further reduction of the loop disturbances to mini-
mize the driving forces that cause oscillation,

2. operation at lower speeds to reduce the oscillatory
response to disturbances.

The first alternative was rejected because it would have
required extensive modification of the loop, and it was
difficult to guarantee that all sources of such disturb-
ances could be reduced to satisfactory levels. Design
calculations showed that the desired flow and head
(3800 liters/min at 30.5 m, or 1000 gpm at 100 ft)
could be obtained by replacing the present 113%-in.-diam
(290-mm) impeller with a 13-in.-diam (330-mm) unit
and operating it at 1500 rpm. A larger impeller is being
machined from an available Hastelloy N rough casting.
Since the larger impeller will be somewhat heavier than
the original one, it will cause a reduction in the critical
speed of the rotating assembly. The estimated critical
speed with the new impeller is 2000 rpm, which makes
the operating speed 75% of the critical speed.
ORNL-DWG 75-12613

140
o
120
. ®* 450-649 gpm TOTAL FLOW 4
100 o 650-849 gpm TOTAL FLOW Ao
a B850 -1049 gpm TOTAL FLOW ‘a
] g
E
;_:; 8o
a
~J a o
-
@
o 60 e
-
w .
®
40 & o
. & o ¢ ¢
mjta O
[ ] o.o ° [ %] ®
20 S o000 "
Y [ ] [ ] [+]
b L ] ® oP [ ] ©
S R Y SN SR
o
1000 1400 1200 1300 1400 1500 1600 1700 1800

SALT PUMP SPEED (rpm)

Fig. 2.3. GSTF pump shaft oscillations.

At a few off-design conditions during some of the
later runs, the pump shaft deflection records showed
random spikes in one direction superimposed on the
relatively uniform oscillations described earlier. These
occurred with higher than normal flow rates in the main
loop or at reduced system overpressure. Since either
increasing the overpressure or injecting gas at the bubble
generator reduced or eliminated these random oscilla-
tions, it was concluded that they were a consequence of
cavitation at the bubble generator. Such cavitation and
the attendant oscillations are not expected to occur at
normal operating conditions.

2.1.2 Salt-Pump Performance Data and Calibration
_of the Variable-Flow Restrictors

The original design of the GSTF provided for varying
the salt-pump speed and/or changing the variable-flow
restrictor settings to obtain different flow rates or pres-

sures needed for future experiments. However, instru-
mentation will not be provided for measuring the
bypass-loop flow rate during salt operation, and only
two salt pressure measuring devices will be installed (at
the salt-pump discharge and at the bubble-separator dis-
charge). Also, since the salt pump was modified and has
a mismatched impeller-volute combination, no perfor-
mance data were available. Therefore, extra pressure
indicators were installed for the water tests, and loop
pressure profiles were obtained at various pump speeds,
flow rates, and variable-flow restrictor settings to evalu-
ate the pump performance.

The calibration of the main-loop variable-flow restric-
tor and of the salt pump at low flow rates was straight-
forward, since, with the bypass loop blanked off, the
total pump flow was measured directly by the main-
loop venturi. However, once the bypass-loop variable-
flow restrictor was installed, the calibration of it and
the pump was complicated. The main-loop variable-flow
restrictor was closed, and the bypass-loop variable-flow
restrictor was calibrated at low flow rates, using the
pump calibration curves established before it was in-
stalled. The main-loop variable-flow restrictor was then
opened to various settings, and the pump calibration
curves were extended by adding the measured flow
through the main loop to the flow through the bypass
loop taken from the bypass-loop variable-flow restrictor
calibration curves. Then using these extended head-
capacity curves for the pump, it was possible to extend
the calibration curves to higher flows.

The pump calibration curves (Fig. 2.4) indicate that at
1770 rpm the pump flow rate will be 970 gpm (3700
liters/min) at 100 ft (30.5 m) of head. The original
design called for 500 gpm (1900 liters/min) through
each loop; however, the bypass flow rate can be reduced
to 470 gpm (1800 liters/min) without compromising
any of the objectives.

To determine the main-loop variable-flow restrictor
setting for normal operation with a flow rate of 500
gpm in the main loop, plots were made of the pump
head vs flow for several settings of the flow restrictor.
From these, a curve was made of pump head at 500
gpm (1900 liters/min) vs settings (Fig. 2.5). A 1.85-in.
(47-mm) setting will give the desired head of 100 ft
(30.5 m) at 500 gpm. '

ORNL —DWG 7512614
T

140 —
| i
—— .
120 1770 rpm ——- S —
' l ;
| { !
N
100 | : :
—_ : W
§- \1500 rpm :
- 80 ™~ :
(=} H
E I
Q
w 60
I
a
=
2
a .
40 —_ ———— .- ,XQ,, —
——
20 T~
0 .
0 200 400 600 800 1000 1200
FLOW (gpm)

Fig. 2.4. Head capacity curves for the GSTF pump.

. . ORNL-DWG 75-12615
5
T l 1
~_§ MAIN VFR (FE-102A) SETTING FOR 500 gpm
2 BY-PASS VFR (FE-104A)
" SETTING FOR 500 gpm
2\ 3 BY-PASS VFR (FE-1044)

SETTING FOR 470 gpm
i

~N

5
|
|
|
|

_
/]
rd

PUMP HEAD {feet of liquid}
/
7

2
2 N
o'
0 ! 2 3 4 5 6

VARIABLE FLOW RESTRICTOR SETTING (in)

Fig. 2.5. Variable-flow restrictor calibrations.

The bypass variable-flow restrictor settings were deter-
mined similarly and found to be 1.85 in. (47 mm) at
500 gpm (1900 liters/min) or 1.70 in. (43 mm) at 470
gpm (1800 liters/min).

2.1.3 Salt-Pump Fountain Flow

The GSTF salt pump is a centrifugal sump pump
having an impeller which rotates in a volute section
which in turn is located in a pump bowl. The clearance
between the impeller and the volute assembly at the
pump inlet allows leakage from the discharge directly to
the pump suction (see Fig. 2.2). A second bypass flow,
called the fountain flow, escapes through the clearances
between the impeller shaft and the volute assembly.
This bypass stream flows into the pump bowl, circulates
downward, and reenters the main stream at the pump
suction. Due to the large liquid holdup and large surface
area in the pump bowl, significant gas-liquid mass trans-
fer can occur in the fountain flow stream, and therefore
its flow rate is important in analyzing mass transfer
processes in the loop. Since the fountain flow is not
measured directly, a method using mass balances on
measured gas flows was developed to determine this
flow rate.

A lumped-parameter model of the GSTF was used to
develop equations from which an expression for the
fountain flow was derived. The system model contains
two major regions: the pump bowl and the primary
loop (main and bypass segments), consisting of a gas
section and a liquid section. Each section was assumed
to be perfectly mixed. The three general time-
dependent equations for a specific gas in a gas mixture
are given in Table 2.1, representing gas mass balances
for the pump-bowl gas section, the pump-bowl liquid
section, and the primary-loop gas section (circulating
voids).

These were simplified by applying the following
assumptions:

1. There is no gas carry-under in the pump bowl, which
implies that (2) the efficiency for separation of bub-
bles from the fountain flow (ef) is unity and Fj, =
Fr(1 — ¥, )+ Fpg,(b) the bubble surface area in the
pump bowl (4, ), the void fraction in the pump bowl
(¥ppg), and the concentration of gas in the pump
bowl (Cs) are nonapplicable or zero.

. Mass transfer equilibrium exists in the primary loop,
which implies that the mass transfer term 4343 (Cs-
KRTC,) is zero.

. Steady-state conditions exist, making all time deriva-
tives zero.

. Fp =0, since there was no gas purge flow during the
experiments.

Therefore, Eq. (3), Table 2.1, reduces to

Fp—F¥; —Q¥,e,=0. (4)
Solving for ¥ ,
v, =F3/(Ff+Q€s)- (5)

By adding Eqgs. (1) and (2), Table 2.1, and simplifying,
Fp¥; Cs + FAl — ¥, )Cs + FggCa
— FC —F{1 = ¥,)C; +FpgCy =0. (6)

By substituting Eq. (5) into Eq. (6), Fr may be ex-
pressed in terms of a quadratic equation:
(Ca — Cg)Ff2 t [(Qe, — Fg + Fpg)Cs — Cy)
+FpCy — FCi ] Fp
Qe [Fge(Cy —C2) —F,C,] =0. (7)

Equation (7) is a general solution for the fountain
flow, which depends upon the gas concentrations in

20

each of the three sections of the model. If it is assumed
that mass transfer equilibrium exists at the gas-liquid
interface in the pump bowl, the gas concentration in the
pump bowl liquid (C,) is related to the corresponding
concentration in the pump bowl gas section (C;) by
Henry’s law. If only one gas is involved (e.g., helium),
C, follows directly from the pump bowl overpressure.
Further, since mass transfer equilibrium was assumed
for the primary loop, the gas concentration in the loop
liquid (for a single gas) follows from the loop average
pressure and Henry’s law. Thus, the fountain flow may
be evaluated from Eq. (7), using other known liquid
flow rates and measurable gas flow rates into and out of
the system. If no mass transfer is assumed to occur at
the gas-liquid interface in the pump bowl, the gas con-
centration in the liquid leaving the pump bowl is the
same as that in the entering liquid (i.e., C; = C,), and
Eq. (7) reduces to
Qe

F.= ]
T (FRCyIF Cy) =1

@)

Since the rate of mass transfer in the pump bowl is
neither infinite nor zero, Eq. (7) will give a low indica-
tion and Eq. (8) will give a high indication of the foun-
tain flow rate. The deviation from the actual fountain
flow rate will depend on how much mass transfer actu-
ally occurs in any experiment. If the loop void fraction
is increased (by increasing the gas input rate), the con-
tribution of mass transfer across the gas-liquid interface
to the flow rate of gas out of the pump bowl will be
reduced relative to the bubble contribution. Therefore,
a plot of the calculated fountain flow vs the reciprocal
of the gas input rate at several different conditions
should give the actual fountain flow rate when extra-
polated to zero (infinite gas flow rate), using either Eq.
(7) or (8).

The preceding equations and approach were used to
calculate the fountain flow for the GSTF pump. Results
from the plot indicated that the curves generated were
not defined well enough to provide accurately the re-
quired extrapolations. The range of fountain flows at
the highest gas input rate at which data were obtained
was 100 to 200 gpm (380 to 760 liters/min).

Even the lower estimated value for the fountain flow
may excessively complicate future mass transfer experi-
ments; so efforts will be made to reduce this flow. Since
the labyrinth clearances cannot be reduced without
incurring metal-to-metal contact between the pump
shaft and the volute, back vanes will be installed on the
top of the impeller to minimize the differential pressure
which drives the fountain flow.
Table 2.1. Gas mass balance equations for computation model of GSTF

Rate of change of gas purge ' bubbles separated : mass transfer of flow of oft-
gas inventory in = flow in + from fountain + dissolved gas across - gas from
_ pump bowl in gas space flow ‘ liquid-gas interface ‘ pump bowl
Equation V, ci'cr‘l = FpCp + Ff\lrLefCJ + hA (C, —KRTC)) - F.C
1 .
Rate of change dissolved gas dissolved gas mass transfer mass transfer dissoived gas
of gas inventory present in ' present in of dissolved of dissolved present in flow
dissolved in = incoming fountain +  flow from — £as across —  gas to bubbles _— from pump
pump bowl liquid flow bubble genera- liquid interface in pump bowl bowl to loop
tor via
bulk salt
dcC separator
Equation V, a’tz = Ff(l - )HC, +  FpoC, - hoA, (C, —KRT(C,) - h,A, (C, —KRTC) - Fp (1 -¥pp)C,
2
Ratio of change flow of flow of bubbles mass transfer bubbles in dissolved gas bubble removed
« of gasinventory gas to from pump bowl of dissolved fountain . in fountain by bubble
in loop voids =  bubble +  toloop +  gas to bubbles —  flow — - flow —  separator
dv generator in loop '
Equation  V, L = Fp +  Fi¥pp + hid, (C, — KRTC,) - Fe¥per R A &) - OV jcg

3 dt

12
2.1.4 Densitometer Studies

The void fraction of the liquid after it leaves the bub-
ble separator must be known in order to evaluate the
bubble-separator efficiency. Densitometer instrumenta-

tion (using a digital voltmeter for readout) was installed

at the loop, and tests were made using the 30-Ci (1.1 X
10'% dis/sec) '37Cs source. The effects of changing
void fraction were simulated by inserting plastic sheets
3 and 6 mils (0.076 and 0.152 mm) thick between the
source and detector and by using the metallic shim slide
containing stainless steel calibration plates 10 to 250
mils (0.254 to 6.35 mm) thick which were designed for
this purpose.

The drift encountered during development testing was
still present. The hourly drift would be equivalent to a
change in bubble-separator efficiency of about 10%

22

during salt operation (assuming a void fraction of 0.3%

at the inlet to the bubble separator). Thus, short-term
tests will be required to determine the bubble-separator
‘efficiencies.

Based on densitometer readings, the bubble-separator

efficiency was greater than 98% at various operating
conditions with water, which is slightly higher than
predicted. :

Nonmenclature

A, = area of liquid-gas interface in pump bowl
A, = bubble surface area in pump bowl

A5 = bubble surface area in loop

C, = gas concentration in pump bowl gas space
C, = gas concentration in pump bowl liquid
C; = gas concentration in the loop bubbles

C, = gas concentration in loop liquid

C; = gas concentration in pump bowl bubbles

Cp = gas concentration in gas purge entering the pump
bowl gas space

F, = flow of total off-gas from pump bowl

Fp = gas flow rate to bubble generator

Fgs = liquid flow from bubble separator via the bulk

salt separator to the pump bowl (assume no bub-
bles)

F¢ = fountain flow (liquid and bubbles)

F, = flow of gas purge into pump bowl gas space

F; =flow of liquid and bubbles from pump bowl to
loop

h, =mass transfer coefficient for gas dissolved in
pump bowl liquid to gas space in pump bowl

h, = mass transfer coefficient for gas dissolved in
pump bowl liquid to bubbles in pump bowl

h, = mass transfer coefficient for dissolved gas in loop

liquid to bubbles in loop liquid

K = Henry’s law (solubility) coefficient
Q = flow of liquid and bubbles to bubble separator
R = universal gas constant
T = temperature
¥, = total gas volume in pump bowl
V', = volume of liquid and bubbles in pump bowl
¥y = volume of circulating liquid and bubbles in main
loop
€s = bubble separator efficiency
er = efficiency for separation of bubbles from foun-
tain flow
WV, =void fraction in loop fluid
W pp = void fraction in pump bowl fluid

2.2 COOLANT-SALT TECHNOLOGY
FACILITY (CSTF)

A. N. Smith

Modifications to the salt cold trap (SCT) were com-
pleted,” and the loop was started up on March 14,
1975, and operated for 1279 hr to check the effective-
ness of the salt mist filter, to obtain data on salt mist
generation rates under different operating conditions,
and to obtain salt and off-gas sample data in preparation
for the tritium tests. Work was completed on design,
fabrication, installation, and checkout of the tritium
addition system, and the tritium test program was
started. At the end of the report period, two tritium
additions had been completed, and plans were being
made for additional tritium addition tests as well as
tests designed to examine how the tritium behavior is
affected by the injection of steam into the salt.

2.2.1 Loop Operation

The SCT flow lines were disconnected from the sys-
tem, and the loop was started up on March 14, 1975.
The loop operated continuously until May 6, 1975,
when it was shut down to permit installation of equip-
ment in the containment enclosure for the tritium tests.
The loop was started again on June 27, 1975, and it was
still in operation at the end of the report period, when
more than 2500 hr of operating time had been logged
without plugging in the off-gas line. This is convincing
evidence that the salt mist filter has been effective, since
the off-gas line had plugged after only 240 hr of opera-
tion before the mist filter was installed.?

The loop is operating at a pump speed of 1790 rpm

~ (estimated salt flow rate, 54 liters/sec) and a pump bowl

2. A. N. Smith, MSR Program Semiannu. Progr Rep. Feb.
28, 1975, ORNL-5047, pp. 25-29.
3. Ibld p. 26.
gas overpressure of 2.67 X 10° Pa (2000 mm Hg abs).
The pump bowl off-gas flow; which consists of helium
containing a few percent of BF3, plus trace quantities
of condensible material, is about 2 liters/min (STP). The
BF; concentration of the off-gas is a function of the
BF; partial pressure in the salt, which in turn is a strong
function of the salt temperature. Except for short
periods of time when special tests required a different
setting, the salt circulating temperature has been main-
tained at 535 to 540°C, at which point the BF, concen-
tration in the off-gas stream is about 2.5% by volume.
The loop off-gas stream, except for a 100-cm?/min
sample stream, is passed through a —72°C cold trap (dry
ice—alcohol bath). Material which is a:dirty white solid
at trap temperature and a dirty brown fluid upon warm-
ing to room temperature, and which is rich in tritium
(about 10°® nCi/g), continues to collect in the cold trap
at a rate of 1 to 10 ng/cm?® (STP) of off-gas. This mate-
rial is believed to be a variable mixture whose composi-
tion depends on the relative partial pressures of BFj;
HF, and H, O over the salt (see Sect. 4.1).

As of 0800 on August 31, 1975, the loop had accu-
mulated 3073 hr of salt circulating time since being
reactivated in December 1974. -

2.2.2 Salt Mist Test

Between March 25, 1975, and April 24, 1975, a series
of tests was run to determine the concentration of salt

23

mist? in the off-gas stream as a function of salt tempera-
ture and BF; flow. For each test, the -loop operating
conditions were set at the desired values, and the off-gas
stream was shunted through a metallic 5- to 9-um filter
which was inserted into the-salt-sample access nozzle on
the pump bowl. The salt mist concentration was calcu-
lated using the gain in weight of the filter and the total
flow of gas. A total of ten tests were carried out. The
test time was normally about 12 to 15 hr, but in two
cases it was shortened to about 3 hr because of the
buildup of a high-pressure drop across. the filter. Pump
bowl pressure was 2.67 X 10° Pa, and total off-gas flow
was 2.1 liters/min (STP). When BF; was added to the
helium entering the pump bowl, the BF; flow was ad-
justed so that the BF, partial pressure in the incoming
gas was the same as the calculated partial pressure of the
BF; over the salt, assuming the eutectic mixture of
NaBF, and NaF. The salt circulating temperature was
controlled at either 535 or 620°C. The observed con-
centrations of mist in: the off-gas (Table 2.2) ranged
from ~100 ng/cm?®-at the lower temperature to as high
as 500 ng/cm?® at the higher temperature. At the lower
temperature; where the expected partial pressure of
BF, in the salt was low, the addition of BF; with the
cover gas was ineffective in reducing the amount of mist
in the off-gas. At the higher temperature (higher BF,

4. Ibid, p. 27.

Table 2.2. CSTF salt mist test, March 1975

Calculated
Salt BF, BF, - .
temperature vapbr flow Salt mist concentration -
: ‘ 3 °C o .
Q) pressure? (cm?®/min, STP) L(ng 5alt/(.m- STP off-gas)
(mm Hg)
620 250 250 198
336
160
Average 230
620 250 0 308
‘ 532
532
Average 455
535 50 50 80
100
Average 90
- 535 50 0 111
88
Average 95

2 Assuming the eutectic composition.
partial pressure in the salt) a significant reduction was
observed in the mist concentration when BF3; was
added with the cover gas. However, it was not reduced
to as low a value as at the lower temperature. These
results, while not completely definitive, suggest that
BF; evolution from the salt may not be the only mist-
producing mechanism in the pump tank; that is, simple
mechanical agitation of the salt in the pump bowl may
also produce some mist. In addition, no data were ob-
tained with excess BF5 concentrations in the cover gas.
Since the installation of the salt-mist filter in the off-gas
line was effective in eliminating the operational prob-
lems in the CSTF caused by the mist, further investiga-
tions of methods to limit or control the mist have been
deferred in favor of experiments to study tritium be-
havior in the system.

2.2.3 Tritium Experiments

The decision to use tritium rather than deuterium as a
test gas® in the CSTF necessitated additional design
effort and a somewhat more elaborate test setup in
order to satisfy applicable radiation safety require-
ments. A conceptual design was prepared for the tritium
addition system, and a preliminary radiation safety
analysis was performed for the proposed test. Engineer-

TRITIUM
TRANSFER
CYLINDER .

400 °«C

—

]

PURIFIER

SAMPLE

i -

HYDROGEN

24

ing design, procurement, fabrication, and installation of
the tritium addition system were completed by the
third week in June 1975. The addition tube and the
addition procedure for tritium are essentially the same
as those devised for the addition of deuterium. The
inner tube of the addition assembly is pressurized with
hydrogen containing a small amount of tritium, and the
gas is allowed to diffuse through the Hastelloy N tube
which forms the lower end of the addition tube, and
which is immersed in the {lowing salt stream (see Fig.
2.6). The Hastelloy N tube is 120 mm long by 12.7 mm
in OD by 10.6 mm in ID, and provision is made to
fasten metallurgical specimens to the upstream face.
The portion of the addition tube immediately adjacent
to the Hastelloy N section is surrounded by an evacu-
ated annulus monitored to check for extraneous tritium
leakage. The hydrogen-tritium mixture is passed
through a purifier (Pd-Ag tube) to remove impurities
such as O,, N,, and H, O which might interfere with
the permeation process. The probe volume, V), (injec-
tion tube plus adjacent tubing), and a calibrated refer-
ence volume, V,, are interconnected and pressurized
with the H,-T, mixture at the start of the test. The two
volumes are then isolated from each other while the
addition is in progress. At the end of the addition

ORNL-DWG 75-12616

7
gz
VACUUM 1 m_ 35°C
VACUUM
ANNULUS~_||
100 °C
INJECTION
TUBE
SALT o
FLOW °32 C&

Fig. 2.6. Tritium addition system for GSTF.
period, the difference in pressure between Vy, and V, is
recorded; then the two volumes are equilibrated and the
final equilibrium pressure is recorded. The initial and
final pressures in Vp, py and p, respectively, the final
equilibrium pressure, p3, and the known volume and
temperature of ¥, are then used to calculate the
amount of gas which permeated the addition probe
according to the equation

(1 —p2)(P, —Pa)x V,
(p3 — P2) RT,

r

n=

where 7 is the number of moles of gas transferred, R is
the molar gas constant, 7, is the"reference volume tem-
perature, and the other symbols are as previously de-
fined. '

During the addition, the amount of extraneous leak-
age is calculated from pressure rise measurementsin the
evacuated annulus, and this quantity is subtracted from

n to obtain the net amount of gas transferred into the

salt. The tritium content of the hydrogen-tritium mix-

ture is determined by mass spectrometer analysis, and

the net amount of added tritium is then calculated.
Tritium (and hydrogen) which enters the salt stream is

25

assumed either to remain in the salt or to leave the salt -

by one of two paths: either by permeation through the
walls of the loop piping or by transfer to the gas phase
in the pump bowl or in the salt monitoring vessel (SMV)
and leaving the loop with the off-gas stream. During and
after an addition, the tritium content of the salt is mon-
itored by taking samples of salt from the salt pool in the
pump bowl or in the SMV, and the tritium content of
the off-gas stream is monitored by taking samples from
the off-gas line at a point about 1 m downstream of the
pump bowl. The off-gas sample stream is passed first
through a water trap to collect chemically combined
(water-soluble) tritium, and then through an oxidizing
atmosphere to convert elemental tritium to tritiated
water, which is collected in a second trap. The tritium
contents of the salt samples and of both the off-gas
samples are determined by a scintillation counting tech-
nique. During the initial tritium addition experiments,
no provision was made for measuring loop wall permea-
tion, so that the tritium lost by this mechanism is
assumed to be the difference between the amount of
tritium added and the sum of the quantities which leave
in the off-gas stream and which remain in the salt.
During the March 14, 1975, to May 6, 1975, oper-
ating period, a number of salt and off-gas samples were
taken to obtain baseline values for tritium concentra-
tion and to shake down and evaluate the sampling tech-
niques. During shutdown of the CSTF in May and June

1975, the tritium addition probe was installed in the
surveillance-specimen access tube, and the final installa-
tion work was done on the tritium addition system. A
stainless steel valve (HV-253A) and some stainless steel
tubing, which were part of the original off-gas sample-
line installation, were removed and replaced with a
Monel valve and Hastelloy N tubing, because it was felt
that the Monel and Hastelloy N would be less likely to
react with the off-gas sample stream. Two 2.5-cm-diam
X 45-cm-long Hastelloy N tubes were filled with salt
from the drain tank and set aside as representative
samples of the salt as it existed prior to the start of the
tritium tests.

On June 27, 1975, the loop was filled dnd salt circula-
tion was resumed. Several additions of hydrogen were
made to check out the operation of the addition system
and to obtain data on permeation rates. A total of 313
cm® (STP) of hydrogen was added in these tests, and
the last addition (31 ¢cm® STP) was made on July 16,
1975. With the addition tube pressurized to 1.37 X 10°
Pa, the measured permeation rate was about 3 em? /hr,
compared with a predicted value of 1.3 cm?/hr. The
first addition of tritium was made on July 17, 1975,
and a second addition, with conditions essentially the
same as for the first addition, was made August 5, 1975.
In each case, salt and off-gas samples were taken during
the tritium addition (about 10 hr) and for about 2
weeks afterward, until sample results indicated that the
tritium levels had returned to their pretest values or had
stabilized. Data for calculation of the amount of added
gas are shown in Table 2.3. A mathematical analysis and
discussion of the sample results are presented in Sect.
1.1.2.

Table 2.3. Tritium addition data for CSTF tests

Test number T1 T2
Date T-17-75 8-5-75
Addition started 1002 0846
Addition ended 2050 1912
Addition time (hr) 10.8 10.4
Initiat pressure, p | (Pa) 1.38 x 10° 1.37 X 108
Final pressure, p2 (Pa) 1.06 x 10° 1.07 x 108
Equilibrium pressure, p3 (Pa) 1.24 X 10° 1.23 X 10°
V, volume (cm?) 155 155
T, temperature ("K) 303 306
Gross permeation rate (cm?® /hr) 3.10 3.33
Average leak rate (cm?/hr) 0.07 0.04
Net permeation rate (cm?/hr) 3.03 3.29
Tritium concentration in mixed 1010 1100

gas (ppm) ‘
Total tritium added {cm?) 0.033 0.038
Total tritium added {(mCi) 85 97

- 2.3 FORCED-CONVECTION LOOPS

W.R. Huntley M. D. Sllverman H. E. Robertson

'The Forced Convection Corrosion Loop Program is
part of the effort to develop a satisfactory structural
‘alloy for molten-salt reactors. Corrosion loop
MSR-FCL-2b is operating with reference fuel salt at
'typic‘:al MSBR velocities and temperature gradients to
evaluate the corrosion and mass transfer of standard
‘Hastelloy N. Addition of tellurium to the salt in
MSR-FCL-2b is planned after baseline corrosion data
are obtained in the absence of tellurium. At this time,
the loop has operated approximately 3000 hr at design
AT conditions with the expected low corrosion rates.

Two additional corrosion loop facilities, designated
MSR-FCL-3 and MSR-FCL4, are being constructed.
They are being fabricated of 2% titanium—modified
Hastelloy N alloy which is expected to be more repre-
sentative of the final material of construction for an
MSBR than standard Hastelloy N.

2.3.1 Operation of MSR-FCL-2b

Loop ‘FCL-2b was operated continuously for about
3000 hr from February to June 1975, under design AT
(565°C minimum, 705°C maximum) conditions. During
this period, standard Hastelloy N corrosion specimens
installed in the loop in January 1975 were exposed to
circulating fuel salt at three different temperatures,
(565, 635, and 705°C). As expected, corrosion rates
were low; the highest value was 0.1 mil/year (3um/year)
at the highest temperature station.

Salt samples taken at intervals have been analyzed for
major constituents, metallic impurities, and oxygen

26

(Table 2.4). Except for an occasional high value for

oxygen or iron, the analyses are relatively consistent
and indicate that the observed corrosion processes have
had very little effect on the concentrations of the
various species present in the fuel salt. Analytical probe
readings for the U*/U3 ratio, indicative of the redox
condition of the salt, have been taken on a weekly basis.
This ratio, which was about 7 X 10? at the beginning of
the corrosion run, rapidly dropped to about 1 X.103
after the first 24 hr of operation. The ratio then gradu-
ally fell to ~v1 X 10% by the end of March (1500 hr
elapsed time), and it has remained at that level during
the latter part of the operation.

After the corrosion specimens were removed for the
3000-hr weight-change measurements, preparations
were made for obtaining heat transfer data on the
Li-Be-Th-U fuel salt (71.7-16-12-0.3 mole %). At this

time' a Calrod electric tubular heater failure was dis-
covered on the pipe line (12.7-mm-OD X 1.1-mm-wall)
which runs from metallurgical station No. 3 to the inlet
of cooler No. 1. After removing the thermal insulation,
about 10 to 20 cm® of salt was found on the loop
piping and the burned-out heater. Grainy material was
present on the heater sheath at three locations directly
opposite peeled-off sections of oxide layer on the
Hastelloy N piping. A small crack ("v5 mm long) was
found on a tubing bend directly under the failed heater.
Whether the heater arced, causing the piping to fail, or
whether the salt leak from the loop caused the heater
burnout is uncertain at this time. Examination of speci-
mens from these regions is continuing.

The fuel salt was drained from the loop into the fill-
and-drain tank after the leak was discovered. Analytical
results on a sample taken from the tank indicated that
no obvious contamination of the fuel salt had occurred.
A new section of piping was installed (approximately
2.4 m, from metallurgical station No. 3 to the inlet to
cooler No. 1)..During the shutdown, several defective
thermocouples and two defective - clam-shell electric
heaters were replaced. Ball valves were refurbished,
numerous small repairs were made, and instruments
were recalibrated. After the thermal insulation had been
replaced, baseline heat loss measurements were made,
with no salt in the loop, in preparation for taking heat
transfer data. The loop was ready for refilling at the end
of July, approximately four weeks after ‘the salt leak
was discovered. After filling the loop, heat -transfer
measurements were -obtained with flowing salt.” The
ALPHA pump speed was varied from 1000 to 4600
rpm, resulting in salt flows of approximately 2.7 to 16
liters/min, which correspond to Reynolds numbers that
vary from 1600 to 14,000. The lower limit for salt-flow
was set to prevent freezing, and the upper limit was
dictated by the power reqt{lred for driving the pump. At
the lowest flow rate, unusual wall temperature profiles
were noted, which probably were caused by entrance
conditions and transitional flow effects. The heat trans-
fer measurements were completed near the end of this
reporting period, and analysis of the data'is in progress.

The stringers containing the Hastelloy N -corrosion
specimens were reinserted in the loop, and AT opera-
tion (565°C minimum, 705°C maximum) was resumed
in order to complete the-originally planned 4000-hr cor-
rosion run. If no unusual corrosion behavior is encoun-
tered in the next 1000 hr of operation, nickel fluoride
(NiF,) additions will be made to the loop in order to
raise ‘the oxidation potential of the salt to a level corre-
sponding to a U*/U% ratio of about 10%, and a new set
of corrosion specimens will be exposed.
27

Table 2.4. Salt sample analysis during MSR-FCL-2b
operation with LiF-BeF, -ThF, -UF,“

Total hours

Trace materials

Date Major components
Sa;lnple mpled , of sa.l.t (wt %) - (ppm) Notes
s (1975) circulaticn L Be , -
when sampled i e Th 9] F Fe Cr Ni O C S
1b 1-17 : 0 78 2.21 43.2 1.11 46.7 101 40 23 <50 11 9.9 Flush salt
2b 1-23 48 . 60 42 5.2
3b 1-28 0 7.99 1.74 43.0 098 46.3 75 70 15 125 29 43 New salt
4b 2-11 177 ' i 137 63 60 15 :
5b 2-18 355 154 64 68 45
6b 2-24 498 98 63 28 48
7b 33 676 7.68 2.36 428 1.05 46.3 147 67 35 45 23 17
8b 3-25 1146 7.01 2.55 43.0 1.04 45.8 256 59 57 <125 14 9
9b 4-16 1647 8.16 229 432 097 45.2 45 70 30 20 78 15
10b 5-12 2197 8.29 2.64 43.0 1.03 45.5 62 85 . 30 60
11ib 6-9 2838 8.23 2.25 42.7 1.00- 445 30 70 25 140
12b 6-23 3173 8.20 2.08 43.3 1.04 45.1 35 15 25 152
13b 7-3 3177 8.30 2.18 430 1.04 452 70 .80 40 30 Fill-and-drain
‘ ) ’ . ) tank
3246 7.28 2.03 1.00 450 45 85 70 58

14b 8-7 45.0

471.7-16-12-0.3 mole %.

2.3.2 Design and Construction of FCL-3 and FCL4

RN

The design work for FCL-3 and FCL-4 was essentially
completed; any changes or revisions which occur during
construction of FCL-3 will also be made on FCL4.

The piping support frame for FCL-3 was installed, and

installation of electrical equipment is proceeding. Con- -

duit lines have been run from the variable-speed motor-
generator set on the ground floor up to the electrical
rack installed on the experiment floor, and a sizable

number of transformers, starters, switches, etc., have
been installed. The instrument panel cabinets have been
positioned, and cable trays are now being installed.
Fabrication of two ALPHA-pump rotary elements and
two pump bowls is 90% complete. A large number of
completed items for both loops (e.g., dump tanks, auxil-
iary pump tanks, cooler housings, blower-duct assem-
blies, electric drive motors, purge gas cabinets, etc.) are
on hand, awaiting installation. Fabrication of the
titanium-modified Hastelloy N tubing for the salt piping
of the loop is in progress.
Part 2. Chemistry

L. M. Ferris

Chemical research and development related to the
design and ultimate operation of MSBRs are still con-
centrated on fuel- and coolant-salt chemistry and the
development of analytical methods for use in these
systems.

Studies of the chemistry of tellurium in fuel salt have
continued to aid in elucidating the role of this element
in the intergranular cracking of Hastelloy N and related
alloys. An important initial phase of this work involves
the preparation of the pure tellurides Li, Te and LiTe,
for use in solubility measurements, loop experiments,
electroanalytical studies, and studies of tellurium redox
behavior in molten salts. Techniques for preparing these
tellurides have been developed, and experimental quan-
tities have been prepared. Spectroscopic studies of tellu-
rium chemistry in molten salts and of the equilibrium
%H,(g) + UF4(d) = UF;(d) + HF(g) have also been
initiated. In work using molten chloride solvents, at
least two light-absorbing tellurium species have been
shown to be present. These species are as yet unidenti-
fied, but have compositions in the range Li,Te to
LiTe, . Preliminary values of the quotients for the above
equilibrium have been obtained, using Li;BeF, as the
solvent. These values are in reasonable agreement with
those obtained previously by other workers.

" A packed-bed electrode of glassy carbon spheres was
constructed, calibrated with Cd?* ions, and used in
experiments with Bi®" ions in LiCI-KCl eutectic. It was
concluded that this electrode was prototypic of one
that could be used for the electroanalysis or electrolytic
removal of bismuth, oxide, and other species in MSBR
fuel salt. Preliminary experiments were also conducted
to evaluate some questions relating to the mixing of fuel

28

and coolant salts. The results suggest that, on mixing
small amounts of coolant salt with large amounts of fuel
salt, the rate of evolution of BF; gas will not be intoler-
ably high and that some oxide can be present in the
coolant salt without effecting precipitation of UQ, or
ThO,. Lattice enthalpies of first-row transition metal
fluorides were calculated to provide a theoretical basis
for evaluating thermochemical data for structural-metal
fluorides. '

Work on several aspects of coolant-salt chemistry has
continued. Analyses of condensates from the Coolant-
Salt Technology Facility (CSTF) indicate that the vapor
above the salt is a mixture of simple gases such as BF,
HF, and H, O rather than a single molecular compound.
Tritium concentrates in the condensates by about a
factor of 10° relative to the salt. Studies of the system
NaF-NaBF4-B,0; at 400 to 600°C show that at least
two oxygen-containing species are present in typical
coolant salt. One species is Na; B; F, O5, while the other
has not yet been identified.

The development of analytical methods for both fuel
and coolant salt was also continued. An in-line voltam-
metric method was used to monitor U*/U3* ratios in
two thermal-convection and one forced-circulation
loops. Two additions of tritium were made at the CSTF.
The salt in the loop did significantly retain tritium, and
the tritium ultimately appeared in the off-gas. Work was
begun on using various electrodes for determining iron
in MSBR fuel salt. Previous work had been conducted
with solvents that did not contain thorium. Preliminary
voltammetric experiments were conducted to identify
soluble electroactive tellurium species in MSBR fuel
salt.
3. Fuel-Salt Chemistry

A.D. Kelmers

3.1 COMPOUNDS IN THE
LITHIUM-TELLURIUM SYSTEM

D.Y. Valentine A.D. Kelmers

It has been demonstrated that tellurium vapor can
induce shallow grain-boundary attack in Hastelloy N
similar to that observed on the surfaces of the fuel-salt
circuit of the MSRE.! However, the actual oxidation
state or states in which tellurium is present in MSBR
fuel salt, an LiF-BeF,-ThF4-UF; mixture, and the
chemical reactions with the Hastelloy N surfaces remain
to be determined. The lithium-tellurium system is being
investigated to determine which Li-Te species can be
present and to synthesize samples of all possible lithium
tellurides. The solubility of these compounds in the fuel
salt will then be determined. In addition, they will be
used in spectrophotometric and electrochemical investi-
gations of tellurium species in melts.

During this report period, samples of Li; Te and LiTes
were prepared. The preparations were made in an
argon-atmosphere vacuum box equipped with an en-
closed evacuated heater which held a molybdenum
crucible. All handling of Li-Te compounds was done in
inert-atmosphere boxes; sometimes the compounds
were sealed under vacuum to minimize oxygen, nitro-
gen, or H,O contamination. Lithium having an oxygen
content of <100 ppm was supplied by the Materials
Compatibility Laboratory, Metals and Ceramics Divi-
sion. Tellurium metal of 99.999+ wt % purity was
obtained from Alpha Ventron Products.

The Li, Te was first prepared by dropping small pieces
of lithium into molten tellurium contained in a molyb-
denum crucible at 550°C. The reaction was extremely
exothermic, emitting fumes and light flashes after each
lithium addition. Solid formation occurred at lower
lithium concentrations than expected from the reported
phase diagram.? Further lithium additions continued to
be absorbed after first melting on the surface of the
solid phase. An amount of lithium necessary to satisfy
the Li, Te stoichiometry was taken up in this manner.
However, because of the loss of vapor and of some solid
material which splashed out of the crucible during the
early additions of lithium, it is doubtful that the stoichi-
ometry was in fact preserved.

The x-ray diffraction pattern showed a single phase,
identified as Li, Te, having a face-centered cubic struc-

29

ture with a lattice parameter of 6.5119 + 0.0002 &.°
The oxygen contamination in the product totaled about
375 ppm. Spectrographic analysis reported 0.5 wt %
molybdenum present. Since the oxygen level and mo-
lybdenum impurities were fairly low, a larger-scale prep-
aration was attempted as well as a direct preparation of
LiTe; by the same method. In both cases the product
was contaminated unacceptably with molybdenum, and
these preparations were discarded. Apparently the first
preparation had affected the surface of the crucible
such that the reaction with molybdenum was acceler-
ated in these subsequent experiments.

The molybdenum crucible was used for one further
preparation, after cleaning and polishing the inside sur-
face. The Li, Te was prepared from the lithium-rich side
of the phase diagram by dropping tellurium into molten
lithium. Since molybdenum is relatively inert toward
lithium,* less reaction with the crucible was expected.
In addition, this preparation could be made at a much
lower temperature. The tellurium was added to the
lithium in small increments with the temperature held
at 250°C. Each of the first additions resulted in a
smooth, quiet reaction with a solid phase forming on
the bottom of the crucible. However, since completion
of the reaction was not visibly apparent, the tempera-
ture of the system was increased above the tellurium
melting point to about 550°C to ensure that unreacted
tellurium was not on the bottom of the crucible. More
additions of tellurium were then made. Above 500°C a
popping noise was heard after each addition of tellu-
rium. After about three-fourths of the tellurium had
been added, the system was mostly solid. As more tellu-
rium was added, the amount of solid in the system
became so great that further additions of tellurium were

1. A..D. Kelmers and D. Y. Valentine, MSR Program Semi-
annu. Progr. Rep. Feb, 28, 1975, ORNL-5047, p. 40.

2. P. T. Cunningham, S. A, Johnson, and E. J. Cairns, J.
FElectrochem, Soc.: Electrochem. Sci. Tech, 120, 328 (1973).

3. X-ray lattice parameters were measured by O. B. Cavin of
the Metals and Ceramics Division. The value 6.5119 + 0.0002 A
measured for Li, Te is in agreement with the value 6.517 A
reported by E. Zintl, A. Harden, and B. Dauth, Z. Elektrochem,
40, 588 (1934). The value 6.1620 + 0.0002 A measured for
LiTe, is in agreement with the value 6.162 A reported in ref. 2.

4. H. W, Leavenworth and R. E. Cleary, Acta Met. 9, 519
(1961).
not covered by the liquid. Subsequent additions pro-
duced light flashes and popping associated with the
highly exothermic reaction as encountered in the pre-
vious preparations of Li, Te. Finally, enough additional
tellurium was added to the system to satisfy the Li; Te
stoichiometry, and the system was allowed to cool to
room temperature. '
Upon crushing the cooled product, four differently
colored ‘substances were distinguishable: gray opaque
material, wine-red to pink opaque material, colorless
translucent crystals, and metallic tellurium. Analyses
were performed separately on each type of material:

1. Gray opaque material. The x-ray diffraction pattern
~revealed Li,Te and LiTe3‘; no other lines were
present. The oxygen level was about 218 ppm. Spec-
trographic analysis indicated the presence of about
0.1 wt % molybdenum.
. Red-hue material. Only a few crystals of all-red
material could be isolated. The remainder of the
red-hue material was ground together with some sur-
rounding gray material. The x-ray diffraction pattern
corresponded mainly to Li;Te. A small amount of
LiTe; was also present: The oxygen content was
reported to be about 275 ppm. Spectrographic
analysis reported <0.01 wt % molybdenum.
Colorless translucent and isolated red crystals. Both
these products gave an x-ray diffraction pattern
corresponding to pure Li, Te with no indication of a
second phase.

To ensure a uniform product, all the 'various colored
materials were recombined and thoroughly mixed. The
Li, Te mixture was then placed in a 2-in.-diam tungsten
boat, which had previously been enclosed in a quartz
bottle. The quartz bottle was then evacuated, sealed,
and heated to 550°C for about 16 hr. The product
obtained after cooling was almost completely cream-
white. However, when the bottle was broken open, the
product began to turn beige upon exposure to the envi-
ronment of the inert-atmosphere box. The product was
then crushed roughly and placed in sample bottles. On
standing in the bottles, the product gradually reverted
to the red-gray color it had been before the heat treat-
ment, with the exception that the product in one bottle
remained beige. The reason for the lack of uniform
behavior is as yet unknown. Some of the darkened
product was returned to the tungsten boat in another
quartz bottle and the heat treatment repeated. It again
turned the cream-white color. The products, both the
light beige and the red-gray color forms, gave x-ray dif-
fraction patterns for a single phase, Li, Te. Analysis of
this Li, Te is given in Table 3.1.

30

Table 3.1. Analysis of Li, Te and LiTe,

Li, Te LiTe,

Li (wt %) 9.5=+0.1 1.7+ 0.1
Te (wt %) 89.9+0.5 984+ 0.5
Li, Te (mole %) . 98.7+ 1.6

LiTe, (mole %) 14+05 829: 5.0
Te {mole %) ) : 17.1 £ 15.0

X-ray diffraction ‘Single phase
740 (£ 10%)
0.05

<0.01

Single phase
275 (£10%)
<0.01
" <0.01

Oxygen (ppm)
Molybdenum (wt %)
Tungsten (wt %)

Red-gray Li, Te was mixed with the amount of tellu-
rium required to satisfy the LiTey stoichiometry. The
mixture - was then sealed in a quartz bottle under
vacuum and heated to 550°C for 2 hr. The hot liquid
was dark metallic gray: On cooling, the solid appeared
bright silver-gray. The x-ray diffraction pattern con-
firmed the presence of a single phase, LiTe;, having a
near body-centered cubic structure with a pseudocell
lattice parameter of 6.1620 + 0.0002 A.> The well-
exposed Debye-Scherrer diffraction patterns suggest
that the structure of this compound is more complex
than previously reported.> Work will continue in an
effort to describe this structure. The oxygen content
was reported to be 275 ppm. Spectrographic analysis
reported no molybdenum or tungsten contamination.
Analysis of this LiTes is also given in Table 3.1.

3.2 SPECTROSCOPY OF TELLURIUM
SPECIES IN MOLTEN SALTS

B. F.Hitch L.M.Toth

A spectroscopic investigation of tellurium behavior in
molten salts has been initiated to identify the species
present in solution and to obtain thermodynamic data
which will permit the determination of the species’
redox behavior in MSBR fuel salt. A previous investiga-
tion> had indicated that Te; is present in LiF-BeF,
(66-34 mole %) on the basis of an absorption band
occurring at 478 nm when LiTe; was the added solute;

however, the work was terminated before these observa-

tions had been fully substantiated. The current work is
an extension of those earlier measurements which

5. C. E. Bamberger, J. P. Young, and R. G. Ross, J. [norg.
Nucl. Chem. 36, 1158 (1974).
should lead ultimately to a measurement of redox equi-
libria such as

LiTes + %H, + SLiF = 3Li, Te + SHF, (1)

% Te, + LiF + 4 H, = LiTe; + HF . (2)
These data should then permit the prediction of tellu-
rium redox chemistry as a function of UF;/UF, ratio.

During the past several months, most of the effort was
devoted to assembly of the apparatus necessary for the
fluoride measurements. This involved fabrication and
assembly of the following: a furnace for the fluoride
studies, diamond-windowed spectrophotometric cells, a
vacuum and inert gas system, and a KHF, saturator
through which H, is passed to generate HF-H, mixtures
of known proportions.

Also during the period of preparatlon some attention
was given to a supporting study in chloride melts.® The
advantages of working in chlorides are:

1. previous ground-work mvest1gat10ns have already
been reported,’ : :

2. chlorides are easier to hold in silica cells without
container corrosion,
3. the greater solubility of the tellurides in chlorides

may reveal greater detail because of more intense
spectra.

Absorption spectra have been measured for Li,Te,
LiTe5, and tellurium solutes as- well as during titrations
of Li, Te with Te, in the LiCI-KCl eutectic at 450 to
700°C. These data indicate that at least two light-
absorbing species are present in molten chlorides con-
taining lithium tellurides with compositions in the range

31

Li, Te to LiTé,~ Furthermore, an exarhination of Te, in -

the LiCl-KCl eutectic has indicated that there is a
second species present besides Te, which is formed at
high temperatures and/or high halide ion activity. More
detailed experiments are anticipated using purer lithium
telluride solutes in the diamond-windowed cell to
demonstrate that the additional species are not related
to impurities from the reagent or silica corrosion.

6. This work has done in cooperation with J. Brynestad of
the Metals and Ceramics Division.’

7. D. M. Gruen, R. L. McBeth, M. S. Foster, and C. E.
Crouthamel, J. Phys. Chem. 70(2),472 (1966).

3.3 URANIUM TETRAFLUORIDE-HYDROGEN
EQUILIBRIUM IN MOLTEN
FLUORIDE SOLUTIONS*

L. O. Gilpatrick L. M. Toth
The equ111br1um '

UF4(d) + %Ha (g) = UF3(d) + HF(g) (1)

is under investigation, using improved methods of analy-
ses and control. The effects of temperature and solvent
composition changes on the equilibrium quotient

[UF3] Pyr .
[UFa](p )/2

Q = (2)

are thq immediate objectives of this work and are
sought to resolve previous discrepancies noted in fuel-
salt redox behavior.® :

The procedure involves sparging a small (approx1-
mately 1 g) sample of salt solution (UF, concentration
of 0.038 to 0.13 mole/liter or 0.065 to 0.22 mole %)
with H, gas at 550 to 850°C until partial reduction of
UF, to UF'3 is observed; HF is added to oxidize the
desired amount of UF; back to UF,. When an equilib-
rium between the HF/H, gas mixture and the UF;-UF,
in solution is reached, a spectrophotometric determina-
tion of the UF; and UF,; concentrations is made. These
data are combined with the analytically determined®
HF/(Hg)l/z ratio to obtain the equilibrium quotient at a
given set of conditions.

The assembly -of the system for this experiment has
been completed, and measurements of equilibrium
quotients, using LiF-BeF, (66-34 mole %) as the sol-
vent, have been initiated. Some "delay has occurred
because of trace water in the HF/H, sparge gas which
was responsible for the hydrolysis of uranium tetra-
fluoride and the subsequent precipitation of UO,. The
problem has been partially alleviated by treatment of
the KHF, saturator, gas supply lines, and spectrophoto-
metric furnace with fluorine at room temperature. How-
ever, back diffusion of water vapor into the furnace
from the exit gas line has also caused substantial solute
losses and has been reduced by using higher HF-H, flow

*This research in support of the MSBR Program was funded
by the ERDA Division of Physical Research
8. L. O. Gilpatrick and L. M. Toth, “The Uranium
Tetrafluoride—Hydrogen Equilibrium in Molten Fluoride Solu-
tions,” MSR Program Semiannu. Progr. Rep. Feb. 28, 1975,
ORNL-5047,p.43.
rates. Together, these modifications have reduced the
solute losses to an acceptable level (2% per day).

Equilibrium has been achieved at 650°C for measured
UF, /UF, ratios of approximately 0.2 X 1072 t0 3.1 X
1072, Although the UF; /UF, values are reproducible at
fixed KHF, saturator temperatures, the analytically
determined HF values are not as yet. Consequently, the
standard error (approximately 50%) in the equilibrium
quotients is still rather high. So far, a value of 0= 107°
has been determined at 650°C, which compares favor-
ably with the previous® value of 9.16 X 1077, Most of
the immediate effort is being devoted to improving the
precision of the HF determination.

Tritium control in an MSBR would be favored by
higher equilibrium quotients. In an MSBR, the
UF;/UF, ratio will probably be fixed by equilibria
involving the structural metals. The tritium inventory
will be established by the tritium production rate and
the various tritium removal processes. If @ is larger than
previously anticipated, the partial pressure of HF would
be higher and the partial pressure of H, would be lower
than previously estimated. Thus, TH would be available
at a lower concentration for permeation through the
heat exchanger to contaminate the coolant loop (and
ultimately the steam system), and a larger proportion of
the tritium would be present as TF, which would be
removed in the helium gas stream.

3.4 POROUS ELECTRODE STUDIES
IN MOLTEN SALTS

H. R. Bronstein  F. A. Posey

Work continued on development of porous and
packed-bed electrode systems as continuous on-line
monitors of the concentrations of electroactive sub-
stances, especially dissolved bismuth, in MSBR fuel salt.
In previous work,'®'! a prototype packed-bed elec-
trode of glassy carbon spheres (100 microns in di-
ameter) was tested in the LiClI-KCl eutectic system.
Linear-sweep voltammetric measurements, carried out
in the presence of small amounts of iron and cadmium
salts, showed that the cell, instrumentation, and auxil-
iary systems functioned successfully and demonstrated

9. G. Long and F. F. Blankenship, The Stability of Uranium
Trifluoride, ORNL-TM-2065, Part II (November 1969), p. 16;
Eq. 6 with xyg, = 0.002.

10. H. R. Bronstein and F. A. Posey, MSR Program Semi-
annu. Progr. Rep. Aug. 31, 1974, ORNL-5011, pp. 49-51.

11. H. R. Bronstein and F. A. Posey, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 44.

32

the sensitivity of this method of analysis. However,
these measurements showed the need for redesign of the
experimental assembly to permit removal and replace-
ment of the cell and addition of substances to the melt.

During this report period the redesigned packed-bed
electrode of glassy carbon spheres was tested again in
LiCI-KCl (58.8-41.2 mole %) eutectic, since the be-
havior of a number of electroactive substances has al-
ready been established in this medium. The packed bed
of glassy carbon spheres was supported on a porous
quartz frit and contained in a quartz sheath. Another
porous quartz frit pressed on the bed from above. A
glassy carbon rod penetrated the upper quartz frit to
provide compaction of the bed and electrical contact
with a long stainless steel rod which was insulated from
the surrounding tantalum support tube. The electrode
assembly was dipped into the melt so that the molten
salt flowed up through the interior of the bed and out
an overflow slot. By this means it was possible to obtain
a reproducible volume of melt inside the packed-bed
electrode.

Voltammetric and coulometric scans of the pure melt
at 395°C showed that the background current was
small. A typical set of current-potential and charge-
potential background curves is shown in Fig. 3.1. A 2-V

ORNL-DWG. 75-11288

11]Tlrlllllillllllllllll =400
TEMPERATURE : ~395°C
REFERENCE ELEGTRODE: Ag/AgCI{~0.1 M ) in —-300
LiCI-KCI {eutectic}/ Pyrex glass
CADMIUM GONTENT : 0.064 coulombs /ml ~3.3 x10™*¢ ~37ppm |
PAGKED-BED ELECTRODE VOLUME : ~1.75mi _
GLASSY CARBON SPHERE DIAMETER: ~10C microns w
-200 2
[=]
5
3
E
—-100 ,
(]
o
<
I
BACKGROUND ©
> 511 —
o - o START
5 o
a | O
E £
2 —E
= - S START
E o = I_’.._
- L & }
E § BACKGROUND N S |
A - cd®-» ca?t
ER
[&] +5 1.1 1 |[ 1 1.1 1 ll |- | | Lol .1 II [ S
1.0 0.5 0.0 -0.5 -1.0 -1.5
ELECTRODE POTENTIAL (w#s Ag/AgCl)  (volis)

Fig. 3.1. Linear-sweep voltammetry and coulometry of cad-
mium in LiCI-KCl (58.8-41.2 mole %) eutectic with a packed-
bed electrode of glassy carbon spheres. Curve A: current back-
ground (sweep rate = 10 mV/sec); curve B: current with Cd**
present (sweep rate = 5 mV/sec); curve C: background charging
curve {(sweep rate = 10 mV/sec); curve D: charging curve with
Cd?* present (sweep rate = 5 mV/sec).
range of electrode potential could be swept without
evidence of significant amounts of oxidizable or reduc-
ible impurities in the melt. For calibration purposes, a
known quantity of cadmium ions (Cd**) was added to
the melt by anodization of molten cadmium metal con-
tained in a specially designed graphite cup which could
be lowered into the melt. The amount of cadmium
added (Fig. 3.1) was monitored by use of an electronic
autoranging coulometer.

Following addition of cadmium, the voltammetric and
coulometric scans indicated that only a small fraction of
the known cadmium content inside the void space of
the packed-bed electrode was being measured. After
removal of the cell assembly, examination showed that
the glassy carbon contact rod had somehow fractured,
possibly due to excessive pressure from the mating
stainless steel contact rod, and resulted in loss of elec-
trical contact with the packed-bed electrode.

The cell assembly was then redesigned and rebuilt to
permit electrical contact to be maintained without
undue pressure and to allow accurate measurement of
the working volume of the packed-bed electrode. The
new design was similar to that of the previous cell,
except that the upper fritted quartz disk was perma-
nently sealed to the surrounding quartz sheath. A small
hole in the center of the disk permitted loading of the
glassy carbon spheres into the electrode assembly and
provided accurate positioning of the glassy carbon con-
tact rod into the bed. Prior to loading of the spheres,
the volume contained between the porous quartz disks
was measured with mercury.

Some voltammetric and coulometric scans in the pres-
ence of cadmium ions are shown in Fig. 3.1. Asin pre-
vious studies in aqueous media with the packed-bed
electrode,'? more accurate analytical results were
obtained on the anodic half cycle (stripping) than on
the cathodic half cycle (deposition). Approximately 40
mC of cadmium was estimated to be within the packed-
bed electrode. The coulometric results shown in Fig. 3.1
are quite consistent with this value. Thus it is possible,
knowing the geometry of a packed-bed electrode, to
estimate the response and sensitivity within reasonable
limits (the accuracy of estimation depends upon void
fraction, the accuracy of the volume measurement, and
other factors). Repeated scans over a period of many
days showed good reproducibility and also established
that diffusion through the quartz frits during the time
of measurement (only a few minutes) has very little
effect on the results.

12. H. R. Bronstein and F. A. Posey, Chem. Div. Annu.
Progr. Rep. May 20, 1974, ORNL4976, pp. 109-11.

33

Another cell was packed with 200-u-diam glassy
carbon spheres and used to obtain the results shown in
Fig. 3.2. In this case a quantity of Bi*" ions had been
anodized into the melt in a manner similar to that used
for cadmium. At the time of these measurements the
same melt had been in use for many weeks. Fig. 3.2
shows voltammetric and coulometric anodic stripping
curves in the vicinity of the anodic peak for stripping of
bismuth which had previously been deposited on the
internal surfaces of the electrode during the cathodic
half cycle. In agreement with observations of others, we
found that volatility of BiCl; precluded close corre-
spondence between added and observed quantities of
bismuth, and that the bismuth peak decreased steadily
with time. The appearance of the bismuth peak suggests
that possibly some alloying of bismuth with the cad-
mium took place.

Other experiments on bismuth reduction and strip-
ping will be carried out in the future in which cadmium,
used for calibration of the cell system, is absent. In
addition, the present apparatus will be used to study the
electrochemistry of lithium telluride in the LiCl-KCl
eutectic. Observations on the tellurium system in the
chloride melt may be useful in interpretation of tellu-
rium behavior in later studies with MSBR fuel salt. The

ORNL-DWG, 75-11289

T T T T T T T T T
TEMPERATURE : 392°C
REFERENCE ELECTRODE : Ag/AgCH(~OiM) in
LiCI-KCI (eutectic)/Pyrex glass i
GLASSY CARBON SPHERE DIAMETER: ~ 200 microns
00 —
O
- 3 A
— - g —— —
¢ < - P4
g F l ,/'—’ ”” ] (E)
g T ~2mC LT 13
= 051 — 2
E =
- 1E
Bio——D Bi3+ —
=N Tw
0O 2+
W i 4@
x | Possibly Fe2+-+ Fe3+ Cd"—Cd 1%
3 1ok from Iron ;mpurity T 5
5mC A
1.5 | | 1 1 | | | | l 1
04 03 02 01 00 -0 -0.2 -03 -0.4 -05 -06

ELECTRODE POTENTIAL (vs Ag/AgCl) {volts)

Fig. 3.2. Linear-sweep anodic stripping voltammetry and
coulometry of bismuth electrodeposited onto a packed-bed
electrode of glassy carbon spheres. Solid lines: experimental
current-potential and charge-potential curves in the region of
the bismuth stripping peak; dashed lines: estimated background
charging curves.
capability of the packed-bed . electrode of glassy carbon
spheres for monitoring electroactive species in molten
salts has been shown to be satisfactory. Consequently,
plans are now under way for design and fabrication of
cells and apparatus for testing the electrode system in
molten fluoride media including MSBR fuel salt. In
bismuth-containing fluoride melts, whether bismuth is
present as Bi® or Li; Bi, or both, it should be possible to
identify and determine the quantities of each species.
The packed-bed. electrode offers hope of removing, as
well as monitoring, dissolved bismuth in the fuel salt
which may be present as a result of the reductive extrac-
tion process for removal of fission products.

3.5 FUEL SALT—COOLANT SALT
INTERACTION STUDIES

A.D.Kelmers D.E. Heatherly

In the alternate coolant evaluation,!® several areas of

potential concern were defined with regard to the appli-
cability of the conceptual design coolant salt [NaBF, -
NaF (92-8 mole %)] for MSBRs. These centered primar-
ily around events associated with off-design transient
conditions, particularly primary heat exchanger leaks,
which would allow intermixing of fuel salt and coolant
salt. If coolant salt leaked into the fuel salt, the quan-
tity and rate of evolution of BF; gas from reaction (1),

NaBF4(d) coolant - BF3(g) + NaF(d)fuel salt » (1)

would determine the transient pressure surges to be en-
countered in the heat exchanger and reactor. Also, pre-
vious work!'* indicated a substantial redistribution of
the ions Li*, Na*, Be**, F~, and BF,” between the result-
ing immiscible two-phase system formed on mixing
Li,BeF4 and NaBF,. The solubilities of UF, and ThF,
have not been measured in such systems; thus the dis-
tribution of uranium and thorium between such phases
and the resulting concentrations are unknown: In addi-
tion, if oxide species were present in the coolant salt,
either deliberately added to aid in tritium trapping or
inadvertently present due to steam leaks in the steam-
raising system, the precipitation of UO, following mix-
ing of an oxide-containing coolant salt with fuel salt has
not been investigated. Therefore, a series of experiments
were carried out to investigate these areas.

13. A. D. Kelmers et al., Committee Report: Evaluation of
Alternate Secondary (and Tertiary) Coolants for the. Molten-
Salt Breeder Reactor (in preparation).

14. C. E. Bamberger, C. F. Baes, Jr., J. P. Young, and C. S.
Sherer, MSR Program Semiannu. Progr. Rep. Feb. 20, 1968,
ORNL-4254, pp. 171-73.

34

The experimental apparatus consisted of a quartz
vessel heated by a quartz furnace so that the volume of
the resulting phases could be observed at temperature
and measured with a cathetometer. The quartz vessel
extended up out of the furnace and was closed with an
O-ring fitting and end plate. A nickel stirring shaft,
driven by a constant-speed dc motor, penetrated the
end plate and during the tests was driven at a speed
adequate to stir the two phases without appreciable
visible dispersion. Access for sample filter sticks was
provided through the end plate as was done also for the
argon inlet and exit lines. A very low argon flow main-
tained an inert atmosphere over the melt during the
experiment.

Predetermined weights of fuel salt [nominal composi-
tion LiF-BeF, ThF,-UF, (72-16-11.7-0.3 mole %)] and
coolant salt [nominal composition NaBF,-NaF (92-8
mole %)] were placed in the quartz vessel and rapidly
heated to 550°C. Bubbles of gas could be observed due
to BF; generation via reaction (1) as soon as the
coolant salt melted during the heatup period. When the
temperature reached 550°C, counted as time zero, stir-
ring was initiated. The volume of the phases was period-
ically determined, and filter-stick samples were taken at
30- or 60-min intervals. ‘

The reaction between the fuel salt and coolant salt
proceeded slowly: approximately 30 to 60 min was
required to complete the visible evolution of BF, gas at
550°C. When the initial coolant salt content was 20 wt
% or less of the total material, no coolant salt phase
remained after approximately 1 hr. All the NaF dis-
solved in the fuel salt phase and all the BF; gas left the
reaction vessel. With larger initial weights of coolant
salt, up to 50 wt %, a small residual volume of coolant
salt phase could be observed after 1 to 3 hr. Severe
corrosion of the quartz reaction vessel occurred at the
interface between the coolant salt phase and the argon
cover gas in the experiments with the larger initial
weights of coolant salt, presumably due to attack of the
quartz by BF; via a reaction such as

2BF;(g) + % Si02(c) =% SiF,4(g) + B, 05(d). ‘ (2)

In experiments 6 and 7, holes were corroded completely
through the vessel wall, and the surface of the stirred
molten coolant salt phase was exposed to air for I to 2
hr at 550°C. ' :

In most of the experiments, samples of the fuel-salt
phase were withdrawn at intervals of 1, 2, 3, and 4 hr.
Samples were also taken after the conclusion of the
experiment; after the melt had cooled to room tempera-
ture the quartz vessel was broken away from the solid
salt. All these samples gave essentially identical analyti-
Table 3.2. Composition of fuel-salt phase and coolant-salt phase after contact at 550°C

Experiment Initial mixture (wt %) Fuel-salt phase (mole %) Coolant-salt phase (mole %)

No. Fuel salt?  Coolant salt? LiF NaF  BeF, ThF, UF, LiF NaF  BeF, ThF, UF, Na, SiF, B,0, NaBF,
2 100 0 68.8 2.3 17.2 11.5 0.24
3 90 10 633 93 167 105  0.24 } .

4 80 20 604 123 16.5 10.6 0.25

5 70 30 544 190 15.5 10.9 0.24 21.2 19.1 4.4 0.18 0.013 d 55.1 0
-6 60 40 508 263 13.1 9.6 0.22 128 36.6 3.8 0.50 0.017 3.1 38.6 4.7
7 50 50 44.0  36.6 8.9 10.4 0.20 94 439 3.9 0.74 0.028 1.9 23.3 16.9

2Nominal composition LiF-BeF, -ThF,-UF, (72-16-11.7-0.3 mole %).
bNominul composition NaBF, -NaF (92-8 mole %).

“No coolant-salt phase remained.

dNot analyzed.

S€
cal values; therefore, the fuel-salt phase analyses (Table
3.2) represent an average of 3 to 5 values. Further sup-
port for the contention that reactions involving the
fuel-salt phase were complete in 60 min or less is shown
by the plots of volume vs time in Fig. 3.3. The coolant-
salt phase volume decreased rapidly for about 30 min
due to reaction (1); thereafter, the volume change was
slower, presumably due to reaction (2).

It was impossible to obtain coolant-salt phase samples
with the filter sticks, both because the phase volume
was small and the salt tended to drain out of the filter
sticks. Therefore, all coolant-salt phase analyses (Table
3.3) were from samples obtained after completion of
the experiment and represent only single values.

The analyses (Table 3.3) show substantial redistribu-
tion of the ions Li*, Na', and Be®* between the two
phases. Thorium and uranium exhibited low solubility
in the coolant-salt phase. Neither NaBF, nor the oxy-
genated fluoroborate compound (represented as B,0,
in the table) was soluble in the fuel-salt phase. Fuel salt
stirred in contact with a coolant-salt phase containing
up to about 50 mole % B, O; showed no precipitation
of UO,. The coolant phase compositions were ex-

ORNL-DWG. 75-13748
| [ I | [

60

W
o

{ml)

H
O

u!t
=
S 301
O
=
20— Coolant Salt Phase
10— —
o l | ] | ]
0 20 40 60 80 100 120
TIME (min)

Fig. 3.3. Volume of coolant-salt phase and fuel-salt phase vs
time in mixing experiment No. 6. Initial mixture was 60 wt %
fuel salt and 40 wt % coolant salt. After heating to 550°C,
stirring was begun and the depth of the two phases was periodi-
cally measured.

36

Table 3.3. Composition of coolant-salt phase expressed
as components of the ternary system MF-B, O, -NaBF,

[nitial mixture Coolant-salt phase

(wt %) (mole %)

Fuel salt Coolant salt MF4 B,O, NaBF,
70 30 47.5 525 0
60 40 64.2 31.9 3.9
50 50 65.3 20.1 14.6

MF = Na — NaBF, + Li+ 2Be + 4Th + 4U + 48i.

pressed (Table 3.3) in terms of the ternary system
MF-B,0;-NaBF,;. The compositions are close to the
glass-forming regions of the ternary phase diagram'3 for
the system NaF-B,0;-NaBF,, where discrete com-
pounds have not been established.

The following pertinent observations can be made:

1. The rate of evolution of BF; gas on mixing was low;
presumably the rate-limiting step is the transfer of
NaF across the salt-salt interface. Thus, in a reactor
system with turbulent flow, the release would be
more rapid; however, these results are encouraging
relative to MSBRs in that very rapid gas release re-
sulting in significant pressure surges was not experi-
enced.

. No tendency was observed for the fuel salt constit-
uents thorium or uranium to redistribute or to form
more concentrated solutions, or to precipitate fol-
lowing mixing of coolant salt into fuel salt. These
experiments do not yield information relative to
mixing fuel salt into coolant salt, since it was
impossible to contain predominantly coolant-salt
phase mixtures in quartz at 550°C. Thus the ques-
tion of uranium (and/or thorium) precipitation as
UF,-NaF complexes, as observed!® in an engineer-
ing loop, remains unresolved.

. Apparently an oxide species forms in the coolant-salt
phase which is more stable than UQO,, since no UO,
precipitation was observed. Thus, large amounts of
oxygenated compounds could be added to the flu-
oroborate coolant salt for the purpose of seques-
tering tritium, since leakage of such a coolant salt
into the fuel salt would not lead to uranium or
thorium precipitation.

15. L. Maya, Sect. 4.1, this report.

16. H. F. McDuffie et al., Assessment of Molten Salts as
Intermediate Coolants for LMFBR s, ORNL-TM-2696 (Sept. 3,
1969), p. 20.
3.6 LATTICE AND FORMATION ENTHALPIES OF
FIRST-ROW TRANSITION-METAL FLUORIDES*

S. Cantor

The primary purpose of this investigation is to provide
a theoretical basis for critically evaluating the thermo-
dynariic data that will be obtained in an experimental
program recently initiated with Division of Physical
Research . funding. In the experiments, free energies of
formation will be deduced from emf measurements of
solid-electrolyte galvanic cells. The first-row transltion
metals include common structural metals (Fe, Ni, Cr)
and other metals (Ti, V)-which may be-used in fission or
fusion reactors. When' these metals are corroded or
otherwise oxidized in fluoride media used in these reac-
tors, metal fluorides are formed; reliable thermo-
dynamic information for these compounds is valuable in
predicting their chemical behavior in the reactor system.

For a metallic fluoride, MF,, (where n is the valence
of the metallic ion), the relationship between lattice
enthalpy AH; and enthalpy of formatron AHy is given
by the equation’

AHy = AHy - AHy+ —nAHp. (1)

The lattice enthalpy is the heat of the reaction

M™(g) + nF(g) =MF,(c) (2)

at 298.15°K. The lattice enthalpy is very similar to the
lattice “‘energy,” the latter being somethat more diffi-
cult to obtain from experimental information: AHy is
the standard heat of formation of MEF, (c) AH,,+ is the
standard enthalpy of formatron at 298.15°K, of the
gaseous cation and electrons (e) formed from the
crystalline metal:

M(c)—>Mn+(g)+.ne'(g). _ o - | (3)

AHF‘ is the standard enthalpy at 298.15°K of a mole
of gaseous fluoride ions formed from the ideal gases,
electrons, and d1atom1c fluorine:

B+~ Flo)- @)

The enthalp1es of formation for reactrons (3) and (4)
are deduced mostly from atomic or molecular data.
AH,+ is obtained by summrng the first n ionization
potentials of M and its enthalpy: of sublimation and
coverting these quantities, where necessary, to
298.15°K; values of AH,+ are given in Tables 3.4 and
3.5. The enthalpy of reaction (4) at 298.15°K is —61.24

*This research'in support of the MSBR Program was funded
by the ERDA Division of Physical Research.

Table 3.4. Standard enthalpies of formation of 3d
divalent fluorides (AHf) and their cations (AH2+),
lattice enthalpies (AHL)

. _AH _AH,. —AH

Fluorlde "(kcal/mole) (kcal/mole)? (kcal/m{;le)
CaF,” . 291.5% 460.3 629.3
ScF, (239)° 540 (657
TiF, (217)¢ 586.6 (681)¢

. VF, (208)¢ . 620.0 (706)¢
CrF, 1869 + § 6353 699 + 5
MnF, 205.4 + 1€ 602.2 685.1 + 1
FeF, 170€ £ 658.3 706 + 5
CoF, 160.5 + 18 682.3 7203 %1
NiF, 157.2 = 0.47 703.5 7382+ 0.4
CuF, 131.2 £ 0.8¢ 732.3 741.0 £ 0.8
ZnF, 182.7¢ 665.1 725.3

Zlonization potentials trom C. E. Moore, NSRDS-NBS 34
(1970); enthalpies of sublimation from ref, 19; valence state
preparation energy corrections from ref 20.

PNBS Technical Note 270-6 (1971).

CEstimated in this investigation.

dNBS Technical Note 270-4 (1969).

eT._N. .R_ezul(hina et al., J. Chem. Then‘nodyn.‘ 6, 890 (1974).

fDerived from Asolid galvanic-cell emf data given in W. H.
Skelton and J. W Patterson J. Less- Common Metals 3l 47
(1973).

EJANAF: Thermochemical Ta bles,
(1971).

ME. Rudzitis et al. ,J. Chem. Eng. Data 12,133 (1967)
'NBS Technical Note 270-3 (1968).

2d ed., NSRDS-NBS 37

kcal per gram-ion; it is based on Popp’s value!” (3.400
eV) for the electron affinity of fluorine, the dissociation
energy (1.58 eV) measured by Chupka and
Berkowitz,'® and the enthalpy difference of F, (ideal
gas), H, 95 — Hy, listed by Hultgren et al.!? :
The lattice -enthalpies of the divalent fluorides are
listed in Table.3.4 and are plotted against atomic num-
ber in Fig. 3.4. The curious double hump has been inter-
preted??.in terms of ligand-field.theory. By this theory,
differences between actual values of AH; and those
lying on a smooth curve-drawn to fit the data of CaF,,
MnF,, and ZnF, are primarily due to ligand-field stabi-
lization energy (LFSE). For this series of compounds,

17. H. P. Popp, Z. Naturforsch. 22a, 254 (1967).

18. W A, Chupka and J. Berkowitz, J. Chem. Phys. 54 5426
(1971)
"'19. R. Hultgren et rrl Selected Values of the Thermo-
dynamic Properties of the Elements, p. 177, American Society
of Metals; Metals.Park, Ohio, 1973.
- 20 P, George and D. S. McClure, p. 381 in Progress in
Inorganic Chemistry, vol. 1, F. A. Cotton, ed., Interscience,
New York, 1959,
38

. ORNL-DWG. 75-i37'49

5] S I o e

~ 700
Q
©
E
~
o
QL
X
-~
I
¥
650
¥
S i T O N T I
d® d' d¢ d3 d* d°> d® d7 d8& qd° d'"

Fig. 3.4. Lattice enthalpies of 3d divalent fluorides. Solid
circles (e) are experimental AH; calculated from Eq..(1); error

. Table 3.5. Lattice enthalpy (a#/,) and standard enthalpy of

formation (AH ;) of 3d trivalent fluorides; standard
enthalpy ol"}ormatlon of 3d cations (AH,.)
: . —AH —AH3+ —AH
: A_Fluonde (kcal/mole) (kcal/mole)? (kca]/mldle)
" ScF, 394.1 1 20 1112 13224+ 2
- TiF, 336.1 + 3.5¢ 1222 1374 + 3.5
“VE, 2959+ 5 1298 . 1409
. CrF, 277¢ 1351.6 1445
" MnF, 238+ 7€ 1381 1436 £ 7
" FeF, 249 + 3¢ " 13654 14313
CoF, 193.8¢ 1455.6 1465.7
NiF, (165)% 1515.5 (1497)8
-CuF, (120¥ 1584 (152088
 GaF, 278M 1389 1483

é[oniiation potentials from C. E. Moore, NSRDS-NBS 34

" (1970); enthalpies of sublimation from ref. 19; valence state
_ preparatlon energy correctlons from ref 20.

€Q. Kubaschewskl et al.,

bT N. Rezukhma et al., J. Chem. Thermodyn 6, 890 (1974)

pp. 334, 354 in Metallurgical Thermo-
chemistry, 4th ed., Pergamon, Oxford, England, 1967.

dDer;ved from solid galvanic-cell emf data given in W. H.

" Skelton and J. W. Patterson, J.. Less-Common Metals 31, 47

bars are uncertainties in AH . Open circles (o) are AH; minus - .
ligand-field stabilization eneIgy. Triangles (A) are. AH; minus -

both LFSE and Jahn-Teller energy. Solid. squares (w) were esti-
mated by adding LFSE plus 3 kcal/mole as an empirical correc-
tion to the smooth curve. ;

the ligands are fluoride ions octahedrally coordinated

(1973).
-ENBS Tebhméai Note 270~ 4 (1969).

fJANAF Thermochemical Tables, 2d ed., NSRDS-NBS 37
(1971).

BEstimated in this investigation.

hNBS Technical Note 270-3 (1968).

‘open circles) above the smooth curve by 5 and 7
_ kcal/mole respectively. Both CrF, and CuF, (and to a

(except for CaF,) to the cation; the octahedral “field” - -

of fluoride ions acts to “stabilize” "the. splitting of
d-electron energy. levels of the metal ions. In a spheri-
cally symmetrical field, the d-electron levels would be
degenerate, that is, be at the same energy. In Fig.. 3.4,
the smooth curve drawn through, for examp]e NiF,
represents. the AH; NiF, would have if the field of

lesser extent, FeF,) are known from crystal structure
data to exhibit major tetragonal departures from octa-

“hedral coordination geometry: This is attributed to the

* Jahn-Teller effect,? which confers an additional stabili-
" zation energy. (Jahn-Teller energy is abbreviated herein
. as JTE.) For CuF, and FeF,, the JTE can be derived

optically (Table 3.6). When LFSE and JTE are addi-

- tively applied to CuF, and CrF;, the lattice enthalpy is
. overcorrected as is shown by the triangles in Fig. 3.4.

fluoride ions around the nickel cation were spherically

symmetrical. Octahedrally coordinated cations with
unfilled, half-filled, or fully filled 3d orbitals will not
have a ligand-field stabilization energy; hence a smooth

The methods outlined above can be applied to predict
AH; for VF, and TiF,. Neither compound would be

- expected to have a significant JTE. The formula used is

curve is drawn through Can(dO) Man(d ), and'

ZnF,(d'%). -

‘spectra. For Fer, NiF,, and CoF,, subtraction of
optically derived LFSE (Table 3.6) from AH[ yields
values (open circles in Fig. 3.4) which are above the
smooth curve by 2 to 3 kcal/mole. Similar LFSE sub-

AH; (kcal/mole) = AH; "+ LFSE+ 3, (5)

~ where AH; ' is the value for the compound lying on the
Values of the LFSE can be deduced from optical

tractions fo; CrF, and CuF, yi_él_d values (dedptéd by °

1972.

smooth curve in Fig. 3.4. The 3 kcal/mole on the right-

“hand side of Eq. (5) is an empirical correction reflecting

pp. 590-93 in Ad-
, Interscience, New York,

21. F. Al Cotton and'G. Wilkinson,
vanced Inorganic Chemistry, 3rd ed.
39

Table 3.6. Ligand-field parameter (10Dq}), stabiliiation energy (LFSE),
' Jahn-Teller splitting (5) and energy (JTE) for the
di- and trifluorides of 3d metals

Electron Fluoride 10Dq LFSE 5 JTE
levels » . (cm ™' )2 (kcal/mole) (cm ™' )? (kcal/mole)
d! " ScF, ' ' (—12)P
Lo TiF, - 16,000¢ —-18.2 910 - 1.7
d* TIF, (10,7004 (-24)
: VF, 15,900¢ -36.4
d? VE, 11,000 (—38)
- CIF, 14,600 ~50.1
d* CrF, " 11,000 —18.9 (-11)8
MnF, . 17,400 ~29.8 - 9000 -12.9
d* FeF, 6,900 ~7.9 . 1400 — 2.7
CoF, 11,400 ~13.0 :
d? . CoF, .- 7,200 ~16.5
NiF, . . 16,200/ -37.0
das NiF, . 7,400 —25.4
CuF, 14,100 _48.4 o
d® CuF, - 7,400 ~15.3 7500 - - —10.7

2Values given in-D. Oelkrug, Struct. Bonding (Berlin) 9, 1-26 (1971) unless otherwise
indicated. ' »

bLFSE estimated very roughly as % of TiF, .

€Based on K, NaTiF, . ,

dEstimated by assuming 10Dq is %, that of TiF, .

€Based on (NH, ), VF,.

TEstimated by Jorgensen’s method: 10Dgq=g X k;g=12,300 cm ', k=0.9.
ERough estimate from JTE of CuF, .

N Based on K, CoF,.

IG. C. Allen and K. O. Warren, Struct. Bonding (Berlin) 9, 107 (1971).
ORNL- DWG. 75-13750

R

1550 | I

1500

1450

B
Q
Q

—AHL {Kcal/mole)

r—

s
by
E
o

I

PO |

ScFy TiFy VF3 Crfy MnF3 FeF3 CoFy NiFy CuF3 (ZnFz) GaFy

dO dl dZ d3 d4 d5 dG d7 d8 d9 d10

Fig. 3.5. Lattice enthalpies of 3d trivalent fluorides. Solid

circles (®) are experimental AHL calculated from Eq. (1); error

bars are published uncertainties in AHp . Open circles (0) are

AHL minus ligand-field stabilization energy. Solid squares (=)
are estimated by adding LFSE to the smooth curve.

the difference between theoretical and thermochemical
lattice enthalpies for NiF, and CoF,. The standard
enthalpy of formation (AHy) for TiF, and VF, is then
obtained from Eq. (1) and is listed in Table 3 .4.

40

Analogous considerations were applied to study AH|
for trivalent fluorides. The data and results are pre-
sented in Table 3.5 and in Fig. 3.5. The double-hump
pattern of the data is evident in Fig. 3.5. Subtraction of
LFSE (given in Table 2.6) yields very satisfactory agree-
ment between theoretical and experimental lattice
enthalpies of VF; and CrF;; the agreement for TiF;
(and for CoF,) is less satisfactory. As may be seen by
the open circle below the curve in Fig. 3.5, subtraction
of LFSE from AH; overcorrects MnF;. This is some-
what surprising, since MnF,, with its 3d* electronic
configuration for Mn**, also has a sizable JTE (Table
2.6). If the JTE were also subtracted, the discrepancy
from the smooth curve would be much greater. In short,
the thermochemical data for MnF; are questionable.

In estimating AH; and AHy for NiF; and CuF,
(Table 3.5), only the LFSE was added to the spherically
symmetrical values (i.e., smooth curve values) of AHy.
In other words, Eq. (5) was applied without the empiri-
cal correction of 3 kcal/mole.

With regard to the AHy of the structural-metal fluo-
rides, the theory, as applied above, suggests that there is
little need to determine AHf for NiF,. Moreover, from
the value of AHy of TiF, obtained in this study, it is
understandable why TiF, has never been prepared as a
pure solid; it can be easily shown that TiF, would
readily disproportionate to TiF; and Ti. However, a
more accurate experimental determination of AHf for
TiF; would be desirable for both practical as well as
theoretical reasons. The same may be said for VF,,

VF;, CrF;, CrF;, FeF,, and FeF;.
4. Coolant-Salt Chemistry
A.D. Kelmers

4.1 CHEMISTRY OF SODIUM FLUOROBORATE
L.Maya W.R. Cahill*

The composition of the condensable fraction of the
vapor phase in equilibrium with molten fluoroborate
can be defined by the system H; OBF,-HBO, ‘H, 0, as
described in the previous report.! The work done ‘dur-
ing this report period was aimed at spectroscopic identi-
fication of the molecular species present. The 1B NMR

as well as IR and Raman spectra of BF3;2H20, '

HBF,(OH),, and of other intermediaté compositions
was obtained. Dihydroxyfluoroboric acid (DHFBA)
participates in éxchange processes which could be dés-
cribed by the following equilibria:

2HBF2(OH)2 = BF3 'VHQO + H3 BO3 ,
BF3'H2 O+ H3BO3 = BF3 '2H2.0 + HB()2

The presence of H3BO3; and HBO, was detected by IR
and Raman spectra, and the pronounced broadening of
the '°F and '!B NMR Signals is an indication of ex-
change processes. The Raman spectrum of DHFBA
indicates that this compound is a tetrahedral molecule.
Exchange processes were not detected for BF3-2H, 0.
This compound appears to be stable at room tempera-
ture. The structural information derived from the
Raman spectrum, which identified BF3-2H,0 as a

© *QRAU summer -participant.

1. L. Maya, MSR Program Semiannu. Progr. Rep. Feb. 28,
1975, ORNL-5047,p.47.

tetrahedral molecule, agrees with the x-ray structural
determination? of this compound.

Additional samples of condensate collected during the
operation of the Coolant Salt Test Facility (CSTF) were
analyzed (Table 4.1). Silicon is present because of
attack on the glass trap used to collect the condensate.
Variations in the chemical composition of the samples
can be interpreted as an indication that the condensed

- material is not a single molecular compound but, rather,

a mixture formed by combination of the simpler gas-
eous. species present in the system, that is, H, O, HF,

~and BFj;. The relatively high tritium content of these

fractions should be noted. Tritium is present in the
system, since some of the Hastelloy N in the loop was
originally used in the MSRE. The condensates show a
tritium concentration factor of about 10° relative to
the salt, suggesting that fluoroborate coolant salt
[NaBF;-NaF (92-8 mole %)] may be an effective means

~of concentrating and conveying tritium out of the

system.

Attempts were made to generate a condensable frac-
tion in laboratory-scale experiments by heating coolant
salt containing up to 200 ppm H as NaBF; OH to 400°C
in a closed system equipped with a cold finger. The
OH™ concentration in the salt decreased to 50 ppm, and
the composition of the condensate in a typical run was
53.2% H, OBF,, 14.8% (H,0), SiF¢ and 32% free water
found by difference. The boron concentration in the
condensed material did not reach as high a level as in

2. W. B. Bang and G. B. Carpenter, 4cta. Cryst. 17, 742
(1964). .

. Table 4.1. Analyses of CSTF trap condensates

Chemical comnposition

b

Tritium content

Sample Operation period Amount? | (Wt %) (mCi/g)
"H,OBF,  HBO, SiF, 2
1 1972 Not avail. 60.4 15.7 Not det. 0.8 to 3.0¢
2 1/14/75 — 1/24/75 100 mg 92.3 0 2.1 5.7
3 3/14/75 — 4/15/759 25¢g - 84.1 12.4 4.0 34
4 4/15/75 — 5/6/75 800 mg 83.0 12.1 0.1 0.6

aApproximate amount, Some of the material remained in the trap.

bDifference from 100% is H, O.

CGiven as a range. Apparently moré than one sample was anatyzed for tritium content. Data from A.
S. Meyer and J. M. Dale, Anal. Chem. Div. Annu. Progr. Rep. Jan. 1974, ORNL4930, p. 28.

dThe loop was not in operation between 1/24)75 and 3/14/75.

41
the CSTF samples, and there was considerable cor-
rosion. Nevertheless, these experiments showed a pos-
sible mechanism for the conversion of dissolved
NaBF, OH into a volatile fraction. o

An apparatus was assembled to measure the vapor
density of BF;:2H,0 and related compounds at ele-
vated tempérgtures to determine the degree of dissocia-
tion of these materials. This work tested the hypothesis
that the condensable materials collected in the opera-
tion of the CSTF are completely dissociated at oper-
ating temperatures (400—600°C) and only combine to
form more complex molecules in the colder parts of the
system. The procedure consists in measuring the pres-
sure developed in a closed system containing a known
amount of BF;+2H, 0 or DHFBA in order to establish
the degree of dissociation according to the equilibrium
described below:

BF3'2H20 =BF3 + 2H20

At this time, volumes in the apparatus have been deter-
mined, and pressure determinations have been made
using argon as a test gas. Initial runs with BF,+2H,0
indicate that this compound may be completely dis-
sociated at 400°C. _

Work on determining the oxide species present in
molten fluoroborate is being continued, and the survey'
of the system NaF-NaBF,-B,0; at 400 to 600°C has
been extended to include IR and x-ray diffraction
analyses in addition to physical and chemical observa-
tions of the behavior of selected compositions. The
observations indicate that there are three main areas in
the system:

1. A region of compositions in which BF; is evolved.
This occurs with compositions having a deficiency,
in terms of equ1molar ratios, of NaF relative to the
B, O3 present.

. A region of compositions in which stable glasses are
formed on cooling. This corresponds to mixtures
containing more than 33 mole % B, 0,.

. A region in which crystalline phases and glasses co-
exist.
decreases with decreasing B, 04 content,

Usually coolant salt [NaBF,-NaF (92-8 mole %)]
contains relatively small amounts of oxide, up to 1000
ppm, and its composition lies within area 3; thus, work
has been directed toward characterizing the oxide
species in this area. At least two species were present;
one formed at the boundary of the glass area (high
oxide content), and the other was Na; B3 F¢ O3, .which
formed in compositions having NaF:NaBF,:B,03 mole
ratios of 2:2:1 and 2:4:1 and was possibly present in

42

The tendency to form glasses on cooling

compositions containing as little as 3 mole % B,0;.
Experiments at the 1.5 to 4.0 mole % B,0; level,
approaching the coolant composition, have been im-
peded by the relatively low sensitivity of IR and x-ray
powder diffraction. The difficulty with IR, using the
KBr pellet method, arises from the fact that, at these
oxide levels, the only band not covered by BF,~
absorptions is the one at 810 cm™!. This band has a
relatively low absorptivity, and it is common to
NaBF;0H, Na,B,F;0, Na;B;F,0,, and possibly
other BOF compounds, although the intensity and line
shape are different for each compound. A more certain
IR identification can be made only when at least two
typical bands can be idéntified (presently observable
only at higher concentrations), as was the case in the
identification of NayB3;F40, at an oxide level corre-
sponding to 14 mole % B,0j;. Difficulties with x-ray
diffraction arise from the low sensitivity of this tech-
nique coupled with the fact that the species have a ten-
dency to form glasses. Raman work on melts is being
planned as the next step in this study.

4.2 CORROSION OF STRUCTURAL
ALLOYS BY FLUOROBORATES

S. Cantor  D.E. Heatherly ~ B.F. Hitch -

Alloys containing chromium in contact with molten
NaBF,-NaF would be expected to form a borlde be-
cause the reaction

(1 +x)Cr(c) + NaBF,(d) + 2NaF(d)
= Na; CrFg(c) + CryB(c) (1)

has a negative standard free-energy change (AG®). At a
temperature of 800°K, AG2y0 = —10 kcal. This value is
based on an estimated standard free energy of forma-
tion (AGJ?) of Nay CrFg of —600 kcal/mole. In reaction
(1), the exact value of x is unknown; however, AG? of
the more stable chromium borides (Cr, B, Crs B; ) is esti-
mated to be —22 kcal per gram-atom of boron.? In
nickel-base alloys, reaction (1) may proceed more
readily because of the probable exothermic nature of
the reaction

Cry B(c) t yNi(alloy) = Cr(alloy) + Nij, B(c). (2)

3. O. H. Krikorian, Estimation of Heat Capacities and other
Thermodynamic Properties of Refractory Borides, UCRL-
51043 (1971).

4. O. S. Gorelkin, A. S. Dubrovin, O. D. Kolesnikova, andN
A. Cherkov, Russ. J. Phys. Chem. 46,431 (1972).
43

Assuming that AG? of N1 B equals ts enthalpy of =

formation,* AG; for reactron (2) is about -
gram-atom of boron.
An experiment to determine the extent of borrde

—3 keal pe r.

formation in the nickel-base alloys ‘Hastelloy N (7% Cr)' :

and Inconel 600 (15% Cr) has been in progress for sev-
eral months. In this experiment, metal speeim'ens are
equilibrated with NaBF,-NaF (92-8 mole %) at 640°C

The 'onnly plausible explanation seems. to be that some

'NaBF, remains on (or in).the metal surface despite the

washing (5—10 "min in boiling water) intended to

‘remove adhering traces of salt. Some of the scatter in
‘the boron analyses. by SSMS is probably due to salt

“contamination of the metal surface. Inconel 600 speci-
- mens scanned by IMMA showed a similar pattern of

under an argon atmosphere and are periodically re- .

moved, washed free of salt, using water, and analyzed-

by spark-sourcé imass spectrometry - (SSMS):and; -less
' ~ routinely, by ion microprobe mass analysis (IMMA). 5

Analysis for boron on specimen surfaces by SSMS sug- '.

gests some boride formation. Hastelloy N specimens

surface inclusions containing B, Na, and F. Unfortu-
nately, IMMA does not provide quantitative analyses for

~ these elements. As yet, the extent of boride formation

cannot be quantified in either Hastelloy N or in Inconel

' 600 by a combination of SSMS and IMMA. Probably,

that had equilibrated for up.to 129 days were found'to -
contain 30 to 1000 ppm B; Inconel 600 specimens con- .

; _'tamed 80. to 2000 ppm B. Control spécimens that had .
_ not been in contact with the molten salt showed 5 to 20

ppm when analyzed by SSMS. Boron in Inconél 600

increased with equilibration time; but, with Hastelloy

* however, reactions (1) and (2) occur to a small extent;

boride is deposited at levels not greater than 500 ppm
on Hastelloy N and not exceeding 1000 ppm on Inconel
600 "after four months of contact with molten

NaBF, NaF

- IMMA was also used to obtam depth profiles of alloy

- ."-constltue.nts through about 5000 layers. In control

N, the data were much more scattered and showed vir- '

‘tually no time dependence.

- Several specimens analyzed by SSMS were also 1nvest1 ,

gated by IMMA. Boron was present w1thm the first few-'
" hundred monolayers of metal, in inclusions also con-

taining sodium and .fluorine in specimens of 2% Ti—
modified Hastelloy N that had equilibrated for 72 days.
“These contamed 150 ppm B as determmed by SSMS.

3 épecimens, elemental concentrations were uniform with
-depth. In equilibrated Hastelloy N, molybdenum was

uniformly distributed throughout the depth explored,
but chromium and titanium concentrations increased

linearly from the surface inward; the iron concentration
appeéared to.decrease slowly with depth. Equ111brated
Inconel 600 showed V1rtually no chromium in the first

; A.SOO Tayers, but chromium increased linearly in the next

5. Spark source mass spectrometry and ion. mlcroprobe mass .
analysis performed by the Analyttcal Chemrstry D1vrsron T

4500 layers; iron and nickel were uniform through the
depth studied. Thus, IMMA indicatés that chromium is

‘selectively oxidized by NaBF,-NaF (92-8 mole %) or by

oxidants contained in this molten mixture.
5. Development and Evaluation of Analytical Methods

- A.S. Meyer

5.1 IN-LINE ANALYSIS OF
MOLTEN MSBR FUEL

B. R. Clark
A.S. Meyer

R.F.Apple
D. L. Manning

Corrosion test loops described previously! have con-
tinued operation with circulating reference fuel carrier
salt, LiF-BeF, ThF, (72-16-12 mole %). No additional
loops have been placed in operation during this report-
ing period, although several are expected-to begin opera-
tion within the next few months, -

Measurements of the U* /U3 ratio in the forced con-
vection loop (FCL-2b) indicate a “‘steady-state’ value of
about 100 (Fig. 5.1). This is somewhat lower? than the

1. H. E. McCoy et al., MSR Program Semmnnu Progr Rep.
Aug. 31, 1974; ORNL- 5011 ,p. 76.

2. A.S. Meyer et al., MSR Program Semiannu. Progr. Rep.
Feb. 28, 1975, ORNL-5047, p. 52. .

ORNL- DWG, 75-12056
|

U4+
log 637
. -,
« ——Loop Down 2 Days

e —=—_oop Down { Day
* ——|oop Down 8 Days
1

I | I
50 t00 150
ELAPSED TIME ({days)

200

Fig. 5.1. U /U ratios in forced convection loop FCL-2b.

apparent steady-state value obtained with the fluoride
mixture, LiF-BeF,-ThF, (68-20-12 mole_%), indicating
a less oxidizing melt. The melt, which started at a ratio
of around 1000, reached this level via a redox process
which' presumably involves reaction with the chromium
in the walls of the véssel or in the specimens. No at-
tempts have yet been made to reoxidize the U3 in the
melt by suifable additions of NiF, or some other oxi-
dant. It is mterestmg that the decrease to a steady-state
value occurred after about 75 days, with a rapid de-
crease in the first 30 days. Previous data from the exper-
imental fuel showed?® a rather stable value near 10* for
about 60 days, until beryllium additions were made to
forcé reduction of the U%*,

'Some of oscillations in the data probably result from
air contammatlon with subsequent oxidation when the
loop was down. This was most prominént with the
experimental melt (68-20-12 mole %) when the U3
ratio was substantially greater than the steady state
value reached at a later date.

Ratios of U**/U3* measured in the two thermal con-
vection loops, NCL 21A and NCL 23, are summarized
in Figs. 5.2 and 5.3’ respectively. No unusual trend is
apparent in the oxidation-state history of the fuel melt
in NCL 21A. This loop was operated for about 240 days
with Hastelloy N corrosion specimens: The curve shows
a rather dramatic rise in the ratio whenever new spéci-
mens are added. This effect is attributed to additions of
moisture and air which partially oxidize U**. A recovery
to lower ratios follows each increase in repetitive fash-
ion.

ORNL-DWG. 75-12057

—1

*8 T i I
=3 § 3
2% 3
R4 g <
. @
sh8E E g ~
,Ig g E
“ & &
% . Zz
DU.D of 2 i
e | s 7 8
v, l £
- l." . ‘
41— .'.-.. “., "ete s —
| 1 | 1 |
50 100 150 200 250 300

ELAPSED TIME (days)

Fig. 5.2. U**/U* ratios in thermal convection loop
NCL-21A.
ORNL-DWG. 75-12055

» | | | 1
o
E.
g -
v -
2r Eg g g o] —
33 » S
2o c o
=< L < R o)
+ £ . B
[& c 8
+ |+ ® o . 8. GEJ <I
< | % w = »
pun | e l- b - o c 4
g Sgb ® @ Q
o eeo * c a £
L -, o N O
° c b @
Q
* S o
- o =
' = | S .
. &
oo o
*ee *
. ¢ - ®we oq o0
| | | : |
50 100 150 200 250
ELAPSED TIME (days)

Fig. 5.3. U**/U?* ratios in thermal convection loop NCL-23.

The U*" in the melt in the Inconel 601 loop, NCL 23,
was rapidly reduced until a U**/U3"* ratio of around 40
was reached. Since then the ratio has continued to
decline, reaching a relatively stable value near 5. The
high level of chromium in the Inconel 601 (23 wt %)
provides a sufficiently active reductant to reduce the
U* more extensively than has been observed in Hastel-
loy N loops; therefore, the greater U3* concentration is
not surprising. '

5.2 TRITIUM ADDITION EXPERIMENTS IN
THE COOLANT-SALT TECHNOLOGY FACILITY

R.F. Apple B.R.Clark A.S.Meyer

One major concern in the development of an MSBR is
the release of tritium to the surroundings. A potential
method for limiting tritium release rates to acceptable
levels involves trapping and removal of the tritium in
the secondary coolant system. This method must be
tested before a complete'understanding is possible of
the manner by which tritium will be retained in an
MSBR. The present series of tritium addition experi-
ments involving sodium fluoroborate will provide data
on this method.

The Coolant-Salt Technology Facility (CSTF) is being
operated for testing the NaBF,-NaF eutectic mixture

with regard to its suitability as a possible secondary
coolant system. In cooperation with loop engineers and
technicians, the Analytical Chemistry Group has been
engaged in experiments to determine the fate and
behavior of elemental tritium added directly to the cir-
culating salt to simulate, at least in part, the predicted
transport of tritium into the coolant system via diffu-
sion through the primary heat exchanger. This section
describes the methodology and results of the first two
experiments.

About 80 mCi of tritium (diluted about 1:1000 with
protium) was introduced into the salt stream over a
period of about 11 hr beginning on July 17. Tritium
concentrations weré measured in the salt and the cover
gas during the addition and for several days thereafter.
Salt samples were collected directly from the pump-
bowl access port with a copper thimble covered with a
copper frit. One-gram samples of the cooled salt were
diluted volumetrically, and aliquots were mixed with a
scintillation emulsion for beta counting.

Cover-gas sampling has proven to be somewhat diffi-
cult. At present a sidestream is being sampled; the diffi-
culty arises from the passage of this stream through a
nickel sampling line that is not completely inert chemi-
cally to cover-gas components. Thus the amount of ele-
mental tritium finally measured may not be an accurate
measure of tritium level within the loop gas system.
More definitive experiments will require the use of inert
precious metals in the sampling system to remove doubt
of chemical alteration of the cover-gas stream composi-
tion by the sampling system.

The off-gas collection train consists of

1. a series of three water scrubber pretraps which serve
to trap BF 3 and any other water-soluble compounds;

. a hot (400°C) copper oxide—filled tube;

. a condensation trap to collect water formed in or

passing through the copper oxide;

a liquid-nitrogen cold trap to remove the last traces

of water;

. a wet test meter to measure the volume of the inert
gas component of the cover gas, that is, helium.

Results of the first injection experiment are summarized
for the off-gas (Table 5.1) and salt (Table 5.2) samples.

Several days after operation had begun, some liquid
collected in the short glass section between the stop-
cock used to divert the gas stream and the first trap in
the analysis train. The liquid was washed from the glass,
counted, and found to contain about 60 uCi of tritium.
This discovery clearly complicates the interpretation of
previously collected samples, since a large portion of
total cover-gas tritium never reached the analysis train.
Furthermore, no conclusion is possible regarding the
chemical state of the tritium in the liquid. The data
suggest that the concentration of elemental tritium

Table 5.1. Tritium content in cover-gas
samples after first tritium addition
in CSTF

46

Tritium in gas
Date

Time | (pCif/ml)
(July)
H, O soluble Elemental
17 1046 2.3 34
1305 0.7 12
1700 1.4 170
1911 1.8 93
2243 1.4 830
18 0100 3.2 420
0805 1.3 61
1000 44 71
19 1037 . 790 62
21 0905 1800 13
1325 S0 13
22 0925 55 58
23 1303 85 29
24 1000 300 22
25 1315 340 26
28 1245 10 2.7

Table 5.2. Tritium content in salt samples
after first tritium addition in CSTF

Date Sample Timé Tritium
(Juty) No. (nCi/g)
17 45 1108 12
46 1308 23
47 1512 ~ 35
48 1718 S5
49 1930 93
50 2145 73
51 2321 112
18 52 1102 32
53 1912 20
19 54 0830 15
SMV 0952 7.5
21 55 1132 1.6
22 56 1400 1.6
23 57 1330 2.0

increased in both the cover-gas and salt samples when
the liquid was washed out between sampling periods
(July 23-25).

A second tritium addition was made August 5. During
this experiment no changes were made in the sampling
apparatus, but that region of the sampling train (des-
cribed above) where liquid had been accumulating was
washed with each sample collection and counted sepa-
rately. The tritium found there was added to the water-
soluble tritium measured in the pretraps. Data for this
experiment are summarized in Tables 5.3 and 5 .4.

An exhaustive analysis of the analytical and sampling
aspects of these data is not warranted at this time, since
several variables which affect the addition, sampling,
and the tritium losses have not yet been established. A
general discussion on the behavior of tritium in the
CSTF is given elsewhere.® The preliminary data are suf-
ficiently encouraging to merit a more extensive investi-
gation into the extent and mechanism of tritium inter-
action with salt and cover-gas components. Plans are
now under way to monitor the tritium  diffusion
through a portion of the loop wall and to measure the
level of active protons in the salt during addition of
tritium. A more intricate cover-gas sampling device (a
probe designed for the salt monitoring vessel or salt
sample port) is being considered and may be fabricated
if no simpler solution to the gas sampling problems can
be found.

3. Reference Sect. 1.1.2, this report.
Table 5.3, Tritium content in cover-gas
samples after second tritium addition

47

in CSTF
Tritium in gas
Date Time (pCi/ml)
(August) :
H, O soluble Elemental
5 0730 96 0.5
0930 110 - 2.2
1130 2,200 10
1330 5,300 20
1530 7,500 27
1830 13,000 34
1920 13,000 39
2100 12,000 40
23135 9,300 39
6 0730 9.300 28
1100 7,000 16
1500 6,500 14
2000 4,700 10
7 0940 3,300 6.8
1415 2,400 4.9
8 0930 1,600 34
9 1450 1,600 5.2
10 1040 1,300 2.8
11 1230 900 2.5
12 1010 570 1.1
13 1010 230 0.8
15 1000 182 1.0
18 0935 76 0.5
21 1010 73 0.5
Table 5.4. Tritium content in salt samples
after second tritium addition in CSTF
Date Sample Time Tritium
(August) No. - (nCi/g)
4 58 1313 0.8
5 39 0954 9.0
60 . 1145 21
61 ! 1350 36
62 1548 52
63 1848 71
64 1932 68
65 2125 71
66 2335 50
-6 67 0830 30
68 1507 19
69 2010 16
7 70 1014 9.5
71 1528 8.1
8 72 1018 3.6
9 73 0916 4.2
11 74 1248 3.9
12 75 1035 1.4
13 76 1048 2.2
15 77 1010 1.4
18 78 1010 0.7
21 SMV 10-A 1308 0.5

5.3 ELECTROANALYTICAL STUDIES OF
IRON(I) IN MOLTEN LiF-BeF, -ThF,
(72-16-12.MOLE %)

D. L. Manning G. Mamantov

Electroanalytical studies in molten fluorides have
particular importance for possible use asin-line analyti-
cal methods for molten-salt reactor streams. Iron(Il) is a
corrosion product present in molten-salt reactor fuels.
We have previously? ® carried out electrochemical
studies of iron(II) in molten LiF-NaF-KF
(46.5-11.5-42.0 mole %), LiF-BeF,-ZrF,
(69.6-25.4-5.0 mole %), and NaBF4-NaF (92-8 mole %).
Since the fuel solvent for the MSBR is a thorium-
containing salt, LiF-BeF,-ThF, (72-16-12 mole %), it is
of interest to conduct voltammetric and chronopotenti-
ometric studies of iron(II) in this fuel solvent. To deter-
mine concentration and/or diffusion coefficients by
linear sweep voltammetry, it is necessary to know
whether the product of the electrochemical reaction is
soluble or insoluble. The measurements discussed below
were done with this purpose in mind.

A voltammogram showing the reduction of iron(II),
Fe?* — Fe?, at a gold electrode is shown in Fig. 5.4. The
circles represent the theoretical shape based on current
functions tabulated by Nicholson and Shain® for a
reversible wave where both the oxidized and reduced
forms of the electroactive species are soluble. Thus,
even though Fe?* is reduced to the metal at gold, the
electrode reaction very closely approximates the
soluble-product case, apparently through the formation
of iron-gold surface alloys. Further evidence that the
Fe** - Fe® electrode reaction at gold conforms to the
soluble product case is illustrated by the chronopotenti-
ograms in Fig. 5.5. The ratio of the forward to reverse
transition times (7,/7,) compares favorably with the

4. D. L. Manning, “Voltammetric Studies of Iron in Molten
LiF-NaF-KF,” J. Electroanal. Chem. 6,227 (1963).

5. D. L. Manning and G. Mamantov, “Rapid Scan Voltam-
metric and Chronopotentiometric Studies of lron in Molten
Fluorides,” J. Electroanal. Chem. 7, 102 (1964).

6. H. W. Jenkins, D. L. Manning, and G. Mamantov, ‘“Elec-
trode Potentials of Several Redox Couples in Molten Fluo-
rides,” J. Electrochem. Soc. 117,183 (1970).

7. F. R. Clayton, Jr., “Electrochemical Studies in Molten
Fluorides and Fluorcborates,” doctoral dissertation, University
of Tennessee, December 1971, p. 82.

8. A.S. Mever et al., MSR Program Semiannu. Progr. Rep.
Aug. 31, 1974, ORNL4728, p. 44.

9. R. S. Nicholson and 1. Shain, “Theory of Stationary
Electrode Polarography,” Anal, Chem. 36, 707 (1960).
ORNL-DWG. 75-11275

CURRENT

H | I ! | |

50 0 -50
POTENTIAL , mV

150 100 -100  -150

Fig. 5.4. Stationary electrode voltammogram for the reduc-
tion of Fe?* at a gold electrode in molten LiF-BeF,-ThF,. Po-
tentiai axis is (£ - E1,)/2. Solid line is experimental. Circles are
theoretical shape for soluble product. Iron(II) concentration
0.027 f; electrode area, 0.25 cm?; temperature, 650°C.

theoretical value of 3 (ref. 10) for the soluble case,
which again points to the formation of surface alloys.

The reduction of Fe?* at a pyrolytic graphite elec-
trode is illustrated by the voltammogram in Fig. 5.6 and
the chronopotentiograms shown in Fig. 5.7. For the
reversible deposition of an insoluble substance where n
= 2, the voltammetric £, — £p/2 = 30.5 mV at 650°C
(ref. 11) is in good agreement with the experimental
value. The chronopotentlometrlc ratio (Tf/'r,) is approx-
imately unity,! ® which also is indicative that Fe?*
reduced to metallic iron without any apparent 1nterac-
tion with the pyrolytic graphite and that all the iron is
stripped from the electrode upon current reversal.
Therefore, iron appears to be reversibly reduced to a
soluble form at gold and to an insoluble material at
pyrolytic graphite. Thus the effect of electrode sub-
strate on an electrochemical reaction is illustrated by
this example.

Chronopotentiograms for the reduction of Fe?* at an
iridium electrode at 518 and 600°C are shown in Fig.
5.8. The ratio at 518°C is approximately unity and is 3
at 600°C, which is evidence that Fe?* reduction at
iridium approximates the insoluble-species case (as with
pyrolytic graphite) at 518°C and the soluble-product
case (as with gold) at 600°C. This change in reduction
behavior with temperature was not as pronounced at
gold or at pyrolytic graphite.

10. W. H. Rienmuth, “Chrono.potentiometric Transition
Times and Their Interpretation,” Anal. Chem. 32,1514 (1960).

11. G. Mamantov, D. L. Manning, and J. M. Dale, “Revers-
ible Deposition of Metals on Solid Electrode by Voltammetry
with Linearly Varying Potential, J. Electroanal. Chem. 9, 253
(1965).

48

The chronopotentiometric transition time 7 for an
electroactive species is given by the Sand equation!?

= (fll/anADl/zC)/ 2i.

Average diffusion coefficients of Fe?* in this melt evalu-
ated from the chronopotentiometric measurements by
means of the Sand equation are approx1mate1y 4.2 X
107°,8.0 X 107%,and 1.5 X 107 c¢m?/sec at 518, 600,
and 700°C respectively. '

5.4 VOLTAMMETRIC STUDIES OF

TELLURIUM IN MOLTEN LiF-BeF,-ThF,

(72-16-12 MOLE %)
D. L. Manning A.S.Meyer G.Mamantov

Tellurium occurs in nuclear reactors as a fission prod-
uct and results in shallow intergranular cracking in
structural metals and alloys.!? It is of interest to char-
acterize this substance electrochemically and ascertain
the feasibility of in situ measurements by electroana-
lytical means. We previously'* carried out preliminary
polarization measurements at a small tellurium pool
electrode in molten LiF-BeF,-ZrF, to establish the
potentials at which tellurium is oxidized and reduced in
the molten fluoride environment. These preliminary
observations indicated that the electrode reactions are
complex.

For tellurium screening studies, J. R. Keiser of the
Metals and Ceramics Division fabricated an experi-
mental cell equipped with viewing ports and electrode
ports for studying the stability of lithium telluride,
Li; Te, in molten LiF-BeF,-ThF,4. The Li; Te was added
as pellets following which voltammograms were re-
corded at gold and iridium electrodes.

As Li,; Te was added to the melt, the voltammograms
became complex and are not yet completely under-
stood. For clarity, pertinent observations at the iridium
and gold electrodes are tabulated separately.

1. Upon adding one 35-mg pellet of Li,Te, a reduc-
tion wave observed at —0.9 V vs the Ir quasireference
electrode (QRE) disappeared. This wave is not yet

12. Paul Delahay, p. 179 ff. in New Instrumental Methods in
Electrochemistry, Interscience, New York, 1964 .

13. H. E. McCoy, “Materials for Salt-Containing Vessels and
Piping,” The Development and Status of Molten-Salt Reactors,
ORNL-4812 (February 1975) p. 207.

14. A. S. Meyer et al., MSR Program Semiannu. Progr. Rep.
Aug. 31, 1974, ORNL-5011, p. 42.
49

CURRENT TIME
(mA) (sec/div)
0.9 || o i
L +0.2
o
= -02
(&l o 10.2
= o +0.2
E i}
0
-0.2
o
~—TIME

Fig. §
0.25 cm® ; temperature, 600°

otential scale, volts vs Ir QRE

ORNL-DWG. 75-11276

500uA

CURRENT

| | |
+0.025 0 -0025

POTENTIAL, V vs Ir QRE

5.6. Stationary electrode voltammogram for the reduc-
tion of iron(l) at a pyrolytic graphite efectrode. Insoluble prod-
uct at 650° theoretical /',’Y E 2 30.5 mV: measured "30
mV. Iron(1l) concentration, 0. 1!167 L electrode area, 0.1 em?.

ORNL-DWG 75-11277

TIME
(sec/div)

CURRENT)

~—TIME

. Cyclic chronopotentiograms for the reduction of iron(11) at a gold electrode. Formality of iron(11), 0.15: electrode area,

identified. The pellets did not melt or dissolve immedi-
ately. Relics of the pellets could be seen on the surface
for several days. The windows of the viewports became
coated with a bluish-gray deposit after a few days.
making viewing of the melt impossible. The bluish-gray
deposit is believed to be tellurium metal. This indicates
that tellurium species added as Li, Te are not stable in
the melt.

2. Voltammograms recorded in molten LiF-BeF,-
ThF, after additions of Liy Te did not reveal any waves
that could be attributed to soluble electroactive tellu-
rium species. Chemical analysis indicated <S5 ppm Te in
the melt.

3. Also at an iridium electrode, a reduction wave was

observed at ~0.45 V vs the Ir QRE, which was reason-
ably well defined at a scan rate of 0.02 V/sec. This wave
is due to Cr** reduction; the wave height increased upon
adding CrF, but did not change upon adding Li, Te. At
our normal scan rate of 0.1 V/sec, the wave was not

well defined, which explains in part why it was not
positively identified on background scans that are nor-
mally recorded at 0.1 V/sec.

4. Voltammetric waves indicative of telluride films on
a gold electrode were observed. However, these waves
disappeared after adding CrF; to the melt. The volt-
ammograms recorded at gold following the Li, Te addi-
tions became complex, and the electrode reactions are
not yet resolved.

Additional voltammetric mgasurements are planned
whereby the supposedly more soluble and stable LiTe;
species will be added to the fuel melt.
50

ORNL
TIME

(sec/div)

CURRENT TIME
(mA) (sec/div)

+0.2

-0.2

+0.2

<-—TIME
Fig. 5.7. Cyclic chronopotentiograms for the reduction of iron(I) at a pyrolytic graphite electrode. Formality of iron(I1), 0.027:
electrode area, (1.1 cm? ; temperature, 650°C; potential scale, volts vs Ir QRE.

TEMPERATURE

RNL-DW
TEMPERATURE 600
TIME REN URRENT TIME CURRENT TIME
< (ma) (sec/dw
2.6 0.1

Fig. 5.8. Cyclic chronopotentiograms for the reduction of iron(Il) at an iridium electrode. Formality of iron(1l), 0.015; electrode
area, 0.2 cm?; potential scale, volts vs Ir QRE.

Part 3. Materials Development

H. E. McCoy

The main thrust of the materials program is the devel-
opment of a structural material for the MSBR primary
circuit which has adequate resistance to embrittlement
by neutron irradiation and to shallow intergranular
attack by fission product penetration. A modified
Hastelloy N containing 2% Ti has good resistance to
irradiation embrittlement; however, it remains to be
shown that the alloy has sufficient resistance to shallow
intergranular cracking. Numerous laboratory tests are in
progress to answer this important question. It may be
necessary to further modify the alloy with rare-earth,
niobium, or higher chromium additions to impart better
resistance to shallow intergranular cracking.

Laboratory programs to study Hastelloy N—salt—
tellurium interactions are being established, including
the development of methods for exposing test materials
under simulated reactor operating conditions. Surface-
analysis capabilities have been improved so that the
reaction products in the affected grain boundaries can
be identified.

The procurement of products from two commercial
heats (8000 and 10,000 1b) of 2% Ti—modified
Hastelloy N continued. All products except seamless

51

tubing were received, and much experience was gained
in the fabrication of the new alloy. The products will be
used in all phases of the materials program.

The work on chemical processing materials is concen-
trated on graphite. Capsule tests are in progress to study
possible chemical interactions between graphite and
bismuth-lithium solutions and to evaluate the mechani-
cal intrusion of these solutions into the graphite. Since
the solubility of graphite in bismuth-lithium solutions
appears to increase with increasing lithium concentra-
tion, a molybdenum thermal-convection loop that con-
tained graphite specimens was run to study mass trans-
fer in a Bi—2.5% Li solution.

Some of the effort during this reporting period was
expended in reestablishing test facilities. Four thermal-
convection loops are in operation in the new loop
facility, which will accommodate at least ten loops. The
mechanical property and general test facility is partially
operational, but numerous test fixtures remain to be
assembled and tests started. An air lock has been added
to the general test facility to make it more functional,
and plans were developed partially for further expan-
sion of the facility.
6. Development of Modified Hastelloy N
" H.E. McCoy

The purpose of this program is the development of

metallic structural material(s) for an MSBR. The current .

emphasis is on the development of a material for the
primary circuit, which is the most important problem at
present. The material for the primary circuit will be
exposed to a modest thermal-neutron flux and to fuel
salt that contains fission products. It is believed that a
modification of standard Hastelloy N will be a satisfac-
tory material for this application. An alloy that contains
2% Ti appears to adequately resist irradiation embrittle-
ment, but it remains to be demonstrated that this alloy
satisfactorily resists shallow intergranular attack by the
fission product tellurium. Small additions of niobium
and rare earths (e.g., cerium, lanthanum) to the alloy
also improve the resistance to shallow intergranular
cracking and likely will not reduce the beneficial effect
of titanium in reducing neutron embrittlement. Increas-
ing the chromium concentration from the present 7% to
a value in the range of 12 to 15% may also be beneficial
in preventing shallow intergranular attack. Currently,
factors associated with production of the 2% Ti—
modified alloy in commercial quantities are being
studied, while smaller heats are being made of Hastelloy
N containing both 2% Ti and additions of niobium and
rare earths. These materials are being evaluated in
several ways.

Two large heats, one 10,000 Ib and the other 8000 1b,
of the 2% Ti—modified alloy have been melted by a
commercial vendor. Product shapes including plate, bar,
and wire have been obtained for use in several areas of
the alloy development program. Tubing is currently
being produced by two independent routes. The various
product forms from the two large heats are being used
to fabricate the salt-contacting portions of two forced-
circulation loops.

Laboratory methods for studying Hastelloy N--salt—
tellurium reactions are under development. Methods
must be developed for exposing candidate structural
materials to simulated reactor operating conditions.
Tests are being run in which specimens are exposed at
700°C to the low partial pressure of tellurium vapor in
equilibrium with tellurium metal at 300°C. Other tests
involve metal tellurides that are either added to salt or
scaled in evacuated quartz vials to provide a source of
tellurium. Scveral experimental alloys have been ex-
posed to tellurium, and the extent of intergranular
cracking was evaluated metallographically. Essential to
this program are adequate techniques for identifying

52

and characterizing the reaction products. Several
methods for the analysis of surface layers are under
development.

Materials that are found to resist shallow intergranular
cracking in laboratory tests will be exposed to fissioning
salt in’ the Oak Ridge Research Reactor TeGen fueled-
capsule series. Three materials (standard Hastelloy N,
Inconel 601, and type 304 stainless steel) were exposed
in this manner during the first TeGen experiment, and
their cracking tendencies closely parallel those noted in
laboratory tests in which these materials were exposed
to tellurium vapor. Fuel pins for a second experiment
have been filled with salt containing 233U and will be
irradiated in the near future.

6.1 DEVELOPMENT OF A MOLTEN-SALT
TEST FACILITY

H. E. McCoy
T. K. Roche

K.W. Boling B.McNabb
J. C. Feltner

When the MSRP was terminated early in 1973, most
of the equipment was reassigned to active programs.
When the MSRP was reactivated a year later, the con-
struction and installation of new equipment .were neces-
sary before testing could begin. Building 2011, acquired
by moving the occupants into a smaller building, had
been used as a mechanical testing area about 12 years
previously and was already equipped with emergency
power and air conditioning. However, numerous im-
provements in the building were necessary in addition
to the acquisition and placement of new equipment.
Although all of the equipment is not operational, this
report will describe the status of the facility.

The building is a two-story structure with nominal
dimensions of 50 X 50 ft. The first floor is quite thick
and more suitable for mounting vibration-sensitive test
equipment. The second floor is of lighter capacity and is
more useful for offices and support activities. There are
two stairways leading to the second floor, but all heavy
items must be brought up by an overhead crane which
extends from the west side of the building. The west
wall had deteriorated, and large doors leading into the
first-floor experimental area made close temperature
control almost impossible. An air lock having the
dimensions 10 X 30 ft was added 1o the west side of the
building which greatly increased the building’s useful-
ness for experimental work. An inoperable emergency
power gencralor, located in a small building on the east
side of Building 2011, was removed and the space
renovated to provide a small shop area.

Figure 6.1 is a photograph of the west side of Building
2011. The air lock which was added is visible on the
left-hand side. The crane for transporting materials

the second floor is also shown. Figure 6.2 is a view of
the north side. Gas storage racks, the new emergency
power generator, and the small shop area (left side) are
evident.

Figure 6.3 shows the equipment layout for the first
floor. Some of the equipment in the southwest corner is
used by the Analytical Chemistry Division for deter-
mining the concentration of oxygen in liquid-metal
samples. A neutron generator is located beneath the salt
storage area and is used for oxide activation analyses.
This analytical capability is quite unique and will likely
be maintained. The 14 lever-arm creep machines on the
north side are for testing in an air environment, the 8
It envi-

machines in the next row are for testing in a s
ronment, the 8 machines in the next row are for testing
in an air environment. Five strain cycle machines are

located in the southeast corner and will operate with

Fig. 6.1. Molten-Salt Test

53

test specimens in a salt environment. The temperature
and strain readout equipment is centrally located. Salt
storage facilities and salt charging equipment are used in
conjunction with the tests operating in salt environ-
ments.

The equipment layout on the second floor is shown in
Fig. 6.4. Six dead-load creep machines for testing in an
air environment are located on the south wall. The
tube-burst equipment is only partially installed, and its
installation is not considered a high-priority item. The
annealing facility consists of ten furnaces having various

temperature and environmental capabilities. A

eparate
laboratory in the northwest corner is used for experi-
N

The other facilities on the second floor include offices,

ments involving tellurium—Hastelloy interactions.
a data storage and processing area, an instrument repair
shop. and general storage

A view of some of the 22 lever-arm creep machines

for testing in an air environment is shown in Fig. 6.5
6.6. All
are in operation. The control cabinet shown in Fig. 6.6

and a closeup is shown in F of these machines

contains the instrumentation for two creep machines.

Photo 2256—75

“acility (Building 2011) from the west side. The newly constructed air lock is on the left.
&

fi' Photo 2254-75

Fig. 6.2. Molten-Salt Test Facility (Building 2011) from the north side. Features of interest
left, the emergency generator on the left. gas storage rack in the center. and the newly constructed ai lock on

the shop

ORML-DWG 7314086

,,,,,,,,, L

14 CREEP MACHINES:

[ ]

Fig. 6.3. Equipment layout for the first floor of Building 2011,

N
o]
3

55

S~ ORNL- DWG 75-14085
T
IA!R COND. | I ]
- STORAGE
§| INsT ARE A OFFICE OFFICE OFFICE
REPAIR —
SHOP HOOD
§ 5 \‘:{\J
&) |
B 2
\ DATA PROCESSING
w
S8 \TQ—METAL REACTION AREA T
% STUDIES
3
DESK — \
| v C
H FILE - I N
\L CABINET PAR s/ ANNEALING
T
! ) CABINETS PARTS CABINETS FACILITY
OVERHEAD i
CRANE AN WORK METAL STORAGE
. BENCH
TWORK |
BENCH
PRESSURE & TEMP 9 TUBE BURST SYSTEMS AR
_ “_‘ READOUT (6 AR & 3 SALT) \ COND.
ENCLOSED GAS
STARWELL || poraBLE

CHARGING

RIG/

6 DEAD LOAD CREEP MACHINES

WORK BENCH

— IO I

l WORK |
T1| eencu T

2nd FLOOR L

10 Ft. 1

BLDG. 2011 I 1

Fig. 6.4. Equipment Jayout for the second floor of Building 2011.

The frame is of welded steel construction. The lever
arms have two sets of knife-edge pivots so that the
weight on the back of the arm is multiplied by factors
of 6 or 12. The pull rods and extensometers are cur-
rently arranged for testing small specimens having gage
dimensions 1'% in. long by ‘4 in. in diameter. Pull rods
and extensometers for. larger specimens (required for
code testing) were also fabricated and can be used in the
same machines.

The specimen deformation can be determined by the
dial gage or by a transducer which measures the deflec-
tion of the dial gage shaft. This transducer signal is con-
veried to a dc signal by the instrumentation in the
bottom of the control cabinet on the right (Fig. 6.6)
and is printed at another location. The electronic circuit
will also accommodate averaging transducers which will
be used on more precise code work. The instrumenta-
tion in the bottom of the control cabinet also has a
module for measuring load from a load cell (not shown)
which [its in the bottom on the creep machine. The
specimen is hecated by a resistance-wound furnace

having a maximum (empcrature capability of 1200°C.
The temperature is measured by up to four Chromei-
Alumel (£%% accuracy) thermocouples located at
various positions on the specimen gage length. The
signal from onc of these thermocouples is used by the
Leeds and Northrup type 80 proportioning controller to
control the furnace temperature. Switches within this
unit activate an alarm shown in the upper left corner of
the control cabinet (Fig. 6.6) if the temperature varies
more than #6°C from the control temperature. This
alarm unit activates a local light and bell alarm as well as
causing an alarm to sound in the Shift Opecrations
Office. A sccond thermocouple is tied to an over-
temperature monitor (lower left side of control cabinet
in Fig. 6.6). This monitor is set 10 to 15°C above the
control temperature and will interrupt power to the
furnace, The monitor must be reset manually. The
furnace is powered by a solid-state power supply de-
signed by T. Hutton of the Instrumentation and Con-
trols Division. The unit incorporates a digital Variac
which allows power settings of 0,10, 25, 50, 75, and
Photo 225575

Fig. 6.5. General view of salt-environment creep machines on the left and air-environment creep machines on the right.

100% of line voltage. Power is pulsed through the unit
as called for by the Leeds and Northrup controller.

The six dead-load creep machines on the second floor
are quite similar to the lever-arm creep machines just
described. As shown in Fig. 6.7 these machines are not
in operation, but the construction work is complete.
Since the load is applied directly to the bottom of the
specimen, the equipment is limited to specimen stresses
of about 20,000 psi. However, the frames can be con-
verted to the lever-arm type.

Two salt environment creep machines are shown in
Fig. 6.8. The frames and control instrumentation are
the same as for the air environment machines shown in
Fig. 6.6. The primary modification is a stress unit which
can be immersed in salt. Four load-bearing rods run
from the bottom of the specimen to a flange near the
top of the frame. A rod from the lever arm passes
through a seal in the flange to the top of the specimen.
Thus, weights placed on the back of the lever arm place
the specimen under tensile stress, with the pulling force
being transferred back to the flange. No rods protrude

far below the bottom of the specimen, and a salt con-
tainer can easily slip over the stress unit. This container
seals against the underside of the flange. Extensometer
rods for measuring the strain pass through seals in the
flange, and strain can either be measured by a dial gage
or a transducer. A 4-in.-diam furnace fits over the salt
container. There are several openings through the flange
into the container for gas lines and ball valves for elec-
trochemical probes and for making additions to the salt.
Figure 6.9 shows the salt-creep machine which is in ser-
vice. The salt was transferred into the pot on the right
in the salt preparation facility at Y-12. The transfer pot
was placed in a furnace, after which the transfer pot and
the receiver vessel were heated to about 600°C before
the salt was transferred by applying argon pressure to
the transfer pot. The temperature was stabilized in the
creep chamber, and a stress was applied to the test spec-
imen.

One of the five strain-cycle units is shown in Fig.
6.10. The test specimen is a 1-in.-OD tube with a re-
duced gage section having a length of 1 in. The tube is

57

Photo 2253-75

Fig. 6.6. Close-up of two air-environment creep machines with associated instrumentation.

welded in place and stressed by a rod which extends
from the bottom of the specimen to a piston above the
specimen. The piston is moved by applying air pressure
to either side, resulting in a tensile or compressive force
on the specimen. The specimen assembly is immersed in
salt while it is being stressed. Extensometer rods extend
through the top flange to measure the strain. These rods
move transducers whose signals are recorded on the
bottom instrument. Switches inside the recorder can be

adjusted to change the stress from tensile to compres-
sive when the strain reaches certain values. The test can
also be controlled on a time basis and the strain re-
corded. Other modes of control are also possible. This
type of test is to study the rate of crack propagation
through thin-walled tubes of varying composition in the
presence of tellurium. Installation of the equipment has
been completed, and test specimens are welded in place.
The tests will be started as manpower becomes avail-

able. These machines will be used for alloy screening,
and more precise work will be done on MTS equipment
to be procured at a later date

The main data collection station is on the first floor
(Fig.
contains switches, a digital readout, and a single point

6.11). The upper part of the cabinet on the left

recorder for temperatures from about one-third of the
machines. The bank of switches in the top of the right-
hand cabinet is for selecting strain ranges for each

machine. The strain readings from each machine are
printed out on one of the multipoint recorders. A data
logging instrument is located in the middle of the right-
hand cabinet. This instrument will print out 100 points
on the designated frequency (usually 1 hr). This is suf-
ficient capacity to print out one temperature and one
strain reading for each piece of equipment. Two other
temperature measuring stations are located elsewhere on
the first floor.

J Photo 2247-75

Fig. 6.7. General view of air-environment dead-load creep machines.

The annealing area is on the second floor (Fig. 6.12)
The two furnaces on the Jower level have environmental
control and are used for short-term anneals. Eight other
furnaces are used for long-term anneals in which the
samples are encapsulated.

Figure 6.13 shows a typical area in the second-floor
laboratory used for tellurium—Hastelloy N studies. The

59

equipment includes a quartz encapsulation apparatus,
special gradient furnaces for annealing the capsules,
equipment for measuring gas-metal reaction rates, and a
general-purpose hood.

The tube-burst equipment is on the second floor (Fig.
6.14). There are nine test stations with each station
having four test positions. This equipment is pressurized

Photo 2241-75

Fig. 6.8. Close-up of two salt-environment lever-arm creep machines. The salt chamber on the left-hand machine seals against the
horizontal flange and the furnace is raised a proportionate distance. The temperature control and strain-measurement instrumenta-
tion are shown on both sides of the creep machines.

60

Photo 224875

Fig. 6.9. Lever-arm salt-environment creep machines in operation. The salt chamber and furnace have been raised. Salt was

transferred by argon pressure from the vessel on the right into the test chamber. The cabinet on the left contains switching and
temperature readout instrumentation for several creep machines

Photo 224375

Fig. 6.10 Close-up of a salt-environment strain cycle machine and associated instrumentation. The test specimen is a 1-in.diam
tube welded on the bottom to a rod and on the top to a heavy-walled tube. The rod passes through the tube, and alternating tensile
and compressive stresses are imposed on the specimen by the actuator (piston-cylinder combination). The instrumentation is used to
control and record the stress-strain-time history.

Photo 226075

Fig. 6.11. The cabinet on the left is one of several readout stations for temperature. Chromel-Alumel sensors from several creep
machines are run to this cabinet. The switches make it possible to read each thermocouple individually on the digital unit at the top
of the cabinet. One point at a time can be recorded on the Azar recorder. The cabinet on the right contains a data logging unit for
recording strain and temperature on all the machines on the first floor. The three recorders in the bottoms of the two cabinets record

strain data from all machines.

63

Fig. 6.12. Photograph of heat-treating facility. The two lower furnaces have controlled argon environments. Eight other furnaces
(all not visible) have air environments and are used for long-time anneals.

Fig. 6.13. Typical view of general-purpose laboratory used primarily to study tellurium-metal reactions. The hood on the left is
used to clean salt from specimens tested in salt environments.

Fig. 6.14. General view of tube-burst testing equipment used to stress tubular specimens by internal pressure. The front panels
contain only the pressure-related equipment. The furnaces where the test specimens are located and their associated control
instrumentation are behind the pressure panels.

by a pump with a vent pressure of 14,400 psi. The
pump and the associated reservoir cylinders have been
approved for operation, and the individual test stations
will be put in service on a low-priority basis.

Immediate emphasis will be placed on getting all
equipment into operation. Longer-term objectives will
include procurement and installation of an MTS fatigue
machine and possible expansion of the first floor to
accommodate additional creep machines.

6.2 PROCUREMENT AND FABRICATION
OF EXPERIMENTAL ALLOYS

T. K. Roche R. E. McDonald
B. McNabb J. C. Feltner

6.2.1 Production Heats of 2%
Ti—Modified Hastelloy N

One of the more promising alloys at present for the
primary circuit of an MSBR is 2% Ti—modified
Hastelloy N. Progress has been made in the scale-up of
this alloy with the production of two large heats, one
10,000 1b and the other 8000 lb, by a commercial
vendor. The analysis of the heats was reported pre-
viously.! These heats were used to establish processing
parameters for producing plate, bar, and wire; and more
recently emphasis has been placed on processing seam-
less tubing. Mill products from these heats are being
tested in the general alloy development program and
used in the construction of two forced-circulation loops
for studying the compatibility of the alloy with fuel
salt.

As reported previously, several fabrication problems
were encountered with the first heat (heat 2810-4-7901
or 74-901; 10,000 1b) in that it was prone to cracking
during hot-working operations, particularly during hot
rolling of the plate. However, with the aid of Gleebie
evaluation tests which defined the hot-working tempera-
ture-range of the heat to be between 1090 and 1177°C,
plate products were successfully rolled. A second prob-
lem was the susceptibility of the heat to cracking during
the annealing treatment following cold drawing in the
production of bar and wire products. This problem was
partially solved by either flexing the drawn product in
straightening equipment prior to annealing at 1177°C or
by lowering the intermediate annealing temperature to
1121°C.

Because a considerable amount of the first heat was
consumed in establishing processing parameters, a

1. T. K. Roche, B. McNabb, and 1. C. Feltner, MSR Pro-
gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp.
60-63.

65

second heat was produced (heat 8918-5-7421 or
75-421; 8000 1b) for conversion to tubing, bar, and
wire. The hot-forging behavior of this heat was quite
good as confirmed by Gleeble data, which showed a
very broad hot-working temperature range of 930 to
1260°C. Approximately one-half of this heat was forged
and turned to 4.5-in.-diam bar for conversion to seam-
less tubing by the vendor. Also, a forged bar 4 X 4 X 60
in. was produced for conversion to tubing by an alter-
nate route. The balance of the heat was converted to
the following products, which have been received:
6-in.-diam bar (630 1b), 0.5-in.-diam bar (292 ft),
0.312-in.-diam bar (996 ft), 0.125-in.-diam wire (405
1b), and 0.094-in.-diam wire (338 Ib).

For making products in the range }%-in.-diam bar
through % ,-in.-diam wire, forged bar was hot rolled to
about l-in.-diam bar, and an attempt was made to con-
vert this material by cold drawing to final sizes with
intermediate annealing treatments. This routing proved
satisfactory until a diameter of 0.395 in. was reached,
but annealing cracks, as experienced with heat 74-901,
were encountered to some degree during processing of
the 0.312-in.-diam bar and the wire products. For
example, during a run involving about 850 Ib of stock,
about 2% of the product was lost due to cracking during
annealing after the material was drawn from 0.395 in. in
diameter to 0.312 in. in diameter. The bar was mechani-
cally flexed prior to annealing, a technique used to
minimize cracking in material from the first production
heat of the alloy (heat 74-901). The annealing cracks
were observed to run parallel to the longitudinal axis of
the bar. Examination of a transverse section of the
cracked 0.312-in.-diam stock showed that the cracks
were intergranular in nature and up to 0.065 in. deep in
the section examined (Fig. 6.15).

It has been possible to reproduce the annealing crack
phenomc'non on a laboratory scale. Samples of
0.5-in.-diam bar of each of the two production heats
were cold drawn to 0.395 in. in diameter (37% reduc-
tion) and annealed at 1177°C. Heat 74-901 developed
longitudinal cracks; heat 75-421 did not. These results
are consistent with the vendor’s observation that heat
74-901 is more susceptible to the cracking problem.
Since the cracking can be reproduced on a laboratory
scale, it may be possible to more fully characterize the
problem and define fabrication parameters necessary for
its prevention.

Of the two routes being pursued for the procurement
of seamless tubing, one by the commercial vendor in-

- volves trepanning forged and turned bar stock to 4.5-in.

OD X 0.5-in. wall, cold tube reducing (or pilgering) the
material in three steps to 2.0-in. OD X 0.187-in. wall,
66

annealed at 1065°C. Etched with glyceria regia. 50X
followed by cold drawing to final sizes of 1.0-, 0.75
0.5-, and 0.377-in. OD X 0.035- to 0.072-in. wall. This
route for tubing production depends upon the efforts of
two other vendors, one for trepanning the bar and the
other for drawing to final sizes. The trepanning opera-
tion has been completed and resulted in six tube hol-
lows each approximately 6 ft long. Each of these pieces

was processed through the first tube reducing pass to a
3.75-in. OD X 0.375-in. wall (36.7% reduction) with no
difficulty. From this point, work was confined to one
tube hollow to determine its response to in-process
annealing at 1121°C and water quenching followed by
further tube reduction. Annealing of the hollow be-
tween each tube reducing operation was preceded by
the annealing of a sample which was then liquid-
penetrant inspected to determine any evidence of crack-
ing. With this procedure the hollow was taken through
the remaining two tube reducing steps and three anneal-
ing treatments with no major problems. A few shallow
surface flaws did develop, but these were readily condi-
tioned from the product. Therefore, on hand at present
is approximately 24 ft of 2.0-in.-OD X 0.187-in.-wall
stock which will be scheduled with the redraw vendor
for processing to final sizes. The tube reduction of the
remaining five hollows will also proceed.

Y-133872

MICRONS
L 50x —L
INCHES

T
0.0

1

200 400

=

~diam bar of 2% Ti-modified Hastelloy N (heat 75-421). Bar was cold drawn 37% and

The second route for obtaining seamless tubing in-
volves hot extrusion of tube shells at ORNL followed
by cold drawing to size by an outside source. Starting
stock was the forged bar of the alloy, 4 X 4 X 60 in.
(Fig. 6.16). The bar was machined into six billets, each
of which measured 3.950 in. in OD by approximately
9.5 in. long and had a 45° tapered nose for extrusion
out of a 4.060-in-diam press container through a
conical die. Two of the billets were drilled 0.812 in. in
1D and four were drilled 1.0 in. in ID to accommodate a
mandrel and to allow for a slight variation in extrusion
ratio. A glass coating, which was molten at the extru-
sion temperature, was applied to the billets and served
as the primary lubricant. Additional lubrication was
provided by Fisk-604 grease that was applied to the
tooling. Five of the billets were extruded at 1200°C and
one at 1250°C. Low extrusion rates (ram speeds) were
used to prevent a sufficiently large temperature increase
that the incipient melting temperature of the alloy
would be exceeded: otherwise serious cracking could
result. This problem was encountered during the devel-
opment of standard Hastelloy N, but was solved by con-
trol of extrusion rate.

The results of extruding the Ti-modified
Hastelloy N billets, in the order performed, are pre-

2%

sented in Table 6.1. For the first three tube blanks
produced (extrusions 1603, 1604, and 1605). the low
rate of extrusion caused mandrel failure due 10 exces-
sive heating of the tooling. However, the length of tube
blanks obtained with various extrusion ratios and ram
speeds suggested that the combination of tooling used
for extrusion 1604 (extrusion ratio of 9.4:1) with an
extrusion speed approximately equal to that of extru-
sion 1605 (2.5 in./sec) should produce a complete tube
blank. This was the case for extrusions 1608 and 1609.
In an attempt to reduce the force required to make
these extrusions, the final tube blank (extrusion 1611)

67

was extruded at a slightly higher temperature, 1250°C.
A rather long length of good extrusion was obtained,
but mandrel failure again occurred due to the low extru-
sion speed.

Visual inspection of the tube blanks showed the OD
surfaces to be quite good as illustrated in Fig. 6.17,
which shows the leading end of extrusions 1608 and
1609. On the other hand, boroscopic examination ol
the IDs by the outside vendor performing the redraw
operation revealed flaws which were subsequently
removed by gun drilling. These flaws, shown typically in
Fig. 6.18, are believed to be caused by inadequate lubri-

Photo 1273-75

Fig. 6.16. Forged bar, 4 X 4 X 60 in., of 2% Ti-modified Hastelloy N (heat 75-421). Stock for extrusion billets to produce

seamless tubing

Table 6.1. Conditions and results of

tube blank extrusions of 2% Ti-modified

Hastelloy N (heat 8918-5-7421)¢

Diediameter 24! © g iidon  Ram speed Force (tons)
Extrusion diameter h Results
(in.) ) ratio (in./sec) Maximum  Running

1603 1.475 0.812 10.4:1 1.5 1230 1 Good OD surface; 22 in.
of tube blank extrusion
before mandrel failure

1604 1.625 1.0 9.4:1 1.5 1160 1100 Good OD surface; 45 in.
of tube blank extrusion
before mandrel failure

1605 1.475 0.812 10.4:1 25 1290 1290 Good OD surface; 40 in.
of tube blank extrusion
before mandrel failure

1607 1.625 10 9.4:1 Stalled; operational
error

1608 1.625 1.0 9.4:1 36 1290 1290 Good OD surface; 70 in.
tube blank extrusion

1609 1.625 Lo 9.411 36 1290 1290 Good OD surface; 70 in.
tube blank extrusion

1611 1.625 1o 9.4:1 1O 1000 900 Repeat of extrusion 1607;
£00d OD surface: 55 in.
of tube blank extrusion
before mandrel failure

@Notes: Container diam: 4.060 in.

Extrusion temp: 1200°C, except extrusion 1611 at 125
Lubrication: glass on billets, Fisk-604 grease on tooling.

Photo 1536—75

Ji07 oy

Fig. 6.17. Tube-blank extrusions of 2% Ti—modified Hastelloy N (heat 75-421). These are the leading ends of the two extrusions
and were photographed in the as-extruded condition

v-133870

600 700
- T

L
INCHES 0.020 0.025

MICRONS
100X

L

T

100 200
0.005 0.010

Fig. 6.18. Typical defects on the inside diameter of extruded tube blanks of 2% Ti-modified Hastelloy N (heat 75421).
Longitudinal section. Etched with glyceria regia. 100X.

cation during extrusion. Therefore, several additional

extrusions are planned to test this assumption and to:

evaluate other lubricants. The billets will be prepared
from the 6-in.-diam bar fabricated from the same com-
mercial heat used for the previous billets.

The products of the six extrusions (Table 6.1) were
sent to an outside vendor for redrawing to finished
tubing. More effort was required than anticipated due
to several factors: the conditioning required to clean up
the surfaces of the extrusions, experimentation with
both plug and rod drawing techniques to establish a
workable processing schedule, and more frequent and
longer intermediate annealing treatments than antici-
pated for the alloy. The vendor believes that a satisfac-
tory drawing schedule has been developed and is pro-
ceeding with processing of the remaining extrusions.
The vendor’s preferred process involves rod drawing and
sinking operations requiring about 18 to 20% deforma-
tion per pass, with intermediate annealing treatments at
1177°C followed by water quenching. Quality control
steps after each process step include light etching fol-
lowed by visual inspection of the OD and boroscopic
inspection of the ID for defects. It is believed that the
yield of tubing from the redrawn ORNL extrusions will
be sufficient for at least one of the two forced-
circulation loops now under construction.

A 6-ft length of cold-worked 0.75-in.-OD X
0.072-in.-wall tubing was received as product from the
development work required to establish a drawing
schedule. The tubing is being evaluated by nondestruc-
tive techniques. Liquid-penetrant inspection of the QD
showed no defects. Silicone rubber replication together
with radiographic inspection indicated the presence of
relatively shallow cracklike indications on the 1D over a
4-ft length. Metallographic examination of a small
sample from the end of the 6-ft section showed the
defects to be a maximum of 0.028 in. deep. The tube
will be annealed and inspection will be repeated,
including examination by an eddy-current technique. In
view of the number of process variations to which the
tube was subjected in developing a drawing schedule,
the quality of the OD and that of portions of the ID is
encouraging.

6.2.2 Semiproduction Heats of 2% Ti—Modified
Hastelloy N Containing Niobium

To provide stock for a more complete characteriza-
tion of niobium-titanium-modified Hastelloy N, eight
50-1b heats and one 2500-1b heat (Table 6.2) are being
prepared by a commercial vendor. Niobium additions to
the 2% Ti—modified Hastelloy N base are of interest for
enhancing resistance to tellurium embrittlement, and

69

Table 6.2. Nominal composition of semiproduction heats
of 2% Ti—modified Hastelloy N containing niobium

Base: Ni—12% Mo—7% Cr-2% Ti—0.07% C

‘ Addition of the indicated element? (%) Heat
Alloy - size
Fe Si Mn Nb (I1b)

1 - 1.0 0.1 0.2 0.85-1.15 2500
2 1.0 0.1 0.2 04-0.6 50
3 1.0 0.1 0.2 1.35—-1.65 50
4 1.0 0.1 02  1.8-22 50
5 3.0-5.0 0.1 0.2 0.85-1.15 50
6 1.0 0.1-0.2 0.2 0.85-1.15 50
7 1.0 0.1 0.2-0.5 0.85-1.15 50
8 3.0-5.0  0.1-0.2 02-05 0.85-1.15 50
9 1.0 0.1 0.2 0.85-1.15 50

21ndividual values denote maximum concentration.

niobium levels between 0.5 and 2 wt % will be investi-
gated. In addition, the compositions of four of the
alloys were chosen to investigate different levels of the
residual elements, Fe, Mn, and Si. These results will be
important because of the beneficial effects of the resid-
ual elements upon oxidation resistance and will allow
greater latitude in scrap recycle.

The nine alloys have been melted and will be proc-
essed to products in the near future. The eight 50-1b
melts will be forged and rolled to Y%-in.-thick plate
approximately 4 in. wide. This material will be used for
weldability, salt corrosion, tellurium compatibility, and
mechanical property tests. The 2500-1b heat will be con-
verted to 1% -, %-, and ¥, ¢-in.-diam bar products and
to a4 X 4 in. round-cornered square bar. About half the
material will be retained in the last form to allow future
capability for producing additional products of sheet,
bar, and tubing.

6.3 WELDABILITY OF COMMERCIAL ALLOYS
OF MODIFIED HASTELLOY N

B. McNabb H. E. McCoy T. K. Roche

Welding at ORNL is generally performed in accord-

~ance with Section 1X of the ASME Boiler and Pressure

Vessel Code.? Basically this requires that a procedure
for welding a material (or class of similar materials) be
developed and that welders demonstrate that they are
qualified to weld by the procedure. The procedure must

2. ASME Boiler and Pressure Vessel Code, Section I1X, Qual-
ification Standard for Welding and Brazing Procedures, Welders,
Brazers, and Welding and Brazing Operators, American Society
of Mechanical Engineers, New York, 1974.
be a written document including the essential variables
associated with making the weld and must be backed by
test reports, including bend and tensile tests which show
that the weld is sound. A welder can then be qualified
to use the procedure by making a weld which is sub-
~jected to bend tests to show that it is sound. This is a
very simplified view of the process used to develop and
maintain high welding standards, and the ASME Boiler
and Pressure Vessel Code, Section IX,? should be con-
sulted for more detail. The Plant and Equipment Divi-
sion maintains a weld test shop under the supervision of

D. R. Frizzell to implement the process, and this shop is -

frequently assisted by the Inspection Engineering
Department.

Procedures were previously developed for joining
Hastelloy N to Hastelloy N (WPS-1402), and Hastelloy
N to the austenitic stainless steels (WPS-2604), but it
was necessary to demonstrate whether these procedures
apply equally well to 2% Ti—modified Hastelloy N.
Therefore, %-in.-thick test plates of 2% Ti—modified
Hastelloy N (vendors heat 2810-4-7901, designated
ORNL heat 74-901) were prepared and welded as fol-
lows: autogenous welds with 74-901 weld wire, welds
with 2% Ti—modified Hastelloy N weld wire (vendor’s
heat 8918-5-7421, designated ORNL heat 75-421),
welds to standard Hastelloy N heat NI-5075 with 2%
Ti—modified Hastelloy N heat 75-421 weld wire, and
welds of type 304 stainless steel heat 18024 with Inco
82T heat NX59138-D weld wire. Each weld was sub-
jected to visual, dye penetrant, x-ray, and metallo-
graphic examination, and to two tensile and four side-
bend tests. These tests were conducted in accordance
with the ASME Boiler and Pressure Vessel Code, Sec-
tion [X,? and all of the above-mentioned welds passed
the tests. The tensile specimens were machined from the
test plates with the weld in the reduced-section gage
length and were tested in a Baldwin tensile machine. All
Hastelloy N welds exceeded the required minimum
100,000 psi ultimate tensile strength except the weld of
2% Ti—modified Hastelloy N to type 304 stainless steel,
which ruptured in the type 304 stainless steel base
metal at 93,300 psi. The side-bend specimens were % X
% X approx 8 in. long with the weld in the center, and
were bent around a 1-in. radius in a guided bend fixture.

The chemical analyses of the various materials
involved are shown in Table 6.3. The side-bend speci-
mens are % in. thick X % in. wide, bent around a 1-in.-
radius mandrel in a guided bend test. The specimens
were macroetched in a solution of HCI-20%
HNO;--20% H,O to delineate the weld and heat-
affected zones. Figure 6.19 is a macrophotograph of
side-bend specimens of standard Hastelloy N heat

70

NI-5075 %-in.-thick plate welded with standard
Hastelloy N weld wire heat NI-5101. There were no
flaws in the specimens after bending; and visual, dye
penetrant, Xx-ray, and metallographic examination
before bending showed that the welds were sound. This
weld was made and tested to recertify the welder and to
update the welding procedure specification. Figure 6.20
is a macrophotograph of side-bend specimens of 2%
Ti—modified Hastelloy N (top) (heat 74-901) welded to
standard Hastelloy N (bottom) (heat N1-5075) with 2%
Ti—modified Hastelloy N (heat 75-421) weld wire by
welding procedure specification WPS 1402. There were
no flaws in the welds, and the strain markings delineate
the weld areas. .

Figure 6.21 is a macrophotograph of 2% Ti—modified
Hastelloy N plates (heat 74-901) welded with 2% Ti—
modified Hastelloy N (heat 74-901) weld wire on weld-
ing procedure specification WPS 1402. There were no
flaws in the welds, and the specimens were macroetched
to delineate the weld areas. Figure 6.22 is a macro-
photograph of the same heat of 2% Ti—modified
Hastelloy N (74-901) welded with 2% Ti—modified
Hastelloy N (heat 75-421) weld wire by specification
WPS 1402. There were no flaws in the welds and the
specimens. Figure 6.23 is a macrophotograph of side-
bend specimens of type 304 stainless steel %-in. plate
(heat 18024) (top) welded to 2% Ti—modified Hastel-
loy N (heat 74-901) (bottom) with Inco 82T (heat NX
59138-D) weld wire by welding procedure specification
WPS 2604. There were no flaws in the welds, and the
specimens were macroetched to delineate the weld
areas. .

Although special welding procedures were prepared
for the joining of standard to 2% Ti—modified Hastelloy
N with 2% Ti—modified Hastelloy N filler wire (WPS
1403) and for joining stainless steel to 2% Ti—modified
Hastelloy N with Inco 82T filler wire (WPS-2606), the
parameters used in making these welds were identical to
those used in procedures WPS 1402 and WPS-2604
developed for standard Hastelloy N. Thus, we believe
that the test welds adequately demonstrate that stan-
dard and 2% Ti—modified Hastelloy N have equivalent
welding characteristics and that procedures WPS 1402
and WPS 2604 can be used for both materials.

Supplies of standard Hastelloy N weld wire were
depleted over a period of time, and additional wire was
purchased to the materials specifications MET-RM-
304B. However, the weldability test was performed by
ORNL. Plates of standard Hastelloy N (heat 5067) 1Y% ¢
X 4 X 10 in. were welded with the new heat of weld
wire from Teledyne Allvac (heat 9725), using welding
procedure specification WPS 1402, and were accepted
Table 6.3. Compositions of standard and modified Hastelloy N alloys used in weldability studies

Heat ' Concentration (%)
number Mo Cr Fe Mn Si C Ti Al w Y Co Cu P S B Ni Nb+Ta Other
N1-5075 163 6.7 4.0 044 0.58 0.06 0.01 0.01 0.04 028 0.07 0.0l 0.003 0.006 0.001 Balance
N1-5067 173 74 4.0 05 043 0.06 001 0.0l 0.06 028 0.08 0.0l 0.004 0.007 0.006 Balance
N1-5101 164 69 39 045 062 0.06 0.0l 0.01 0.06 034 0.10 0.01 0.001 0.009 0.007 Balance
474-901 129 7.0 007 0.02 0.04 006 1.8 0.1 0.02 0.02  0.02 0.003 0.002 0.001 Balance
475421 119 7.1  0.06 0.12  0.04 0.07 1.9 0.12 <0.06 0.02 <0.01 0.004 <0.002 <0.001 Balance
INCO 82-T¢ 200 3.0 3.0 05 0.1 0.75 0.5  0.03 0.015 67.0 2.5 0.5
Type 304 19.0 Balance 2.02 1.00 0.08? 0.045%  0.0300 10.0
554

2Nominal Composition.

bMaximum.

14
72

Fig. 6.19. Bend specimens of standard Hastelloy N (heat N1-5075) joined with standard Hastelloy N filler wire (heat 5101).

Fig. 6.20. Bend specimens of 2% Ti-modified Hastelloy N (heat 74-901) and standard Hastelloy N (heat N1-5075) joined with
Ti—modified Hastelloy N filler wire (heat 75-421).

73

¥—133301

Fig. 6.21. Bend specimens of 2% Ti—modified Hastelloy N (heat 74-901) joined with 2% Ti—modified Hastelloy N filler wire
(heat 74-901).

v-133208

Fig. 6.22. Bend specimens of 2% Ti-modified Hastelloy N (heat 74-901) joined with 2% Ti-modified Hastelloy N filler wire
(heat 75-421).

74

Fig. 6.23. Bend specimens of type 304 stainless steel and 2% Ti-modified Hastelloy N (heat 74-901) joined with 82T filler wire.

as passing all tests, including one all-weld-metal tensile
test and four side-bend tests with no flaws in the welds.
This material has been made available for general pro-

ject use.

6.4 STABILITY OF VARIOUS MODIFIED
HASTELLOY N ALLOYS IN THE
UNIRRADIATED CONDITION

T.K.Roche H. E. McCoy

The stability of Nb-, Ti
Hastelloy N with respect to intermetallic precipitation,
known as aging, is being studied. It is known that addi-

J. C. Feltner

and Al-containing modified

tions of these elements are desirable, respectively, for
enhancing resistance to tellurium-induced intergranular
cracking, for improving resistance to radiation embrit-
tlement, and for deoxidizing the alloy during melting.
However, beyond certain levels these elements can cause
aging reactions by the precipitation of gamma prime
[Ni3(ALTi)] or NizNb, which, in turn, causes hardening
or strengthening and loss of ductility. Therefore, studies
are in progress for defining the amounts of Nb, Ti, and
Al which can be added to Hastelloy N and still maintain
a reasonable degree of stability. The stabilities of a
number of alloys, including laboratory, semiproduction,
and production heats, having varying amounts of the

elements in question are being determined by hardness,
tensile properties, creep-rupture properties, and micro-
structural evaluation.

The first approach to evaluating stability has been the
determination of the room-temperature hardness of the
various alloys before and after heat treatment at 650,
704, and 800°C for periods up to 1000 hr. The data for
alloys held at these temperatures for 100 hr were re-
ported previously,” and during the present period the
1000-hr data presented in Table 6.4 were obtained. The
data for alloys which show significant hardening relative
the as-annealed condition are blocked off in the
table.

The hardness of the various alloys after a 1000-hr
aging period follows the same pattern noted for the
100-hr aging period. However, the previously reported
niobium concentrations of the niobium-containing
alloys were low due to an analytical error; the correct
values are shown in Table 6.4, The present data show
that with low aluminum concentrations (0.05 wt %),
70 as
concluded earlier) can be tolerated in 2% Ti—modified

to

niobium contents approaching 2% (rather than

3. T. K. Roche, D. N. Bra eltner, MSR Pro-
gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp.
71-76.

75

Table 6.4. Hardness of various heats of modified Hastelloy N?
after aging 1000 hr at 650, 704, and 800°C

Data in blocks indicate significant
hardening relative to the
as-annealed condition

Composition (wt %)

Rockwell B Hardness

Heat - b Aged 1000 hr
: Nb Ti Al - C Annealed” -

650°C 704°C 800°C
474-557 2.14 0.02 0.04 81.5 87.8 88.8 87.4
472-503 1.94 0.09 0.06 84 4 89.8 88.2 89.1
474-901 1.8 0.10 0.06 79.4 85.7 86.3 85.6
471-114 1.96 0.12 0.05 79.2 82.9 84.6 84 .4
427 2.4 0.18 0.014 74.7 78.4 77.3 77.9
428 2.417 0.16 0.064 825 85.2 86.1 86.0
474-533 2.17 048 0.05 81.0 85.7 86.2 84.9
474-534 2.09 0.53 0.08 89.3 92.5 90.4 90.9
429 2.4 0.35 0.017 76.9 94.8 85.8 78.8
430 . 2.5 0.34 0.073 88.6 100.1 88.9 91.2
431 ‘ 2.5 0.74 0.016 78.6 97.8 98.4 92.8
432 2.35 0.69 0.057 87.4 103.4 103.4 97.8
425 048 1.9 0.08 0.037 80.1 84.1 85.8 85.2
421 1.04 1.9 0.07 0.048 87.3 86.5 87.8 86.7
424 1.34 1.8 0.10 0.063 88.6 90.2 88.9 90 .4
418 1.92 2.0 0.05 0.058 89.1 904 90.0 90.8
420 1.90 1.8 0.15 0.055 88.7 101.5 91.6 91.3
435 1.42 2.3 0.15 0.04 88.6 105.0 102.1 90.3
438 1.80 24 0.13  0.05 91.8 106.6 105.1 96.8
433 1.89 2.2 0.33 0.024 84 .8 104.3 102.1 88.1
434 1.86 2.2 0.32 0.061 93.3 107.0 103.8 95.7
441 2.52 2.2 0.15 0.05 93.1 108.8 107.6 104.1
442 3.0 2.2 0.14 0.052 95.0 109.8  109.6 107.2

4Base: Ni—12% Mo—7% Cr.
b1 hrat 1177°C.

Hastelloy N before aging occurs. However, the tolerance
for niobium decreases with small increases in the tita-
nium and aluminum contents. It must be emphasized
that the hardness data were obtained on unstressed
spécimens, and that creep-rupture tests now under way
indicate that lower concentrations of niobium are toler-
able when the alloys are subjected to stress. These re-
sults are described later in this section. .

The effect of aging on room-temperature and
elevated-temperature tensile properties is being deter-
mined for most of the alloys. Limited room-
temperature results have been obtained after a 100-hr
aging period at 650 and 800°C (Table 6.5). There is
good correlation between the tensile data and the pre-
viously reported hardness data. As would be expected,
alloys which age harden during a given thermal treat-
ment (data underlined in Table 6.5) also show an in-
crease in strength and a corresponding decrease in duc-
tility relative to the solution-annealed condition.

In the case of the titanium-aluminum-modified
Hastelloy N alloys the hardening phase is most likely
gamma prime, and this phase has been identified in heat
430 (2.5% Ti + 0.34% Al) after 100 hr at 650°C.3

Stress-rupture results for the titanium-aluminum-
modified Hastelloy N alloys at 650 and 704°C are
shown in Tables 6.6 and 6.7 respectively. Again, there is
very good correlation between age-hardening behavior
as determined from short-term hardness data and the
corresponding rupture life in the stress-rupture tests.
Also, the effect of temperature upon aging can be seen
by comparing the data sets for heats 429 and 430 at
650 and 704°C. At the lower temperature both of these

‘alloys hardened, but neither hardened at 704°C. Since

hardening in these alloys is due primarily to the forma-
tion of gamma prime, these results suggest that the
gamma prime solvus temperature is located between
650 and 704°C for these two alloys. This observation
and the other aging data in Tables 6.4, 6.5, and 6.6 were

76

Table 6.5. Room-temperature tensile data for various heats of Nb-Ti-Al
modified Hastelloy N7 in annealed and aged conditions

Underlined data indicate significant strengthening relative to the as-annealed condition

Tensile properties

H Composition (wt %) © Conditi Strength (10° psi) Hardness,
eat ondition i
Nb - Al C rEe—— . Elongation (%) Rp
. Yield
tensile
474-901 1.8 0.10 0.06 Annealed | hr, 1177°C 114.0 44.2 66.7 79.4
Aged 100 hr, 650°C 116.8 50.0 559 81.8
Aged 100 hr, 800°C 119.0 51.2 537 85.1
428 2.47 0.16 0.06 Annealed 1 hr, 1177°C 124.5 49.0 56.2 82.5
Aged 100 hr, 650°C 125.3 50.5 52.7 84.2
Aged 100 hr, 800°C\ 123.7 51.1 48.1 - 84.7
430 2.5 0.34 0.07 Annealed 1 hr, 1177°C 125.6 50.0 54.1 88.6
Aged 100 hr, 650°C 1348 63.4 437 95.7
Aged 100 hr, 800°C 126.8 53.1 50.0 88.6
424 1.34 1.8 0.10 0.06 Annealed 1 hr, 1177°C 137.2 52.7 51.2 88.6
Aged 100 hr, 650°C 138.5 52.5 47.7 89.3
Aged 100 hr, 800°C 137.7 54.7 45.8 89.9
435 1.42 2.3 0.15 0.04 Annealed 1 hr, 1177°C 128.4 51.0 58.6 88.6
Aged 100 hr, 650°C 169.0 89.3 35.3 102.5
Aged 100 hr, 800°C 1339 54.5 53.8 87.3
442 3.0 2.2 0.14 0.05 Annealed 1 hr, 1177°C 139.8 60.7 58.2 95.0
) Aged 100 hr, 650°C 190.5 108.7 25.1 108.1
Aged 100 hr, 800°C 169.4 _ 83.8 39.9 105.6

9Base: Ni—12% Mo-7% Cr.

Table 6.6. Stress-rupture data for various heats of Nb-Ti-Al
modified Hastelloy N? at 650°C and 47.0 x 10° psi

Composition (wt %) Rupture Total

) ) Age harde
Heat life strain £ raens

Nb Ti Al C ) %) at 650°CP
474901 . 1.8 0.10 0.06 395.0 27.0 No
474-533 2.17 0.48 0.05 465.0 28.0. No
427 2.4 0.18 0.014 86.6 21.7 No
428 247 0.16 0.064 115.0¢ 7.3 No
429 2.4 0.35 0.017 957.2 16.3 Yes
430 : 2.5 0.34 0.073 1698.0¢ 7.3 Yes
431 2.5 0.74 . 0.016 2309.0 10.1 Yes
432 2.35 0.69 0.057 4171.0¢ 2.1 Yes
425 0.48 1.9 0.08 0.037 1430.1 23.7 No
421 1.04 1.9 0.07 0.048  2007.09 7.3 No
424 1.34 1.8 0.10 0.063 3953.0¢ 6.5 No
418 1.92 2.0 0.05 0.058 - 3955.0¢ 3.2 No
420 1.90 1.8 0.15 0.055 3951.0¢ 1.9 Yes
433 1.89 2.2 0.33 0.024 4169.0¢ 1.0 Yes
434 1.86 2.2 0.32 0.061 3955.0¢ 0.9 Yes

9Base: Ni-12% Mo—7% Cr.
bBased on hardness measurements on aged, unstressed specimens,
CTest still in progress.

dTest discontinued prior to fracture.

used to estimate the gamma prime solvus temperature
boundaries at 650 and 704°C as a function of aluminum
and titanium concentrations. Alloys with compositions
lying below the proposed boundaries in Fig. 6.24 are
stable at the indicated temperature, and those above
will precipitate gamma prime.

With the addition of niobium to titanium-aluminum-
modified Hastelloy N, defining or predicting stable com-
positions becomes more complex. There is evidence®
that mechanical stress significantly affects age hardening
of these alloys, which is not uncommon. The room-
temperature hardness data for annealed and aged speci-
mens of the various Nb-Ti-Al modified Hastelloy N
alloys suggest a tolerance of about 2% Nb in a 2%
Ti—0.5% Al modified Hastelloy N base (heat 418) be-
fore aging occurs. Further increases in the aluminum,
niobium, and titanium contents lead to age hardening at
650°C, then at 704°C, and finally -as high as 800°C
when the niobium content is increased to about 2.5%
with 2.2% Ti and 0.15% Al (heat 441). If the broad
assumption is made that the three elements are equally
effective in promoting an age-hardening reaction, and a
plot is made of the total atomic percent of these ele-
ments {(at. % Nb + Ti + Al) in the various alloys against
the increase in hardness (AR g} caused from aging 1000
hr at 650°C, a curve is obtained (Fig. 6.25). A sharp
break, indicative of appreciable aging, occurs between
3.8 and 3.9 at. % (Nb + Ti + Al). Adding the variable of
stress to aging response and plotting the parameter of
minimum creep rate from creep tests at 650°C and 47.0
X 103 psi (Table 6.8) against total atomic percent (Nb +
Ti + Al) results in the curve shown in Fig. 6.26. The
break now is indicated between 2 and 3 at. % (Nb + Ti +
Al). The creep rate of alloys containing up to about 2
at. % (Nb + Ti + Al) is 1.5 to 3.0 X 107°%/hr. Three
heats (70-835, 69-648, and 69-344) in the 2 to 3 at. %

77

region which are high in niobium and low in titanjum
are known to age upon creep testing at 650°C,* and
exhibit creep rates around | X 107>%/hr. One heat
(425) with the reverse combination, low in niobium and

4, H. E. McCoy, MSR Program Semiannu. Progr. Rep. Feb.
29, 1972, ORNL4782, pp. 167-69.

ORNL-DWG 75-13743

3.0
-_.. P
- N 4
. \\
25 |— h 430 ® 431
428 <
% 427" 429 ® 432
i 74-533 @
® 74-557 74-534 @ .
20 ——971-114 N 704 °C T
~ 72-503 | | Ty |
& ® 74-901 " 650 °C
-
o
Woys
<
o)
Q
i '
NO
1.0 4
0.5
0
0 0.2 0.4 06 0.8 10

Al CONTENT (%)

Fig. 6.24. Proposed boundaries separating stable from un-
stable alloys of Ni—12% Mo—7% Cr + Al and Ti with respect to
gamma prime precipitation at 650 and 704°C. Alloys above the
lines will form gamma prime and those below will not (see
Table 6.4 for the compositions of the various alloys).

Table 6.7. Stress-rupture data for several heats of titanium-aluminum
modified Hastelloy N? at 704°C and 35.0 X 10° psi

Composition (wt %) Rupture TOtfll Age hardens

Heat . life strain o b
Ti Al C (hr) (%) at 704°C

474-901 1.8 0.10 0.06 193.2 394 No
474-533 2.17 048 0.05 196.0 42.0 No
427 2.4 0.18 0.014 82.0 234 No
428 2.47 0.16 0.064 201.8 61.0 No
429 2.4 0.35 0.017 200.6 22.7 No
430 2.5 0.34 0.073 212.4 55.8 No
431 2.5 0.74 0.016 2938.3 6.8 Yes
432 2.35 0.69 0.057 3611.5 13.7 Yes

Base: Ni—12% Mo—7% Cr.

b Results of hardness measurements taken on aged, unstressed specimens.
78

ORNL-DWG 75-43714

20
0433

A Rg HARDNESS

4| — o425

3.5 4.0 4.5 5.0

at. % (Nb + Ti+Al}

Fig. 6.25. Change in hardness of unstressed specimens of
various heats of Nb—Ti—Al-modified Hastelloy N after aging
1000 hr at 650°C (see Table 6.4 for the compositions of the
various alloys).

high in titanium, does not seem to show much aging.
Alloys containing 3 to S at. % (Nb + Ti + Al) age appre-
ciably with creep rates of about 1 X 1073%/hr or less.

The above results indicate that alloys containing
approximately 2.5 at. % or less (Nb + Ti + Al) will be
satisfactory from the aging standpoint. Such an alloy
would be represented by the composition on a weight
percent basis of 0.5% Nb—1.5% Ti—0.1% Al. Alloys
having concentrations of Nb + Ti + Al above the 2.5 at.
% range will be more susceptible to aging.

Future work will include evaluation of microstructure
for a number of these specimens to confirm present
conclusions, evaluation of additional alloys to test the
indicated boundaries for stable compositions, and an
extension of data analysis to determine whether a quan-
titative relationship can be derived that separates the
relative effects of the individual elements, Nb, Ti, and
Al, on the stability of alloys of this type.

6.5 MECHANICAL PROPERTIES OF
TITANIUM-MODIFIED HASTELLOY N ALLOYS
IN THE UNIRRADIATED CONDITION

T. K. Roche J.C. Feltner B. McNabb

Several tests were completed or are in progress to
determine the mechanical properties of recently re-
ceived heats of 2% Ti—modified Hastelloy N in the unir-
radiated condition. These alloys include two production
heats (74-901 and 75-421) and six semiproduction heats
(74-533, 74-534, 74-535, 74-539, 74-557, and 74-558).

Table 6.8. Comparison of hardness changes? and creep behavior? of several heats
of Hastelloy N¢ modified with Nb, Ti, and Al

Composition Hardness, Rp
Minimum creep rate
Heat wt % at. % Annealed? 1000 l‘;lr Change (%/hr)
Nb Ti Al C Nb + Ti + Al at650°C
237 1.03  0.04 <0.05 0.84 1.5x 1072
63 2.5 <0.01 0.13 1.66 1.1 x10-?
181 1.85 0.50 <0.01 0.045 1.86 3.1 % 1072
69-648 1.95  0.92 0.05  0.043 2.55 7.0x 107¢
69-344 1.7 0.77 0.24  0.10 2.63 26X 1073
70-835 2.6 0.71 - 0.10 0.053 2.82 6.0x 1074
425 048 1.9 0.08 0.037 2.90 80.1 84.1 4.0 7.8% 1073
421 1.04 1.9 0.07 0.048 3.24 87.3 86.5 —0.8 1.3x 1073
424 1.34 1.8 0.1 0.063 3.38 88.6 90.2 1.6 7.1x107*
418 1.92 2.0 0.05 0.058 3.91 89.1 90.4 1.3 22x 107
420 1.90 1.8 0.15  0.055 3.87 88.7 101.5 12.8 1 x10°°
433 1.89 22 0.33  0.024 4.75 84.8 104.3 19.5 2 X10°°
434 1.86 2.2 0.32  0.061 4.70 93.3 107.0 13.7 3 x10°%

9 Alloys aged for 1000 hr at 650°C and hardness measured in unstressed condition.

bCreep tested at 650°C and 47.0 X 103 psi.
CBase: Ni—12% Mo—7% Cr.
41 hrat 1177°C.
79

ORNL-DWG 75-13742

6
> 0433
¢ 434
~d
4 RS
S €420 N~ [* 48
+ o~
= S+ 4 e424
+ T'|'--°\421
L
A 70-835 ¢ RN ARG
3 ' L' 69-344 TN
_ g
. 69-648 oY
© \
2 \ p 181
® 63\ '
\
4 \\
4 IR
237
\
0
100 2 5 1074 2 5 10° 2 5 {072 2 5 40!

MINIMUM CREEP RATE (% /hr)

Fig. 6.26. Minimum creep rate of various heats of Nb—Ti—Al-modified Hasfelloy N tested at 650°C and 47.0 X 10° psi (see

Table 6.4 for the alloy composition).

Four of the six semiproduction heats contain small
additions of rare earths, lanthanum, cerium, and misch
metal. The compositions of these alloys were chosen to
study the effectiveness of rare-earth additions for mini-
mizing the extent of shallow intergranular cracking.
Each of the six semiproduction alloys was the prodxct
of a 120-1b double-melted (vacuum induction plus elec-
troslag remelt) heat produced by an outside vendor. The
chemical analysis of the alloys was reported pre-
viously.® The mechanical-property studies include the
determination of room- and elevated-temperature
tensile properties and creep-rupture properties in air at
650, 704, and 760°C. These data serve as a reference for
comparison with the properties of standard and other
modified Hastelloy N alloys both in the unirradiated
and irradiated conditions. A

The principal effort during this report period was
directed toward completing the creep-rupture data on
the above heats. Tests are being run at three stress levels
for each of the three test temperatures. Most of the
tests were completed, with the major exception being
heat 75-421, the 8000-b production heat, for which
specimens are being prepared. Specimens of the other

5. T. K. Roche, B. McNabb, and J. C. Feltner, MSR Pro-
gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 61
and 65.

heats were obtained from swaged rod and were annealed
for 1 hrat 1177°C prior to test.

Figures 6.27 through 6.29 are plots of rupture time as
a function of stress at 650, 704, and 760°C, respec-
tively, for the 2% Ti-modified Hastelloy N heats and
are compared with plots for a previous heat (471-114)
of the same alloy and standard Hastelloy N. Minimum

creep rates measured from these tests at the three tem-

peratures are shown in Fig. 6.30 as a function of stress.
As concluded previously, the more recent heats of 2%
Ti—modified Hastelloy N are essentially equivalent in
strength to the earlier heat, and there is no significant
effect resulting from the addition of rare earths to the
2% Ti—modified alloy. As determined from past work
and confirmed by the recent tests, the modified alloy
exhibits longer rupture lives than standard Hastelloy N
at the three temperatures.

Additionally, the first of eight creep machines capable
of tests in molten fluoride fuel salt was put into opera-
tion. A specimen of heat 474-533 has been in test at
650°C and 30.0 X 10® psi for slightly over 1300 hr.
Data are not available, as yet, on this same heat in air,
but a comparison is made in Table 6.9 with an air test
of an earlier heat (471-114) of the same nominal com-
position.

The data appear to be falling within a normal scatter
band for alloys with the same nominal composition that
80

ORNL—DOWG 75-437M1

© M P TTHI
60 N.[2% Ti-MODIFIED HASTELLOY N (HEAT 471-114)
b | F
8 -51.
\\ \A (17.8 —51.0)
~ 50 SN~ ||
@ | F | Hrowme]| 4 a (19.8~42.9)
Y
5 STANDARD HASTELLOY N T™NL |1
Z 40 roQ mmal (23.3 - 42.8) |
b I ' | nt ~d
w \ o
o ¢ 474-533 \\\
= | @ 474-534 A
© 30 I 4 474-535 N
A 474-539
- o 474-557
20 |— ® 474- 558 . °
o 474- 901 TEST TEMPERATURE : 650 °C
S L] L []]
10 2 5 102 2 5 10% 2 5 104

RUPTURE TIME (hr)

Fig. 6.27. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastelloy N at 650°C. (Ranges
of rupture strain indicated in parentheses.)

ORNL-DWG 75-13710

60
50
."\-
iy
~ 40 ~+.| 2% Ti-MODIFIED HASTELLOY N (HEAT 471-114)
@ “T-h__ [ T TTTT
" S {39.4-62.8)
g \L\ e Lo
30 = = — (37.2-47.6)
L
@ T T~ ™ 0 (36.8-47.6)
= ¢ 474-533 ] \“‘*-I l ‘ l
20 ® 474-534 "‘-..._‘ Tt T —T1 1
5 474-535 ~{q=-STANDARD HASTELLOY N
A 474-539
0 474-557
10 |—w® 474-558
o 474-901 I ] ]
TEST TEMPERATURE : 704 °C
. T Y
10! 2 5 10° 2 5 103 2 5 104

RUPTURE TIME (hr)

Fig. 6.28. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastetloy N at 704°C. (Ranges
of rupture strains indicated in parentheses.)
81

ORNL- DWG 75-13709

35 ‘
\\
30 \{,2% Ti-MODIFIED HASTELLOY N (HEAT 471-114)
N
25 ~ o (47.2-57.5)
N \\
Z ~L L
“t 20 < eo‘mto-(m.g-ss.g)
Q h N
— '\‘
- \ N
;) n
w 15 STANDARD HASTELLOY N-ZP o a—0-(257-54.2)
& \\ N
Z R N ~
0 0 474-533 N
® 474-534 3
a 474-535 N
& 474-539
5 o 474-557
o 474-901 TEST TEMPERATURE 76C°C
] TN N
10! 2 5 102 2 5 10° 2 5 104

RUPTURE TIME (hr)

Fig. 6.29. Stress-rupture properties of several heats of 2% Ti—modified Hastelloy N and standard Hastelloy N at 760°C. (Ranges
of rupture strains indicated in parentheses.)

ORNL-DWG 75-13708

80
70 — o 474-533
® 474-534 -
-~ HEAT
s 474 -535 1 LA 47114
60 |- & 474-539 T
0 474 —557 PP i
® 474 —558 I~
~ 50 | © 474-901
@
=}
Lt p
o P le
40 — g 471114 ]
1 -
& T 704°C
— L]
n P - 4
30 = — 1" Jar-na
b 1 6’_’
20 . £7600C
-
10 = 2% Ti—MODIFIED HASTELLOY N
(47t-114)
O L L
1073 2 5 1072 2 5 107! 2 5 10°

MINIMUM CREEP RATE (%/hr)

Fig. 6.30. Creep properties of several heats of 2% Ti—modified Hastelloy N at 650, 704, and 760°C. Solid lines are for heat
71-114, and the dashed lines indicate bands which contain the data for the other modified alloys.
Table 6.9. Creep data for specimens
of two heats of 2% Ti—modified
Hastelloy N in salt and air
environments?

Accumulated strain (%)

Time
(hr) 474-533 471-114
(fluoride fuel) (air)
500 1.3 2.4
1000 2.8 4.4
1300 3.5 5.5

9Tests run at 650°C and 30.0 X 102 psi.

are tested under similar conditions, and there is no
indication that corrosion by the molten fluoride fuel
salt represents a significant factor.

6.6 POSTIRRADIATION CREEP PROPERTIES
OF MODIFIED HASTELLOY N

H.E.McCoy T.K.Roche

Postirradiation creep tests are in progress on speci-
mens from five experiments that were irradiated in the

82

poolside of the ORR. Each experiment contains 102
miniature creep specimens in an instrumented facility in
which temperatures can be measured and controlled by
supplying heat from auxiliary heaters. Only 12 in-cell
creep machines are available for postirradiation creep

_testing; hence the testing proceeds rather slowly. The

most recent tests have concentrated on (1) the prop-
erties of six 125-lb semiproduction heats that contain
2% Ti and low concentrations of rare earths, and (2) the
properties of several alloys containing both niobium and
titanium.

The results of tests completed to date on the six heats
that contain titanium and rare-earth additions and the
10,000-lb commercial heat that contains titanium are
summarized in Table 6.10. Previous tests at tempera-
tures of 650 and 704°C showed that the creep prop-
erties of these heats are about equivalent.® The rupture
life at 650°C and 40.0 X 10 psi varied from 1200 to
1800 hr, and the rupture life at 704°C and 35.0 X 103
psi varied from 170 to 200 hr. Final conclusions con-

6. T. K. Roche, ). C. Feltner, and B. McNabb, MSR Pro-
gram Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 78.

Table 6.10. Postirradiation creep properties of titanium-modified Hastelloy N
at 650°C after irradiation at the indicated temperature?

Irradiation Minimum Rupture Total fracture
Alloy Test temperature Stress. creep life strain Composition?
3 positi
number CO) (10° psi) rate (hr) %)
(%/hr)
474-533 R-1912 650 40.0 0.025 231.1 7.2 2.17% Ti, 0.48% Al
R-1908 650 47.0 0.050 111.8 7.8
704 40.0 58.2 6.6
R-1929 760 35.0 0.035 202.2 7.8
474-534 R-1913 650 40.0 0.021 572.2 16.2 2.09% Ti, 0.53% Al, 0.013% La
R-1909 650 47.0 0.087 66.0 6.9
704 40.0 58.8 3.5
R-1930 760 35.0 0.0085 73.9 2.8
474-535 R-1915 650 40.0 0.016 767.1 13.6 2.13% Ti, 0.55% Al, 0.04% rare earth
R-191] 650 47.0 0.099 93.0 7.5
R-1922 704 40.0 0.023 467.7 12.3
R-1926 771 35.0 0.019 6454 13.9 )
474-539 R-1914 650 40.0 0.020 601.8 10.8 1.93% Ti, 0.20% Al, 0.03% Ce
R-1910 650 47.0 0.11 73.9 9.7
R-1921 713 40.0 0.031 429.5 16.5
R-1925 774 35.0 0.0080 1220.0 11.4
474-557 R-1920 671 47.0 0.045 217.4 10.6 2.14% Ti, 0.02% Al
R-1923 713 40.0 0.044 186.2 9.5 .
R-1927 771 35.0 0.019 434.5 8.3
474-558 R-1916 650 47.0 0.092 74.8 79 2.05% Ti, 0.02% Al, 0.02% La
R-1924 716 40.0 0.021 179.3 8.6
R-1928 795 35.0 0.012 42.5 1.1
474901 R-1936 650 47.0 0.069 171.6 13.6 1.80% Tt, 0.10% Al
R-1937 704 47.0 0.15 33.2 5.2
R-1907 732 47.0 0.14 52.5 74

“All specimens annealed 1 hr at 1177°C prior to irradiation for ~1100 hr to a thermal fluence of N3 X 107° neut.rons/cm2 .

Ib'Alloy nominal base composition of Ni—12% Mo—-7% Cr—0.05% C.
cerning the postirradiation properties (Table 6.10) are
not possible, because the test matrix has not been com-
pleted. Specimens irradiated at 650°C and tested at
650°C have rupture lives that are about half those of
the unirradiated specimens, but there are no differences
in the properties of the various heats that are considered
significant in view of the limited data. The properties of
all heats are considered good after irradiation at 650°C.
After irradiation at 704°C and testing at 650°C and
40.0 X 10* psi, the rupture lives of heats 474-533 and
474-534 appear to be lower than those for the other
heats by a factor of 3. In all cases the rupture life and
the fracture strain were lower after irradiation at 704°C
than at 650°C. After irradiation at "760°C and testing
at 35.0 X 10° psi at 650°C, the rupture life varied from
43 to 1220 hr and the fracture strain from 1.1 to 11.4%

83

respectively. Thus, differences in creep behavior of
these alloys likely become progressively more important
as the irradiation temperature is increased.

The fracture strains of the various heats appear to
show significant trends with increasing irradiation tem-
perature. Heats 474-533 and 474-557 have good frac-
ture strains (6 to 10%) which do not decrease appreci-
ably with increasing irradiation temperature. The frac-
ture strains of heats 474-535 and 474-539 are in the
range of 10 to 16% and do not change appreciably with
irradiation temperature. Alloys 474-534 and 474-558
show decreasing fracture strains with increasing irradia-
tion temperature. The behavior of alloy 474-901
appears to be unique in that it shows a marked drop in
fracture strain as the irradiation temperature is in-
creased from 650° to 704°C. However, this effect may

Table 6.11. Postirradiation creep properties of several
modified Hastelloy N alloys at 650°C?

Minimum Total
Rupture
Alloy? Test Stiess' creep life fractgre
number (10° psi) rate (hr) strain
(%/hr) (%)
428 R-1948 47.0 0.043 222.1 11.9
474-533 R-1908 47.0 0.050 111.8 7.8
R-1912 40.0 0.025 231.1 7.2
430 R-1947 47.0 <0.0049 972.0¢ 4.8¢
432 R-1946 47.0 <0.00051 972.0¢ 0.5¢
431 R-1945 47.0 <0.0024 972.0¢ 2.3¢
424 R-1919 35.0 ~0 316.0¢
40.0 0.00024 140.64
47.0 0.00033 452.64
55.0 0.0066 . 949 49 7.84
424 R-1944 63.0 0.0096 3554 10.4
420 R-1918 35.0 ~0 532.24
40.0 ~0 140.59
47.0 0.00021 450.99
55.0 0.00037 695.99
63.0 0.00080 33434
70.0 0.0062 277.6¢ 4.3d
420 R-1943 63.0 <0.0017 1068.0¢ 1.8¢
418 R-1917 35.0 0.00002 65244
40.0 0.00007 140.64
47.0 0.00014 452.34
55.0 0.0040 794 .84 5.4d
418 R-1942 63.0 0.0022 559.2 7.2
434 63.0 <0.085 12.9 1.1
433 R-1949 63.0 <0.0011 660.0¢ 0.75¢

2All specimens annealed 1 hr at 1177°C prior to irradiation. Irradiation carried
out at 650°C for approximately 1100 hr to a thermal fluence of ~3 X 102°

neutrons/cm?.

bSee Table 6.4 for detailed chemical analyses.

CTest still in progress.

d3tress increased on the same specimen in the increments shown.
84

Table 6.12. Summary of information relative to metallurgical stability
of several compositions of modified Hastelloy N

Age hardens at 650°C based on indicated parameter

Hardness, N Postirradiation Composition (wt % osition (at. %
Heat? unstressed Unirradiated creep behavior ? ) Wi ) Compos ; ar 7)
. creep . . ’ Nb Ti Al Nb + Ti + Al
specimen, 5 behavior’ irradiated at
1000-hr anneal 650°C for ~1100 he?
428 No No No 2.47 0.16 3.57
474-533 No No No 2.17 0.48 392
430 Yes Yes Yes 2.5 0.34 4.02
432 Yes Yes Yes 2.35 0.69 4.63
431 Yes Yes Yes 2.5 0.74 4.94
424 No Yes Yes 1.34 1.8 0.10 3.38
420 Yes Yes Yes 1.90 1.8 0.15 3.87
418 No Yes Yes 1.92 2.0 0.05 3.91
434 Yes Yes Yes 1.86 2.2 0.32 4.70
433 Yes Yes Yes 1.89 2.2 0.33 4.75

2See Table 6.4 for detailed chemical analyses.
bFrom Table 6.4.

CFrom Table 6.6 based on considerations of data on rupture life and total strain.

dFrom Table 6.11.

be related to the strain rate. In summary, these sparse
data suggest that the fracture strains of alloys con-
taining only titanium and those containing titanium
plus cerium remain at adequate levels as the irradiation
temperature is increased, while the fracture strains of
the two alloys (474-534 and 474-558) that contain lan-
thanum do not. Additional specimens were irradiated
and tested to check this important point.

Section 6.4 of this report deals in detail with the
metallurgical stability of alloys containing Nb, Ti, and
Al in the unirradiated condition. Some of these alloys
have been irradiated, and limited test results are avail-
able (Table 6.11). The alloys were annealed for 1 hr at
1177°C prior to irradiation for about 1100 hr at 650°C.
The anneal at 1177°C should have dissolved most of the
alloying elements, and the subsequent period at 650°C
may have resulted in the formation of gamma prime.
Precipitation of this embrittling phase also strengthens
an alloy; hence the postirradiation creep tests should
show whether significant quantities of gamma prime
were formed. As discussed in Sect. 6.4, precipitation of
this phase may be strain induced, and a detailed analysis
of the creep data will be required to determine whether
the gamma prime formed in the specimens during irradi-
ation or whether it formed as the specimens were
stressed initially. _

The data from Table 6.11 and information from Sect.
6.4 are summarized in Table 6.12, which shows that the

same conclusions are reached with regard to aging of
creep specimens in the unirradiated and irradiated con-
ditions. However, hardness measurements on unstressed,
unirradiated specimens fail to be a good indication of
aging in alloys containing niobium. Alloys having a com-
bined titanium and aluminum content as high as 3.57
at. % had excellent postirradiation properties. Alloys
with higher combined concentrations are quite strong,
but no conclusion can be made about their fracture
strains. All of the alloys containing Nb, Ti, and Al are
quite strong, and can take considerable strain before
fracturing. Alloy 434 has a low fracture strain, while no
conclusion can be drawn relative to alloy 433.

The alloys containing Nb, Ti, and Al which have been
evaluated thus far are likely too highly alloyed even
though some of the fracture strains are acceptable. Less
highly alloyed materials are being irradiated.

6.7 MICROSTRUCTURAL ANALYSIS OF
TITANIUM-MODIFIED HASTELLOY N

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

The first part of this section presents the results of
microstructural studies of two titanium-modified
Hastelloy N alloys, 472-503 (designated as 503) and
471-114 (designated as 114). Previous analyses of these
same two alloys dealt with their microstructures after
aging” and after postirradiation creep tests.® In the
present investigation the microstructures of both alloys
were analyzed in an attempt to explain some unusual
postirradiation creep results in specimens that were
given a slightly higher solution annealing treatment
before irradiation. The analysis showed that many of
the test specimens were quite inhomogeneous and that
the poor creep properties could in some cases be related
to the inhomogeneities. This finding prompted a study
aimed at producing more homogeneous Hastelloy N
alloys. The problem is being approached by reducing
the carbon content of the alloy and by giving careful
attention to the fabrication parameters. The results of
initial experiments to fabricate homogeneous alloys are
presented in the second part of this section,

6.7.1 Microstructural Analysis of
Alloys 503 and 114

Postirradiation creep tests. The results of creep tests
on specimens of alloys 503 and 114, which had been
previously irradiated in the ORR at 760°C, are given in
Fig. 6.31. The creep tests were conducted at 650°C at a
stress level of 35.0 X 107 psi. This particular series of
specimens was designed to show the effect of solution
annealing temperature on the postirradiation creep rup-
ture life of the materials. The solution anneal was a 1-hr
heat treatment and was given to all specimens before
they were irradiated. As seen in Fig. 6.31, the 503 speci-
men given the standard 1 hr at 1177°C solution anneal
demonstrated good creep rupture life, while the 114
specimen given the same treatment had a comparably
short lifetime. However, with an increase of only
~30°C in annealing temperature, alloy 503 had a
greatly reduced lifetime, while alloy 114 showed
marked improvement. It was considered unlikely that
these results could be caused by changes in solution
annealing temperature alone, and other possible expla-
nations were sought. It is important to note that despite
the apparent instability in creep behavior, the properties
of the 2% Ti—modified alloys are generally good. The
problem is thus to determine why some specimens have
poor properties. A solution to this problem was sought
by carefully analyzing the microstructures of the two
alloy 503 and 114 specimens described above.

7. D. N. Braski, J. M. Leitnaker, and G. A. Potter, MSR
Program Semiannu. Progr. Rep. Aug. 31, 1974, ORNL-5011,
pp. 62—-68.

8. D. N. Braski, J. M. Leitnaker, and G. A. Potter, MSR
Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047,
pp. 83-90.

85

ORNL-DWG 75-3454

1600 T
: STRESS LEVEL = 35,000 psi
IRRADIATION TEMPERATURE = 760°C
1400
1200 R
N |
w 1000
-
- 503 ‘
« 3—’1
2 800 ~
u_ -
z / - | \ .
- .
< R
o |
x T /
« N \
400 « f \
g / \11a
2] \
200 | } \.\
/
\
P U
0 — B
1000 1100 1200 1300

SOLUTION ANNEALING TEMPERATURE (°C)

Fig. 6.31. Creep-rupture life of 2% Ti—modified Hastelloy
N alloys 503 and 114 at 650°C after irradiation in ORR for
1200 hr.

Transmission electron microscopy. Samples were pre-
pared for transmission electron microscopy (TEM) by
electropolishing small transverse sections of the tested
creep specimens in perchloric acid solutions. Figures
6.32 and 6.33 show electron micrographs representative
of 503 and 114 specimens respectively. Figure 6.324
shows an area near a grain boundary in the 503 speci-
men annealed at 1177°C. The microstructure was ob-
served to contain MC-type carbides — both in the grain
boundary and in the form of small platelets. Disloca-
tions were nearly always found to be associated with
the MC platelets. The 503 specimen annealed at 1204°C
had similar features (Fig. 6.32b). In both specimens the
MC platelets were concentrated near the grain bound-
aries. This suggests that the element or elements (prob-
ably titanium) making up the MC-type carbide in both
specimens were not uniformly distributed throughout
the matrix. Figure 6.33 shows electron micrographs of
the 114 specimens annealed at 1177°C (Fig. 6.33a) and
1204°C (Fig. 6.33h). These specimens also contained
fine MC-type carbides, but instead of forming platelets
they precipitated out on stacking faults. The stacking
fault - precipitates initiate from dislocations associated
with preexisting or primary MC carbides and grow along
(111) planes.® The primary MC carbides (the dark

9. J. M. Silcock and W. J. Tunstall, “Partial Dislocations
Associated with NbC Precipitation in Austenitic Stainless
Steel,” Phil. Mag. 10, 360—89 (1964).
spherical particles in Fig. 6.33) were not dissolved dur-
ing the solution anneal. Although not shown in these
micrographs, a nonuniform distribution of MC precipi-
tates was also observed near many grain boundaries in
both 114 specimens. However, in this case, a denuded
zone was observed rather than a higher concentration of
MC at the grain boundaries. This implies that at least

Fig.
1177°C. (b) Solution anneal of 1 hr at 1204°C.

86

one of the elements (probably titanium) needed to form
MC-type carbides remained in the volume surrounding
the primary carbides. Since most of these primary car-
bides were located within grain interiors, the areas near
grain boundaries were free of carbides. Transmission
electron microscopy has revealed several interesting dif-
ferences in carbide morphology between specimens of

6.32. Specimens of alloy 503 after 1200 hr in ORR at 760°C and creep testing at 650°C. (a) Solution anneal of 1 hr at

YE-11275
=R

Fig. 6.33. Specimens of alloy 114 after 1200 hr in ORR at 760°C and creep testing at 650°C. (a) Solution anneal of 1 hr at
1177°C. (b) Solution anneal of 1 hr at 1204°C.

the two titanium-modified Hastelloy N alloys. The
investigation also showed two different types of non-
uniform MC-type carbide distributions near grain
boundaries in both alloys. Unfortunately, it was not
possible to identify differences in microstructure by
TEM that could be directly related to the wide variation
in creep properties with changes in solution annealing
temperature.

Metallography. The same two pairs of 503 and 114
specimens studied by TEM were also examined in the
Radiation Metallography Laboratory at ORNL. Polished
longitudinal sections of 503 and 114 specimens were
examined near the fracture surfaces. Micrographs of
these areas are shown for 503 and 114 in Figs. 6.34 and
6.35 respectively. The 503 specimen annealed at
1177°C (Fig. 6.34a), which had a long creep rupture
life, had many intergranular cracks along its outer sur-
face. Several cracks or evidence of grain boundary sepa-
ration was also observed within the specimen interior.
Carbide stringers were found to be distributed uni-
formly across the entire section of the sample. The car-
bide stringers are composed of numerous MC-type par-
ticles which lie in lines parallel to the primary working
direction in fabrication. Some carbide stringers are more

clearly illustrated in Fig. 6.36 by the higher magnifica-
tion micrographs taken of the 503 specimen. The 503
specimen annealed at 1204°C had only a few cracks
(Fig. 6.34b), but one of them was apparently able to

87

propagate quite rapidly, leading to early fracture. Note
also the striking differences in microstructure of the
surface region, which extends to a depth of ~0.012 in.
This region was free of carbide stringers and had a grain
size of at least twice that seen in more central areas.
Quite similar microstructural features were also ob-
served for the 114 specimens having poor creep prop-
erties. The 114 specimen annealed at 1177°C, which
had a short creep rupture life, also had a
~0.0124n.-thick carbide-free layer (Fig. 6.352). How-
ever, a large grain size was not observed in the layer as
seen in the 503 specimen (Fig. 6.34b). It was also found
that surface cracks never penetrated beyond the
carbide-free layer. The 114 specimens annealed at
1204°C (Fig. 6.35h) demonstrated better creep proper-
ties and had a microstructure much like that shown for
the better SO3 samples (Fig. 6.35¢). The carbide-free
layer found in the 503 and 114 specimens with poor
creep properties may have been caused by decarburiza-

tion during the solution anneal at 1177°C in argon.
Decarburization layers of ~1 mm (0.039 in.) were
observed in Inconel 617 after creep tests at 1000°C for
127 hr in helium containing 500 ppm oxygen.'® The
lack of carbides in the surface layer may have, in turn,

10. Y. Hosoi and S. Abe, “The Effect of Helium Environ-
ment on the Creep Ripture Properties of Inconel 617 at
1000°C." Met. Trans. 6A, 117178 (June 1975).

| R-70167

0.25mm

Fig. 6.34. Microstructure of spec

nens of alloy 503 near fracture surfac

after 1200 hr in ORR at 760°C and creep testing at

650°C. (a). Solution anneal of 1 hr at 1177°C. (h) Solution anneal of 1 hr at 1204”C

88

oot

on_|

025mm

Fig. 6.35. Microstructure of specimens of alloy near fracture surfaces after 1200 hr in ORR at 760°C and creep testing at 650°C.
(@) Solution anneal of 1 hr at 1177°C. (b) Solution anneal of 1 hr at 1204°C.

encouraged grain growth in the 503 specimen (Fig.
6.35b) during the solution anneal. It is unclear as to
why only certain specimens have the carbide-free layers
when all specimens were supposedly fabricated in the
same way.

The results of the metallographic and TEM examina-
tions cannot be used to fully explain the mechanisms
which led to the early creep failure of two of the speci-
mens studied. However, we have shown that a number
of microstructural inhomogenieties exist in the 2% Ti
modified Hastelloy N alloys, including carbide-free sur-
face layers, large-grain-size surface layers, generalized
carbide stringers, and nonuniform distributions of
MC-type carbide near grain boundaries. Some of these
inhomogenieties appeared to affect the results of
mechanical tests and may also influence other impor-
tant tests, such as those relating to tellurium attack.
Consequently, a study was initiated to produce Hastel-
loy N alloys with more homogenous microstructures.

6.7.2 Homogeneous Hastelloy N Alloys

The problem of producing Hastelloy N alloys with
homogeneous microstructures is being approached in
two ways. The first is to reduce the carbon content in
the alloy to ensure that all of the MC-type carbides are
dissolved during the solution annealing treatment. If all

the carbides could be held in solution during fabrica-
tion, the formation of carbide stringers might be elimi-
nated. The second approach is a detailed evaluation of
the fabrication process. This latter effort is primarily
aimed at identifying the steps at which the different
inhomogenieties are introduced and finding suitable

alternate processing methods to remove the inhomo-
genieties. A definite concern throughout the entire
study is that any successful fabrication changes also be
compatible with commercial practices.

Carbon content. The first series of experiments was
designed to eliminate carbide stringers by reducing the
carbon content of the alloy. Thermodynamic calcula-
tions using data from previous experiments indicated
that all of the carbides should dissolve at 1177°C in
alloys with carbon contents of less than 0.045 wt %.
Therefore, two alloys, 451 and 453, both with a nomi-
nal Hastelloy N composition (13 wt % Mo, 7 wt % Cr,
bal Ni) and 1.94 wt % Ti, were cast into l-in-diam
ingots having carbon contents of 0.017 and 0.035 wt %
respectively. The fabrication schedule called for the cast
ingots to be hot swaged at 1177°C from a l-in. to a
0.430-in. diameter and then to be annealed at 1177°C
for 1 hr. The rods were further reduced to a 0.337-in.
diameter by cold swaging, annealed at 1177°C for 1 hr,
and cold swaged to a final diameter of 0.250 in. One-
inch-long samples were then cut from each alloy rod,

89

R-69236

Fig. 6.36. Carbide stringers in specimen of alloy 503 after irradiation at 760°C and creep testing at 650°C.

encapsulated in quartz under an argon atmosphere, and
aged at 760°C for 116.5 hr to precipitate the carbides.
After aging, the carbides in alloy 453 (0.035% C) were
extracted electrochemically in a methanol—10% HCl
solution. Consecutive extractions produced the profile
shown in Fig. 6.37 of wt % carbide precipitate through
the thickness of the sample. The profile for alloy 453 is
considerably more uniform than those observed for
alloys 503 and 114 specimens aged at 750°C for 1000
hr. The difference may not be entirely due to a reduc-
tion in carbon content, because the 503 and 114 speci-
mens were swaged from bars cut from %-in.-thick plate,
not from drop-cast ingots. (Carbides are fairly uni-
formly distributed in the grain boundaries of the 2-Ib
laboratory ingots, while they appear as stringers in the
J-in. plate.) Metallographic examination of the aged
451 and 453 samples (Fig. 6.38) showed that the reduc-

tion in carbon content did not eliminate the carbide
stringers. However, the stringers were finer and more
evenly distributed than those observed previously (Fig.
6.34b). Carbide-free surface layers were observed in
both specimens; a typical surface layer in a heavily
etched 453 sample is shown in Fig. 6.39. The depth of
the carbide-free surface layer was ~0.003 in.
Fabrication. One of the most critical steps in fabri-
cating Hastelloy N alloys with respect to its effect on
microstructure is the solution anneal. Electrochemical
extractions on an as-swaged alloy 453 (0.035% C) sam-
ple showed that a moderate number of carbide particles
("0.2%) was present in the microstructure after proc-
essing. It is suspected that the sample was not ade-
quately annealed at 1177°C prior to the final cold swag-
ing operation. That is, the annealing time was too short
or the annealing temperature was actually less than

ORNL-DWG 75-3455R

0 T

ESTIMATED
ERROR w
08 o
w ! \ ‘\Qs. H
: "CI \\ o
L os ,‘ C &
S / w
8 2N / ~. |
o e — :
‘; 453w T~d |
02 E
3

0
O 0025 0050 0075 0100 0425

DISTANCE FROM CENTER OF SAMPLE (in.)

Fig. 6.37. Distribution of carbides through the specimen
thickness of titanium-modified Hastelloy N alloys. Alloys 503
and 114 solution annealed at 1177°C and aged at 650°C for
1000 hr. Alloy 453 swaged and aged at 760°C for 116.5 hr.

90

1177°C. Therefore, the response of titanium-modified
Hastelloy to solution annealing at 1177°C was studied
as a function of time at temperature.

Samples of alloy 451 (0.017% C) were annealed in
argon at 1177°C for 15 min to 8 hr. The samples were
cleaned electrochemically for 6 hr to remove any sur-
face effects, and the carbides were electrochemically
extracted, separated, and weighed. The results of this
experiment are plotted in Fig. 6.40. Only extremely
small amounts of precipitates were present in samples
annealed for 2 hr or more. At times less than 2 hr, there
was some scatter in the data, but, in general, slightly
more precipitate was extracted. These results indicate
that 30 to 60 min are needed in addition to the stan-
dard 1-hr solution anneal at 1177°C to dissolve the car-
bides completely. Micrographs of sections from each of
the samples of alloy 451 from the first annealing series
are shown in Fig. 6.41. Little grain growth was observed
between the 15-min and 1-hr anneals, while slight grain
growth was evident after 2 hr at 1177°C. As expected,
rather extensive growth occurred at the longer annealing
times of 4 and 8 hr.

Y-133206

®)

0.010in.
0.25mm

Fig. 6.38. Microstructure of titanium-modified Hastelloy N alloys 451 (0.017% C) and 453 (0.035% C) after cold swaging and

aging at 760°C for 116.5 hr. (a) Alloy 451. (b) Alloy 453.

Although most of the effort in this study has been
directed toward elimination of carbide stringers in the
alloys, experiments are also under way to determine the
cause of the carbide-free layers. In one experiment, sam-

Y-133201

Fig. 6.39. Carbide-free surface layer in sample of alloy 453
(0.035% C).

91

ples will be examined metallographically before and
after a solution anneal at 1177°C. This is a reasonable
starting point, because carbide-free layers are found
along the reduced section of machined tensile speci-
mens; that is, any effects of hot or cold swaging would
have been removed by machining in that area. Finally,
there is the consideration of fabrication practice, which
may, in fact, be the key to producing a homogeneous
alloy. A number of relatively minor changes in the way
the material is handled may have dramatic effects on
the resultant microstructure. Duplicate ingots of alloys
451 and 453 are available and will be fabricated to
0.250-in.-diam rod, with special attention paid to the
fabrication parameters. Metallographic samples will be
cut from the work piece throughout the processing so
that the associated microstructural features may be
observed.

6.8 SALT CORROSION STUDIES

J.R.Keiser J.R.DiStefano  E.J. Lawrence

The corrosion of both nickel- and iron-base alloys by
molten fluoride salts has been the subject of extensive
research for many years. Results show that impurities
such as FeF,, NiF,, and HF in the salt react with con-
stituents of the alloys, but corrosion from these sources
is limited by the supply of reactants. The strongest
oxidant of the normal constituents of fuel salt is UF4,
and of the major constituents of most iron- and nickel-
base alloys, chromium forms the most stable fluoride.
Consequently, the major corrosion reaction between

ORNL-DWG 75-142243

0.2
.t R
o 4st ANNEALING RUN
o 5 2nd ANNEALING RUN
= o0 3rd ANNEALING RUN
= 0 4th ANNEALING RUN
& o
o
* ) ©
3 A\
[} T
ole 2 1 T =4
o] ! 2 3 4 5 6 7 8

TIME AT 1477 °C (hr)

Fig. 6.40. Amount of carbides extracted from alloy 451 as a function of time at solution annealing temperature of 1177°C.

92

0.0101n.
it

@ (e)

Fig. 6.41. Micrographs of samples of alloy 451 after solution annealing at 1177°

1 hr;(d) 2 hr; (e) 4 hriand (f) 8 hr.

nickel- or iron-base alloys and molten-salt reactor fuel
salt has been found to be

2UF4(d) +Cr(c) 2 2UF5 (d) + CrF, (d) .

Because the equilibrium constant for this reaction has a
small temperature dependence, temperature gradient
mass transfer can occur and results in continuous re-
moval of chromium from the hotter sections of a
system and a continuous deposition of chromium in the
cooler sections.

The experiments described in this section are being
conducted to determine the corrosion rate of various

(f)

C for different times. () 15 min; (b) 30 min; (c)

salt-alloy systems under controlled test conditions. The
variables include composition of the alloy, oxidation
potential of the salt, temperature, and exposure time.
All loops incorporate electrochemical probes to measure
the concentrations of uranium and transition-metal flu-
orides. The systems used to conduct these experiments
include one forced circulation loop operated by person-
nel in the Reactor Division and three thermal convec-
tion loops. Five additional thermal convection loops
have been constructed and are being prepared for opera-
tion. The status of these eight thermal convection loops
is summarized in Table 6.13.

93

Table 6.13. Status of thermal convection loop tests on August 31, 1975

' Loop Insert

Loop :
numbet’ material specimens Salt type Status Purpose
21A Hastelloy N Hastelloy N MSBR fuel salt Operating 1. Analytical method development?
6969 hr 2. Baseline corrosion data
3. Tellurium mass transfer studies
23 Inconel 601 Inconel 601 ‘ MSBR fuel salt Qperating 1. Baseline corrosion data for
6035 hr high-chromium atloy under MSBR
conditions
2. UF, -graphite reaction
31 Type 316 Type 316 Li, BeF, Operating 1. Baseline corrosion data
SS SS 248 hr 2. Effect of oxidation potential
. of salt on corrosion rate
18B Hastelloy N Modified MSBR fuel salt In preparation 1. Screening test loop for
Hastelloy N modified alloys
24 Hastelloy N Nb-Ti modified MSBR fuel salt In preparation 1. Baseline corrosion data for
Hastelloy N modified alloys
2. Tellurium mass transfer
25 Hastelloy N Nb-Ti modified MSBR fuel salt In preparation 1. Baseline corrosion data for
Hastelloy N modified alloys
2. Tellurium mass transfer
27 Type 316 Type 316 MSBR. fuel salt. In preparation 1. Oxidation potential studies
S8 SS 2.. Tellurium mass transfer
and other iron- 3. Effect of tellurium on
base alloys mechanical properties of
specimens
29 Hastelloy N Standard and MSBR coolant salt In preparation 1. Analytical method development
modified 2. Baseline corrosion data
Hastelloy N 3. Tritium transport data

2 All eight thermal convection loops are equipped with electrochemical probes.

6.8.1 Fuel Salt Thermal Convection Loops

Two thermal convection loops, NCL 21A and NCL
23, have been operating with MSBR fuel salt (LiF-
BeF,-ThF,;-UF,, 72-16-11.7-0.3 mole %) to obtain
baseline corrosion data. NCL 21A is a Hastelloy N loop
with specimens of the same material. As with all ther-
mal convection loops, eight specimens are inserted in
the hot and the cold legs. The 16 specimens are re-
moved periodically for visual examination and weighing.
The results of the weight change measurements are
shown in Fig. 6.42. The corrosion rate of the hottest
specimen in this loop is somewhat higher than has been
observed in other Hastelloy N systems (see Sect. 6.8.2
discussion of FCL-2b). The higher corrosion rate of
loop 21A relates to the relatively high oxidation poten-
tial of the salt in this loop (U**/U" about 10*). How-
ever, assuming uniform removal of material, the cor-
rosion rate of the hottest specimen was 0.24 mil/year,
which is within acceptable limits. This loop will con-
tinue to be used to obtain corrosion data for Hastelloy
N in salt with a relatively high oxidation potential.

Loop NCL 23 is constructed of Inconel 601 and has
specimens of the same material. A loop was built of
Inconel 601 because of this alloy’s resistance to grain
boundary penetration by tellurium. Since the alloy con-
tains 23% Cr, there was concern about its ability to
resist attack by molten fluoride salt. The corrosion rate
of Inconel 601 in fuel salt was determined from weight
measurements of the 16 specimens of loop 23, and the
results are shown in Fig. 6.43. All specimens lost
weight, and the loss shown by the hottest specimen was
very large. The material lost by the hottest specimens
did not result in uniform removal of the surface, but
resulted in the formation of the porous surface struc-
ture shown in Fig. 6.44. As shown in Fig. 6.45, electron
microprobe examination of this specimen showed high
thorium concentration in the pores. The only known
source of thorium was the salt which contained ThF,,
so it is very likely that the salt penetrated the pores.
Continuous line scans with the microprobe indicated a
depletion of chromium near the surface. Figure 6.46
shows the results of analysis for Ni, Cr, and Th. This
figure clearly shows the chromium concentration gra-
94

"ORNL-DWG 75-12246

{
e
o - 566 °C
N o .
o O » -
E 635°C
] )
z - A
<1
I .
%) \
E -2 A \\k
g_l \
704°C
_3 \
0 1000 2000 3000 4000 5000

SPECIMEN EXPOSURE TIME (hr)

Fig. 6.42. Weight changes of Hastelloy N specimens from
loop NCL-2IA exposed to MSBR fuel salt at the indicated -
temperature.

ORNL-DWG 75-12245

0

-4 P, e ——
w \ 635°C
g -2 ™
o \
E N
W 3 N
prd
3 ~.
o4 N
- \
I . .
5 g <
[¥1]
S \

-6

704°C
-7
o - - 500 1000 1500 2000

SPECIMEN EXPOSURE TIME (hr)

Fig.- 6.43. Weight changes of Inconel 601 -specimens from
loop NCL-23 exposed to MSBR fuel salt at the indicated tem-
perature.

dient and ‘provides further evidence of the presence of
thorium in the pores. Deposits such as those shown in
Fig. 6.47 formed on the specimens in the cold leg, and
the deposits’ were identified by microprobe analysis as
chromium. - This compatibility test of Inconel 601 in
MSBR fuel salt shows a relatively high corrosion rate,
and it is doubtful that this alloy would be suitable for
use in an MSBR under the conditions of this test.

The lower limit for the U**/U3" ratio in an MSBR will
likely be determined by the conditions under which the
reaction

4UF, +2C23UF, + UG,

proceeds to the right. Because the salt in loop NCL 23 is
strongly reducing with a U**/U3" ratio of less than 6, it
was decided to try to reproduce the results of Toth and

Gilpatrick,!! which predicted that at temperatures
below 550°C and U*/U*" ratios below 6 the UC,
would be stable. However, graphite specimens exposed
to the salt for 500 hr did not show any evidence of
UC,. The specimens used were made of pyrolytic graph-
ite, and it is likely that the high density of the material
limited contact of the salt and graphite. The experiment
is being repeated with a less dense graphite.

6.8.2 Fuel Salt Forced Circulation Loop

Hastelloy N forced circulation loop FCL-2b has been
operated during this reporting period to gather baseline
corrosion data under conditions where the Uy
ratio was relatively low (see Sect. 2.3). Eighteen Hastel-
loy N specimens were exposed to MSBR fuel salt with a
U# /U ratio of about 100. The specimens were re-
moved at predetermined intervals for visual examination
and weighing, and the weight changes are shown in Fig.
6.48. Six specimens were held at each of three tempera-
tures: 704, 635, and 566°C. Of the six specimens at
each temperature, three were exposed to salt having a
velocity of 0.49 m/sec and three to salt having a ve-
locity of 0.24 m/sec. No effect of salt velocity on the

- corrosion rate was found, so each data point represents

the average weight loss of the six specimens. The weight
loss of the specimens at the highest temperature corre-
sponds to a uniform corrosion rate of 0.11 mil/year.
Uniform corrosion at this rate is acceptable and well
within the limits which can be tolerated in an MSBR.

Following termination of the ~3200-hr corrosion
experiment, FCL-2b was to be used to make heat trans-
fer measurements. This operation has been delayed,
because a salt leak developed and a section of the %-in.-
diam Hastelloy N tubing had to be replaced (see Sect.
2.3). Examination of the tubing in the vicinity of the
leak is under way. : ‘

Further corrosion measurements will be made in ‘this
loop with the U* /U ratio at about 103. Additions of
NiF, to the salt will be made to raise the U**/U3* ratio
to the desired level. »

6.8.3 Coolant Salt Thermal Convection Loops

Thermal convection loop NCL 31 is constructed of
type 316 stainless steel and contains LiF-BeF, (66-34
mole %) coolant salt. The 16 removable corrosion speci-
mens are also made of type 316 stainless steel. The
maximum temperature of the loop is 639°C, and the
minimum temperature is 482°C. The initial objective of

11. L. M. Toth and L. O. Gilpatrick, The Equilibrium of
Dilute UF, Solutions Contained in Graphite, ORNL-TM-4056
(December 1972).
95

Y —129632

60 MICRONS
500X
INCHES

\
<
O

o O )¢ a

Fig. 6.44. Microstructure of Inconel 601 exposed to MSBR fuel salt at 704°C for 720 hr. As polished

20
1

Y-131294

’
Backscattered Electrons

ThMa X-Rays

Fig. 6.45. Electron beam scanning images of Inconel 601 exposed to MSBR fuel salt for 720 hr at 704°C.

Y-131219

96

Ni-~3000 COUNTS FULL SCALE
CR--3000 COUNTS FULL SCALE.
TH--1000 COUNTS FULL SCALE

i v
POy PRI, Y
L

{10 MICRONS © ;|

T &

Fig. 6.46. Microprobe continuous line scan across corroded area in Inconel 601 exposed to MSBR fuel salt for 720 hr at 704°C.
97

Y-131514

2 [
gxu
gOT
5go
=S
o
@0’
o
¥

g
Q10

Fig. 6.47. Microstructure of Inconel 601 exposed to MSBR fuel salt at 566°C for 720 hr. As polished.

ORNL-DWG 75-12244

"¢ os

<

3 566°C
oo

2 5 o
z [635°c
i

:7051 s
3 704°C
$ -0 ==

0 500 4000 4500 2000 2500

SPECIMEN EXPOSURE TIME (hr)

3000 3500

Fig. 6.48 Weight changes of Hastelloy N from loop FCL-2b
exposed to MSBR fuel salt at the indicated temperature.

this loop is to provide baseline corrosion data on a com-
mercial iron-base alloy. The loop has been in operation
for 248 hr.

6.9 CORROSION OF HASTELLOY N AND
OTHER ALLOYS IN STEAM
B.McNabb  H. E. McCoy

The corrosion resistance of several heats of standard
and modified Hastelloy N and other iron-, nickel-, and

cobalt-base alloys is being evaluated in the unstressed
condition in the TVA Bull Run Steam Plant. Two heats
of standard Hastelloy N tubing (N15095 and N15101)
are being evaluated in the stressed condition from 28.0
X 10® to 77.0 X 10° psi.

The method whereby the specimens are stressed is
shown in Fig. 6.49. The wall thickness of the gage sec-
tion of the specimens was varied from 0.010 in. (77.0 X
10° psi) to 0.030 in. (28.0 X 10 psi) to produce the
desired stress range. The %-in.-OD capillary tube con-
nects the annulus between the two tubes to the con-
denser. When the inner tube ruptures, steam passes
through the capillary, and a rise in temperature of a
thermocouple attached to the capillary indicates rup-
ture. Time to rupture can be taken directly from the
multipoint recorder and plotted vs stress for design pur-
poses. Data of this type for periods as long as 11,000 hr
were reported previously.'?

A photograph of the specimen holder (Fig. 6.50)
shows the ten instrumented stressed specimens, the four
uninstrumented stressed specimens in the filter basket,
and the unstressed sheet specimens bolted to the speci-

12. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 94-101

98

’ — —
STEAM SUPPLY
3500 psig, 1000 °F,
16-17 1bs/min

ORNL-DWG 68-3995R2 .

PRESSURE TRANSDUCER™

SAMPLE HOLDER

SAMPLE VESSEL SUPPORT AND
/1 THERMOCOUPLE
9 weLL
=S
FILTER/

ale \-MULTIPOINT

RECORDER
SAMPLE HOLDER
| FLOW RESTRICTER REDUCED WALL

THICKNESS

]

| EVENT
CONDENSER }:

WATER QUT

RETURN TO CONDENSATE STORAGE
<

—

CAPILLARY
TUBE

TUBE BURST SPECIMEN { TYP. 10)

B
bz

WATER IN

Fig. 6.49. Schematic of double-walled tube-burst specimen.

men holder. The filter basket bolts to the small flanges
on each side of the sheet specimens (shown exposed), so
that the specimens are covered and the flow of steam is
directed over the specimens rather than around them.
The steam enters the specimen chamber near the middle
of the stressed specimens in front of the unstressed
specimen holder and is directed lengthwise over the two
stacks of 2-in.-long X %-in.-wide X 0.035-in.-thick sheet
specimens. The steam passing over the specimens flows
through the Neva-Clog filter to prevent scale from enter-
ing the flow restricter orifice or the remainder of the
steam system. The steam is condensed and returned to
the condensate storage vessel. No specimen has lost any
scale so far, but some of the Croloy-type alloys are
beginning to develop blisters, a prelude to scaling. The
oxide on all Hastelloy N specimens is thin and ad-
herent, with no evidence of scaling. Some of the un-
stressed Hastelloy N specimens have been exposed to
steam for 19,000 hr at 538°C and 3500 psig. Several
alloys were included in this study, and, as reported pre-
viously,'* they displayed a wide range of oxidation
rates. Several obeyed the parabolic rate law, Aw -
Kt where Aw is the weight change in mg/ecm?, ¢ is
the time in hours, and K is a constant. Figure 6.51 is a

log-log plot of weight change in mg/cm? as a function
of time in hours. Note the sudden increase in the rate of
weight change, with each alloy gaining approximately
0.5 mg/cm® over the last 4000 hr. This probably indi-
cates deposition of some substance on the specimens at
a rate that was equal for all specimens. We noted pre-
viously that fine particles of iron oxide that was en-
trained in the steam had deposited on the specimens,
but this deposition occurred at a much lower and con-
stant rate.

The increased rate of weight gain for the specimens
was discussed with Bull .Run engineers. The Bull Run
facility has had several instances of condenser tube leaks
in the last year of operation, whereas in previous years,
few if any condenser leaks occurred. The cooling water
in the condensers is at higher pressure than the condens-
ing steam to prevent back pressure on the turbines, and

. when a leak occurs, untreated cooling water is intro-

duced into the steam system hot well. Continuous
monitoring of silicon in the four hot wells (condensed

13. H. E. McCoy and B. McNabb, Corrosion of Several fron-
and Nickel-Base Alloys in Supercritical Steam at 1000°F,
ORNL-TM-4552 (August 1974).
99

Photo 224475

Fig. 6.50. Photograph of the steam corrosion chamber after 19,000 hr of exposure. Features to note are the stressed but

uninstrumented specimens in the filter (foregound), the two groups of unstressed specimens, and the ten instrumented stressed

specimens. The stressed specimens have an outside diameter of 1 in. and a length of 3

steam wells) indicates a condenser leak when the silicon
level increases, and the leaking condenser can be iso-
lated and repaired. The condensed steam (and any cool-
ing water introduced by condenser leakage) passes
through demineralizers and is monitored again, with sili-
con and other impurities being held below acceptable
limits before the condensate is returned to the steam
system. Even though care is taken to prevent excessive
amounts of impurities in the steam system, the facility
is evidently operating with a different level of impurities
than had been experienced before condenser problems
developed. Some evidence of sodium silicate, as a black-
ish gray deposit, has been observed on some safety-valve
seats, and this is possibly the material that has deposited
on the specimens. The oxide on most of the specimens
is black or gray, and no changes in its appearance were
noticed during routine examination and weighing of the

specimens. When the specimen holder is removed for
the next scheduled examination, an effort will be made
to determine the composition and nature of the deposit.
A scheduled replacement of condenser tubes is planned
by Bull Run engineers in the near future to eliminate
the problem of condenser leaks.

Some of the alloys represented in Fig. 6.51 lost
weight initially before gaining at an accelearated rate
during the last 4000 hr. These alloys were Hastelloy X,
Haynes alloy 188, and Inconel 718, and they contain
approximately 20% Cr. Other investigators have re-
ported weight losses due to loss of chromium in steam
at high temperatures. It is probable that these alloys
would have continued to lose weight if the steam con-
ditions had not changed; new specimens of some of the
alloys will be inserted in the test facility when steam
conditions improve.

ORNL-DWG 75-15519

10
N
\\\
AN
R CROLOY 1-9% Cr
2 3
- s
NE 1 )LJ j
\ ol ~— O o
o N I
£ —— /1
05 1 + HASTELLOY N
E / porser = sl
T q% ] 3
[ . 4
E 0.2 / IAW K| o5
5 IAW Kt
o AW =Kt ul
; 0.1 l : ol
7% 1
> 1t o INcoLoY 800
003 X1/ 1 J 1T & HASTELLOY X
\ / Py o H-188
® INCONEL 600
0.02 |— \ a4 347 STAINLESS STEEL
N / < INCONEL 718
001 5 S II4 L L 11 .
10° 2 5 10 2 5 10
TIME (hr)

Fig. 6.51. Corrosion of several alloys in steam at 1000°F
(538°C) and 3500 psi.

6.10 OBSERVATIONS OF REACTIONS IN
METAL-TELLURIUM-SALT SYSTEMS

J. Brynestad

- Several criteria must be met for a good screening test
system for the tellurium corrosion of Hastelloy N:

1. The tellurium activity must be appropriate, repro-

ducible, and known.

. The tellurium must be delivered uniformly over the
sample surfaces and at a rate sufficient to prevent
excessive testing times. ‘

. Preferably, the system should operate under invari-
ant conditions during the test run.

. The system must be relatively cheap, simple, and
easy to operate.

In the MSBR the productlon of tellurium per time
unit will quickly reach a constant value, and in due time
a steady state will be reached where tellurium is re-
moved from the melt at a rate that equals the rate at
which tellurium is produced.

100

In addition to reacting with the material of which the
primary circuit is constructed, tellurium could be re-
moved by several means which include the following:

1. The processing system: Since the MSBR is to be

equipped with a processing system to remove fission

products, tellurium might be effectively removed

from the salt by appropriate measures.

. The gas phase: If the gas phase is contacted with a
getter such as chromium wool, the tellurium activity
in the melt might be kept close to that defined by

the equilibrium
YaCryTeg(s) @ Tey(g) + 7 Cr(s)

This activity is sufficiently low that Hastelloy N
would not be attacked.

. A “‘getter” immersed in the salt melt: Obvious dis-
advantages of this arrangement would be the prob-
lems of mass transport in temperature gradients and
the lack of a candidate material.

Until the steady-state condition in an MSBR is more
clearly defined, it is impossible to state the likely tellu-
rium activity. It is only known that in the MSRE, stan-
dard Hastelloy N was embrittled (probably by tellu-
rium). In the MSRE the steady-state tellurium
activity — if ever reached — probably was defined by gas
phase removal and was likely rather high.

Until the steady-state situation in the MSBR is de-
fined, it must be assumed that one must deal with the
MSRE condition, under which standard Hastelloy N is
embrittled. In order to define this condition, we have
tested several systems with defined tellurium activities
with regard to their behavior toward Hastelloy N:

1. equilibrium mixture of Cr; Te; (s) + Cr; Tes(s),

2. equilibrium mixture of Ni3Te,(B,, 41 at. % Te) +
NiTeq 775(y, v 43.7 at. % Te),

equilibrium mixture of Cry Teq(s) + Crs Teg(s),
equilibrium mixture of Ni3 Te, (s) + Ni(s).

3.
4.

The systems are arranged in sequence of decreasing Te,
activity, as determined by isopiestic experiments.
Typical corrosion experiments were conducted at
700°C for 250 to 1000 hr. The arrangements were by
isothermal gas phase transport of Te, in previously
evacuated, sealed-off quartz ampuls; by embedding the
specimens in the mixtures; and, in the Cr, Te;-Cr; Te,
and Cr3Te,s-CrsTeq cases, by transport in molten salt.
The most pertinent results are as follows:

1. Hastelloy N samples exposed to NizTe,(s) + Ni(s)
(system 4) did not show intergranular cracking. This
is promising, because if one can establish a steady-
state condition in which the tellurium activity is
lower than that defined by this system, standard
Hastelloy N will not be embrittled.

. Systems 1 and 2 have tellurium activities that are too
high. These systems corrode Hastelloy N severely
under all the experimental arrangements used.

. System 3 [Cr3Tes(s) + Crs Tes(s)] shows promise as
a tellurium-delivery method in molten salt, since it is
sufficiently corrosive to cause intergranular cracking
of Hastelloy N but does not form reaction layers.

o

It is of value to note that the system Cr,Teg(s) +
Cr(s) has a tellurium activity that is much lower than
the system NizTe,(s) + Ni(s). This system also is prom-
ising, since high-surface chromium might be used as a
tellurium getter in the gas phase. Experiments are under
way to measure the tellurium activities of the above
systems.

6.11 OPERATION OF
METAL-TELLURIUM-SALT SYSTEMS

J. R. Keiser
J. R. DiStefano

1. Brynestad
E.J. Lawrence

The discovery of shallow intergranular cracking of
Hastelloy N parts of the Molten-Salt Reactor Experi-
ment which were exposed to fuel salt led to a research
effort which identified the fission product tellurium as
the probable cause of the cracking. Experiments showed
that Hastelloy N specimens which had been electro-
plated with tellurium or exposed to tellurium vapor
exhibited shallow intergranular cracking like that of
specimens exposed in the MSRE. Subsequently, a pro-
gram was initiated to find an alloying modification for
Hastelloy N which would enhance its resistance to tellu-
rium. The resistance of these modified alloys to crack-
ing is measured by exposing specimens to tellurium
vapor, deforming them, and then evaluating their sur-
faces by metallographic and Auger methods. However.
the chemical activity of tellurium in these experiments
is significantly higher than it was in the MSRE. In order
to simultaneously expose specimens to the combined
corrosive action of molten fluoride salt and tellurium at
a more realistic chemical activity, a method is being
sought for adding tellurium to molten salt in a manner
that would simulate the appearance of tellurium as a
fission product. Experiments have been started that will
permit evaluation of several methods to determine
whether they will produce the desired conditions.

6.11.1 Tellurium Experimental Pot 1'%

Tellurium experimental pot 1 was built to evaluate
the use of lithium telluride as a means for adding tellu-

101

rium to salt. This pot (Fig. 6.52) allows tellurium to be
added periodically in the form of salt pellets containing
a measured amount of lithium telluride. Three viewing
ports permit observation of the pellets after their addi-
tion to the salt. Electrochemical probes are inserted
through Teflon seals and are used to detect and measure
the concentration of certain species in the salt. The salt

14. The lithium telluride for this experiment was prepared
by Valentine and Heatherly. The electrochemical measurements
were made by Meyer and Manning

Y—129592

Access for
Additions,
Sampling,
and
Electro-
chemical
Probes

Viewing
Port

Reaction
Vessel

Fill or
Dump
Vessel

Fig. 6.52. Lithium telluride experimental pot No. 1.

used is LiF-BeF,-ThF, (72-16-12 mole %), and its tem-
perature is maintained at 650°C.

In the first experiment, tellurium was added as the
lithium telluride Li, Te, which was prepared by the
Chemistry Division (Sect. 3.1). Initially, two pellets
containing a total of 0.070 g of Li, Te were added to
600 m] of salt. Electrochemical examination of the salt
by members of the Analytical Chemistry Division gave
no indication of the presence of tellurium (Sect. 5.3).
Subsequently, three more Li, Te pellets were added, and
a sample of the salt was taken for chemical analysis.
Three additions of CrF, totalling 1.82 g were then
made, followed by the addition of three more Li, Te
pellets. A final addition consisting of 0.2 g of BeO was
made.

Results can be summarized as follows:

. The Li, Te pellets did not melt and disappear imme-
diately; some evidence of the pellets remained on the
salt surface for the duration of the experiment.

. No electrochemical evidence of a soluble tellurium
species was detected.

. Following the Li, Te additions, visibility through the
view ports was limited by a bluish-grey deposit that
was subsequently identified as being predominately
tellurium.

. Chemical analysis of the salt sample taken after the
addition of five Li, Te pellets showed that the tellu-
rium content was less than 5 ppm.

The conclusion is that use of the compound Li,Te
does not provide an adequate means for adding tellu-
rium to MSBR fuel salt. However, it is thought that
LiTe; is much more soluble in MSBR fuel salt than is
Li, Te. Considerable difficulty was encountered by the
Chemistry Division in synthesizing LiTe; (Sect. 3.1),
but material is now available and the experiment will be
repeated with LiTe; as the source of tellurium.

6.11.2 Chromium Telluride Solubility Experiment

The addition of a soluble chromium telluride, either
Cr,Te; or CryTes, represents another method for
adding tellurium to molten MSBR fuel salt. If an excess
of chromium telluride is maintained, the activity of
tellurium in solution in the salt will be constant pro-
vided the temperature is not changed and no changes
are made in the salt composition. To determine whether
there is a temperature at which either of these chro-
mium tellurides can provide a reasonable amount of
tellurium in solution, an attempt was made to deter-
‘mine the solubility of the chromium tellurides as a func-
tion of temperature. '

102

A Hastelloy N pot was filled with the salt LiFF-BeF, -
ThF, (72-16-12 mole %) and the temperature con-
trolled at 700°C. After a sample of the salt had been
taken, Cr; Te, was added and a small Hastelloy N sheet
specimen was inserted into the salt. After 170 hr the
specimen was removed and after 250 hr a salt sample
was taken. The temperature was then lowered to 650°C,
and a day later a salt sample was again taken. This
sequence was then repeated at 600°C. Next, the salt
temperature was raised to 700°C, Cr,Te; was added,
and another Hastelloy N specimen inserted. The speci-
men was removed and salt samples were taken under the
same time-temperature conditions as discussed above.

The two Hastelloy N specimens were weighed and
submitted for Auger examination. No weight changes
were detected, but evidence of tellurium in the grain
boundaries was found (Sect. 6.12). The results of the
chemical analysis of the salt sample are shown in Table
6.14. Tellurium concentrations at 700°C were not as
high as was expected, but the fact that some tellurium
was in solution is demonstrated by the tellurium found
on the specimens. Additional Cr,Te; was added to the
solution, and two salt samples were taken. Preliminary
results indicate that the solution may not have been
saturated when the first series of salt samples was taken,

Following the solubility measurements, two tensile
specimens were exposed to the salt-Cr, Te; solution.
Both specimens, one of regular Hastelloy N and one of
2.6% Nb—0.7% Ti—modified Hastelloy N, showed a
weight increase after 500 hr exposure at 700°C. After
room-temperature tensile testing, the regular Hastelloy
N specimen was observed to have significantly more and
deeper cracks than did the modified Hastelloy N speci-
men (Sect. 6.14).

Table 6.14. Results of chromium telluride
solubility experiment

Tellurium and chromium content
of salt samples (ppm)

i f
Sampling Background After After
temperature (no Te) Cr,Te, Cr,Te,
C0) © addition addition
700 Te 5 Te <5 Te <5
Crd4 Cr75 Cr90
650 Te 15.1 Te 7.5
Cr 105 Cr 120
600 Te <5 Te <5
Cr? Cr 88

2]nsufficient sample.
103

The experimental assembly is being used to expose
standard Hastelloy N specimens to salt containing
Cr, Te; to obtain data on the extent of attack at 700°C
as a function of time.

6.11.3 Tellurium Experimental Pot 2

When a technique for introducing tellurium into salt
at an acceptable chemical activity has been developed, a
method will be needed for exposing a large number of
specimens to salt-tellurium solutions. A large experi-
mental pot has been constructed for this purpose. The
pot has a stirring mechanism, facilities for introduction
of electrochemical probes, and sufficient accesses to
allow simultaneous exposure of a large number of speci-
mens. Operation of the system will begin when a satis-
~ factory tellurium addition technique is available.

6.12 GRAIN BOUNDARY EMBRITTLEMENT OF
HASTELLOY N BY TELLURIUM

R. E. Clausing L. Heatherly

Auger electron spectroscopy (AES) is a powerful
technique for studying grain boundary embrittlement of
Hastelloy N by tellurium. The recent development of
the technique to permit AES analysis using a small-
diameter ("v5-u) electron beam to excite the Auger elec-
trons of a specimen surface has made truly microscopic
analysis possible.!> The development of techniques for
scanning the beam and the development of electronic
data processing equipment have continued to be a cen-
tral part of our efforts. As the techniques improve, our
ability to see the details of the tellurium embrittlement
process improves dramatically. We can now not only
provide a qualitative image of the elemental distribution
on intergranular fracture surfaces at a magnification of
several hundred times, but we can also provide a semi-
quantitative elemental analysis as the beam is scanned
along a line across the sample. However, it is not pres-
ently practical to provide a quantitative analysis along a
line across an intergranular fracture surface, since Auger
intensities at each point on a rough surface vary accord-
ing to topography. This effect can be corrected in prin-
ciple by a normalization technique, but data for each
point must be normalized individually, and the present
equipment cannot handle the volume of data required.
The data presented below are typical of several samples
of tellurium-embrittled Hastelloy N that were examined
recently. These samples are being studied in various

15. R. E. Clausing and L. Heatherly, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 104.

parts of our previously outlined efforts to understand
the tellurium embrittlement of nickel-based alloys. The
sample chosen for the present discussion demonstrates
our state-of-the-art capabilities and limitations and at
the same time provides some new insights into the
nature of the tellurium embrittlement of Hastelloy N.

A sample of Hastelloy N that had been exposed to
tellurium vapor at low partial pressure for 500 hr at
700°C was fractured in the AES system, and the result-
ing fracture surface was analyzed using Auger electron
spectroscopy. The fracture surface is shown in Fig.
6.53. The scanning electron micrographs, made by
Crouse, reveal that intergranular fracture occurred along
the edges of the sample and that the central region
failed in a ductile manner. One fairly large area of
ductile shear can be seen. Three types of Auger data
presentations are used below: imaging, line scans, and
selected area analyses. The first is qualitative, while the
second and third are progressively more quantitative.

Figure 6.54a is an image obtained using the scanning
beam in the AES system and the absorbed sample cur-
rent to produce the image contrast. It is similar to the
scanning electron micrograph (1), but, because of the
larger electron beam and the different method for pro-
ducing the image contrast, the resolution in Fig. 6.54a is
poorer, and some distortion is evident. Nevertheless, it
is relatively easy to correlate the features shown in Fig.
6.54a with those in Fig. 6.53a. Figure 6.54b 1s an image
of the same area shown in Fig. 6.54z but with image
contrast produced by the tellurium Auger signal. A care-
ful comparison of the areas of high tellurium concentra-
tion with the areas of intergranular fracture shows that
a good correlation exists between the two. No tellurium
can be detected in the regions of ductile or shear frac-
ture. Figure 6.55 is a series of line scans showing the
peak-to-peak intensity of the Auger signals for nickel,
molybdenum, chromium, and tellurium as the electron
beam was scanned along the path shown by the bright
line in Fig. 6.544. Some of the observations that can be
made are: (1) The intensities of the Auger signals are
influenced considerably by topography; that is, some
features, such as the shear region between feature Y and
the tellurium-embrittled region below it, show lower
Auger emission for all elements (this dependence on
topography accounts for much of the jagged nature of
the line scan). (2) The tellurium concentration is quite
high in the region of intergranular fracture near each
original surface. (3) There is a definite tendency for the
concentration of molybdenum to be higher in the
regions of intergranular fracture. (4) The nickel and
chromium concentrations are in approximately the
same ratio throughout the scan.

=
0t

350

300

200X
INCHES 0.010

MICRONS

0.005

0003

15 MICRONS 25
2000X

°
3
3
3
z
]
8
8
S

10

Fig. 6.53. Scanning electron micrographs of fracture surface of Hastelloy N sample exposed to tellurium vapor at 700°C for 500
hr showing the region examined by Auger electron spectroscopy. (@) 200X, (h) 500X, (¢) 1000X, (d) 1000X, (¢) 2000X. The lower
half of (c) shows an area of ductile shear.

105

¥-133510

Crack

Te Embrittled

Feature X
Feature Y

Te Embrittled

Y—-133512

eature X
eature Y

Te Embrittled

125 MICRONS 375
150X
0.005 INCHES ~ 0.015
Fig. 6.54. (a) Scanning electron micrograph obtained in the AES apparatus using absorbed sample current to produce image
contrast. The bright vertical line shows the path of the line scan and identifies regions in Fig. 6.55. (b) Image of the same region at

the same magnification using the tellurium Auger signal to produce contrast. The embrittled grain boundary regions next to the
original sample surfaces are obvious. (c) Same image as (a) with the regions analyzed and reported in Table 6.15 identified.

106

ORNL-DWG 75-15427

\_ Te (483 eV)

Mo (22 eV)

WAV

Cr (529 eV)
—
AREA
NO.

\ Ni (848 eV)

3
v

ORIGINAL b |
EMBRT'FTLED X

Te
EMBRITTLED

-

Fig. 6.55. Auger signal intensities for scans along the path indicated in Fig. 6.54a. The vertical axis is displaced and the vertical
scales arbitrarily varied to permit a qualitative comparison of the variations of Ni, Cr, Te, and Mo as a function of distance along the
scan line. The zones and features identified along the horizontal axis are also identified in Fig. 6.54a and c. The AES analysis of the

regions labeled area 1, area 2, and area 3 are given in Table. 6.15.

Another observation based on the detailed examina-
tion of this and other samples is that the tellurium con-
centration in the grain boundary is not a monotonically
decreasing function as one proceeds inward from the
original surface. On nearly all of the embrittled samples
examined thus far, the tellurium concentration is uni-
formally high throughout the embrittled area as, for
example, is shown on the right in Fig. 6.55. (The signal
intensity on the left is strongly influenced by topo-
graphy. If this effect were removed by a normalization
process, this area would have a more nearly uniform
composition similar to that on the right.) The high, rela-
tively uniform tellurium concentration in the embrittled
regions suggests that either a particular grain boundary
phase of fixed composition may exist or that the tellu-
rium atoms fill all of the appropriate grain boundary
sites in the embrittled region. Sputtering this fracture
surface (and those of similar samples) to a depth of a
few-atomic layers (3 to 10) reduced the tellurium con-
centration to below 1 at. %, showing that the tellurium
is concentrated very sharply in the grain boundary. It is

therefore unlikely that the tellurium present in the grain
boundary exhibits the properties of a bulk telluride.
The molybdenum concentration remained high during
sputtering operation, indicating that the concentration
of molybdenum is high in the bulk phase, perhaps in a
phase that has precipitated in the grain boundary.

Table 6.15 shows quantitative selected-area analyses
made in the three regions of the sample indicated in Fig.
6.54c¢. The compositions have been normalized to equal
100 at. % in each row. The three rows for each area are
obtained from one Auger spectra, but some elements
were ignored in the first two rows to make changes in
the relative amounts of the other elements more
obvious. These results confirm the above conclusions
and show (1) that tellurium is present in relatively large
amounts in the embrittled regions and (2) that area 3,
which is near the extreme of the depth to which the
tellurium penetrated, contains about as much tellurium
as area 2, which is located near the center of the upper
embrittled region. Regions 2 and 3 are both enriched in
molybdenum and carbon as indicated in the line scans.
107

Table 6.15. Composition of regions on the fracture surface
of a Hastelloy N sample exposed to tellurium
for 500 hr at 700°C

) Composition (at. %)b
Region? _
Ni Mo Cr Te C g S S
Composition in the lower region 70 19 11
area 3 (intergranular fracture) 64 17 10 9
40 11 6 6 33 2 1 1
Composition in central region . 75 16 9
area 1 (ductile fracture) 75 16 9
61 13 8 13 1 1 3
Composition in the upper - 64 25 12
region area 2 (intergranular 58 23 11 8
fracture) 42 17 8 6 23 1 1 2

2 Areas identified in Fig. 6.54c¢.

bThe composition in each row is normalized to equal 100 at. %. The three rows
for each region are from the same data, but are normalized so as to make changes
in relative amounts of the elements more obvious. For convenience and
consistency in reporting data we assume the AES spectra (Paul Palmberg et al.,
Handbook of Auger Electron Spectroscopy, Physical Electronics Industries, Inc.,
Edina, Minn., 1972) are accurate and directly applicable to our data. Elemental
sensitivities are taken directly from the spectra presented in the handbook with
no attempt to correct for chemical effects, line shape, matrix effects, escape
depth, or distribution of elements as a function of depth in the sample. The
analyzer used is Varian model 981-2707, operated with an 8000-eV electron beam

energy.

The above results suggest the need for a detailed
examination of the causes and effects of the high mo-
lybdenum and carbon contents in the grain boundary
region and also an examination of the implications that
the presence of a two-dimensional tellurium-rich grain
boundary phase may have on the time dependence of
tellurium penetration into the alloy.

6.13 X-RAY IDENTIFICATION OF REACTION
PRODUCTS OF HASTELLOY N EXPOSED TO
TELLURIUM-CONTAINING ENVIRONMENTS

D. N. Braski

" Hastelloy N and several modifications of the alloy
have been exposed to tellurium to determine their rela-
tive susceptibilities to intergranular cracking. Different
methods for exposing samples to tellurium have also
been studied in an attempt to develop a suitable screen-
ing test for the alloy development program. Some speci-
mens were exposed directly to tellurium vapor at
700°C, while others were subjected to attack by nickel
or chromium tellurides at 700 and 750°C respectively.
This section presents the results of x-ray diffraction
analyses of reaction products produced during the tests.
Knowledge of the reaction products aids in evaluating a

given method of tellurium exposure and may provide
information relating to the mechanisms of intergranular
cracking. ‘

A number of Hastelloy N tensile specimens and flat
x-ray samples were exposed to tellurium vapor at 700°C
for 1000 hr in an experiment conducted by Kelmers
and Valentine.!® The specimens were positioned in the
top portion of a long quartz tube having'a small amount
of tellurium at the bottom. The tube was evacuated,
backfilled with argon, and placed in a gradient furnace
with the specimens at 700°C and the tellurium source at
440°C. With this arrangement, tellurium vapor diffused
upward through the tube at a rate dependent on the
temperature difference between the specimens and the
tellurium (~0.05 mg Tefhr).!® At the end of 1000 hr
exposure, the specimens were covered with a very fine,
hairlike, deposit similar to that observed previously in
creep tests at 650°C.!7 The results of x-ray diffraction
analyses on these deposits are given in Table 6.16. The
first alloy listed is standard Hastelloy N, while the other
three have titanium and niobium additions. The main

16. A. D. Kelmers and D. Y. Valentine, MSR Program Semi-
annu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 40—-41.

17. R. E. Gehlbach and H. Henson, MSR Program Semiannu.
Progr. Rep. Aug. 31, 1972, ORNL4832, pp. 79-86.
108

Tabie 6.16. X-ray diffraction results for specimens exposed for

1000 hr at 700°C
Wt % alloying Method of
Heat number additions to nominal tellurium Surface reaction products
Hastelloy N composition exposure

405065 None Kelmers-Valentine Ni,Te, + Cr,Te,
experiment?

472-503 2.16% Ti Kelmers-Valentine Ni, Te,
experiment?

470-835 0.71% Ti + 2.6% Nb Kelmers-Valentine Ni, Te, + Cr,Te,
experiment?

180 1.84% Nb Kelmers-Valentine Ni, Te,
experiment?

474-533 2.0% Ti Brynestad? — low Ni, Te, + unidentified substance
Te activity exposure
Ni, Te, (s) + Ni(s)

405065 None Brynestad: Ni, Te, + NiTe, 4,

LiCl + Cr, Te,

2A. D. Kelmers and D. Y. Valentine, MSR Program Semiannu. Progr. Rép. Feb. 28, 1975, ORNL-5047, pp. 40—-41.

by, Brynestad, MSR Program Semiannu. Progr. Rep. Feb. 28, 1975, ORNL-5047, p. 102.

reaction product was Ni;Te,, which was detected on
the surfaces of all four. alloys. (Ni; Te, was found on
earlier samples exposed for shorter times in the same
apparatus.'®) X-ray lines which could be indexed as
Cr3 Tey were also found on standard Hastelloy N and on
the alloy modified with 0.71% Ti plus 2.6% Nb. The
Cr3Tes interplanar spacings and relative intensities were
calculated by H. L. Yakel, Metals and Ceramics Divi-
sion, from the crystallographic data in ref. 18. The pres-
ence of Cr3Te, in the reaction layer is reasonable, be-
cause both chromium and tellurium were detected pre-
- viously on Hastelloy N exposed to nickel telluriddes by
electron microprobe analysis.'”> In addition, chromium
tellurides were previously identified by x-ray diffraction
on Hastelloy N exposed to tellurium vapor.!?
Brynestad®® exposed 2% Ti—modified Hastelloy N
specimens to a low tellurium activity (Ni; Te, + Ni mix-
ture) at elevated temperatures. The specimens were first
placed in a quartz tube and the Ni;Te, + Ni powder
mixture packed around the specimens. The tube was
then sealed off under vacuum and placed in a furnace at
700°C for 1000 hr. The reaction products obtained in
this test also contained Ni; Te,, but the remaining four
lines could not be satisfactorily indexed to any of the

18. A. F. Bertaul, G. Roult, R. Aleonard, R. Pauthenet, M.
Chevretou, and R. Jansen, “Structures Magnetiques de Cr, X,
(X =8, Se, Te),” J. Phys. Radium 2(5),582-95 (1964).

19. D. N. Braski, O. B. Cavin, and R. S. Crouse, MSR Pro-
gram Semiagnnu. Progr. Rep. Feb. 28, 1975, ORNL-5047, pp.
105--09.

20. J. Brynestad, MSR Program Semiannu. Progr. Rep. Feb.
28, 1975, ORNL-5047, p. 102.

Ni, Cr, or Mo tellurides. The unusually broadened x-ray
diffraction peaks suggest that a complicated telluride,
such as Ni-Cr-Te, may have been formed. In another
tellurium experiment, Brynestad exposed a standard
Hastelloy N tensile specimen to a melt of LiCl contain-
ing Cr; Te; (solid) at 750°C. Some Cr, Te; dissolved in
the LiCl melt and reacted with the Hastelloy N. After
146 hr the tensile specimen was removed, and the flat
surface on one end was analyzed by x-ray diffraction.
The results (Table 6.16) showed that Ni3Te, and
NiTey o were produced.

In summary, these tests have shown that the primary
reaction product between Hastelloy N and tellurium
near 700°C is Ni;Te,. X-ray lines corrsponding to
Cr3 Teq were also present in patterns from the surfaces
of several Hastelloy N alloys exposed to tellurium vapor
at 700°C. Exposure of Hastelloy N to tellurium at low
activities (Ni;Te, + Ni mixture) may have produced
some complicated Ni-Cr-Te compounds in addition to
Ni3 Te, as evidenced by the unusually broadened X-ray
lines.

'6.14 METALLOGRAPHIC EXAMINATION OF
SAMPLES EXPOSED TO
TELLURIUM-CONTAINING ENVIRONMENTS

H.E.McCoy B.McNabb I.C. Feltner

Several samples of modified Hastelloy N were exposed
to tellurium-containing environments. They were de-
formed to failure at 25°C, a procedure which forms
surface cracks if the grain boundaries are brittle; a
109

metallographic section of each was prepared to deter-
mine the extent of cracking. These tests have two objec-
tives. The first is to develop a method for exposing
samples to tellurium to produce a reaction rate com-
parable to those anticipated for an MSBR. This rate is
thought to be a flux of tellurium of about 10'° atoms
cm™? sec™'. The second is to compare the cracking
tendencies of various alloys of modified Hastelloy N.

A new technique developed for measuring the extent
of cracking is more nearly quantitative than that used
previously, In the new technique a mounted and polished
longitudinal section of a deformed specimen is viewed
on a standard metallurgical microscope. The eyepiece
has a filar which can be moved to various locations in
the field being viewed. The filar is attached to a trans-
ducer which produces an output voltage that is a func-

tion of the location. The output signal is interfaced with
a small computer which will, on command, compute
crack lengths and several statistical parameters. The
information is displayed on a teletypewriter. The
cracked edge of the mounted specimen is scribed every
0.1 in., and the operator measures all cracks in succes-
sive 0.1-in. intervals until at least 30 cracks have been
measured. The computer then calculates and displays
the average crack length, the maximum crack length,
the standard deviation, and the 95% confidence interval.
A typical scan requires about 10 min and is consider-
ably faster than other methods used thus far.

The experimental conditions associated with the ten
experiments to be discussed in this report are summa-
rized in Table 6.17. The chemical compositions of the
alloys studied are given in Table 6.18. In all cases the

Table 6.17. General description of Te—Hastelloy N exposures

Experiment E . ¢ Exposure Alloys General
Designation Xpenmenters conditions included® comments
75-1 Brynestad LiCi + Cr, Te, 405065 Heavy reaction layers
for 146 hr at ~750°C
75-2 Kelmers, Te vapor for Whisker growth, evidence
Valentine 1000 hr at 700°CP 405065, of inhomogenous reaction
470-835, with Te
472-503, -
180
753 Brynestad 250 hr at 650°C 405065, Heavy reaction layers
packed in Cr, Te, 474-534,
474-535
754 Brynestad 200 hr at 700°C 405065 Heavy reaction layers
. packed in Cr;Te,
75-5 Brynestad, 504 hr at 700°C 405065 Reaction layers
Keiser in salt + Cr, Te, 470-835
75-6 Brynestad 1000 hr at 700°C 405065 No visible reaction layers
with vapor above
Cr,Te,
75-7 Brynestad 1000 hr at 700°C 405065 Shallow reaction layers
' with vapor above
Cr, Te,
75-8 Brynestad 1000 hr at 700°C with 405065 No visible reaction layers
vapor above 8, + 1,
nickel tellurides
75-9 McNabb, 250 hr at 700°C in 405065,471-114,474-534, No visible reaction layers
McCoy vapor above Te at 474-535, 600600, 62, 63,
300°C 181, 237,295, 297, 298,
303, 305, 306, 345, 346,
347,348,21543,469-344,
469-648,469-714, 470-786,
470-835
75-10 McNabb, 250 hr at 700°C in 405065, 21543, 345, 348, No visible reaction layers
McCoy vapor above Te at 411,413,421,424,425

300°C

ZSee Table 6.18 for chemical compositions.

bA. D. Kelmersand D. Y. Valentine, MSR Program Semiannu. Progr. Rep. Feb: 28, 1975, ORNL-5047, pp. 40-41.
110

Table 6.18. Chemical analyses of nickel-base alloys
used in tellurium cracking studies (weight percent)

Heat number Mo Cr Fe Mn C Si Ti Nb Al QOther

62 11.34 752  a 020 0.042 001 a 1.9 a

63 11.45 733 4 020  0.135 0.01 a 2.5 a

180 11.2 70 0 0.040 022  0.046 001  <0.02 184 a

181 11.5 684  0.054 023  0.45 0.0l 050 185 & 0.03 W

237 12.0 6.7 43 049 0032 4 0.04  1.03 <0.05

295 114 8.06  4.02 028  0.057 <002 <002 085 a 0.05 W

296 11.5 8.09  3.96 0.28  0.059 <0.02 <002 1.2 a 02 W

297 12.0% 700 a0b 020 0.06 0.02 024 057 a

298 12.0° 7.00  agb 0.2 0.06 002 <001 20 a

303 12.00 700 aob 020 0062 002 049 084 4

305 11.2 825  4.16 022  0.072  0.09 088 13 a

306 10.6 8.04  3.11 0.18 0065 027 001 055 4

345 11.0 7.1 3.8 026 0052 022 002 045 4

346 11.0 6.7 3.7 0.18  0.05° 0.48 002 049 4

347 120 76 4.3 025  005® 047 <002 088 4

348 12.0 7.2 0.07 0.19  0.05° 047  <0.02 062 a4

411 12.0¢ 700 4 020 0050 4 a 115  a

413 12.00 700 4 020 0052 o 100 113 a4

421 12.00 700 4 028 0052 219  1.04 007

424 12.00 7.0 4 026 0052 4 1.8 134  0.10

425 12.00 700 4 026 o0 ¢ 1.9 048  0.08

405065 16.0 7.1 4.0 055  0.06 057 <001 & <0.03

472-503 12.9 679 0089 <00l 0066 0089 216 005  0.09

471-114 12.5 7.4 0.062 002 0058  0.026 175  a 0.07

474-534 11.66 712 006 <001  0.08 0.03 209 a 053  0.14 W,
0.013 La

474-535 11.79 730 005 <001  0.08 0.03 213 @ 055  0.10W,
0.010 La,
0.03 Ce

600600 1600 8.0 0.19 0.27

(Inconel 600)

469-648 12.8 6.9 0.30 034 0043 @ 092 195 4

469-714 13.0 8.5 0.10 035 0013 a 080 160 a .

470-835 12.5 7.9 0.68 060 0052 « 071 260 a 0.031% Hf

470-786 12.2 7.6 0.41 043 0.044 a 0.82 0.62 a 0.024 Zr

469-344 13.0 74 40 056  0.11 2 077 1.1 a 0.019 Zr

421543 0.04 0.08 0050 0019 0.7 0.02

12.4 7.3

9Not analyzed, but no intentjonal addition made of this element.

bNot analyzed, but nominal concentration indicated.

sample was a small tensile specimen ¥% in. in diameter X
1% in. long having a reduced section ' in. in diameter
X 1% in. long. All specimens were annealed 1 hr at
1177°C in argon prior to exposure to tellurium. The
results of crack measurements and data resulting from
the tensile tests at 25°C that were used to open the
embrittled grain boundaries are shown in Table 6.19.
Experiment 75-1 was run by Brynestad and involved a
sample of standard Hastelloy N that was immersed in
LiCl saturated with Cr,Te; for 146 hr at 750°C. The
specimen formed a heavy reaction layer (Table 6.17)
but lost weight (Table 6.19). Figure 6.56 shows that the
reaction was rather extensive, with some obvious grain

boundary penetration which resulted in extensive crack
formation in the deformed section. The extent of reac-
tion in this experiment was higher than anticipated for
an MSBR, and therefore it is not believed that the
experimental conditions employed constitute a good
screening method.

Experiment 75-2 was run by Kelmers and Valentine,
and the detailed results were described previously.?! All
samples lost weight in this experiment (Table 6.19).
Although the samples had more reaction product on

21. A. D. Kelmers and D. Y. Valentine, MSR Program Semi-
annu. Progr. Rep. Feb, 28, 1975, ORNL-5047, pp. 40-41.
Table 6.19. Intergranular crack formation and tensile 'pr_Operties” of samples exposed to tellurium and strained to failure at 25°C

s X
Experiment Cracks/unit length Depth (u) Stapdfard cor?fifidzbnce Weightb Yield U‘l::':;:e Fracture l_Jnifor‘m Fract‘ure Rtaduc!ion
number Heat number Crackeli Crack Mo deviation interval change streSS' stream suess‘ elongau?n strain in area

racksfin. racks/fem  Average aximum () o (mg) (10° psi) (10 pi) 10° psi)  (10° psi) (%) %)
75-1 405065 310 122 101.1 187.2 46.3 16.6 -1.7 48.9 [14.3 111.8 384 395
75-2 405065 320 126 46.3 75.7 12.8 4.5 -1.0 51.7 124.7 117.8 40.4 41.8 42.1
470-835 167 66 26.7 70.9 15.6 44 -2.1 571.7 142.0 136.0 42.5 44.0 353
472-503 300 118 533 97.0 264 9.6 -1.8 56.9 133.8 123.0 38.9 40.4 41.0
180 63 25 243 60.2 £3.2 6.1 -4.7 54.5 124.2 115.8 3422 39.¢6 23.7
75-3 405065 240 95 27.2 429 8.0 2.7 —287 46.7 116.0 109.0 40.2 424 48.4
474-534 220 87 14.5 276 4.5 1.4 —245 56.6 114.0 109.0 41.1 442 52.7
474-535 230 91 20.3 344 54 1.6 -231 459 110.6 99.7 47.4 50.6 543
754 405065 410,570 161,224 69.3,59.1 118.1,123.5 283,25.1 8.9,6.7 —788, —631 44.8,44.0 103.7,100.9 99.3,96.8 32.2,31.0 33.2,324 48.6,46.0
75-5 4035065 370 146 42.5 ©69.0 159 5.2 -60.1 50.9 121.2 117.3 36.5 37.2 28.2
470-835 180 71 331 547 10.8 36 -58.8 55.6 131.2 126.7 37.0 37.9 384
75-6 405065 215 85 85.2 148.3 294 9.0 +0.02 51.6 121.3 117.2 404 42.0 339
15-7 405065 520 205 40.0 106.6 249 6.9 -25.5 524 124.9 118.5 38.7 394 333
75-8 405065 360 142 59.5 92.0 219 7.3 -36.3 52.2 123.8 117.6 40.0 41.4 329
75-9 405065 360 142 41.6 75.3 15.1 5.0 -7.6 529 127.5 122.0 39.5 41.1 394
471-114 185 73 374 639 13.3 4.4 +8.4 479 113.4 105.0 49.6 54.1 46.6
474-534 360 142 225 435 10.7 36 =27 56.9 126.9 1124 43.3 45.5 488
474-535 240 95 20.1 371 9.1 . 37 -0.9 525 1224 113.3 46.3 48.3 479
600600 265 104 17.8 325 6.6 1.8 +3.0 39.9 101.9 79.4 321 37.5 57.2
345 100 39 256 64.1 11.3 36 +0.7 61.7 119.9 107.7 © 391 41.8 46.3
346 160 63 20.7 44 8 10.0 35 +0.4 55.1 1289 111.7 42.1 44.7 47.1
348 9 4 9.6 i5.9 3.1 2.1 -0.3 54.8 123 .4 107.5 42.5 459 503
347 270 106 18.8 373 6.6 1.8 +2.9 58.8 131.2 116.2 40.1 431 46.3
306 450 177 21.0 379 7.3 2.2 +0.8 60.0 130.0 119.2 376 40.1 42.8
303 430 169 14.6 26.8 6.1 1.9 +1.3 56.2 122.3 1100 471 495 49.0
297 _ 310 122 18.0 41.0 6.2 2.2 +2.3 61.1 129.2 117.2 353 378 48.5
295 25 10 10.1 17.9 36 23 -2.5 53.5 122.1 1145 43.3 46.2 44 9
231 85 33 14.8 385 8.5 2.9 -0.9 58.4 122.6 108.5 43.5 46.5 494
305 360 142 16.9 32.7 7.3 24 +0.5 53.6 126.5 117.3 46.8 489 47.7
298 330 130 17.3 30.7 6.4 2.2 +0.4 72.3 123.6 1104 42.7 499 49.7
181 220 87 16.7 38.6 8.7 37 -7.4 549 127.1 118.8 454 45.6 395
62 30 12 29.3 146.0 44.5 25.7 -5.9 47.9 117.5 110.1 49.7 50.7 42.8
63 440 173 12.7 26.7 53 . 1.6 -89 58.1 133.4 121.7 333 38.0 423
469-648 330 130 14.8 433 7.4 26 +0.2 58.9 135.7 127.3 43.7 459 443
469-714 83 33 124 504 12.9 4.5 +0.08 49.9 121.7 116.3 53.7 53.3 43.6
470-835 93 37 8.6 2477 5.0 19 -0.03 55.9 137.5 1304 46 .4 47.7 43.9
470-786 32 13 13.8 578 14.5 7.3 -03 49.5 116.7 107.3 50.6 52.7 40.8
469-344 360 142 9.1 17.1 38 1.3 +0.07 58.7 138.8 129.0 38.0 39.7 414
421543 85 33 8.8 296 48 1.7 —1.2 454 110.6 99.1 54.7 579 53.3
75-10 405065 340 133 17.7 48.5 7.8 1.9 +0.04 53.1 1326 122.0 40.9 364 40.9
- 425 300 118 27.1 548 11.2 4.1 +0.5 51.8 126.0 117.9 46.7 48.5 42.1
425 240 95 227 51.0 124 36 +0.02 51.8 126.3 1163 46.3 42.0 47.9
421 340 133 244 60.1 10.8 2.6 -0.03 53.0 129.7 122.3 43.7 45.6 42.9
424 430 169 26.7 50.8 84 26 -0.03 57.4 1393 132.7 44.2 45 .8 383
298 208 82 8.7 206 34 0.74 +0.1 53.6 125.8 1120 47.5 51.2 52.7
295 26 10 10.6 29.5 6.9 38 +0.2 534 122.1 115.7 42.4 44 6 47.5
348 34 13 16.1 36.2 7.2 3.5 +54 49.8 1219 107.1 44.8 476 54.7
345 13 5 89 14.1 33 3.0 +0.01 59.1 119.8 107.9 44.3 474 454
413 80 32 10.5 25.7 53 1.5 +0.08 52.0 1269 115.5 46.9 494 44.0
411 2 9 14.8 339 9.2 5.1 +14 42.1 1136 97.9 51.9 54.2 536
421543 17 7 109 237 6.1 39 +2.9 47.3 1139 101.4 523 55.7 53.7

9dTensile test run at 25°C at a strain rate of 0.044 min~'

bTotal weight of specimen 6.1 t0 6.2 g.

It
112

Y—133193

(a)

(b) ¥

Y—-133195

Fig. 6.56. Standard Hastelloy N (heat 5065) exposed to LiCl saturated with Cr, Te, at 750°C for 146 hr. (a) Edge of unstressed
portion of specimen, () edge of stressed portion of specimen. As polished. 100X.

one end than the other, the extent of cracking seemed
reasonably uniform. Typical photomicrographs of the
four materials are shown in Fig. 6.57. Alloys 405065
(standard) and 472-503 (2.16% Ti) formed extensive
cracks, but alloys 470-835 (0.71% Ti, 2.60% Nb) and
180 (1.847% Nb) were considerably more resistant to
cracking.

In experiment 75-3, three samples were packed in
CryTey granules for 250 hr at 650°C. The samples
formed heavy nonadherent reaction products and lost
weight (Table 6.19). All three materials formed exten-
sive cracks (Table 6.19, Fig. 6.58), with the depth of
cracking being slightly less in the two modified alloys
(474-534 and 474-535) than in standard Hastelloy N
(heat 5065). However, the extent of reaction is too high
under these conditions for the results to be meaningful.

In experiment 754, duplicate samples of standard
Hastelloy N (405065) were packed in granules of
CryTes and heated 200 hr at 700°C. The samples
formed nonadherent reaction films and lost weight dur-
ing the test (Table 6.19). The reaction layer and the

intergranular cracking produced during stressing are
shown in Fig. 6.59. Again, the reaction rate was un-
reasonably high for use of the exposure conditions as a
screening test.

In experiment 75-5, Brynestad and Keiser exposed
two specimens to MSBR fuel carrier salt (containing no
uranium) that was saturated with Cr,Te;. The ex-
posure was for 504 hr at 700°C. These samples formed
reaction layers but lost weight (Table 6.19). As shown
in Fig. 6.60 both materials formed reaction layers, but
in heat 470-835 (0.71% Ti, 2.60% Nb) there appeared
to be less penetration of the reactants along the grain
boundaries. The standard Hastelloy N had regions where
layers of grains dropped out during the exposure. The
number and depth of cracks in the stressed portion of
the samples were less for heat 470-835 than for stan-
dard Hastelloy N, but both materials formed extensive
intergranular cracks.

Since the samples packed in the various tellurides
reacted extensively, several experiments were run in
which the samples and the telluride were separated in

113

Y—-134367

Y—134365

(b)

Y —134362

(c)

Y —134369

0.010in

0.25mm

Fig. 6.57. Specimens from experiment 75-2 which were exposed to a low partial pressure of tellurium for 1000 hr at 700°C. (a)
Heat 405065, () heat 472-503 (2.16% Ti), (¢) heat 470-835 (0.71% Ti, 2.6% Nb), (d) heat 180 (1.84% Nb). As polished. 100X

Y-134623

-

] v-134361

(e)

v 134360

0.010in
025 mm
Fig. 6.58. Specimens from experiment 75-3. Packed in Cr, Te, granules for 250 hr at 650°C and deformed to fracture at 25°C.

(@) Heat 405065, unstr , (b) heat 405065, stressed, (c) heat 474-534 (2.09% Ti, 0.013% La), unstressed, (d) heat 474-
stressed, (¢) heat 474-535 (2.13% Ti, 0.01% La, 0.03% Ce), unstressed, (f) heat 474-535, stressed. As polished. 100X.

115

(®

Y 134626

Y—134625

Fig. 6.59. Specimen of standard Hastelloy N (heat 405065) from experiment 754. Packed in Cr, Te, granules for 200 hr at
700°C and deformed to fracture. (a) Edge of unstressed shoulder, (b) edge of stressed gage length. As polished. 100X

the reaction capsule. In experiment 75-6, standard
Hastelloy N was reacted with the vapor above Cr3Te, at
700°C for 1000 hr. The specimen gained a small
amount of weight (Table 6.19), did not form an obvious
reaction layer (Fig. 6.61), but did form extensive inter-
granular cracks (Fig. 6.61, Table 6.19). Experiment
75-7 was run in the same way, but CryTe; was used.
The sample lost weight, formed a surface reaction prod-
uct, and formed intergranular cracks when strained
(Table 6.19, Fig. 6.62). In experiment 75-8 the source
of tellurium was two nickel tellurides, 8, and v, . The
specimen lost weight, did not form a visible surface
reaction product, and did form intergranular cracks
(Table 6.19, Fig. 6.63). From these experiments it was
concluded that the tellurium activity produced by
Cr3 Tes was likely that best suited for screening studies.

Experiment 759 included 25 alloys which were
exposed to tellurium vapor at 700°C for 250 hr. The
weight changes covered a range of +8.4 to —7.4 mg,
with no obvious correlation between weight change and
crack depth or number (Table 6.19). These specimens
were sealed in four different capsules for exposure to
tellurium, and there were differences in the extent of
discoloration of the samples. These differences are
likely associated with slight differences in the extent of
reaction due to variation of the temperature of the
tellurium metal in the various capsules. Thus, it is ques-
tionable as to how far one should carry the analysis of
the data from this experiment.

One further problem concerning data analysis which
applies equally well to all data sets is the basis that
should be used for comparison. The number of cracks
and their average depth are two very important param-
eters. However, it is possible that a specimen can have a
large number of shallow cracks (e.g., heat 63, Table
6.19) or a few rather deep cracks (e.g., heat 62, Table
6.19). The formation of intergranular cracks of any
depth is important, because this may indicate a ten-
dency for embrittlement. The depth of the cracks is
important, because this is a measure of the rate of pene-
tration of tellurium along the grain boundaries. How-
ever, for a relatively short test time [test 759 (250
hr)], the formation of numerous shallow cracks may be
indicative of a near-surface reaction which will not lead
to rapid penetration with time. Obviously, longer-term
tests are needed to determine the rate of penetration of
tellurium into the metal.

On the basis of number of cracks formed, the alloys in
experiment 75-9 which formed less than 40 cracks per
centimeter were 345, 470-835, 421543, 237, 469-714,
470-786, 62, 295, and 348. The alloys forming cracks
with an average depth of 7 u were 63, 469-714,
295, 348, 469-344, 421543, and 470-835. Several of
the alloys appear good on the basis of both criteria.
These alloys all contain niobium and several contain
niobium and titanium. Another parameter used for
comparison was the product of the number of cracks
and the average crack depth. The alloys from experi-

116

Y—132811

(a)

Y-132812

Y-132813

(c)

Y_132814

(@
0.010in
0.25mm
Fig. 6.60. Samples from experiment 75-5. Samples exposed to fuel salt saturated with Cr, Te, for 504 hrat 700°C and strained
to fracture at ". (a) Standard Hastelloy N, unstressed shoulder, (b) standard Hastelloy N, stressed gage length showing region

where grains were lost during salt exposure, (c) heat 470-835 (0.71% Ti, 2.60% Nb), unstressed shoulder, (d) heat 470-835, stressed
portion. As polished. 100x.

117

(a)

(b) =

Y—132881

Y—132882

0.25mm

Fig. 6.61. Standard Hastelloy N (heat 405065) from experiment 75-6. Samples exposed to the vapor above Cr, Te, at 700°C for
1000 hr and strained to failure. (¢) Edge of unstressed portion, (b) edge of stressed portion. As polished. 100X,

ment 75-9 are ranked on this basis (Table 6.20). Stan-
dard Hastelloy N is significantly different from all other
heats on this basis. There are large variations among the
other heats, but it is difficult to pick out general trends
on the basis of niobium and titanium concentrations.

Several typical photomicrographs of samples from
experiment 75-9 are shown in Fig. 6.64. No reaction
films were visible on any of these specimens. The pic-
tures show clearly the wide range of cracking experi-
enced by the various heats.

The mechanical property data show small, but signifi-
cant, variations in the yield and ultimate tensile stresses
of the various heats (Table 6.19). The higher stresses are
generally associated with the alloys containing the
higher amounts of niobium and titanium. However, the

high fracture strain and reduction in area for all heats
indicate that only very small (if any) amounts of gamma
prime formed during the 250 hr at 700°C.

In experiment 75-10, steps were taken to ensure that
the specimens were at a uniform 700°C and that the
tellurium was at 300°C. The weight changes were very
erratic and show no correlation with the number of
cracks or the depth of crack formation (Table 6.19). A
sample of heat 425 was included in each of the two
capsules used in this experiment to obtain some idea of
reproducibility. The reproducibility was reasonably
good. Samples of alloys 405065, 298, 295, 348, and
345 were included in experiments 759 and 75-10.
Heats 405065, 298, and 345 in experiment 75-9
cracked more severely than in experiment 75-10. Alloy

(a)

Y-132816

® - de i : ;

0.010 in.
0.25mm

Fig. 6.62. Standard Hastelloy N (heat 405065) from experiment 75-7. Samples exposed to the vapor above Cr, Te, at 700°C for
1000 hr and strained to failure. (¢) Edge of unstressed portion, (b) edge of stressed portion. As polished. 100X .

Y-132818

Fig. 6.63. Standard Hastelloy N (heat 405065) from experiment 75-8. Samples exposed to the vapor above g, + v, nickel
tellurides at 700°C for 1000 hr and strained to failure. (a) Edge of unstressed portion. (b) edge of stressed portion. As polished
100%

119

Table 6.20. Ranking of materials from experiment 75-9

Product of number of cracks

and average depth Heat Concentration (%)?
<C“‘Cks X micmns) number Ti Nb Other
cm
5907 405065
3717 306 0.55 0.27 Si
3195 474-534 2.09 0.013 La
2730 471-114 1.75
2467 303 049 0.84
2400 305 0.88 1.3
2249 298 2.0
2197 63 25
2196 297 0.24 0.57
1993 347 0.88 047 Si
1924 469-648 0.92 1.95 ‘
1910 474-535 2.13 0.04 La + Ce
1851 600600 0.27 15 Cr '
1453 181 0.50 1.85
1304 346 0.48 0.49
1292 469-344 0.77 1.7
998 345 0.45 0.22Si
488 237 1.03
409 469-714 0.80 1.60
352 62 1.9
318 470-835 0.71 2.60
290 421543 0.7
179 470786 0.82 0.62
101 295 0.85
38 348 0.62 0.47 Si

@%ee Table 6.18 for detailed chemical analyses.

348 was less severely cracked in experiment 75-10, and
heat 295 reacted similarly in both tests. Such differ-
ences emphasize the importance of duplicating test
results before making important conclusions.

The alloys in experiment 75-10 that formed <32
cracks/cm were 413, 34, 295, 411, 421543, and 345.
Those with average crack depths <10.9 u were 421543,
295, 413, 345, and 298. Again, this ranking is of ques-
tionable value, because alloy 298 had the shallowest
cracks of the 12 specimens, but formed a large number
of cracks. In an effort to combine the factors of number
and depth of cracks, the two factors were multiplied
and the alloys ranked as shown in Table 6.21. There is a
“very large step between alloys 425 and 298, and the
better alloys appear to be ones containing from 0.45 to
2.0% Nb with titanium additions of 1% or less.

The tensile data show small variations, but do not
show evidence of embrittlement due to gamma prime
formation during 250 hr at 700°C (Table 6.19). In spec-
imens from experiment 75-10 the wide range of crack-
ing behavior is apparent (Figs. 6.65 and 6.66).

These tests have shown that several methods are avail-
able for exposing metal specimens to tellurium. The
metal telluride, CryTe4, has an activity most consistent
with our estimates of tellurium activity in an MSBR.
Specimens can be exposed to salt containing Cr3 Te4 or
exposed to vapor above the compound. Tellurium metal
at about 300°C has a vapor pressure of about 1 X 107
torr and appears to provide a tellurium activity com-
parable to that expected in an actual MSBR. The speci-
mens exposed thus far show that niobium is effective in
reducing the extent of intergranular embrittlement of
Hastelloy N.

6.15 EXAMINATION OF TeGen-1

B.McNabb  H. E. McCoy

The TeGen series of capsules was designed for study-
ing the effects of tellurium and other fission products
on metals. The fuel capsule is a %-in.-OD X 0.035-in-
wall X 4-in.-long tube segment of the metal under

120

Y 132820

(a)

Y—132838

(b)

Y—132864

(c)
.00
0.25mm
Fig. 6.64. Comparison of intergranular cracking in Hastelloy N type alloys exposed to partial pressure of tellurium of 10 * torr
for 250 hr at 700°C and strained to fracture at 25°C. (a) Standard Hastelloy N (heat 405065) (142 cracks/cm, av depth 41.6 w), (b)

modified Hastelloy N containing 0.85% Nb (alloy 295) (10 cracks/cm, av depth 10.1 u), (¢) modified Hastelloy N containing 0.82%
Ti, 0.62% Nb (alloy 470-786) (13 cracks/cm, av depth 8.6 ). 100X

Table 6.21. Ranking of materials from experiment 75-10

Product of number of cracks

and average depth Alloy Concentration (%)
(L‘*‘x ,,,,wm) number Ti Nb Other
cm
4512 424 1.8 1.34
3245 421 2.19 1.04
3198 42 1.98 0.48
2328 405065
2157 425 1.98 048
713 298 2.0
336 413 1.0 113
209 348 0.62 047 Si
133 411 L1s
106 295 0.85
76 421543 0.7
45 345 045 0.22 Si

@See Table 6.18 for detailed chemical analyses.

121

Y -134586

(a)

Y—134583

(b)

Y—134589

(c)

Y—134574

Y —134588

(e)

Y—134572

(£)
0.010in
0.25mm
Fig. 6.65. Strained specimens from experiment 75-10. Specimens were exposed for 250 hr at 700°C to the vapor above tellurium
metal at 300°C. (@) Alloy 424, (b) alloy 421, (¢) alloy 425, (d) alloy 405065, () alloy 425, () alloy 298. As polished. 100X

122

Y 134582

(a)
w

(b)

Y-134579

(c)

Y—-134570

(d)

Y —134591

(e)
w

(£)

0.25mm

Fig. 6.66. Strained specimens from experiment 75-10 exposed for 250 hr at 700°C to the vapor above tellurium metal at 300°C.
(a) Alloy 413, () alloy 348, (c) alloy 411, (d) alloy 295, (e) alloy 421543, (£) alloy 345. As polished. 100X.

123

study. The capsule is partially filled with the MSRE-
type fuel salt and irradiated in the ORR to produce
fission products.

The first experiment of this series involved fuel pins
made of Inconel 601, standard Hastelloy N, and type
304 stainless steel, and the irradiation time was such
that the amount of tellurium produced per unit area of
metal in contact with salt was equal to that at the end
of operation of the MSRE. Some of the details of the
postirradiation examination were described pre-
viously.>? A typical fuel pin is shown schematically in
Fig. 6.67. The segments marked “A” were subjected to
tensile tests, using the fixture shown in Fig. 6.68. The
mechanical property data obtained from the rings and
the results of limited metallographic examination were
reported previously.>> More detailed metallographic
studies have been completed during this report period.
The segments marked “B” were used for chemical
studies. The salt from each segment was analyzed, and
the fission product distributions on the tube surface and
a short distance into the tube were determined from
two successive leach solutions. The first leach used a
“verbocit” solution (sodium versenate, boric acid, and
sodium citrate) which should have dissolved only resid-
ual salt from the metal surface. The second solution was
aqua regia, and the time was sufficient to remove about

22. B. McNabb and H. E. McCoy, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1975, ORNL-5047, pp. 123-36.

TYPE DESCRIPTION AND USE
A Y6 in. RING FOR MECHANICAL
PROPERTIES
8 Y4 in. FOR LEACH (2 STEP)
c SECTION TO BE RETAINED

Fig. 6.67.

1 mil of the tube. Both solutions were subjected to
various chemical procedures to analyze for various
nuclides and elements. These results are partially ana-
lyzed, and the results for tellurium will be discussed.
The tube segments marked “C” were retained for pos-
sible future studies.

6.15.1 Metallographic Observations

Photomicrographs of the three materials in the un-
deformed condition are shown in Fig. 6.69. Numerous
voids were present near the surface of the Inconel 601
specimen to a depth of about 0.2 mil. Voids were likely
caused by the removal of chromium from the alloy via
reaction with UF,; in the salt. The Hastelloy N section
shows no evidence of chemical reaction with the salt.
The type 304 stainless steel shows some grain boundary
attack to a depth of about 0.5 mil. This was likely
caused by selective removal of chromium along the
grain boundaries. The features in the type 304 stainless
steel appear much like shallow cracks and may have
influenced the number of cracks that were observed in
stressed samples of this material.

Composite photomicrographs of the Inconel 601 rings
after straining to failure are shown in Fig. 6.70. Rings 2
and 4 from near the salt-vapor interface exhibit some
evidence of attack, but the other samples are almost
entirely free of indications of chemical reaction.

Photomicrographs of the deformed rings from the
Hastelloy N capsule are shown in Fig. 6.71. The count

ORNL-DWG 75-9062

END CAP

DN

AND SALT LEVEL

DI

Schematic diagram of individual fuel pin showing the locations of test specimens.
124

ORNL- DWG - 75+ 17506 metallographic sample includes the fracture and an
adjacent segment. Since the fracture occurred at dif-
ferent locations, the metallographic specimen contains
varying amounts of inhomogeneously deformed mate-
rial. For example, Fig. 6.71¢ includes a very small seg-
ment of homogenously deformed material, whereas Fig.
6.71f includes a relatively long segment. As shown by
the photomicrographsin Fig. 6.71 and the data in Table
6.22, specimens from the vapor region (2-A-1), the salt-
vapor interface (2-A-2), and the bottom of the salt
(2-A-16) cracked most severely. Three samples from
other locations formed shallower cracks. It is not
known whether these differences are significant.

Typical photomicrographs of deformed rings from the
type 304 stainless steel capsule are shown in Fig. 6.72.
These specimens, located on the inside surface, had shal-
low cracks with an average depth of about 0.4 mil
(Table 6.22). These cracks were rather uniformly distri-
buted in the samples from all four locations. As noted
in Fig. 6.69, the unstressed specimen also contained
cracklike features having a maximum depth of about
0.5 mil. Hence, the cracks in the stressed specimens may
simply be the result of further opening of features that
are likely related to corrosion.

6.15.2 Chemical Analyses for Tellurium

The tube segments designated B-1, B-2, and B-3 in
Fig. 6.67 were subjected to several types of chemical
analyses, but only the results for tellurium have been
analyzed in sufficient detail to report at this time. The
results for the three pins are shown in Table 6.23. The

Table 6.22. Summary of crack frequency and
depth information for rings from TeGen-1
fuels after straining to failure at 25°C

G Crack .

Fig. 6.68. Fixture for tensile testing rings from TeGen fuel =~ Specimen frequency Crack depth (mils)
pins. number (cracks/in.) Average Maximum
of crack frequency shown in Table 6.22 was made in an Hastelloy N
effort to detect significant differences in cracking — 2-A-1 480 0.80 2.0
among the various specimens. These counts are subject g:fij :‘158 (l) 'éo f;
to numerous ?roplems, the main one -bemg the inhomo- 2-A-5 480 0.58 12
geneous distribution of strain within the sample. In 2-A-8 330 046 1.0
deforming the ring specimens in the fixture shown in 2-A-16 380 1.4 2.5

Fig. 6.68, the small portions of the ring located between

the two parts of the fixture likely deformed very uni- Type 304 stainless steel

formly, but this length is very short relative to the total ~ 3-A-2 160 0.49 1.2
length. The part of the ring that contacted the fixture g:fig’ gég 8§Z ig
likely deformed in some areas but was restrained in 3.A-16 202 0.37 1.0

other areas by surface friction from the fixture.' The
125

(a)
R-70188
: % = . »
(b) - -

() e
i 20 4‘0 60 MICRONS 100 120 140
— 500X ———L
0.00t INCHES 0.005

Fig. 6.69. Undeformed rings (sample No. 9) from each TeGen fuel pin near the middle of the fuel salt. (a) Inconel 601, (b)
Hastelloy N, (c) type 304 stainless steel. As polished. 500X

126

R—70185 R—70182 |

(b)

Fig. 6.70. Samples from Inconel 601 fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and deformed to
failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (¢) Location A-1, (b) location A-2, (c)
location A-4, (d) location A-5, (e) location A-8, (f) location A-16.

127

R—70180

Fig. 6.71. Samples from Hastelloy N fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and deformed to
failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (¢) Location A-1, (b) location A-2, (c)
location A4, (d) location A-5, (e) location A-8, (f) location A-16.
128

R—70177

Fig. 6.72. Samples from type 304 stainless steel fuel pin from TeGen-1. Rings taken from the locations shown in Fig. 6.67 and
deformed to failure at 25°C. Portion of specimen exposed to fuel salt is on the lower side of each figure. (@) Location 2A, (b)
location 4A, (¢) location 8A, (d) location 16A.

129

Table 6.23. Chemical analyses for '? 7™Te and ' ?® MTe nuclides in the three fuel pins from TeGen-1

Sample Type Concentration of 2 7MTe Concentration of ' **MTe
Pin identification i - chemistry B . z "
ocation sample? dpm total or dpm/g g/lem? or gfg dpm total or dpm/gb g/cm? or gfg€
No. 1 — Inconel 601 1B1 A <22 X107 d <9.5 % 108 d
1B1 B 4.65x 10* 257%x 107 237 x 10° 446X 10°°
1B2 A <2.5 % 10’ d <2.37 x 107 d
1B2 B 1.59 X 10° 8.75x 10" 6.00 x 10 1.13x 1077
1B2 C 7.85 x 10° 1.14 x 107° 7.44 X 10°¢ 4.5%x 1077
1B3 A <53 x 108 d <1.7 X 10°¢ d
1B3 B 6.27 x 10® 345x 10°°® 3.06 x 10® 5.76 x 10°*
1B3 C 247 x 107 3.58x10°° 1.63 x 107 9.9x 10°°
No. 2 — Hastelloy N 2B1 A <6.0 X 10¢ d <8 64 X 10¢ d
2B1 B 7.49 x 10° 38 %107 4.97 x 10* 094 X 1077
2B2 A <8.5 x 107 d <54x107 d
2B2 B 295 x 10* 1.52x 107 2.02 x 10* 3.76 x 10°®
2B2 C 6.78 X 107 6.58x 10°° 5.94 x 107 24x10°°
2B3 A <54 x 10° d <5.8x 10°¢ d
2B3 B 8.19 x 10° 4.22x10°® 8.05 x 10° 1.51x 1077
7 2B3 C 3.48 x 107 269X 10°° 3.93 x 107 1.61x 1078
No. 3 — type 304 - 3B1 A 5.52x 10° 3.04x10°® 9.59 x 10° 1.80x 1077
stainless steel iB1 B 1.31 x 10® 0.72x 1078 1.83 x 10® 3.44 x 10°®
iB2 A 9.54 x 107 6.29 X 10°° 3.04 X 107 8.51x 10°*
iB2 B 2.50 x 10° 1.66 X 107* 3.60 x 107 10.1 x 10°°
3B2 C 9.00 X 107 1.31 x10°® 10.62 x 107 6.48x 107
3B3 A 1.66 X 107 L.I0X 107 5.42 X 10° 3.3% 107
3B3 B 1.27 x-10°® 0.84 x 107 3.15 x 107 1.9x 1078
3B3 C 5.38 x 107 7.8x 107° 3.23 x 107 197 x 1073

4A denotes 100 cm® solution obtained by leaching the metal sample in “‘verbocit” (sodium versenate, boric acid, and sodium citrate}. B denotes
100 cm? solution obtained by leaching the metal sample in aqua regia to remove about 1 mil of metal. C denotes 100 cm® solution obtained by

dissolving about 1 g of salt in nitric acid (8 M) saturated with boric aci

d.

bCounts for individual nuclides given in disintegrations per minute (dpm) total for chemistry sample types A and B and dpm per gram of salt
for type C chemistry sample. These counts are laboratory numbers and subject to several corrections which have not been made.

“These concentrations are expressed as grams of the particular nuclide per cm?® of metal surface for chemistry sample types A and B and as
grams of nuclide per gram of salt for chemistry sample type C. The values have been corrected back to the conclusion of the irradiation.

dConcentration throught sufficiently low to be ignored.

sample numbers énding with 1 (i.e., 1B1,2B1,and 3B1)
designate the material that came from the fuel pin
wall exposed to the gas space above the salt. The sample
numbers ending with 2 designate material that came
from the fuel pin exposed to the fuel salt just below the
salt-gas interface, and the sample numbers ending with 3
designate material that came from the portion of the
fuel pin exposed to fuel salt near the bottom of the
capsule.. Solutions were prepared for analysis by leach-
ing metal samples of each tube in “verbocit™ to remove
residual salt (type A solution in Table 6.23), leaching
the rings in aqua regia (type B solution in Table 6.23),
and dissolving about 1 g of salt removed from the metal
rings in nitric acid (type C solution in Table 6.23).
These solutions were counted to determine the amounts

of 1*"MTe and '?°™MTe present. The direct results of-

these analyses are presented in Table 6.23, but cannot
be interpreted directly because a number of corrections
have not been made. The data have been corrected as
well as possible to reflect the concentration of each
nuclide at the end of irradiation. The concentrations for
the leaches from the metal specimens are expressed as
grams per square centimeter of tube wall exposed to the
fuel salt, and the concentrations for the salt samples are
expressed as grams per gram of salt.

The ORIGEN code was used by Kerr and Allen to
predict the concentrations of tellurium isotopes that
should have been present. These calculations have been
used extensively in the subsequent analysis of the data.
Table 6.24 compares the quantities of '*7"Te and
129mTe found in the three fuel pins with those pre-
dicted to be present by the ORIGEN calculations. For

130

each fuel pin the one sample taken of the tube in the
gas space was assumed to be typical of that region, and
the two samples from the salt-covered parts were aver-
aged to obtain a typical value for the salt-covered
region. As shown in Table 6.24, generally about 20% of
the 127MTe and '*°"Te was found. The percent of
tellurium found in the Inconel 601 capsule was appreci-
ably higher due to the higher amount found on the
salt-covered metal surfaces.

There are several possible explanations why the con-
centrations of '>7MTe and '2°"Te found are only
about 20% of those produced. One possibility is that
the amounts calculated are too high. This appears not to
be the case, but the calculations will be checked further.
The most likely explanation is that the acid leach was
not sufficient to remove all of the tellurium from the
wall. The tube segments were suspended in the acid
with the inside and outside surfaces of the tube wall
exposed, as well as the cut surfaces on each side of the
%-in. tube segment. Based on the weight changes ob-
served and the assumption of uniform metal removal,
the thickness of metal removed appears to be about 0.8
mil. Since the cracks extended deeper than 0.8 mil in
the Hastelloy N, the tellurium likely penetrated deeper
than did the leaching solution. However, the cracks in
the other two materials were very shallow, and the

0.8-mil dissolution should have recovered a higher frac-
tion of the tellurium if one can equate the depth of
cracking to the depth of tellurium penetration. The
results in Table 6.24 show no evidence of a systematic
variation in the percent recovered from the three tubes. -
Several possible explanations for the apparent discrep-
ancy in the quantities of tellurium generated and that
actually found are being investigated, but none appears
reasonable at this time.

The concentrations of 27" Te and '?° " Te found in
the salt can be used to predict upper limits for the
solubility of tellurium in fuel salt under these condi-
tions. The '27™MTe nuclide concentration in the salt
ranges from 1,14 X 107° to 1.31 X 1078 g per gram of
sait (Table 6.23). The ORIGEN calculations were used
to estimate the ratio of 127" Te to total tellurium, and
this ratio was used to convert the above concentrations
of 127MTe to total tellurium concentrations of 0.07 to
0.83 ppm. Similarly, the concentration of '??™Te
ranged from 4.5 X 107° to 6.48 X 107® g per gram of
salt, and these correspond to total tellurium concentra-
tions of 0.08 to 1.13 ppm. The low values in both cases
were noted in the Inconel 601 pin, and the higher values
were observed in the type 304 stainless steel pin. The
concentrations in the Hastelloy N pin were only slightly
less than noted for the type 304 stainless steel pin. The

Table 6.24. Amount of Tellurium in various locations of
fuel pins from TeGen-1 (g)

Inconel 601 Hastelloy N Type 304
Location stainless steel
127}nTe l‘z9mTe lZTmTe IZBmTe lITmTe 129mTe
Salt 4,1 x10°°® 1.3x 1077 1.0x 1077 3.5 x 1077 1.8 1077 7.4 % 1077
Metal-vapor space 1.4X 1077 24 x 1077 21x1077  50x 1077 20X1077  1.2x10°¢
Metal-salt covered 1.7x 10°¢ 2.3 x10°° 78 x 1077 26 X 10°¢ 4.4 %1077 39x 1077
Total found 1.9x 10°¢ 277 %x10°¢ 1.1 x 10°¢ 345%X 10°¢ 8.2x 1077 2.3x 10°¢
Total formed 3.62X 107¢ 1.34 X 10°% 40x10°¢ 1.48x 1075 3.6x10°¢ 1.34 X 10°°¢
" Percent found 52 20 28 23 23 17
Table 6.25. Concentration of tellurium in various locations
of fuel pins from TeGen-1 (108 g/cm?)
Inconel 601 Hastelloy N Type 304
Location stainless steel
12‘1Te 129Te 127Te 129Te lz‘?Te l‘nge
Metal-vapor space, Bl 2.57 446 3.86 94 3.76 21.4
Metal-salt location, B2 8.75 11.3 1.52 3.76 2.29 1.86
Metal-salt location, B3 3.45 5.76 422 15.1 0.95 1.03
Average if total 11.2 41.3 12,3 45.6 11.2 41.3

yield evenly distributed

131

higher chromium concentration of the Inconel 601 may

have caused the lower tellurium concentration in the
fuel salt. :

The concentrations of '*7”Te and '*°MTe are
expressed in Table 6.25.in terms of grams per unit sur-
face .area. There appear to be significant variations
within each capsule, but there is no consistency be-
tween the various pins. The high value for '**”Te in
the vapor space of the type 304 stainless steel pin is
likely anomalous, since the '?7”Te is not as high.
Thus, at this time we conclude that the tellurium is
distributed uniformly over the entire surface area of the

pin.

6.16 SALT PREPARATION AND FUEL PIN
" FILLING FOR TeGen-2 AND -3

M. R. Bennett = A. D. Kelmers

The purpose of this portion of the TeGen activity is
to prepare purified MSRE-type fuel salt containing
233U and to then transfer a known quantity of this salt
into fuel pins for subsequent irradiation in the ORR.
One batch of purified salt will be prepared and used, in
two filling operations, to fill two sets of six fuel pins
each, identified as TeGen-2 and TeGen-3. Similar activi-
ties in 1972 to fill the fuel pins used in experiment
TeGen-1 have been previously described.?® To MSRE-

type fuel carrier salt containing LiF-BeF,-ZrF,
(64.7-30.1-5.2 mole %), sufficient 233 UQ, and 238 UF,
were added to produce a final composition of LiF-BeF, -
ZrF,-*33UF,-?38UF, (63.08-29.35-5.07-1.00-1.50
mole %) after.hydrofluorination to reduce the oxide
content. The uranium will be reduced by hydrogen, or
by beryllium if necessary, to a U* content of 1.0 to
1.8%, and a measured volume of salt will be transferred
into the fuel pins. The design permits obtaining a pre-
determined volume in the pins by flushing through an
excess salt volume and then blowing back the salt in the
upper portion of the pins to leave a predetermined vol-
ume. .

The equipment in Bfiilding 4508 used previously for
this work was reactivated and modified where appropri-.
ate. A safety summary and step-by-step operating proce-
dure have been prepared and approved. During the lat-
ter part of this report period the salt components were
charged to the salt purification vessel, and a 36-hr
hydrofluorination at 600°C was. completed. Both fil-
tered and unfiltered samples were obtained after hydro-
fluorination .in- copper filter sticks. After analytical
results indicating satisfactory removal of oxide have
been received, hydrogen reduction of about 1% of the
UF, will be carried out.

. 23. R. L. Senn, J. H. Shaffer; H. E. McCoy, and P. N.
Haubenreich, MSR Program Semiannu. Progr. Rep. Aug. 31,
1972, ORNL-4832, pp. 90-94.
7. Fuel Processing Materials Development

J. R. DiStefano

The processes that are being developed for isolation of
protactinium and removal of fission products from
molten-salt breeder reactors require materials that are
corrosion resistant to bismuth-lithium and molten fluo-
ride solutions. Past experience has indicated that, al-
though their solubilities in bismuth are low, iron-base
alloys mass transfer rapidly in bismuth at 500 to 700°C.
"The most promising materials for salt processing are
molybdenum, Ta—10% W, and graphite. Molybdenum
has been tested in a wide range of bismuth-lithium solu-
tions for up to 10,000 hr and has shown excellent com-
patibility. Thermodynamic-data and literature reports
indicate that molybdenum will also be compatible with
molten fluoride mixtures.

Ta—10% W also has excellent compatibility with
bismuth-lithium solutions, but tests are required to
measure its compatibility with molten fluoride salts. A
thermal convection loop has been constructed of
Ta—10% W, and a test with LiF-BeF,-ThF,-UF,
(72-16-11.7-0.3 mole %) will be started during the next
reporting period.

Graphite has shown excellent compatibility with both
bismuth-lithium solutions and molten salts. Although
no chemical interaction between bismuth-lithium solu-
tions and graphite has been found, the liquid-metal solu-
tion tends to penetrate the open porosity of graphite.
Recent tests have evaluated the extent of penetration as
a function of structure of the graphite and the lithium
concentration of the bismuth-lithium solution. Dynamic
tests of graphite with bismuth-lithium have thus far
been limited to quartz loop tests circulating Bi—0.01 wt
% (0.3 at. %) Li. During the report period a test was

H. E. McCoy

completed in which graphite samples were exposed to
Bi—2.4 wt % (42 at. %) Li in a molybdenum thermal
convection loop for 3000 hr at 600 to 700°C.

7.1 STATIC CAPSULE TESTS OF GRAPHITE
WITH BISMUTH AND
BISMUTH-LITHIUM SOLUTIONS

J. R. DiStefano

Samples of graphite with varying densities and pore
diameters were exposed to Bi—0.17 wt % (4.8 at. %) Li
and Bi—3 wt % (48 at. %) Li in capsule tests for 3000 hr
at 650°C. Two of the graphites (Table 7.1) were pitch
impregnated to increase their densities and reduce their
pore sizes.! The relatively high densities of these graph-
ites indicate that impregnation was effective, but the
pore size distribution in the samples shows that some of
the larger pores were unfilled or only partially filled.
Specimens were graphite rods 6 mm (0.24 in.) X 38.1
mm (1.5 in.) long that were threaded into an ATIJ
graphite holder. The specimens and holder fit into a
graphite capsule which contained the bismuth-ithium
solution (Fig. 7.1). The entire assembly was sealed in a
stainless steel outer capsule by welding in argon. Sam-
ples exposed to Bi—0.17 wt % (4.8 at. %) Li showed
little evidence of penetration except in low-density
areas (Fig. 7.2). Samples exposed to Bi—3 wt % (48 at.
%) Li were penetrated more uniformly, and the depth

1. All graphites were fabricated by C. R. Kennedy of the
Carbon and Graphite Group, Metals and Ceramics Division,
ORNL.

Table 7.1. Penetration of graphite by bismuth-lithium solutions

in capsule tests for 3000 hr at 650°C

Maximum pore

Graphite Bul}( Range.of diameter that Penetration (mils)
. e s density pore diam contributes 10% - - -
identification (g/em?) (1) to total porosity Bi-0.17% Li ~ Bi-3% Li
()
33-6K 1.84 0.1-1 1 0 5
44-25K49 1.84 0.1-2 1.2 0-172 8
33-38K 1.80 0.1-2 1.5 05> 5
44-26K9 1.80 0.1-3.5 1.5 0-2b . 8
44-23K 1.59 0.1-4.5 4.5 0—2b 15
ZImpregnated.
bNonuniform, penetration in one or two areas only.

132
133

ORNL-DWG 75-14509

rt—— STAINLESS STEEL
OUTER CONTAINER

ATJ GRAPHITE
CAPSULE

GRAPHITE ROD
SPECIMEN
(0.24-in-0CIAM x
1.5-in. LONG)

Bi—-Li

INCHLS |

Fig. 7.1. Graphite (bismuth-lithium) capsule test assembly.

of penetration increased with increasing pore size and
decreasing density. Results from previous tests have
been inconclusive as to the effect of lithium concentra-
tion in bismuth on penetration of graphite. In the cur-
rent series all graphites were penetrated to a greater
extent by Bi—3 wt % (48 at. %) Li than by Bi—0.17 wt
% (4.8 at. %) Li. Tests of 10,000 hr duration with these
graphites are continuing. -

7.2 THERMAL GRADIENT MASS TRANSFER
TEST OF GRAPHITE IN A MOLYBDENUM LOOP

J. R. DiStefano

Although graphite has low solubility in pure bismuth
(less than 1 ppm at 600°C), capsule test results have
shown that higher carbon concentrations are present in
Bi—2 wt % (38 at. %) Li and Bi—3 wt % (48 at. %) Li
solutions after contact with graphite. To avoid the
joining problems associated with fabrication of a graph-
ite loop, a molybdenum loop was constructed, and
interlocking, tabular graphite specimens were suspended

in the vertical hot- and cold-leg sections.? In addition to
mass transfer of graphite from hot- to cold-leg areas,
penetration of graphite by bismuth-lithium and mass
transfer between graphite and molybdenum were evalu-
ated.

7.2.1 Weight Changes

The loop (CPML4) circulated Bi—2.4 wt % (42 at. %)
Li for 3000 hr at 700°C (approximately) maximum
temperature and 600°C minimum temperature. Weight
changes in the graphite samples are given in Tables 7.2
and 7.3. After the bismuth-lithium solution was drained
from the loop, the samples were removed and weighed
(“after-test” column in Tables 7.2 and 7.3). Sub-
sequently, they were cleaned at room temperature in
ethyl alcohol and in an H, O—HNO, (100 ml H,0-30
ml 90% HNO,) solution to remove bismuth-lithium
adhering to the surfaces of some samples. Samples from
the cold leg were weighed and then kept in air for two

days prior to the alcohol treatment. After soaking in

alcohol, these samples showed larger weight gains than
the after-test weight gains, and this is attributed to reac-
tion of lithium in the sample with moisture in the air
during the two-day period. All samples showed large
weight gains (33—67%), and gains in hot-leg samples
were, on the average, larger than those in the cold-leg
samples.

7.2.2 Compositional Changes

Graphite samples were analyzed before and after
treatment with H, O—HNO,, and the results are shown
in Table 7.4. These results indicate that bismuth was
primarily responsible for the large weight increases and
that samples picked up molybdenum, but treating them
with H,O-HNO; completely removed the molyb-
denum. An electron-beam microprobe analysis of a
graphite sample before acid cleaning showed that
molybdenum was present on the outer surface of the
specimen (Fig.-7.3). Chemical analyses of other graphite
samples after acid cleaning are shown in Table 7.5.

Analyses of bismuth-lithium samples from the loop
are shown in Table 7.6. For sampling, the hot leg was
sectioned so that one sample came from the surface that
was in contact with the molybdenum tube wall while
the other sample was taken from the interior of the
section, away from the wall. The concentration of
carbon in the melt was highest in the sample from the
hot leg, and both molybdenum and carbon concentra-

2. 1. R. DiStefano, MSR Program Semiannu. Progr. Rep.
Feb, 28, 1975, ORNL-5047, pp. 140—-41.
o =1.8g/cm’ 1.8g/cm
~75% of pores = 1 diam ~25% of pores = 1-4y diam
8i-0.17 Li

Fig. 7.2. Penetration of graphite as a function of structure of graphite and lithium in bismuth. Conditions: 1000 hr at 650

135

Table 7.2. Weight increases in ATJ graphite
hot-leg samples from CPML-4

Weight (g)

Weight
:Sfiglei Befaore After A’fte‘r A'fte.r increase
test test .cleanmg . cleaning @ %
in alcohol in H, O-HNO,
59 0.4563 0.8244 0.8122 0.7006 0.2443 54
7 0.5220 0.9706 0.9656 0.8298 0.3078 59
9 0.4994 0.9551 0.9336 0.8298 0.3304 66
10 0.4800 0.9596 09819 0.7944 0.3144 66
11 0.4326 0.8339 0.8303 0.7102 0.2776 64
12 0.4594 0.9302 0.9263 0.7669 0.3075 67
13 0.4753 0.9796 0.9764 0.7686 . 0.2933 62
16 0.4632 09175 09152 0.7658 0.3026 65
17 04742 0.9099 0.9058 0.7694 0.2952 62
18 4.5070 0.9005 0.8956 0.7975 0.2905 57
19 0.4709 0.8189 0.8142 0.7168 (.2459 52
20 0.5369 -0.9184 0.9149 0.8107 0.2738 51
21 0.5163 0.8652 0.8629 0.7626 0.2453 48
22 0.5184 0.8499 0.8459 0.7591 0.2407 46
23 0.5389 0.9103 0.9066 0.7475 0.2067 38
240 0.5405 0.8545 0.8497 (.7235 0.1830 33
4Tgp of hot leg; temperature: 680—700°C.
bBottom of hot leg; temperature: 600—620°C.
Table 7.3. Weight changes in ATJ graphite
cold-leg samples from CPML-4
Weight (g)
After Weight
Sample Before After ’ standing in . Af.ter , increase
number test test air for two days cleaning in © %)
and then cleaning H, O-HNO,
in alcohol
S 274 3.4614 0.7469. 0.7622 0.6408 0.1794 39
28 0.4821 0.8284 0.8412 0.6260 0.1439 30
29 0.4717 0.7793 0.7913 0.6433 0.1716 36
30 04776 - 0.7136 0.7239 0.6291 0.1515 32
31 0.4848 0.7209 0.7315 0.6396 0.1548 32
32 0.4827 0.7629 0.7744 0.6647 0.1820 38
33 0.4788 0.7193 0.7300 0.6331 0.1543 32
34 0.4708 0.7566 0.7672 0.6536 0.1828 39
35 0.4624 0.7382 0.7485 0.6290 0.1666 36
36 0.4823 0.7496 0.7598 0.6343 0.1520 32
37 0.4667 0.7331 0.7432 .0.6239 0.1572 34
38 04713 0.7513 0.7619 0.6418 0.1705 36
39 0.4745 0.7498 0.7603 0.6506 0.1761 37
40 0.4628 0.7285 0.7390 0.6233 0.1605 35
41 0.4725 0.7458 0.7553 0.6419 0.1694 36
42 0.4800 0.7427 0.7525 0.6526 0.1726 36
43b 04766  0.7280 0.7401 0.6478 01712 36

2Top of cold leg; temperature: 660—680°C.
bRottom of cold leg; temperature: 620—630°C.
136

Table 7.4. Chemical analyses of graphite tions were higher than were found previously in quartz

exposed to bismuth-lithium solution loop tests circulating Bi-0.01 wt % (03 at. %) Li.?

ina mlybdeaim lodp Quartz loop test 11 contained molybdenum samples,

and analysis of the bismuth-lithium solution after test

Sample number Condition ;‘v% showed that it contained 2.5 ppm molybdenum. Quartz
: : loop 8 contained samples of three different grades of

4 (hot leg) Uncleaned 46 0.6 0.1 graphite, and the bismuth-ithium solution contained 10

4 (hot h‘;» Acid cleaned 03 <0.01  to 15 ppm carbon after the test.

26 (cold leg) Uncleaned 43 1.2 O e e e

26 (cold leg) Acid cleaned 40 04 <0.01 3. 0. B. Cavin and L. R. Trotter, MSR Program Semiannu.

Progr. Rep. Feb. 28, 1971, ORNL4676, p. 228.

Y-133056

40 pm

BACKSCATTERED ELECTRONS

Bi Mg X-RAYS Mo LG X-RAYS

Fig. 7.3. Electron beam scanning images of graphite exposed to bismuth. View (4) is backscattered electron picture of sample
surface; dark material is graphite and bright material is bismuth and molybdenum as indicated by (8) and (C)

Table 7.5. Chemical analyses of
graphite samples? from CPML-4

Sample ] Concentration (wt %)
Location
number Bi Li Mo
1 Hot leg 37 0.35 <0.01
9 Hot leg 42 0.44 <0.01
15 Hot leg 46 0.47 <0.01
19 Hot leg 36 0.35 <0.01
24 Hot leg 23 0.25 <0.01
27 Cold leg 36 0.35 <0.01
31 Cold leg 21 0.25 <0.01
36 Cold leg 19 0.26 <0.01
42 Cold leg 23 0.26 <0.01
?Samples acid cleaned in H, O-HNO, prior to analysis.
Table 7.6. Analysis of bismuth-lithium
solution from CPML-4
Concentration?
Sample location | C Mo Li
(ppm) (ppm) (%)
Hot leg (core)? 43 17 2.3
Hot leg (surface)® 106 102 3.0
Cold leg (core)? 24 14 1.6

20n a weight basis.
binterior sample.

€Surface sample in contact with molybdenum tube wall.

7.2.3 Microstructural Changes

Selected graphite samples from hot- and cold-leg
regions are shown in Fig. 7.4. The white phase distrib-
uted throughout the samples is bismuth. These samples
were acid cleaned and it is evident that bismuth was
dissolved from the area near the surface. Molybdenum
samples from hot- and cold-leg regions are shown in Fig.
7.5. Surface layers measuring 0.015 to 0.025 mm
(0.6—1 mil) thick were found on the hot-leg sample. In
some areas there was a single layer, while a double layer
was found in other areas. Electron-beam microprobe
analysis indicated the single layer and/or outer layer to
be primarily molybdenum. This layer was much harder
than the base metal (1000-1200 DPH compared with
about 200 DPH), indicating that it is probably Mo, C.
Where there is a double layer the outer layer appears to
be Mo, C, but the inner layer is primarily bismuth. One
explanation is that the Mo,C layer cracked and/or
spalled, allowing the entry of bismuth which did not
drain when the test was terminated. The molybdenum
sample from the cold leg also exhibited a surface layer

less than 0.1 mil thick that was similar in composition
to that found in the hot leg. Samples of molybdenum
from hot and cold legs have been submitted for chemi-
cal analysis.

7.2.4 Discussion of Results

The principal objective of this experiment was to eval-
uate temperature-gradient mass transfer of graphite in
bismuth containing a relatively high concentration of
lithium. However, mass transfer data were obscured by
the gross pickup of bismuth by the graphite samples.
Previous capsule and quartz loop tests with ATJ graph-
ite had indicated much less intrusion of the graphite by
bismuth than occurred in the molybdenum loop test.
This suggests that the permeability of ATJ graphite to
bismuth-lithium does not depend simply on the po-
rosity of the graphite. It is generally accepted that some
fraction of the pores in graphite is effectively sealed off
and contributes nothing to flow. Therefore, the con-
nected pore system controls the permeability, The
shape of the connected pores influences the type of
flow and the length of the path the fluid takes through
the sample. For a nonwetting liquid, the external pres-
sure forcing the liquid into the pores, AP, must over-
come the surface tension of the liquid. This defines a
critical pore radius, 7., and until the pressure exceeds
the value given by

AP =2y cos 8/r, (1)

where 7 is the surface tension and 8 the wetting angle,
the pore cannot support flow. Thus, for a given AP, 7. is
the minimum pore size that will be penetrated. In both
the metal and quartz thermal convection loops AP is
determined by the argon overpressure (<1 atm) and the
height of bismuth-lithium solution above the sample,
and these were essentially the same in both types of
tests. Temperature affects both ¢ and @0, but all of the
tests were operated under similar time-temperature-AT
conditions. Graphite samples used in the quartz loop
tests had almost four times the surface area of the tabu-
lar specimens used in the current test, but they were
almost three times as thick. The larger surface area of
the quartz loop specimens should have increased the
relative amount of bismuth-lithium intrusion, but the
greater thickness of these samples would reduce the per-
centage increase. ATJ graphite samples from the quartz
loop tests increased in weight by 0.1 to 0.6 wt %, far
less than the 30 to 67 wt % increases noted in samples
from the current loop test. Thus, specimen geometry
alone does not seem to explain the differences noted.
However, the surface tension o and wetting angle § were
Hot Lea No. 10 Hot Leg No. 13

Fig. 7.4. Graphite samples after exposure to Bi—2.4% Li for 3000 hr in CPML-4.

Y-132704
Hot Leg No, 23

600 700
i -8 B

il

"
z
S
&
o
H

100 200

8€T

Y-133405

60 MICRONS

——————1—500x—L

40

S == P
e o _— '
= N = — B3 -
M S
= 9 -
et e > 3 3 S
= i —= HE :
L ~ e —— S AA = = - -
Top of Hot Leg: 700°C g Top of Hot Leg: 700°C
5
g

0.001

Bottom of Hot Leg: 600°C

specimens.

probably different because the lithium concentrations
of the bismuth-ithium solutions were different, and
molybdenum was present in the current test. It is pos-
sible that the presence of molybdenum on the surfz
of the graphite had a marked effect on the contact angle
0. In an earlier series of tests, the bismuth content of

e

graphite specimens was much higher when they were
tested in molybdenum capsules instead of graphite cap-

Bottom of Cold Leg: 620°C

. Molybdenum tube wall from thermal convection loop CPML-4 that circulated Bi-2.4% Li and contained graphite

sules.* Accordingly, data on the wetting of graphite by
bismuth containing lithium and other constituents of
processing solutions would be useful for predicting the

resistance of graphite to penetration.

4. J. R. DiStefano and O. B. Cavin, MSR Program Semiannu.
Progr. Rep. Feb. 28, 1975, ORNL-S047, pp. 137 -39,

Part. 4. Fuel Processing for Molten-Salt Reactors

J. R. Hightower, Jr.

The activities described in this section deal with the
development of processes for the isolation of protac-
tinium and for the removal of fission products from
moltensalt breeder reactors. Continuous removal of
these materials is necessary for molten-salt reactors to
operate as high-performance breeders. During this
report period, engineering development progressed on
continuous fluorinators for uranium removal, the metal
transfer process for rare-earth removal, the fuel recon-
stitution step, and molten salt—bismuth contactors to
be used in reductive extraction processes. Work on
chemistry of fluorination and fuel reconstitution was
deferred to provide experienced personnel for the prep-
aration of salt for the TeGen-2 and -3 experiments
(Sect. 6.17).

The metal transfer experiment MTE-3B was started.
In this experiment all parts of the metal transfer process
for rare-earth removal are demonstrated using salt flow
rates which are about 1% of those required to process
the fuel salt in a 1000-MW(e) MSBR. This experiment
repeats a previous one (MTE-3) to determine the
reasons for the unexpectedly low mass transfer coeffi-
cients seen in MTE-3. During this report period the salt
and bismuth phases were transferred to the experi-
mental vessels, and two runs with agitator speeds of 5
rps were made to measure the rate of transfer of neo-
dymium from the fluoride salt to the Bi-Li stripper solu-
tion. However, in these runs the fluoride salt was en-
trained at low rates into the LiCl, which resulted in
depletion of the lithium from the Bi-Li solution in the
stripper. Fuel-salt entrainment was unexpected, since no
entrainment was seen in experiment MTE-3 under (as
far as can be determined) identical conditions. The
measurement of mass transfer coefficient in these first
two runs was not compromised by the entrainment. The
measured mass transfer coefficients were lower than

predicted by literature correlations, but the values are
comparable to those obtained from experiment MTE-3.

Mechanically agitated nondispersing salt-metal con-
tactors of the type used in experiment MTE-3B are of
interest, because entrainment of bismuth into the fuel
salt can be minimized, because very high ratios of bis-
muth flow rate to salt flow rate can be more easily
handled than in column-type contactors, and because
these contactors appear to be more easily fabricated
from molybdenum and graphite components than are
column-type contactors. Attempts were made to meas-
ure entrainment rates of fluoride salt in bismuth and
entrainment rates of bismuth in fluoride salt under con-
ditions where the phases were not dispersed and under
conditions where some phase dispersal was expected.
These measurements were made in the 6-in.-diam
(0.15-m) contactor installed in the Salt-Bismuth Flow-
through Facility. The results indicate that mild phase
dispersal with its concomitant high mass transfer coeffi-
cients might be allowable in the reductive extraction
processes. We are continuing development of methods
for measuring mass transfer coefficients in mercury-
water systems to learn how to scale up contactors which
would be used with salt and bismuth.

A nonradioactive demonstration of frozen salt cor-
rosion protection in a continuous fluorinator requires a
heat source that is not subject to attack by fluorine in
the fluorinator. To provide such a heat source for future
fluorinator experiments we have continued our studies
of autoresistance heating of molten salt. During the
report period we have completed new equipment for
studying autoresistance heating of molten salt in a flow

* system similar to a planned continuous fluorinator ex-

140

periment; three preliminary runs have been made with
the equipment. The design was started for a facility for
developing continuous fluorinators, and equipment is
being installed for an experiment to demonstrate the
effectiveness of frozen salt for protection against fluo-
rine corrosion.

The uranium removed from the fuel salt by fluorina-
tion must be returned to the processed salt in the fuel
reconstitution step before the fuel salt is returned to the
reactor. An engineering experiment to demonstrate the
fuel reconstitution step is being installed. In this experi-
ment gold-lined equipment will be used to avoid intro-
ducing products of corrosion by UF¢ and UF;. Alterna-
tive methods for providing the gold lining include elec-
troplating and mechanical fabrication. The choice be-
tween the two depends on availability of gold from
ERDA precious-metal accounts and the price of gold
from the open market. Instrumentation for the analysis

141

of the vessel off-gas streams has been installed and is
being calibrated.

Future development of the fuel processing operations
will require a large facility for engineering experiments.
A design report is being prepared to define the scope,
estimated design and construction costs, method of
accomplishment, and schedules for a proposed MSBR
Fuel Processing Engineering Center. The building will
provide space for preparation and purification of salt
mixtures, for engineering experiments up to the scale
required for a 1000-MW(e) MSBR, and for laboratories,
maintenance areas, and offices. The estimated cost of
this facility is $15,000,000; and authorization is pro-
posed for FY 1978.
8. Engineering Development of Processing Operations

* J. R. Hightower, Jr.

8.1 METAL TRANSFER
PROCESS DEVELOPMENT

H. C. Savage

During this report period the salt and bismuth solu-
tions were charged to the process vessels of the metal
transfer experiment MTE-3B.! Two experiments were
completed in which the rate of removal of neodymium
from molten-salt breeder reactor fuel salt (72-16-12
mole % LiF-BeF,-ThF,) was measured.

The MTE-3B process equipment (Fig. 8.1} consisted
of three interconnected vessels: a 14-in.-diam (0.36-m)
fuel salt reservoir, a 10-in.-diam (0.25-m) salt-metal con-
tactor, and a 6-in.-diam (0.15-m) rare-earth stripper.
The salt-metal contactor is divided into two compart-
ments interconnected through two 0.5-in.-high (13-mm)

1. H. C. Savage, Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. 20, ORNL-
TM-4870 (in preparation).

by 3-in.-wide (76-mm) slots in the bottom of the
divider. Bismuth containing thorium and lithium is cir-
culated through the slots. Thus fluoride fuel salt was in
contact with the Bi-Th in one compartment, and LiCl
was in contact with the Bi-Th in the other compart-
ment. The stripper contains lithium-bismuth solution
(595 at. %) in contact with the LiCl. Mechanical agita-
tors having separate blades in each phase in the con-
tactor and stripper were used to promote mass transfer
across the three salt-metal interfaces. The fluoride fuel
salt was circulated between the reservoir and contactor
by means of a gas-operated pump with bismuth check
valves. The LiCl was circulated between the stripper and
contactor by alternately pressurizing and venting the
stripper vessel.

The bismuth-thorium phase was circulated between
the two compartments of the contactor by the action of
the agitators, and no direct measurement of this flow
rate was made during the experiment; however, meas-
urements made in a mockup using a mercury-water
system indicated that the Bi-Th circulation rate between

ORNL-DWG-71-147-RI

‘ AGITATORS
VENT LEVEL
ELECTRODES LEVEL
ELECTRODES
FLUORIDE
" SALT PUMP VENT
—— E —— [ —
ARGON
SUPPLY
ARGON
SUPPLY
125
V2 -7 AV 77 Loin | -
s - . 4 b - < /
\‘\—ucr~/’4
72-16-12 mole % Bi-Th \\ Li-Bi
LiF'Ber‘ThF4
FL%%?PE SALT - METAL RARE EARTH
RESERVOIR CONTACTOR STRIPPER

Fig. 8.1. Flow diagram for metal transfer experiment MTE-3.

142
the two compartments should be high enough to keep
the concentration of rare earths in both compartments
essentially the same.> This was found to be the case in
the two experiments in MTE-3B.

In this experiment, neodymium is extracted from the
fuel carrier salt into the thorium-bismuth solution.
Next, the neodymium is extracted from the thorium-
bismuth into molten LiCl, and finally, the neodymium
is stripped from the LiCl into bismuth-lithium alloy.

Operating variables in the experiment are:

. the flow rate of the fluoride fuel salt between the
fuel salt reservoir and the contactor,

. the flow rate of the lithium chloride salt between the
contactor and the stripper vessel,

. the degree of agitation of the salt and bismuth
phases in the contactor and stripper,

. the amount of reductant (lithium) in the bismuth
phase in the contactor.

The operating temperature of the system is Vv650°C.
Overall mass transfer rates for representative rare-earth
fission products are determined by adding the rare earth
to the fluoride fuel salt in the reservoir and observing
the rate of transfer of the rare earth across the three
salt-bismuth interfaces as a function of time by periodic
sampling of all phases.

2. H. O. Weeren and L. E. McNeese, Engineering Develop-
ment Studies for Molten-Salt Breeder Reactor Processing No.
10, ORNL-TM-3352 (September 1974) pp. 57-59.

143

During the course of the experiments the concentra-
tions of neodymium in each phase were determined by
counting the 0.53-MeV gamma radiation emitted by
147Nd tracer added to the neodymium originally in the
fuel salt. This provided a rapid method for following the
transfer rate. More accurate data necessary for calculat-
ing the overall mass transfer coefficients at each of the
three salt-metal interfaces were obtained by analyzing
samples of the salt and bismuth phases for total neo-
dymium via an isotopic dilution mass spectrometry
technique. Use of this technique allows measurement of
neodymium concentrations as low as 0.01 ppm (wt).

8.1.1 Addition of Salt and Bismuth Phases to Metal
Transfer Experiment MTE-3B

The quantities of salts and bismuth charged to the
process vessels of experiment MTE-3B are listed in
Table 8.1. All internal surfaces of the carbon-steel ves-
sels were hydrogen treated at 650°C for 7 hr to re-
move.any oxides prior to the addition of the salt and
bismuth solutions. The auxiliary charging vessels used in
the additions were also hydrogen treated. Subsequently
a purified argon atmosphere was maintained in all the
vessels to prevent oxide contamination (via ingress of air
or moisture) of the vessels and process solutions.

The charging vessels were 10-in.-diam (0.25-m)
carbon-steel vessels of about 22 liters (0.022 m®) in
volume equipped with .electric heaters for melting the
salts and bismuth. Nozzles and access ports were pro-

Table 8.1. Quantities of salts and bismuth for experiment MTE-3B

Volume? at Weicht
Material Vessel 650°C (kg) g-moles
" (liters) E
Fluoride fuel salt? Reservoir - 29.4 97.0 1535
(72-16-12 mole % LiF-Bel', -ThF,) ‘ '
Fluoride fuel salt Contactor 3.1 10.2 161
(72-16-12 mote % LiF-BeF,-ThF,)

Bismuth-thorium [~1500 ppm (wt) Th, Fluoride salt .29 27.6 132
~50 ppm Li] side of contactor '
Bismuth-thorium [~1500 ppm (wt) Th, LiCl side of 3.5 33.8 161

~50 ppm Li] contactor
Lithium chloride Contactor 2.9 4.3 101
Lithium chloride Stripper 3.8 5.6 132
Bismuth-5 at. % lithium in " Stripper 4.3 41.8 200

stripper

2Densities at 650°C: fluoride fuel salt = 3.30 g/cc; LiCl = 1.48 g/ce; Bi = 9.66 g/cc.

bmole weight = 63.2 g.

144

vided for the addition of the salts and bismuth, argon
and hydrogen purge gas lines, and lines required to
transfer the salt and bismuth phases into the process
vessels.

Bismuth, hydrogen treated in the charging vessel to
remove oxides, was the first material to be added to the
contactor. The fluoride fuel salt was then contacted in
the charging vessel (using argon sparging) with a
bismuth—0.15 wt % thorium solution (50% of Th satu-
ration) for several days prior to transfer into the fuel-
salt reservoir and the fluoride salt compartment of the
contactor. Thorium metal (0.1197 kg) was then added
to the 61.4 kg of bismuth in the contactor. This quan-
tity of thorium is about 50% of the amount that would
be soluble and was calculated to produce a lithium con-
centration of 40 ppm (wt) in the thorium-bismuth
phase in the contactor based on previously reported
data® on the distribution of thorium and lithium be-
tween molten bismuth and fluoride fuel salt.

Following the additions of bismuth to the contactor
and the fluoride fuel salt (72-16-12 mole % LiF-BeF,-
ThF,;) to the contactor and fuel-salt reservoir, a new
charging vessel was installed for makeup and charging of
the bismuth—-5 at. % lithium to the stripper and the
LiCl to the contactor and stripper. First, bismuth was
added to the charging vessel and was hydrogen treated
to remove oxides by sparging with hydrogen at ~600°C
(873°K) for ~7 hr. The charging vessel contained 67.87
kg of bismuth to which was added 0.120 kg of lithium
metal to produce the bismuth—5 at. % lithium for the
stripper. Part of the bismuth--5 at. % lithium solution
(41.8 kg) was then transferred into the stripper vessel.

Thorium metal (0.109 kg) was added to the 26 kg of
bismuth-lithium solution remaining in the charge vessel,
and 15.88 kg of LiCl that had been oven dried at 200°C
(473°K) was added to the charge vessel. The bismuth-
lithium-thorium and LiCl phases were sparged with

argon, using a gas-lift sparge tube, for four days. The"

LiCl was then transferred into the LiCl side of the con-
tactor and the stripper vessel.

The salt and bismuth solutions were filtered through
molybdenum filters [~30 u (3.0 X 107> m) in pore
diameter| installed in the transfer lines during transfer
from the charging vessels into the MTE-3B process
vessels.

8.1.2 Run Nd-1

For the first run in MTE-3B, 3300 mg of NdF; (2360
mg of Nd) was added to the 97 kg of fluoride fuel salt
(72-16-12 mole % LiF-BeF,-ThF,) in the fuel salt reser-
voir on June 6, 1975, The neodymium contained 72.2
mCi of '*7Nd tracer (r,, = 11 days) at the time of

addition. The neodymium concentration in the fuel salt
in the reservoir was calculated to be 24 ppm (wt), which
approximates that expected in the fuel salt of a single-
region 1000-MW(e) MSBR. Neodymium was chosen as
the representative rare-earth fission product for the first
series of experiments in MTE-3B for several reasons:

1. results could be compared with those obtained using
neodymium in the previous experiment,* MTE-3,

2. "7Nd tracer used for following the rate of transfer
of neodymium has a relatively short half-life (11
days), which would prevent excessive levels of radio-
activity in the experimental equipment as additional
neodymium, containing '*7Nd, was added to the
fuel salt during the experiment.

3. neodymium is one of the more important trivalent
rare-earth fission products to be removed from
MSBR fuel salt.

An attempt was made to start the first run (Nd-1) on
June 9, 1975. However, a malfunction in the electronics
of the speed control unit for the stripper-vessel agitator
prevented startup. After this unit was repaired, run
Nd-1 was started on June 15, 1975, and the scheduled
period of operation (100 hr) was completed on June 20,
1975. Operating conditions of run Nd-1 were: 650 to
660°C (923 to 933°K), 5 rps agitator speeds in both
contactor and stripper, fluoride salt flow rate of 35
cc/min (5.8 X 1077 m3/sec), and LiCl flow rate of 1.2
liters/min (2.0 X 107° m?3/sec).

After 100 hr of fluoride salt and LiCl salt circulation,
the fluoride salt circulation was stopped and the run
was continued for 16 hr. This was done to observe the
expected larger decrease in the concentration of neo-
dymium in the smaller amount of fluoride salt in the
contactor (10.2 kg), as compared with the 107.2 kg
contained in both the contactor and reservoir. These
data would provide a more accurate measure of the rate
of transfer of neodymium across the fluoride salt—
bismuth—thorium interface.

Finally, the circulation of LiCl was also stopped. The
agitators in the contactor and stripper vessels were then
operated for 24 hr over a three-day period (8 hr each
day) to allow the salt and bismuth phases to equilibrate
in an attempt to determine neodymium distribution
coefficients between the phases.

3. L. M. Ferris, “Equilibrium Distribution of Actinide and
Lanthanide Elements Between Molten Fluoride Salts and Liquid
Bismuth Solutions,” J. Inorg. Nucl. Chem. 32, 2019-35
(1970).

4, Chem. Technol. Div. Annu. Progr. Rep. March 31, 1973,
ORNL-4883, p. 25.

The experimental equipment operated satisfactorily
throughout run Nd-1. All operating variables were main-
tained at desired conditions. Results obtained during
run Nd-1 are discussed in Sect. 8.1.4.

8.1.3 Run Nd-2

Run Nd-2 was done with the same operating condi-
tions as run Nd-1 except for run duration (139 hr in-
stead of 100 hr). Prior to run Nd-2, 3590 mg of NdF,
(2580 mg of Nd) containing 101 mCi of '*7Nd tracer
was added to the 97 kg of fuel salt in the reservoir.
Including the neodymium remaining in the fuel salt at
the end of run Nd-1, estimated to be 18 ppm (wt), the
neodymium concentration in the fuel salt in the reser-
voir at the start of run Nd-2 is estimated to be 45 ppm.
The neodymium concentration in the fuel salt in the
contactor is estimated to be 9 ppm at the start of run
Nd-2. We are uncertain of the amounts of neodymium
in the other phases at the beginning of run Nd-2, as
discussed in Sect. 8.1.4.

Run Nd-2 was started on July 13, 1975, and was ter-
minated on July 19, 1975, after 139 hr of operation.
During the first 50 hr of operation the rate of transfer
of neodymium into the lithium-bismuth phase in the
stripper appeared to be about the same as observed
during run Nd-1, based on counting of the !*7Nd tracer
in samples taken at regular intervals. After about 60 hr
of operation, the transfer of neodymium into the
bismuth-lithium phase in the stripper suddenly stopped,
and it was observed that neodymium was being ex-
tracted from the bismuth-lithium phase in the stripper
into the LiCl in the stripper and contactor. During the
run a significant decrease in the emf between the strip-
per vessel and the contactor occurred (from 160 mV
to ~25 mV over a 30-hr period), indicating loss of lith-
ium reductant from the bismuth-lithium phase. The run
was terminated after 139 hr of operation, when it be-
came clear that useful information could no longer be
obtained and it appeared that fluoride salt was being
entrained into the LiCl in the contactor.

8.1.4 Discussion of Results

Subsequent investigation and results of chemical anal-
yses of samples of the salt and bismuth phases indicate
“that fluoride fuel salt was being entrained into the LiCl
in the contactor throughout both runs Nd-1 and Nd-2.
Estimates of the amount entrained are shown below:

Estimated amount of fluoride salt
transferred into LiCl

Nd-1 (104 hr) Nd-1 and -2 (301 hr)

Basis of estimate

0.292 kg Fluoride in LiCl phase
Thorium in Bi-Li phase
Increase in LiCl level in

stripper

0.607 kg
0.400 kg

145

Based on fluoride analyses of LiCl samples taken during
run Nd-1, the entrainment of fluoride salt appears to
have occurred at a relatively constant rate throughout
the run. The total amount of neodymium which trans-
ferred into the Li-Bi phase in the stripper during run

-Nd-1 is estimated to be 300 mg. The amount of neo-

dymium contained in the entrained fuel salt is estimated
to be 6 mg. Thus, most of the neodymium which trans-
ferred into the Li-Bi in the stripper vessel was by mass
transfer rather than as a result of entrainment.

The reason for the observed entrainment is not clear

at present. One explanation is that the 5.0-rps agitator

speed is sufficient to cause entrainment (entrainment of
fluoride salt into the chloride salt occurred in the pre-
vious experiment MTE-3 at 6.7 rps, but not at 5.0 rps).
Experiments are in progress for determining whether
this explanation is correct, and results to date indicate
that entrainment does not occur at 3.3 rps. Further
experiments in MTE-3B will depend on determining the
reason for the unexpected entrainment of fluoride salt
into the LiCl. However, it appears feasible to continue
rare-earth mass transfer experiments in MTE-3B by
removing the LiCl (contaminated with fluoride salt) and
the Li-Bi solution from the stripper vessel, after which
purified LiCl and Li-Bi solution will be added to the
system.

The main purpose of the metal transfer experiment is
to measure mass transfer coefficients for the rare earths
at the various salt-metal interfaces in the system and to
determine whether a literature correlation® (based on
studies with aqueous-organic systems) which relates
many transfer coefficients to the agitator speed and
other physical properties of the system is applicable to
molten salt—bismuth systems. Data obtained from run
Nd-1 have been analyzed, and estimates have been made
of overall mass transfer coefficients for neodymium at
the three salt-metal interfaces. Even though entrainment
of fluoride salt into LiCl occurred during run Nd-1, it is
believed that the mass transfer rate for neodymium was
not significantly affected. The concentration of fluoride
in the LiCl at the end of run Nd-1 was ~1.3 wt % or
0.03 mole fraction. Based on previous studies,’® the dis-
tribution coefficient, D;,, for neodymium between the
molten bismuth-thorium solution and LiCl (mole frac-
tion Nd in bismuth/mole fraction Nd in LiCl), would be
decreased by ~20%, while the distribution coefficient
for thorium would be decreased by a factor of V150.
This would result in a decrease in the separation factor

5. 1. B. Lewis, Chem. Eng. Sci. 3,248-59 (1954).

6. L. M. Ferris et al., “Distribution of Lanthanide and
Actinide FElements Between Liquid Bismuth and Molten
LiCl-LilF and LiBr-LiF Solutions,” J. lnorg. Nucl. Chem. 34,
313-20 (1972).
146

between neodymium-thorium from ~10* to ~102%,
with an increase in the amount of thorium transferred
into the LiCl.

The rate of transfer of neodymium across the three
salt-metal interfaces was determined by analyses of sam-
ples taken throughout the run. Two analytical methods
were used: (1) counting of the 0.53-MeV gamma radia-
tion emitted by the !*7Nd tracer and (2) isotopic dilu-
tion mass spectrometry. Based on counting of the
147Nd tracer, a material balance of the neodymium of
>95% obtained at the end of run Nd-1 indicated that
about 13% of the neodymium originally added to the
fuel salt reservoir had been transferred into the Li-Bi
solution in the stripper.

The counting technique is a rapid method and pro-
vided information on the rate of transfer while the runs
were in progress. However, it does not provide the
accuracy required for calculation of the overall mass
transfer coefficients, particularly in the Bi-Th and LiCl
phases in the contactor and stripper vessels in which the
neodymium concentrations are usually less than 1 ppm
(wt). The isotopic dilution analysis is capable of accu-
rately determining neodymium concentration down to
~0.01 ppm (wt), and results obtained from the isotopic
dilution technique for the Bi-Th and LiCl phases were
used for calculations of the overall mass transfer coeffi-
cients.

Values for distribution coefficients for neodymium at
the three salt-metal interfaces were measured at the end
of run Nd-1 for comparison with those calculated from
the data of Ferris®>” (Table 8.2). The experimental
values are in reasonable agreement with the calculated
values in the absence of fluoride contamination;indica-

7. L. M. Ferris et al., “Distribution of Lanthanide and
Actinide Elements Between Molten Lithium Halide Salts and
Liquid Bismuth Solutions,” J. lnorg. Nucl. Chem. 34,2921-33
(1972).

Table 8.2. Distribution coefficients? for neodymium
in experiment MTE-3B, run Nd-1

Salt-metal interface Calculated? Experimental
Fluoride salt—Bi—Th 0.006 0.013
LiCl-Bi-Th 0.94 0.64
LiCl-Li-Bi 3.5x 107 >1x 10°

?Distribution coefficient = (m.f. Nd in bismuth)/(m.f. Nd in
salt).

bConditions: 650°C (923°K), Li concentration in Bi—Th = 40
ppm, Li concentration in Li—Bi = 5 at. %, no fluoride in LiCl
phase.

ting that the distribution coefficients for neodymium
were not seriously affected by the entrainment of the
fluoride salt into the chloride.

Data obtained during run Nd-1 were analyzed by
simultaneous solution of seven time-dependent differen-
tial material-balance equations (Fig. 8.2) that relate the

ORNL DWG. 7I-13962R2

FLUORIDE LiCl
SALT SAl

Li-BISMUTH
dx
i _ _ —_
dXz _ _& _
Vza— z - Killi()(2 DA)+ F1 (X1 Xz)

dxs _ X3 _ -
Vagp = + KA (%58 - Fy (X3,
X4 = — KyA, (X5DgXe) * Fp (X3~ Xg)

ng:—i =+ KzAz (X4" DBX5)-F3(X5-XG)

dx X

Vegi~ =~ KaAs (Xe"fiz)*':flxs'xe)
dx Xz

Vgl T+ K3A3(X5—55)

V = MOLAR VOLUME OF EACH PHASE
F = FLOW RATE, MOLES /SEC.

D, ,Dg,D. = RARE-EARTH DISTRIBUTION COEFFICIENTS,
MOLES/MOLE

A = AREA AT EACH INTERFACE, CM°

,,K.= RARE - EARTH OVERALL MASS TRANSFER
COEFFICIENT, CM/SEC

X,., = RARE- EARTH CONCENTRATION IN EACH
PHASE, MOLES/CM?®

EQUATIONS USED TO CALCULATE MASS TRANSFER

COEFFICIENTS FOR METAL TRANSFER EXPERIMENT
MTE-38B

Fig. 8.2. Equations used to calculate mass transfer coeffi-
cients for the metal transfer process experiment. V = volume of
each phase; x = rare-earth concentration; ¢+ = time; 4 = mass
transfer area; F = flow rate; D = rare-earth distribution coeffi-
cient; K = overall mass transfer coefficient.
rate at which the rare earths are transferred through the
several stages to the distribution coefficients of the rare
earth, the mass flow rates, and the mass transfer coeffi-
cients at each salt-metal interface.® The set of equations
was solved, using a computer program, by selecting
values for the mass transfer coefficients which.resulted
in the best agreement between the experimental data on
rate of change of neodymium concentration in all
phases in the system and the calculated values. Several

trial-and-error iterations were required, using adjusted .

values of mass transfer coefficients, until a “best-fit”
solution was obtained.

The final calculated results for run Nd- 1 are shown in
Table 8.3, where the values for the overall mass transfer
coefficients are given and are compared with values cal-
culated by the correlation of Lewis.® The coefficients
are lower than predicted and are similar to results
obtained in the previous experiment, MTE-3.* Final
analytical results for run Nd-2 are not yet available.
However, for the first 50 hr of operation, the rate of
accumulation of neodymium in the Li-Bi solution in the
stripper appeared to be similar to that observed in run
Nd-1. The significance of these absolute values of mass
transfer coefficient cannot be assessed until the scaling
laws in this type of contactor are known.

8. L. E. McNeese, Engincering Development Studies for
Molten-Salt Breeder Reactor Processing No. 11, ORNL-TM-
3774 (in preparation). ~

147

8.2 SALT-BISMUTH
CONTACTOR DEVELOPMENT

C. H. Brown, Jr.

Mechanically agitated nondispersing salt-bismuth con-
tactors are -being . considered for the protactinium
removal step and the rare-earth removal step in the
reference MSBR processing plant flowsheet. These con-.
tactors have several advantages over packed column
salt-bismuth contactors:

1. they can be operated under conditions that minimize
entrainment of bismuth to the fuel salt returning to
the reactor,

2. they can be fabricated’ more economically from
graphite and molybdenum components,

3. they can handle more easily large flow-rate ratios of
bismuth and molten salt.

Experimental development of stirred interface contac-
tors is being carried out in two different systems, a
facility in which molten fluoride salt is contacted with
bismuth containing a dissolved reductant and a system
in which mercury and an aqueous electrolyte phase are
used to simulate b1smuth and molten salt. These two
systems and the development work performed during
this report period are-described in Sects. 8.2.1 and 8.2.2.

Table 8.3. Overall mass transfer coefficients? for neodymium.
in metal transfer experiment MTE-3B, run Nd-1

K, (mm/sec)b

K, (mm/sec)b

K, (mm/sec)b

Measured®  Predicted value  Measured¢ Predicted value Measured¢ Predicted value
value (%) value (%) value (%)
0.0035 39 0.25 20 0.13 : 2.5

2Based on the neodymium in the salt phase.

1 1 1 .
b—— =— +-————at fluoride salt—Bi~Th interface.
! 1+DB t LiCl-Bi—Th interf:
= Ty T d -Bi1-—- erface.

K. k. &, 1t Ly i—Th interta

1 1 1 ‘ o

— =t , at LiCl- Li—Bi interface,
K, k, kDo : ,

, = individual mass transfer coefficient, fluoride salt to bismuth,

k

k, = individual mass transter coefficient, bismuth to fluoride salt,

k, = individual mass transfer coefficient, bismuth to lithium chloride,

k, = individual mass transfer coefticient, lithium chloride to bismuth,
where

DA = distribution coefficient between fluoride salt and bismuth,
DB = distribution coefficient between chloride salt and bismuth,
D = distribution coefficient between chloride salt and lithium-bismuth,

CAgitator speed is 5.0 rps.
8.2.1 Experiments with a Mechanically Agitated
Nondispersing Contactor in the Salt-Bismuth
Flowthrough Facility

Operation of a facility has continued in which mass
transfer rates are being measured between molten LiF-
BeF,-ThF, (72-16-12 mole %) and molten bismuth in a
mechanically agitated nondispersing contactor. The
equipment consists of a graphite-lined stainless steel
vessel, salt and bismuth feed and receiver vessels, and
the contactor vessel. In the first of these the salt and
bismuth phases are stored between runs. The other
vessels allow for treatment of the phases with HF and
H,. The contactor consists of a 6-in.-diam carbon-steel
vessel containing four l-in.-wide vertical baffles. The
agitator consists of two 3-in.-diam stirrers having four
noncanted blades. A %-in.-diam overflow at the inter-
face allows removal of interfacial films, if present, with
the salt and metal effluent streams. During a run the salt
and bismuth phases are fed to the contactor by con-
trolled pressurization of the respective feed tanks; the
phases return to the receiver vessels by gravity flow. A
detailed description of the facility and operating pro-
cedures has been previously reported.” A total of nine
mass transfer runs have been completed to date along
with one hydrodynamic run intended to determine the
amount of entrainment of one phase into the other at a
series of different agitator speeds. Results from the nine
mass transfer runs have been previously reported.’ ~!2
The experimental procedure for, and résults obtained
from, the hydrodynamic run and treatment of the salt
and bismuth with HF and H, are discussed in the re-
mainder of this section.

Experimental operation during the hydrodynamic
run. The hydrodynamic run was performed with salt
and bismuth flow rates of 150 and 140 cc/min
respectively. The agitator was operated at three dif-
ferent speeds during the run, 250, 310, and 386 rpm. At
250 and 310 rpm, three sets of unfiltered salt and bis-
muth samples from the contactor effluent streams were
taken at 4-min intervals. Three sets of unfiltered ef-
fluent samples were also taken with the agitator operat-
ing at 386 rpm, but the samples were taken at 2-min
intervals.

To avoid contamination of the sample contents with
extraneous material, the sample capsules were cleaned
of foreign matter by the following procedure: Gross
amounts of salt or bismuth were first removed with a
file; then the sample capsule was polished with emery
cloth, and finally the capsule was washed with acetone.

The sample capsules were then cut open with a tubing
cutter, and the contents of each sample were drilled out

148

and visually inspected for the presence of one phase in
the other. No such evidence of gross entrainment was
found. In some of the salt samples, small flecks of metal
were noticed which were probably small pieces of the
sample capsule produced during the drilling operation.
The contents of each sample were then sent to the Ana-
lytical Chemistry Division for determination of bismuth
present in the salt samples and beryllium present in the
bismuth samples. It is assumed that any beryllium
present in the bismuth is indicative of entrained fluoride
salt. The results of these analyses are given in Table 8.4.
The bismuth concentration in the salt samples shows a
general decrease with increasing stirrer speed, with very
low values occurring at the highest stirrer speed. It also
seems evident that the bismuth concentration in the salt
phase may have been a function of the run time, since

- after the fourth sample the bismuth concentration re-

mained at a relatively constant value of 50 %11 ppm,
which is quite different from the values reported for the
first four samples, which ranged from 1800 to 155 ppm.

These results are significantly higher than those of
Lindauer,'®> who saw less than 10 ppm of bismuth in

9. J. A. Klein et al., Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. 19, ORNL-TM-
4863 (July 1975) pp. 21-38.

10. C. H. Brown, Jr., Engineering Development Studies for
Molten-Salt Breeder Reagctor Processing No. 21, ORNL-TM-
4894 (in preparation).

11. J. A_ Klein, Engineering Development Studies of Moiten-
Salt Breeder Reactor Processing No. 18, ORNL-TM-4698
(September 1974) pp. 1-22. ‘

12. C. H. Brown, Ir., Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. 20, ORNL-TM-
4810 (in preparation).

13. R. B. Lindauer, Engineering Development Studies of
Molten-Salt Breeder Reactor Processing No. 17, ORNL-TM-
4178 (in preparation).

Table 8.4. Analyses of salt and bismuth
samples taken during the hydrodynamic run

Agitator speed Bisample Bein Bi Salt sample Biin salt

(rpm) number (ppm) number (ppm)
250 428 215 437 1800
250 429 125 438 205
250 430 215 439 155
310 431 85 440 270
310 432 910 441 53
310 433 442 34
386 434 110 443 64
386 435 175 444 54
386 436 50 445 43

fluoride salt in contact with bismuth in several different
contacting devices. It is likely that sample contamina-
tion is a contributing factor to the high bismuth concen-
trations measured. Three possible sources of sample
contamination have been reported:!?

1. contamination by withdrawing samples through a
sample port which has been in contact with bismuth,

2. contamination during sample handling and in the
analytical laboratory by the use of equipment rou-
tinely used for bismuth analyses,

3. contamination from a low-density bismuth-containing
material which may be floating on the salt surface.

Since no maximum permissible rate of bismuth entrain-
ment in the fuel salt going to the bismuth removal step
or in the salt returning to the reactor from the fuel
processing plant has been set, it is difficult to assess the
significance of these results. However, the bismuth con-
centrations in the salt do not seem to be inordinately
high at the highest stirrer speed, and it seems likely that
some degree of phase dispersal might be tolerated in
order to achieve higher mass transfer rates.

The beryllium concentrations in the bismuth samples
at each agitator speed show both high and low values
with no discernable dependence on agitator speed.
These results agree well with previously reported data’®
for beryllium concentration in the bismuth phase during
mass transfer runs in this system at agitator speeds of
124, 180, and 244 rpm. Previous experiments with
water-mercury and organic-mercury systems suggest
entrainment of the light phase into the heavy phase at
an agitator speed of about 170 rpm. The concentration
of beryllium in the bismuth phase is not significantly
different from previous results observed at lower agita-
tor speeds. The effect of entrained fluoride salt in the
bismuth would be most detrimental in the metal trans-
fer process, where fluoride salt in the chloride salt phase
decreases the separation factors between thorium and
the rare-earth fission products.

H,-HF treatment of salt and bismuth. The mass trans-
fer runs completed to date in the salt-bismuth contactor
have all been performed under conditions where the
controlling resistance to mass transfer is in the inter-
facial salt film. One final mass transfer run will be per-
formed in which the bismuth-film mass transfer coeffi-
cient is measured. In preparation for this run, the salt
and bismuth in the graphite-lined treatment vessel were
treated with HF diluted with H, to oxidize the reduc-
tants present in the bismuth phase. The procedure used
was essentially that reported previously.!® The salt and
bismuth at "600°C were sparged with 25 scfh of 30%
(mole) HF for 9 hr. The HF utilization decreased from

149

75% at the beginning of treatment to 35% during the

~ final 2 hr of treatment. Analysis of the salt and bismuth

phases before and after treatment with HF and H, in-
dicated that essentially all of the reductant in the bis-
muth phase was oxidized by hydrofluorination. The
uranium distribution ratio decreased from 740 moles/
mole prior to the treatment to 0.03 mole/mole after the
HF-H, treatment.

8.2.2 Experiments with a Mechanically Agitated
Nondispersing Contactor Using Water and Mercury

We have continued development of a mechanically
agitated nondispersing two-phase contactor, using an
aqueous electrolyte and mercury to simulate molten
salts and bismuth.

As previously reported,'® we have investigated the
feasibility of using a polarographic technique for
measuring electrolyte-film mass transfer coefficients in
this type of contactor. During this report period we
have

1. tested three different anode materials,

2. produced cathodic polarization waves corresponding
to the reduction of Fe®*, complexed with excess
oxalate ions, at the mercury surface,

3. obtained and calibrated a slow-scan controlled-
potential cyclic voltameter,

4. examined the quinone-hydroquinone redox couple
as a possible alternate to the Fe®" -Fe** couple now
being used.

Modifications to experimental equipment. With the ex-
ception of the tests made with the quinone-hydroquinone
redox couple, all tests made during this period were
performed with the equipment previously described.!®
The equipment consists of the 5 X 7 in. Plexiglas con-
tactor used in previous work with the water-mercury
system. The mercury surface in the contactor acts as the
cathode in the electrochemical cell. The cathode is elec-
trically connected to the rest of the circuit by a %-in.-
diam stainless steel rod electrically insulated from the
electrolyte phase by a Teflon sheath. The anode of the
cell is suspended in the aqueous electrolyte phase and
consists of a metallic sheet formed to fit the inner
perimeter of the Plexigas cell. The current through the
cell is inferred from the voltage drop across a 0.1-§2 *
0.5%, 10-W precision resistor. The signal produced

14. B. A. Hannaford et al., Engineering Development Studies
for Molten-Salt Breeder Reactor Processing No. 3, ORNL-TM-
3138 (May 1971) p. 30.

15. C. H. Brown, Jr., Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. 22, ORNL-TM-
4041 (in preparation).
across the resistor is recorded as the y coordinate on a
Hewlett-Packard x,y plotter. The x coordinate on the
plotter is produced by the potential difference between
the mercury surface and a standard calomel electrode
(SCE) suspended in the electrolyte phase.

Prior to studies made on the quinone-hydroquinone
system, a slow-scan controlled-potential cyclic volta-
meter (potentiostat) was obtained from the Analytical
Chemistry Division to replace the Hewlett-Packard dc
power supply previously used. The cyclic voltameter is a
three-electrode instrument which controls the potential
between the mercury surface and a standard calomel
reference electrode while passing a current between the
auxiliary electrode and the mercury surface. Voltages
can be scanned between +2 V vs SCE and —2 V vs SCE
at a scan rate up to 1 V/min. The potentiostat can carry
a current of up to 2.5 A between the auxiliary and

mercury electrodes.

Experiments with the Fe®* -Fe* system. The electro-

lyte used for all the experiments performed during this
report period was nominally 0.001 M Fe?* obtained
from ferrous sulfate, 0.00025 M Fe3* obtained from
ferric sulfate, and 0.8 M potassium oxalate. The oxalate
ions form a stable complex with both the Fe3* and Fe?*

150

which facilitates measurements of the Fe®' reduction
wave directly.

Three anode materials have been tested:. copper, iron,
and gold; and satisfactory polarization waves were pro-
duced with all three materials. However, the copper and
iron reacted with the electrolyte solution. This addi-
tional side reaction caused poor reproducibility in the
data and could also possibly alter the properties of the
solution. To avoid this problem, an anode was fabri-
cated by plating gold on a 0.0625-in.-thick sheet of
nickel, which was formed to fit the inner perimeter of
the electrochemical cell. '

Shown in Fig. 8.3 is a polarogram measured with the
electrolyte described above in the 5 X 7 in. Plexiglas
contactor using the gold anode with phase volumes of
about 1.8 liters each and no agitation. The cell current
is plotted as a function of the mercury surface potential
vs the SCE. The current increases from zero at zero
applied potential to a relatively constant value at an
applied potential of about —0.35 V vs SCE. In this
region continuous electrolysis is taking place in the cell,
corresponding to reduction of Fe(C,0,);* ~ at the mer-
cury cathode. In the region of applied potential from
—0.35 V vs SCE to —0.80 V vs SCE, the cell current

ORNL DWG 75- 8425

| | | I I ] |
3 —
(] ‘ 2 | _ []!’ 1 -
| — In i.l“
[
z
w
@
«
=
o
1=
EVZ = 0.245 Y
0 L | I | l | I
-0. -0.2 -03 -0.4 -05 -0.6 -07 -0.8
VOLTAGE vs SCE
Fig. 8.3. Cathodic polarization wave for Fe(C,0,),> ~measured in the 5 X 7 in. rectan-

gular contactor with no agitation.
increases only a small amount; here, the current is
limited by the rate of diffusion of the Fe(C;04)3>~ to
the mercury surface, where this ion is reduced. The dif-

fusion current can be related to the mass transfer .

coefficient through the electrolyte film as previously
discussed.*®

The half-wave potential is defined as the potential at
which the current is equal to one-half the limiting value.
Figure 8.3 shows the measured half-wave potential for
the ferric oxalate complex. The half-wave potential of
—0.245 V measured in the contactor agrees well with
the value reported in the literature of —0.24 V vs SCE
for the reduction of ferric oxalate.'®

Under ideal conditions, the diffusion current is
directly proportional to the polarized electrode surface
area and the bulk concentration of the limiting ion. To
determine that the mercury surface was actually being
polarized, two tests were performed. First, the anode
surface area was decreased by about 48%. This had no
effect on the magnitude of the diffusion current, indi-
cating that the mercury surface (cathode) was polarized
rather than the anode surface. In the second test, the
‘concentration of the ferric ion was doubled, but no
concomitant increase in diffusion current was seen.
Since the diffusion current is directly proportional to
the concentration of the limiting ion (Fe®"), the current
should have doubled. The only explanation for this
behavior is that the Fe’ had been reduced by some
contaminant in the system, possibly present in the mer-
cury. This would have caused ferric ions to be present at
only a very low concentration during cell operation, due
to electrolytic oxidation of the ferrous iron.

To eliminate the possibility of reductant being present
in the mercury, a supply of purified mercury was ob-
tained from the Analytical Chemistry Division. A test
was performed using the purified meré:ury and an elec-
trolyte having the same nominal Fe®" and Fe* concen-
trations given above. Preparation of the electrolyte was
completed in the absence of oxygen to preclude pos-
sible oxidation of Fe?* to Fe*. Again, the anode sur-
face area was decreased with no discernable decrease in
the diffusion current, indicating -that the mercury sur-
face was polarized. An increase of the Fe* concentra-
tion from ~0.25 mM to ~0.5 mM resulted in an in-
crease in the diffusion current by a factor of 2, indi-
cating that the wave being measured was the ferric ion
reduction wave. However, the half-wave potential was
measured to be —-0.75 V vs SCE, which is about three
times the reported value.

To calculate the aqueous-film mass transfer coeffi-
cient from polarographic data, the bulk concentration
of the oxidized species must be accurately known. The

151

method for this measurement is a polarographic tech-
nique using the dropping-mercury electrode. Samples of
the electrolyte used in the second of the two tests men-
tioned above were analyzed for Fe¥ and Fe?®' by this
method. Results indicated that the Fe®* and Fe®* con-
centrations were 1.7 and 0.28 mM respectively, which is
in poor agreement with the expected values of 0.50 mM
Fe* and 1.0 mM Fe**. One possible cause for the poor
agreement is that the Fe?* was oxidized to Fe** during
the period when the solution was held in the sample
bottles. However, this was not expected, since the elec-
trolyte had been sparged with argon to remove dissolved
oxygen, and the sample bottles were purged with argon
to remove air.

To aid in determining if the reported analytical results
were in error due to analytical technique or to method

" of solution preparation, two standard solutions were

prepared and sampled for analysis. One solution was

. prepared to contain 56 ug/ml Fe®", and the other solu-

tion was prepared to contain 56 ug/ml Fe**. Both solu-
tions were 1 M in K,C, 0,4 -H; 0. Subsequent analytical
results indicated that both solutions had essentially the
same concentrations of Fe® and Fe?", 50 and 27 ug/ml
respectively. Further investigation will be necessary to
determine the correct method for preparing and/or
analyzing iron oxalate solutions.

Experiments with the quinone-hydroquinone system.
A possible alternate to the Fe* -Fe?* system for meas-
uring electrolyte-phase mass transfer coefficients is the
reversible reduction of quinone to hydroquinone at the
mercury cathode.

The reaction under consideration is

C6H4O2 +2H++2€—2C6H4(0H)2 . (1)

Since hydrogen ion as well as quinone is a reacting
material, a strong buffer must be present to serve as a
supporting electrolyte. The buffer causes the H™ con-
centration to be essentially constant across the inter-
facial electrolyte film, because the rate at which the
buffer equilibrium is established is relatively rapid com-
pared with the quinone diffusion rate.!”

A qualitative test was made with the quinone system
to determine whether acceptable polarization waves
could be measured and to determine whether the qui-
none electrolyte is inert to mercury. The electrolyte was
0.01 M hydroquinone and 0.005 M quinone with a 0.05
M phosphate buffer at a pH of 7.0. Satisfactory polari-

16. [. M. Kolthoff and 1. J. Lingane, p. 484 in Polorography,
Interscience, New York, 1946.

17. C. A. Lin et al., “Diffusion-Controlled Electrode Reac-
tions,” Ind. Eng. Chem. 43, 2136—43 (1951).
152

zation waves were obtained in a small cell with a large
copper anode and a mercury pool cathode. The electro-
lyte was chemically inert to mercury during the tests.
The color of the quinone electrolyte changed from a
light yellow to deep brown within several hours. This
phenomenon is due to the decomposition of quinone by
ultraviolet light. Further studies in the 5 X 7 in. contac-
tor will be done to determine whether this system is
suitable for mass transfer measurements.

8.3 CONTINUOUS-FLUORINATOR
DEVELOPMENT

R. B. Lindauer

Continuous fluorinators are used at two points in the
reference flowsheet for MSBR processing. The first of
‘these is the primary fluorinator, where 99% of the
uranium is removed from the fuel salt prior to the re-
moval of *23Pa by reductive extraction. The second
point is where uranium produced by decay of 232 Pais
removed from the secondary fluoride salt in the protac-
tinium decay tank circuit. These fluorinators will be
protected from fluorine corrosion by frozen-salt layers
formed on the internal surfaces of the fluorinator which
are exposed to both fluorine and molten salt. To keep
frozen material on the walls while maintaining a
molten-salt core in the fluorinator, an internal heat
source is necessary to support the temperature gradient,
Heat from decay of the fission products in the salt will
be used in the processing plant. However, to test
frozen-wall fluorinators in nonradioactive systems,

another internal heat source which is not attacked by
fluorine is needed. Since electrolytic or autoresistance
heating of molten salt has proven to be a feasible means
for providing this heat source, studies of autoresistance
heating of molten salts are continuing. A conceptual
design was made for a continuous fluorinator experi-
mental facility (CFEF) to demonstrate fluorination in a
vessel protected by a frozen-salt film. Design was com-
pleted and installation was begun of a fluorine disposal
system in Building 7503 which uses a vertical spray
tower and a recirculating KOH solution. Installation was
completed of equipment to demonstrate the effective-
ness of a frozen-salt film as protection against fluorine
corrosion in a molten salt system.

8.3.1 Installation and Initial Operation of
Autoresistance Heating Test AHT4

Equipment for autoresistance heating test AHT4 was
installed in cell 3 of Building 4505. In this system (Fig.
8.4) molten LiF-BeF,-ThF, (72-16-12 mole %) is cir-
culated by means of an argon gas lift from a surge tank
to a gas-liquid separator from which the salt flows by
gravity through the autoresistance electrode, through
the test vessel, and returns from the bottom of the test
vessel to the surge tank. The test vessel (Fig. 8.5) used
in experiment AHT-3 was decontaminated, equipped
with new cooling coils, heaters, and thermocouples, and
reinstalled for experiment AHT 4.

The test vessel is made of 6-in. sched-40 nickel pipe
with a 44-in.-long (1.1-m) cooled section from the elec-
trode to below the gas inlet side arm. The cooled sec-
tion is divided into five separate zones, each with two

ORNL DWG 75-498%

GAS-LIQUID T
SEPARATOR

_—

FREEZE
VALVE

HEAT
FLOWMETER

» OFF-GAS

ARGON
AUTORESISTANCE :
HEATING POWER
SUPPLY

TEST
VESSEL

Fig. 8.4. Flowsheet for autoresistance heating test, AHT-4.
154

parallel coils through which an air-water mixture flows. -
The gas outlet section above. the salt level has an in-
creased diameter for gas-salt disengagement and is made
of 8-in. sched40 pipe. The surge tank has a 46-in.-long
(1.2-m), 6-in:-diam (0.15-m) section to provide submer-
gence for the gas lift. The upper section of the surge
tank is 24 in. (0.61 m) in diameter and provides suffi-
cient capacity to contain the salt inventory for the en-
tire system. The gasliquid separator is an 8-in.-diam
(0.20-m) conical-bottom vessel with baffles and York
mesh in the upper part for gas-liquid disengagement. In
the heat flowmeter the salt is heated by an internal
cartridge heater, and the flow rate is calculated from the
‘heat input and the temperature rise of the salt stream.

 The system is started up by heating the equipment
and lines to 600°C (873°K). The argon gas lift is
started, and initially the salt flow rate is determined by
the decrease in surge-tank liquid level. After the salt
levels in the tank, separator, and test vessel are constant,
cooling of the test vessel is started. The resistance be-
tween the high-voltage electrode and the test vessel
walls is checked periodically by applying a low voltage

to the electrode and measuring the current. As cooling
progresses, this resistance will increase until the point is
reached where heat can be produced in the salt at a
significant rate (several hundred watts) without causing
a reduction (shorting) of the resistance.

The 80-iter salt batch was charged to the surge tank;
and after minor modifications to the heating system,
operation was started. Four preliminary runs were
made, lasting from 4 to 12 hr (from the time the gas lift
was started until plugging occurred). In the first run,
plugging apparently occurred in the electrode when the
liquid level.in the separator fell too low to provide suf-
ficient head for flow to the test vessel.

Salt flow in the second run was much smoother, and
circulation continued for 11 hr without adjustment of
the gas lift. During this time the test vessel was being
cooled,-and the salt flow rate slowly decreased by 7%,
from 450 to 425 cm? /min. This was probably caused by
an increase in salt viscosity, a buildup of frozen salt in
the test vessel, or a combination of the two. The steady
salt flow rate and higher salt temperature (>873°K and
20—30°K higher than in run No. 1) kept the electrode
from freezing, but the heat supply at the bottom of the
test vessel was insufficient to keep the salt outlet from
freezing, which terminated run 2. The resistance be-
tween the high-voltage electrode and the vessel wall
increased from 0.01 to 0.08 £, but autoresistance heat-
ing was not attempted. The vertical portion of the test

section had been cooled to 639°K (solidus temperature,
623°K).

Before the third run, the output of the powerstat con-
trolling the test vessel bottom heaters was increased by
44% to keep the salt outlet above the freezing point.
The run was terminated by salt freezing in the elec-
trode. This resulted from too low a salt flow rate (the
heat flowmeter was inoperative because of a burned out
heater) and too low an initial temperature (723°K vs
823°K in the second run) in the vertical section of the
side arm through which the electrode passes.

The fourth run was started with some heat on the
vertical section of the side arm. This section was un-
heated previously. As cooling progressed, the bottom
heaters on the test vessel were inadequate at the salt
flow rate being used. Increasing the salt flow rate pre-

‘vented freezing at the bottom of the test vessel. After

7% hr of operation, the liquid levels in the separator and
test vessel started to increase, indicating salt flow prob-
lems both at the inlet and exit of the test vessel. Al-
though the salt resistance had only increased from 0.01
to 0.03 € and the average test vessel wall temperature
(in the cooled zone) was 658°K, autoresistance heating
was started. This freed the plug in the electrode, allow-
ing salt flow from the separator to the test vessel, and
the increased flow raised the test vessel bottom temper-
ature, and flow resumed from the test vessel. However,
salt flow rates were erratic for the next 2 hr, and 9% hr
after the start of the run the test vessel level started to
rise, indicating a frozen salt restriction in the vessel. It
was decided to try to transfer the molten salt from the
test vessel to the surge tank before complete plugging
occurred. This was done successfully, and 5.6 liters of
salt was transferred to the surge tank. After cooling,
radiographs were taken of the test vessel by the Inspec-
tion Engineering Department, using a 35-Ci '®2Ir
source. In the test section of the test vessel, radiation
penetration was insufficient to permit measurement of
the film thickness. The bottom of the vessel between
the salt outlet and the gas inlet was free of salt as ex-
pected, and the radiograph of the top of the vessel
showed a 25-mm-thick ring of salt above the normal
liquid level. This is salt deposited on the colder pipe
wall by the action of the gas bubbling through the salt.
Calculations from the volume of salt transferred indi-
cated an average film thickness of 45 mm (a 65-mm-
diam molten core). The salt resistance at the end of the
run was 0.18 £2, and the maximum autoresistance heat-
ing used was 450 W.

The main problem seems to be the forming of a uni-
form salt film. Near the electrode where the hot molten
salt enters, cooling is much slower than in the vertical
section above the gas inlet. It is probably in the vertical
section where the salt film becomes too thick and re-
stricts the salt flow.

8.3.2 Designof a Continuous-Fluorinator
Experimental Facility (CFEF)

The purpose of the CFEF is to measure the perfor-
mance of a continuous fluorinator which has frozen-
wall corrosion protection in terms of uranium removal.
The uranium which is not volatilized, but is oxidized to
UF,, will be reduced back to UF, in a hydrogen reduc-
tion column. The facility will be used to obtain operat-
ing experience and process data, including fluorine utili-
zation, reaction rate, and flow-rate effects; and to
demonstrate protection against corrosion, using a frozen
salt film. : . o

The facility will be installed in a cell in Building 7503
to provide beryllium containment. The system will con-
tain about 8 ft* (0.23 m?®) of MSBR fuel carrier salt
(72-16-12 mole % LiF-BeF,-ThF,) containing 0.35
mole % uranium initially. The salt will be circulated

“ through the system at rates up to 50% of MSBR flow’

rate (6.7 X 107> m?>/sec). Because of the short fluori-
nator height (1 to 2 m) the amount of uranium volatil-
ized will be between 80 and 95% per pass. The variables

155

of salt flow rate, fluorine flow rate, and fluorine con-
centration will be studied by measuring the UF, con-
centration in the fluorinator off-gas stream and by sam-
pling the salt stream after reduction of UFs to UF,.
The fluorinator will have two fluorine inlets to provide
data for determining the column end effects. Reduction
of UFs will be carried out in a gas lift in which hydro-
gen will be used as the driving gas and also as the reduc-
tant. If additional reduction is required, this can be

done in the salt surge tank. The surge tank is designed
" to provide sufficient salt inventory for about 10 hr of

fluorination with 95% uranium volatilization per pass.
About 99% of the uranium should have been removed
from the salt batch after this period of time.

The facility flowsheet is shown in Fig. 8.6. Salt will
enter the fluorinator through the electrode in a side arm
out of the fluorine path. The electrode flange will be
insulated from the rest of the fluorinator, and the auto-
resistance power will be connected to a lug on the
flange. The salt will leave at the bottom of the fluori-
nator below the fluorine inlet side arm. The fluorinator
wall will be cooled by external air-water coils to form
the frozen salt film which will serve the dual purpose of
preventing nickel corrosion and of providing an electri-
cally insulating film for the autoresistance current.
Below the fluorine inlet the fluorinator wall will not be
cooled, and the molten salt will complete the electrical

ORNL-DWG 73-8426

: . TO OFF-GAS
TO FLUORINE DISPOSAL SYSTEM
: GAs-Liquio| || |-
P HF
SAMPLER SEPARATOR TRAP
HEAT i
UFO TRAPS F‘LOWMETER E
]
l A
I
FLUORINE o
]
AUTO sunslr; REDUCTION
N MN
(DRESISTANCE TANK coLvS
POWER GAS-LIFT
SUPPLY —_ PUMP
FREEZE VALVE
‘| FLUORINATOR
HYDROGEN
=
| | FREEZE vALvE
|
DRAIN [
TANK :
i
!

Fig. 8.6. Continuous fluorinator experimental facility flowsheet.
FLUORINE -
CONTAINING GAS

69 FE) (FE

156

ORNL DWG 75-15057

TO HOT OFF-GAS

SYSTEM
PHOTOMETRIC
ANALYZER
KOH
SURGE
TANK

Fig. 8.7. Fluorine disposal system.

circuit to the vessel wall. Since all of the uranium will
not be volatilized from the salt, there will be some UFg
in the salt at the bottom of the fluorinator. The fluori-
nator bottom, exit line, and reduction column will be
protected from the highly corrosive UF; by gold lining
or plating. The molten salt containing UF5 will enter
the bottom of the column where the salt will be con-
tacted with hydrogen. The hydrogen will enter through
a palladium tube, which will result in the formation of
atomic hydrogen and greatly increase the reduction rate
to UF,. The hydrogen reduction column will also act as
a gas lift to raise the salt to a gasdiquid separator. The
salt will then flow by gravity to the fluorinator through
a salt sampler, surge tank, heat flowmeter, and electrical
circuit-breaking pot. Off-gas from the separator which
contains HF and excess hydrogen will pass through an
NaF bed for removal of the HF. Uranium hexafluoride
from the fluorinator will also be removed by NaF. Mass
flowmeters before and after the NaF beds will be used
to continuously measure the UF, flow rate.

8.3.3 Fluorine Disposal System for Building 7503

The CFEF (Sect. 8.3.2) will be the first test of the
frozen-wall fluorinator using fluorine. For the disposal
of the excess fluorine, a vertical scrubber is being in-

stalled in Building 7503. A flow diagram of the system
is shown in Fig. 8.7. The scrubber is a 6-in.-diam.
8-ft-high (0.15- by 2.4-m) Monel pipe with three spray
nozzles in the upper half of the vessel. The surge tank
contains 200 gal (0.95 m*) of an aqueous solution con-
taining 15 wt % KOH and 5 wt % KI. This equipment is
designed to be able to dispose of one trailer of fluorine
(18 std m*) at a flow rate of 1.2 scfm (9 X 107 std
m?/sec). The KOH solution will be circulated through
the spray nozzles at a total flow rate of 15 gpm (0.001
m?*/sec). The fluorinator off-gas stream will flow cocur-
rently with this stream. The scrubber exit stream passes
through a photometric analyzer for monitoring the
efficiency of the scrubber.

8.3.4 Frozen-Wall Corrosion
Protection Demonstration

Equipment has been installed for demonstrating that a
frozen salt film will protect a nickel vessel against fluo-
rine corrosion by preventing the NiF; corrosion prod-
uct film from being dissolved in the molten salt. A small
vessel containing 6 X 107 m? of molten LiF-BeF,-
ThF, (72-16-12 mole %) will be used for the demon-
stration (Fig. 8.8). The fluorine inlet consists of three
concentric tubes which provide a path for an air coolant

ORNL DWG. 75-8949

FLUORINE IN —1

L

-~ ARGON OUT

L

-—— _ARGON IN

————=———— ARGON PURGE

[————h FLUOQRINE OUT

i

SALT LEVEL—»

Fig. 8.8. Frozen salt protection demonstration test vessel.

stream that will be used for freezing a salt film on the
outside of the outer tube. The wall of the inner tube
through which the fluorine will flow is 31 mils (0.79
mm) thick. The inner tube of the fluorine inlet will not
be protected from corrosion. The vessel wall is also
unprotected but is 280 mils (7.11 mm) thick. Fluorine
will be passed at a low flow rate (830 mm?/sec)
through the salt until failure occurs, which is expected
in less than 100 hr at the tip of the probe near the
gas-liquid-solid interface. Wall thickness measurements
before and after the demonstration will show to what
extent the salt film afforded protection.

A flow diagram for the system is shown in Fig. 8.9.
The argon back pressure will be recorded to provide an
indication of corrosive failure. Failure of the tube below
the salt film will allow some salt to leak into the argon
cooling annulus. The salt will be entrained up into the
cool portion of annular space, causing a restriction to

157

the argon flow. The system will be designed such that
the fluorine flow is terminated automatically when
either a low argon pressure is detected in the annulus or
when a high argon back pressure occurs.

8.4 FUEL RECONSTITUTION
ENGINEERING DEVELOPMENT

R. M. Counce

The reference flowsheet for processing the fuel salt
from an MSBR is based upon removal of uranium by
fluorination to UFg as the first processing step.'® The
uranium removed in this step must subsequently be re-
turned to the fuel carrier salt before its return to the
reactor. The method for recombining the uranium with
the fuel carrier salt (reconstituting the fuel salt) consists
in absorbing gaseous UFy into a recycled fuel salt
stream containing dissolved UF,; according to the reac-
tion

UFg(g) + UF4(d) = 2UF5(d) . 2)

The resultant UF; would be reduced to UF, with
hydrogen in a separate vessel according to the reaction

UFs(d) + %2H, (g) = UF4(d) + HF(g) . ()

Engineering studies of the fuel reconstitution step are
being started to provide the technology necessary for
‘the design of larger equipment for recombining UF,
generated in fluorinators in the processing plant with
the processed fuel carrier salt returning to the reactor.
During this report period, equipment previously de-
scribed'? was fabricated and has been installed in the
high-bay area of Building 7503. This report describes
instrumentation for off-gas analysis, including a prelimi-
nary calibration curve, and two alternatives for pro-
viding corrosion-resistant gold linings for equipment to
be installed later.

The nickel reaction vessels presently installed will be
used to test the salt metering devices and gas supply
systems. After the initial shakedown work is completed,
the UFg absorption vessel, H; reduction column,
flowing-stream samplers, and associated transfer lines
will be replaced with gold or gold-lined equipment.
Gold is being used because of its resistance to corrosion
by UF¢ gas and UF; dissolved in the salt.

18. Chem. Technol. Div. Annu. Progr. Rept. Mar. 31, 1972,
ORNL-4794,p. 1.

19. R. M. Counce, Engineering Development Studies for
Molten-Salt Breeder Reactor Processing No. 19, ORNL-TM-
4863 (July 1975) pp. 38—42.
ORNL OWG 75-8948

FCV-3 '
£, ! 100 cc/MIN HASTINGS MASS FLOWMETER
2 .&]_(FI)
5 PSIG | ‘
i V_g _____________________________
! .
o VALVE CLOSES ON 1+ CELL 3B
y : HIGH Ar BACK }  FIRST FLOOR 8,";2
! PRESSURE [
g ettt tdied ke 0s150 " | (OUTSIDE
2l IN. WATER | | : - BLDG)
w I _——_=Jd ! V-5 /@
a8 ! I 1!
w i
Qv 0-30 ! 1
- q 1 ' _ - b
s} = : PSIG e i V-2 His v-4
g 8 1 I-CFM i )
- I
@ @ 1 xe © ®
I 0-15
L___! | V-3
4 PSIG | J
(s) v-ejg V-7 : /®
ARGON  \ o I /@
20 PSIG L
AR | ' ]
l ./@
> . -
© IE—(9 |

DELIVERY

|
| VESSEL
|
I

ACTIVATED ALUMINA
TRAP

TEST
VESSEL

Fig. 8.9. Frozen salt protection demonstration flowsheet,

8.4.1 Instrumentation for Analyzing Reaction
Vessel Off-Gases

The equipment for the second phase of the experi-
ment will consist of a feed tank, a UF, absorption ves-
sel, an H, reduction column, flowing-stream.samplers, a
receiver tank, NaF traps for collecting excess UFg and
for disposing of HF, gas supplies for argon, hydrogen,
nitrogen, and UF,, and means for analyzing the gas
streams from the reaction vessels (Fig. 8.10). The equip-
ment will be operated by pressurizing the feed tank
~with argon in order to displace salt from the feed tank
to the UF, absorption vessel. From the UF¢ absorption
vessel, the salt flows by gravity through a flowing-
stream sampler into the H, reduction column. From the
H; reduction column the salt {lows by gravity through a
flowing-stream sampler to the receiver tank. Absorption
of gaseous UF¢ by reaction with dissolved UF, will
occur in the UFg absorption vessel, and the resultant
UFs will be reduced by hydrogen in the H, reduction
column. The effluent salt is collected in the receiver
tank for return to the feed tank at the end of the run.

The off-gas from the absorption vessel and the reduc-
tion column will be analyzed for UF4 and for HF re-
spectively.

The respective off-gas streams will be continuously
analyzed with the use of the Gow-Mac gas density
balance. A sample stream is taken from the main off-gas
stream and passed through the balance for analysis (Fig.
8.11). These analyses will be used in determining the
efficiencies of UF4 absorption and H, utilization.

The efficiency of UF¢ absorption will be determined
by metering UF¢ and Ar to the UF, reaction vessel and
determining the UF¢ content in the vessel off-gas, using
a model 11-373 Gow-Mac gas density cell.2® The H,
utilization will be determined similarly. Hydrogen will
be metered to the H, reduction column, and the
column off-gas will be analyzed for H, content, also
using a model 11-373 Gow-Mac gas density cell. The
Gow-Mac cell, commonly used as a gas chromatograph

20. Gow-Mac Instruments Company, 100 Kings Road,
Madison, New Jersey. . :
159

ORNL DWG 74- {1{666R3

]
| —
UFg HF
ANALYS!S| IANALYSIS HF UFg
TRAP TRAP
1 —
UFg —@—— | SAMPLE =— 55—
PORT
; \e—————p BUILDING
= OFF - GAS
# SYSTEM
SAMPLE
PORT

UFg
ABSORPTION
VESSEL

H2
REDUCTION
COLUMN

g ——
v ME ST wTw-

FEED
R RECEIVER

Hp

Fig. 8.10. Flow diagram of equipment used in the second fuel reconstitution engineer-

ing experiment.

ORNL DWG 75-8309
OFF- GAS

EAIECARS® B0 ’
v BUBBLER
REACTION |
VESSE : __GAS
SSEL oECaTY VENT
. CELL

BUBBLER

ROTAMETER
REFERENCE
GAS

Fig. 8.11. Schematic diagram of fuel reconstitution -engineering experiment off-gas
system.

detector, provides a continuous signal which varies
directly with the density of the sample gas, allowing
continuous analysis of the sample gas stream with accu-
racies of 3 to 4%.2! Because the detector elements are
not exposed to the sample stream, the gas density cell is
useful in analyzing corrosive gas mixtures.

Nitrogen and argon will be reference gases for the gas
density cells used for analyzing the off-gas from the
UF¢ absorption vessel and the H, reduction column
respectively. The response of the gas density cell is
fairly insensitive to changes in the sample gas flow rate
when nitrogen or argon is used as a reference gas.*? To
measure varying Ar-UFg and H,-HF ratios with the gas
density detectors, it is necessary to control the refer-
‘ence gas flow rate precisely. However, high precision is
not required for controlling the sample gas flow rate.
The reference gas flow rates are controlled sufficiently
by rotameter and separate gas supply systems. A satis-
factory means for providing reproducible sample flow
rates has been developed. The sample stream is taken
from the main off-gas stream (Fig. 8.11) and flows
through a capillary tube, the gas density detector, an
NaF trap to remove the corrosive constituent (UF, or
HF), and a bubbler to provide a constant downstream
pressure. The pressure upstream from the capillary is
maintained at a higher constant value by means of a
similar bubbler in the off-gas line downstream from the
NaF trap. The NaF traps provide sufficient volume in
the lines so that small pressure fluctuations from bub-
bles in the process vessels and in the bubblers are effec-
tively damped out. The flow rate is not constant
(although it is reproducible), because, as the concentra-
tion of the sample gas changes, its viscosity changes,
producing changes in sample flow rate under the prevail-
ing conditions. These flow rate changes superimposed
upon concentration changes in the sample stream to the
gas density detector result in a nonlinear response of the
gas density detector to changes in concentration. The
effects are reproducible, however, and a reproducible
calibration can be obtained. Such a calibration was
obtained with mixtures of hydrogen and nitrogen (Fig.
8.12).

For sample gases containing hydrogen and at refer-
ence gas flow rates below a certain critical flow rate,
hydrogen will diffuse countercurrently into the refer-
ence gas stream to the area of the detector elements.

21. J. T. Walsh and D. M. Rosie, J. Gas Chromatogr. 5(5),
23240 (May 1967).

22, C. L. Guillemin and M. F. Auricourt, J. Gas Chromatogr,
1, 24 -29 (October 1963).

160

ORNL DWG 75-8310

100 | E— | — T T T 1 "
GOW MAC GAS DENSITY CELL
90 MODEL I1- 373 A
DETECTOR CURRENT 70mA
w SENSITMVITY: 32mV
Z 80r- DETECTOR TEMR 28°C 7
Sw REFERENCE GAS : ARGON e
S 4o REFERENCE GAS FLOW RATE | 66 “Srin
¥ SAMPLE GAS FLOW RATE :APPROX. 33°C/min ]
2y®
=Z3 50~ ]
x0T
s
=l 40| -
L
Wl
oo
0 30 —
<«
(L)
20F _
1o -
| | 1 | i | 1 1 1
0 50 00
%H,
;00 T T Ll L 5’0 Li L] L L) 8
% N

Fig. 8.12. Calibration curve of Gow-Mac gas density cell
model in fuel reconstitution engineering equipment for H, and
N,.

Due to the high thermal conductivity of H,, the back
diffusion of H, can greatly affect the sensitivity of the
gas density cell. However, if sufficiently high reference
flow rates are maintained, this problem can be over-
come.

8.4.2 Design of the Second Fuel Reconstitution
Engineering Experiment

The design of equipment for the second fuel reconsti-
tution engineering experiment (FREE-2) is continuing.
The equipment for FREE-2 will be similar in design to
the equipment for experiment FREE-1 except for the
addition of an intermediate liquid-phase sample port be-
tween the UFg absorption vessel and the H, reduction
column (Fig. 8.10). In addition, all vessels and transfer
lines exposed to dissolved UF, with the possible excep-
tion of the receiver tank, will be gold or gold lined.
Gold sheet 0.010 in. (0.25 mm) thick is on hand for the
fabricated liner of the UF, absorption vessel. Two alter-
natives exist for lining the H, reduction column and the
receiver vessel: interior gold plating or a fabricated gold
liner,

The minimum plating thickness that would probably
provide a pinhole-free lining is approximately 0.005 in.
(0.13 mm). The minimum thickness for a fabricated
gold liner in vessels of this size is approximately 0.010

in. (0.25 mm). Fabricated gold liners are economically
competitive with gold plating in the thicknesses men-
tioned, because gold sheet is available at ERDA
precious-metal account prices, approximately
$34.99/troy oz ($1.13/g), and gold in commercial gold-
plating solutions is. available only at market prices of
about $164/troy oz ($5.27/g) as of June 18, 1975.
Some comparisons important in the choice between
interior gold plating or fabrication of a gold liner are:

1. the technology involved in fabricating a welded gold

vessel is available, while some technology would

need to be developed for interior plating of vessels

having a- high length/diameter ratio such as the H;

reduction column,

. the time involved in both approaches is approxi-
mately the same,

. the plating will be difficult to inspect, and there will
be no guarantee of pinhole-free coverage, while dye
penetrant examination of welded joints is available
for a fabricated liner.

. Because it is unclear whether there is sufficient gold in
the ERDA precious-metals account for lining the re-
ceiver tank liner, gold plating is favored. There is the

161

additional alternative of not lining the receiver tank, .

since corrosion of the receiver vessel by UF; in the salt
could be tolerated, and corrosion products could be re-
moved by hydrogen reduction and filtration between
runs. ' :

8.5 CONCEPTUAL DESIGN OF A MO-LTEN-SALT
BREEDER REACTOR FUEL PROCESSING
ENGINEERING CENTER

D. L. Gray* J.R. Hightower, Jr.

A conceptual design is being prepared to define the
scope, estimated final design and construction costs,
method of accomplishment, and schedules for a pro-
posed MSBR Fuel Processing Engineering Center
(FPEC). The proposed building will provide space for
the preparation and purification of fluoride salt mix-
tures required by the Molten-Salt Reactor Program, for
intermediate- and large-scale engineering experiments
associated with the development of components re-
quired for the continuous processing capability for an
MSBR, and for laboratories, maintenance work areas,
and offices for the research and development personnel
assigned to the FPEC.

*ORNL Engineering Division.

The project will consist of a new three-story engineer-
ing development center approximately 156 ft (47.5 m)
wide by 172 ft (52.4 m) long. The building will have a
gross floor area and volume of 54,900 ft> (5100 m?)
and 1,218,000 ft® (34,500 m?), respectively, and will
be constructed of reinforced concrete, structural steel,
concrete block masonry, and insulated metal paneling.
The building will be sealed and will be operated at nega-
tive pressures of up to 0.3 in. of H, O.(75 Pa) to provide
containment of toxic materials. The FPEC will be
located in the 7900 area approximately 300 ft (91 m)
west-southwest of the High Flux Isotope Reactor. The
engineering center will contain:

1. Seven multipurpose laboratories built on a 24 X 24
ft (7.3 X 7.3 m) module, for laboratory-scale experi-
ments requiring glove boxes and walk-in hoods;

A high-bay area, 84 X 126 ft (25.6 X 38.4 m),

equipped with a 10-ton (9000-kg) crane, for large-

scale development of processes and equipment for
fuel processing at the pilot-plant level;

. A facility for preparing and purifying 16,000 kg per
year of fluoride salt mixtures needed for the
Molten-Salt Reactor Program;

. Support facilities, including counting room, process

control rooms, change rooms, lunch and conference

room, and data processing room,;

Fabrication and repair shop, decontammatlon room,

and clean storage areas; _

A truck air lock to prevent excessive ingress of out-

side air during movement of large equipment items

into and out of the high-bay area;

Two S-ton (4500-kg) service elevators, one inside the

building to service the regulated areas and one out-

side to service the clean areas and to move filters to.
filter housings on the third floor and roof;

General service and building auxiliaries, including

special gas distribution systems, liquid and solid

waste collection and disposal, and filtered air-
handling and off-gas scrubbing facilities.

The experimental program planned for the building
involves large engineering experiments that use 238U,
232Th, Be, hazardous gases (F,, H,, and HF), molten
bismuth, and various fluoride and chloride salts. Ini-
tially, radioactivity will be limited to that necessary for
low-level beta-gamma tracer experiments. The labora-
tory area can later be upgraded, if desired, for use with
alpha-emitting materials at levels up to 1 kg of **°Pu.

The laboratory area will consist of seven 24 X 24 ft
(7.3 X 7.3 m) modular-type laboratories and a general-
purpose room. Bench-scale experiments of the type now
performed in Buildings 4505, 3592, and 3541 will be
carried out in these laboratories. Problems encountered
in the large-scale experiments can be studied via small
subsystems. Inert-atmosphere glove boxes will provide
space for examination of samples removed from both
the large and the small experiments. The laboratory area
will be maintained at a negative pressure of 0.3 in. of
H, 0 (75 Pa).

The high-bay area will be the main experimental area
where large engineering experiments will be performed.
Experiments will involve circulating molten mixtures of
LiF-BeF,-ThF,, lithium chloride, and molten Bi-Li
alloys. The experiments will also use elemental fluorine,
hydrogen fluoride, hydrogen chloride, and hydrogen
gases as reactants and will use purified argon for purg-
ing. Excess fluorine, hydrogen fluoride, and hydrogen
chloride will be neutralized in a caustic scrubber using
KOH solutions, and the cleaned and filtered off-gas will
be ducted to a building exhaust system. The experi-
mental equipment and components will be housed in
steel cubicles with floor pans which can contain any salt
spill. The cubicles will be maintained at a negative pres-
sure with respect to the high-bay ambient. The high-bay
area can be supplied with up to 45,000 cfm (21.2
m?/sec) of air. This air can be from recirculated inside

162

air or fresh air from the outside. The high-bay exhaust
system will be designed for 30,000 cfm (14.2 m?/sec) at
floor level and 50,000 cfm (23.6 m3/sec) at the roof
framing level. All exhaust ducts will contain fire barriers
upstream from the double HEPA filter banks.

The salt preparation and purification area will consist
of a 25-ft-wide by 35-ft-long by 14-ft-high (7.6 X 10.7
X 43 m) raw materials storage room, a 22 X 22 X

28-ft-high (6.7 X 6.7 X 8.5 m) room for weighing and

blending the salt constituents, and a 40-ft.-wide by
45-ft-long by 28-ft-high (12.2 X 13.7 X 8.5 m) room for
melting, H,-HF treating, and filtering the fluoride salt
mixtures. This facility should be capable of producing
16,000 kg per year of fluoride salt mixtures, using the
batch processing method in use at the facility at Y-12.

The estimated cost for the FPEC is $15,000,000, of
which $5,200,000 provides for inflation during the
three years required for design and construction of the
building.

The design is essentially complete, and the conceptual
design report is scheduled to be issued in September
1975. Authorization for this project will be proposed
for FY 1978,
Part 5. Salt Production

9. Production of Fluoride Salt Mixtures for MSR Program Research and Development

F. L. Daley

A salt production facility is operated by the Fluoride
Salt Production Group for preparation of salt mixtures
required by experimenters in the MSR Program. The
group is responsible for blending, purifying, and pack-
aging salt of the required compositions.

Much of the salt produced is used in studies on Hastel-
loy N development in which the concentrations of
metal fluorides, particularly nickel, iron, and chromium,
are important study parameters. It is thus desirable to
use salt in which the concentrations of these metal
fluorides are low and also reproducible from one salt
batch to the next. Oxides are undesirable salt contami-
nants primarily because of the adverse effect of uranium
precipitation, and also because of the effect of oxides
on corrosion behavior of the salt. Sulfur is another con-
taminant present in the raw materials used for preparing
salt mixtures. Sulfur is quite destructive to nickel-based
alloys at temperatures above 350°C, because a nickel—
nickel sulfide eutectic which melts at about 645°C
penetrates the grain boundaries and leads to intergranu-
lar attack of the metal. The maximum desired levels for
these contaminants in the fluoride salt mixtures are
iron, 50 ppm; chromium, 25 ppm; nickel, 20 ppm; sul-
fur, <5 ppm; oxygen, <30 ppm. Other duties of the
group include procurement of raw materials, construc-
tion and installation of processing equipment, and re-
finement of process operating methods based on results
from operation of the production facility.

When the facility was reactivated during 1974, initial
production was carried out in existing small-scale (8-in .-

*Consultant.

R. W. Horton*

diam) reactors while new large-scale (12-in.-diam) reac-
tors were being installed. Experience with both the
small and large units is summarized in the remainder of
this chapter.

9.1 QUANTITIES OF SALT PRODUCED

The 8-in.-diam reactor was used for production from
startup of the program in early 1974 through the first
three months of 1975. During this period, a total of
nine full-scale batches (315 kg total) were processed and
made available to investigators. Salt from the nine
batches was shipped in a total of 23 containers of ap-
propriate sizes. In general, operation of the 8-in.-diam
reactor proceeded smoothly, and the resulting salt was
of acceptable composition and purity.

Production in the 12-in.-diam reactor was started in
March 1975. Five production runs, each involving about
150 kg of salt, have been carried out. Of the five salt
batches processed, four were suitable for use; most of
the salt from these four runs was used for fuel proc-
essing experiments. In contrast to the earlier runsin the
8-in.-diam reactor, difficulty has been observed in the
12-in.-diam reactor with corrosion of dip lines in the
meltdown vessel and with increasing concentrations of
metallic impurities in the product salt.

9.2 OPERATING EXPERIENCE IN
12-in.<diam REACTOR

Operating data from the five production runsin the
12-in. reactor are summarized in Table 9.1. Analyses of
the resulting salt batches are given in Table 9.2. A
description of the processing operations and conditions

163
164

Table 9.1. Processiflg data derived from
flow rate, time, and titration of inlet and outlet flows

Batch Batch Total HF H, HF HF
size time in in out reacted
number {FS-
rES) kg (hr) (moles)  (moles)  (moles)  (moles)

During hydrofluorination

101 150 12.5 19.53 425.3 17.12 241
102 150 10.25 23.31 273.8 23.31 0
103 150 9.75 18.08 260.3 14.19 3.89
104 150 9.5 28.53 247.3 24.54 3.99
105 112 14.0 37.52 285.7 18.77 18.76
Batch  Total Total H,  Total HF HF in off-gas
Batch .

26 .
number (FS-) S1Z time n

(kg) (hr) {moles)

out (meq per liter of H,)

(moles) Start Finish

During hydrogen reduction

101 150 17.6 472.5 0.026 0.0018 0.0064
102 150 18.6 4995 0.595 0.100 0.050
103 150 244 652.0 1.560 0.625 0.016
104 150 44.0 1204.0 2.180 0.746 0.008

105 112 32.0 857.14 3.290 0476  0.102

Table 9.2. Analyses of 150-kg batches of
LiF-BeF, -ThF, (72-16-12 mole %)
produced in the 12-in. reactor

Analyses

Batch

number (FS-) Li Be Th F

(%) (%) (%) (%)

Fe Cr ‘Ni S 0
(ppm)  (ppm)  (ppm)  (ppm)  (ppm)

Nominal :

72-16-12 7.90 2.28 44.11 45.71
101 _ 7.95 2.52 4392 46.20
102 - 8.64 2.22 42.00 46.00
103 8.11 2.31 43.27 45.74
104 ' 8.39 2.00 43.81 45.32

105 10.46 290 3455 51.25

82 24 17 7.37 <25

75 30 600 8.0 360

60 25 8 2.5 350

85 65 10 9.1 110
820 125 .8 5.76

prevailing during hydrofluorination and hydrogen
reduction is given in the remainder of this section.

9.2.1 Charging and Melting of Raw Materials

The salt produced in the 12-in.-diam reactor has been
of the MSBR fuel carrier salt composition (72-16-12
mole % LiF-BeF,-ThF,;);, production of salt of this
composition will continue, except that some batches
will also contain 0.3 mole % UF,. If the production
schedule permits, an inventory of non-uranium-bearing
salt will be accumulated before beginning the produc-
tion of uranium-bearing salt. The LiF raw material for

the salt production facility is supplied by Y-12 as
needed; the BeF, and ThF, are taken from raw mate-
rials that have been on hand for several years. The only
apparent effect of the long storage time on the raw
materials is an increased moisture content of the BeF,.

The production unit includes two 12-in.-diam, 72-in.-
high, type 304 stainless steel vessels, each of which is
fitted internally with a full-length, open-top copper
cylinder in which the salt is contained. One vessel is
used for batch melting of the raw materials, which are
charged to the meltdown vessel by gravity transfer
through a 2-in.diam pipe; the pipe extends into a
weighing and charging room and is closed by a sealing
165

flange except during the loading operation. The second
unit, the processing vessel, is identical to the meltdown
vessel except for the charging line. Both vessels are
fitted with dip lines for introducing gas to the bottom
of the vessels for mixing or purifying a salt batch, and
both are connected to an off-gas system. Each vessel is
supported in a stainless steel liner, and while in use is
located in a heavy-duty electrical furnace. The receiver
vessel, to which the salt product is transferred,is 12 in.
in diameter, 36 in. high, and is supported similarly in a
furnace adjacent to the processing vessel furnace. Salt
transfer lines from the meltdown vessel to the proc-
essing vessel and from the processing vessel to the re-
ceiver are autoresistance heated via a 24-V power
supply.

The operational sequence includes salt charging, melt-
ing and mixing in the meltdown vessel, and transfer of
the resulting salt to the processing vessel for purifica-
tion. The process steps include hydrofluorination,
hydrogen reduction, and filtration during transfer of the
purified salt from the processing vessel to the receiver
vessel. During both the hydrofluorination and hydrogen
reduction steps, the receiver and processing vessels are
maintained at the same temperature, and the process
gases are passed through the receiver before being fed to
the processing vessel in order to eliminate any oxide
film on the interior of the receiver.

The raw materials are loaded into the meltdown vessel .

by technicians wearing air suits having a supply of
cooled fresh air. The work is carried out in a small,
enclosed room in which containment is maintained by
positive flow of air through the room to a bank of
absolute filters. The appropriate quantities of each of
the raw materials are weighed and charged through a
loading chute directly into the 12-in.-diam meltdown
vessel, which is at room temperature. The larger lumps
of BeF, and occasional lumps of LiF are broken by
hand to facilitate loading and to provide improved mix- '

ing. The ThF, is a fine powder which does not require
size reduction. The charging method leaves much to be
desired; melting would be more rapid and more predict-

‘able if the particle size of the raw materials could be

reduced and all components mixed well before they are
charged to the meltdown vessel.

Some of the more important impurities in the raw
materials are listed in Table 9.3. The values shown are
average values in most cases. The metallic impurities are
satisfactorily low; however, sulfur and possible silicon
contribute to corrosion problems during melting of the
raw materials. The moisture content of the raw mate-
rials is not shown but is an important parameter. It is
believed that hydrolysis of the fluorides during the
initial heating period generates hydrofluoric acid which
subsequently reacts with sulfur- and silicon-containing
compounds in the raw materials to form hydrogen sul-
fide and fluorosilicic acid. The quantities of these mate-
rials produced appear to be dependent on the tempera-
ture at which the meltdown vessel is held during the
initial portion of the melting operation, with H, S pro-
duction being most noticeable at temperatures above
500°C. An acidic compound, which contains silicon and
fluorine, is evolved freely at lower temperatures, in the
125 to 500°C range; however, the extent to which the
material is corrosive to the meltdown vessel is not
known. Analytical data necessary to determine whether
these low-temperature gases contain sulfur-bearing com-
pounds are not available.

The major effect of hydrogen sulfide on nickel com-
porients at temperatures in the range 600 to 700°C is
rapid embrittlement of the nickel. This action has
resulted in breakage of dip legs in the meltdown vessel
at the rate of one dip leg per run. Breakage is observed
to occur in the gas space above the melt, and the broken
dip leg falls to the bottom of the meltdown vessel,
where it is available for further attack by corrosive
materials dissolved in the salt. After melting, batch

Table 9.3. Impurities in raw materials used
in fluoride salt production (ppm)

S Si Fe Cr
Component
Avg Max Avg Max Avg Max Avg Max

LiF 21 44 100 100 20 25 <1 <1
BeF, 300 500 100 100 50 100 20 40
ThF, <100 <100 <10 <10 25 62 11 17
Mixed raw 100 131 47 47 26 56 9 15

materials?

Mixture required to produce salt having composition of 72-16-12 mole %

LiF-BeF,-ThF,.
166

FS-101 was passed through a nickel filter having a mean
pore size of 40 u, but plugging of the filter on sub-
sequent transfers led to its removal from the system.
Transfers from the meltdown vessel are now made after
allowing a period for particulate material to settle.

A stainless steel- dip leg was used in the meltdown
vessel during the melting of batch FS-105 in an attempt
to avoid cracking of the dip leg. The use of stainless
steel was later concluded to be unsuitable because of
the increased concentrations of iron and chromium
observed in the resulting salt product. The dip leg did
not embrittle nor break during melting of the salt, but
extensive corrosion was noted on the submerged por-
tion of the leg. As a result of these observations, a dip
leg of copper and nickel was constructed by placing a
copper sheath over a heavy-wall nickel tube. The nickel
tube provides rigidity, and the copper is used both out-
side and inside of the nickel tube to obtain resistance to
corrosion. The copper sheaths are welded together at
the lower end of the dip leg located in the meltdown
vessel. This combination. of materials is expected to
result in an increased dip leg life and less contamination
of the product salt.

An error was made in charging the ThF,; for batch
FS-105 which resulted in salt that did not have the
desired composition.

9.2.2 Hydrofluorination and Hydrogen Reduction

After a salt batch has been melted in the meltdown
vessel, it is transferred at a temperature of about 750°C
to the processing vessel, where it is sparged with an
HF-H, mixture at a temperature of about 625°C for a
period of about 10 hr. The salt is then sparged with H,
at 700°C; the H, flow rate of 10 std liters/min used
during the hydrofluorination step is continued for 30 hr
to reduce iron and nickel fluorides to their respective
metals.

Progress of the hydrofluorination step is monitored
by determining the HF content of the HF-H, inlet and
exit gas streams by absorption and titration of the HF
in a metered volume of exit gas. When the HF concen-
tration of the inlet and exit streams becomes equal (or
the concentration in the exit stream becomes slightly
higher than that in the inlet stream), contact of the salt
with the HF-H, mixture is stopped. A relatively low
temperature, about 100°C above the salt liquidus tem-
perature, is used to minimize the rate of corrosion of
equipment and to maximize the rate at which oxides are
hydrofluorinated. The hydrofluorination step is fol-
lowed by treatment of the salt with hydrogen at 700°C

to reduce iron and nickel fluorides. The utilization of
hydrogen during this step is low, and large volumes of
H, are required. Since the reduction reaction releases
HF, the concentration of HF in the off-gas stream is
monitored, and hydrogen treatment is stopped when
the HF concentration reaches a low value (about 0.02
meq/liter) and remains constant within the detection
limits of the titration method.

The total gas flows (H, and HF) during the processing
operations are shown in Table 9.1. The values in the
table reflect a steadily increasing quantity (from FS-101
to FS-105) of HF generated by H, reduction of metallic
fluorides that can be ascribed to a buildup of metals
(largely iron and nickel) in the salt heels in the melt-
down vessel and in the processing vessel during opera-
tion. These metals are converted to fluorides during
hydrofluorination, and thus add to the total quantity of
metal fluoride to be reduced during hydrogen treat-
ment.

9.3 SUMMARY

The information presented in the previous sections
indicates that the following factors are important in
producing high-quality salt:

1. Analyses of the raw materials indicate that there will
be no concern with metallic contaminants unless
metallic corrosion products are introduced during
the salt purification or melting steps. :

2. The sequential buildup of metallic impurities in the
salt produced in the 12-in.-diam facility is the result
of corrosion of the equipment. This corrosion can be
minimized by use of copper whenever possible when
equipment is simultaneously .exposed to salt and
process gases. Periodic hydrofluorination and discard
of flush salt in the processing vessel should control
any minor buildup of corrosion products.

3. Although not demonstrated by data shown in this
chapter, it is believed that oxygen contamination can
be held at low levels by maximizing the removal of
moisture from the raw materials before they are
melted and by improving control of the hydro-
fluorination process. A method for measuring the
H, O produced by reaction of HF with oxides in the
starting materials is being tested. This should aid in
determining the proper time at which to terminate
contact of the salt with the HF-H, mixture. Also, it
may be necessary to determine the sulfur content of
the off-gas, since this may be the most difficult con-
taminant to remove from the salt.
MOLTEN-SALT REACTOR PROGRAM

AUGUST 1975

L. E. McNEESE, PROGRAM DIRECTOR

r

*PART-TIME ON MSRP
*"DUAL CAPACITY

AC  ANALYTICAL CHEMISTRY DIVISION
C CHEMISTRY DIVISION
CT CHEMICAL TECHNOLOGY Di1VISION
M&C METALS AND CERAMICS DIVISION
R REACTOR DIVISION
UT  UNIVERSITY OF TENNESSEE

J.R. DISTEFANO"* M&C
TECHNICAL SUPPORT

K. W, BOLING® MEC
C.E. DUNN" MEC
W. H.FARMER"* M&C
J.C.FELTNER M&C
J.L.GRIFFITH® Mac
L. HEATHERLY* M&C
J. W, HENDRICKS® ME&C
T.J. HENSON® MEC
J. D. HUDSON® M&C
E.J. LAWRENCE Mac
E.H.LEE Mac
W. H.MILLER MEC
G.A.POTTER MEC
t.G.RARDON" M&C
R. H.SHANNGON Mac
L.J. SHERSKY" R
W. B.STINES M&C
C. K. THOMAS® M&C
L.R. TROTTER® MEC
C.A.WALLACE" R

[

.J. WOODHOUSE"

MSBR CHEMISTRY

SALT PRODUCTION

R. W HORTON

F.L.DALEY
W.J.BRYAN
A. L. JOHNSON

cT
cT
CcT
CcT

MSBR DESIGN AND DEVELOPMENT MATERIALS MSBR PROCESSING DEVELOPMENT
J. R ENGEL"" R H.E. McCOY"" M&C J.R. HIGHTOWER, JR.** cT
SYSTEMS AND ANALYSIS HASTELLOY N STUDIES CHEMICAL DEVELOPMENT
J.R.ENGEL"*" R H. E.McCOY"* M&C A.D.KELMERS"* [9
E.J ALLEN" R D. N. BRASKI M&C M. R. BENNETT c
H.T.KERR"® R J BRYNESTAD*® M&C
G.T.MAYS R R.E.CLAUSING" M&C ENGINEERING DEVELOPMENT
- R -
D.L.REED R.S.CROUSE ME&C J.R.HIGHTOWER, JR."* cT
J.R.D/STEFANO"* M&C C H. BROWN JR or
SYSTEMS AND COMPONENTS A.L.HEESTAND® M&C o COUNCE ’ oF
DEVELOPMENT C.R.HYMAN R o
J.R. KEISER M&C c.w. KEE’- cr
R. H. GUYMON R B‘ c' LESLIE" M&C R.B. LINDAUER CcT
W.R.HUNTLEY" R o H.C.SAVAGE CcT
B. McNA BB M&C
M. D.SILVERMAN R T K ROCHE MEC J. BEAMS cT
A.N.SMITH R e . R.O.PAYNE cT
R F BENSON R A.C.SCHAFFHAUSER M&C
E L BIDDLE R 1, W.WOO0DS M&C
D. J. FRAYSIER R M . ss
. E. ROBERTSON R CHEMICAL PROCESSING MATERIALS

L.M. FERRIS® c
A.D.KELMERS c
L. MAYA c
B.F.HITCH" c
S.CANTOR* c
L. M. TOTH* C
L.0.GILPATRICK" C
F.A.POSEY" C
H.R.BRONSTEIN c

D.E. HEATHERLY® c
D. Y, VALENTINE C
ANALYTICAL CHEMISTRY

RESEARCH AND DEVELOPMENT

A.S MEYER' AC
R.F. APPLE" AC
B.R, CLARK" AC
J.M.DALE" AC
D.L.MANNING AC
J.P. YOUNG* AC

ANAL YSES

L.T.CORBIN® AC
J.H.COOPER" AC
W.R. LAING" AC
J.A CARTER" AC

CONSULTANT

G. MAMANTOV uT

91
1-10.

11.
12.
13.
14.
15.
l6.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
32.
33.
34,
35.
36.
37.
38.
39.
40.
41.
42.
43.
44,
45.
46.
47.
48.
49.
50.

SRP Director’s Office

Bamb erger
. Beeson

Bronstein
. Brooksbank
.B

L. Crowley
L. Culler

M. Dale

L. Daley

H. DeVan
R. DiStefano

M
E.
R.
C.
C.
H.
J.

M.
M.
E.
A.
C.
D.
J.

M.
R.
C.
H.
R.
C.
J.

Ww.
S.

D.
J.

Ww.
B.
R.
J.

w.
J.

L.
J.

R.
J.

F.
J.

F.
J.

I

W. P. Eatherly

fikk
\

ORNL-5078

UC-76 — Molten-Salt Reactor Technol‘ogy

INTERNAL DISTRIBUTION

51-57.

58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
85.
86.
87.
88.
89.
90.
91.
92.
93.
94.
95.

96 102.

Ha ubenre1ch
Herndon

. Hibbs

. Hightower, Jr.
. Hill

R
. G.
. E.
.M.
. 0.
.R.
. G.
.S,
. H.
. 0.
. N.
. G.
F

=z~

. W. Hoffman
. P. Holz

. R. Huntley
. R. Hyman

W. Kee
. R. Keiser
. L.
.T.

. D. Kelmers

Ker

.R. Lai

. M. Leitnaker
. B. Lindauer
. 1. Lundin

. G. MacPherson
. E. MacPherson
. Mamantov

. L. Manning

goow:zwhsm>o~0w0€wmez“wwwéw€>EFFOOH
<
T
o
=
=

C. L. Matthews
L. Maya
r
103. H. F. McDuffie : 127. G.P. Smith
104. C.J. McHargue - 128. L. Spiewak
105. H. A. McLain 129. J. O. Stiegler
106. B. McNabb 130. R.E. Thoma
107. A.S.Meyer 131. A.J. Thompson
108. R. L.Moore ' 132. L. M. Toth
109. F. H. Neill 133. D. B. Trauger
110. P. Patriarca 134. D.Y. Valentine
111. T. W. Pickel ‘ 135. T.N. Washburn
112. C. B. Pollock 136. A. M. Weinberg
113. F. A. Posey . 137. J. R. Weir
114. H. Postma 138. J. C. White
115—116. H.P. Raaen 139. M. K. Wilkinson
117. D. L. Reed 140. W. R. Winsbro
118. T.K. Roche 141. J. W, Woods
119. M. W. Rosenthal 142. R. G. Wymer
120. H.C. Savage 143. G.T. Yahr
121. C. D. Scott 144. J.P. Young
122. W. D. Shults 145. E. L. Youngblood
123. M. D. Silverman 146—147. Central Research Library
124. M. J. Skinner 148. Document Reference Section
125. A.N. Smith 149—151. Laboratory Records
126. F.J. Smith 152. Laboratory Records, ORNL R.C.

EXTERNAL DISTRIBUTION

153. Research and Technical Support Division, ERDA, Oak Ridge Operations Office, Post Office Box E,
~ Oak Ridge, TN 37830
154. Director, Reactor Division, ERDA, Oak Ridge Operations Office, Post Office Box E, Oak Ridge, TN 37830
155—156. Director, Division of Reactor Research and Development, ERDA, Washington, DC 20545
157—258. For distribution as shown in TID-4500 under UC-76, Molten Salt Reactor Technology category (25
copies — NTIS)

v U.S. GOVERNMENT PRINTING OFFICE: 1976-748-189/236