ORNL-2973.txt

fi GENTRAL RESEARCH LIBRARY

T s T

3 4456 0361518 5

ORNL-2973
UC-81 - Reactors - Power ./X

¢ MOLTEN-SALT REACTOR PROGRAM
QUARTERLY PROGRESS REPORT

§ FOR PERIODS ENDING

‘ JANUARY 31 AND APRIL 30, 1960

CENTRAL RESEARCH LIBRARY
DOCUMENT COLLECTION

LIBRARY LOAN COPY
DO NOT TRANSFER TO ANOTHER PERSON

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

OAK RIDGE NATIONAL LABORATORY
operated by

UNION CARBIDE CORPORATION
for the

U.S. ATOMIC ENERGY COMMISSION
Department of Commerce
Waoshingion 25, D.C.

LEGAL NOTICE

This report was prepared as an account of Government sponsored work. Neither the Uni

nor the Commission, nor any person acting on behalf of the Commission:
A. Makes any warranty or representaf

completeness, or usefulness of the information contained in this report, or that the use of

n, expressed or implied, with respect to the accuracy,

any information, apparatus, method, or process disclosed in this report may not
privotely owned rights; or
B. Assumes any licbilities with respoct o the use of, or for damages resulting from the use of
any information, apparstus, method, or process disclosed in this report.
As used in the cbove, “person acting on behalf of the Commission’” includes any employee or
contractor of the Commission, or employee of such contractor, 1o the extent that such employ
or contractor of the Commission, or employee of such controctor prepores, disseminates, or

provides access to, any information pursuant to his employment or contract with the Commission,
or his employment with such contractor.

ORNL-2973
UC-81 — Reactors - Power
TID-4500 (l5th ed.)

Contract No. W-7405-eng-26

MOLTEN-SALT REACTOR PROGRAM

A Y

QUARTERLY PROGRESS REPORT

For Periods Ending January 31 and April 30, 1960

H. G. MacPherson, Project Cocordinator

DATE ISSUED

AUG 26 1960

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

MARTIN MARIETTA ENERGY SYSTEMS LIBRARIES

EREE IR

3 Y456 03L1518 ¢

MOLTEN-SALT REACTOR FROGRAM QUARTERLY PROGRESS REPORT

SUMMARY

Part 1. Engineering and Component Development

1.1 Component Development and Testing

Tests of molten-salt-lubricated hydrodynamic bearings continued in
bearing test rigs and in the sump-type centrifugal pump. Operation of the
type PKP pump, circulating NaK, with a modified Fulton Sylphon bellows
seal, was suspended after 18,834 hr of continuous operation for examina-
tion of the lower seal. The seal was found to be worn and was replaced.
The MF-type pump has circulated molten salt continuously for more than
25,000 hr without maintenance.

The small frozen-lead-sealed centrifugal pump has circulated molten
salt continuously at 1200°F for 16,500 hr.

Molten salt was circulated at 1200°F in the Remote-Maintenance De-
velopment Facility. Following the hot operation the salt was dumped, and
all maintenance operations were carried out remotely. A color motion pic-
ture with sound, demonstrating the remote removal and replacement of all
major components, was completed.

Twenty-five forced-circulation corrosion loops have been placed in
operation in the MSR corrosion program. Of this number, five INOR-8 loops
and five Inconel loops have now been terminated for metallurgical examina-
tion; the ten loops had operated an average of 10,250 hr each.

A frozen-salt-sealed graphite-to-metal Joint was tested. No leakage
past the frozen joint was experienced although molten salt leaked through
the porous graphite tube at 18 psi pressure.

An in-pile graphite-salt capsule was removed from the MIR after 720
hr. It is awaiting hot-cell examination. A second test continues to op-
erate in the MI'R and has accumulated more than 1000 hr of exposure at MIR

power.

1.2 Engineering Research

The enthalpy of the BULT-14 mixture (LiF-BeF,-UF,-ThF,, 67-18.5-0.5-14

mole %) was determined over the temperature range from 100 to 800°C. TFor

the ligquid phase, the heat capacity varied from 0.367 Btu.1b 1. (°F)~! at
1000°F to 0.302 Btu-1b™t.(°F)”* at 1500°F; the heat of fusion was 80.1
Btu/lb. Liquid heat capacities for the BeF,-containing salt mixtures
studied to date have been summarized. The thermal conductivity for an
INOR-8 rod specimen was experimentally established up to a temperature of
900°F. The results are in reasonable agreement with the thermal conduc-
tivities of several related metal alloys. The surface tension of an NakF-
BeF, (64-36 mole %) mixture was measured over the range from 600 to 800°C.
Comparison with data obtained previously on several LiF-BeF,-ThF,-UF,; mix-
tures shows the current results to be somewhat higher. TImprovements in
the experimental procedure and in the instrumentation have led to in-
creased precision in the maximum-bubble-pressure technique for surface-
tension determination. Liquid densities for the NaF-BeF, mixture obtained
in the course of the surface-tension experiment agree with empirical den-
sity estimates to within 7%.

Additional data have been obtained for heat transfer to BULT-14 flow-
ing within heated Inconel and INOR-8 tubes. Approximately 4500 hr of op-
eration have been accumulated under primarily isothermal (1150°F) condi-
tions. No systematic variation of the heat-transfer coefficient has been
observed during this time. Reanalysis of the data using recent values for
the heat capacity led to an improvement in the comparison between the BULT-
14 results and the general heat-transfer correlations, while adversely af-
fecting the experimental heat balances. The previously reported discrep-
ancy between the heat transfer in the Inconel and INOR-8 tubes has been
resolved. Mean values of the heat-transfer modulus for both sets of data
nov agree closely; the Inconel-tube data, however, exhibit a greater ex-

perimental scatter.

Part 2. Materials Studies

2.1 Metallurgy

Examinations of nine INOR-8 thermal-convection loops, which operated
with fused fluoride mixtures for one year, were completed. Five of the
loops showed no attack, and the other four exhibited very light attack in

the form of shallow subsurface voids. Seven Inconel thermal-convection

loops were also examined, all of which showed attack in the form of inter-
granular voids to depths ranging from 4 to 15 mils.

Examinations of four Inconel forced-convection loops, which operated
for different periods with fluoride salt of the LiF-BeF,-UF, system, re-
vealed heavy subsurface-void formation in regions where locp wall temper-
atures exceeded 1200°F. Maximum attack ranged to 7 mils after 3000 hr of
cperation and to 15 mils after one year of operation.

Metallographic examinations were completed on three INOR-8 forced-
convection loops which operated with fluoride mixtures for periods of one
year or longer; with the exception of one loop, negligible attack was
found. In that one loop, hot-leg attack, in the form of surface roughen-
ing and surface pitting, was found to a maximum depth of 1-1/2 mils, Ex-
amination of the cold-leg regions of this loop revealed the presence of
metal crystals, the composition of which was determined to be predominantly
nickel, The fuel mixture circulated in the lcop showed appreciable oxide
contamination, apparently as a result of H,0 entering the system during
operation.

Chemical and metallographic examinations were cerried out to investi-
gate the thin corrosion film which has appeared in a majority of the long-
term INOR-8 corrosion loops. Qualitative chemical analyses indicated the
film to be composed primarily of nickel, with smaller amounts of iron,
chromium, and molybdenum. Hardness data obtained indicated that the film
was approximately twice as hard as the base metal.

The 48b Ti—48% Zr—4% Be alloy was used to braze an assembly of graph-
ite tubes to an INOR-8 header in order to demonstrate the feasibility of
fabricating such an assembly.

Alloys in the Au-Ni-Ta and Au-Ni-Mo ternary systems, which are re-
sistant to corrosion by molten fluorides and which may be useful in cer-
tain applications, are under study for brazing graphite-to-graphite and
graphite-to-molybdenum joints. Several alloys that melt at about 1300°C
and exhibit excellent flowability have been developed. However, prelimi-
nary thermal-cycling and leak tests, coupled with metallographic examina-
tions, indicate that these alloys are brittle and tend to crack. Thus
alloys with somewhat less flowability but more ductility are being con-

sidered for future study.

Difficulty has been experienced in brazing low-porosity graphite, in
contrast to experience with highly porous grades. More suitable joint de-
signs and/or new, stronger bonding alloys are being investigated to over-
come this difficulty.

Thirty-one different grades of graphite have been tested for their
resistance to permeation by molten fluorides in 100-hr screening tests at
1300°F and pressures of 150 psig. Twenty-four of the grades were manu-
factured as low permeability; and of these, four grades, B-l1, S-4-1IB, GT-
123-82, and CS-112-S, had less than 0.5% of their bulk volume permeated
by molten fluorides in the screening test. It appears that for some
grades, significant decreases in molten-fluoride-salt permeation can be
accomplished by proper pressure reductions. Moderately-low-permeability
grades (R-4 and $S-4) were permeated to the same extent by molten fluoride
salts under 150 psig at 1300°F (704°C) in l-hr exposures as they were in
100-hr exposures.

Grade AGOT graphite was impregnated with molten bismuth and subse-
quently subjected to a standard fuel-permeation test. Results showed that
the bismuth pretreatment suppressed the fuel pickup, but the bismuth was
not completely retained in the graphite pores during the fuel-permeation
test.

,@( Two tests have been conducted to remove from graphite the oxygen con-
tamination that causes a portion of the uranium of LiF-BeF,-UF, (62-37-1
mole %, fuel 130) to precipitate as UO, when this fuel is exposed to the
graphite at 1300°F (704°C) in a vacuum. A 50-hr exposure of grade AGOT
graphite to hydrogen preheated to 1300°F (704°) failed to detectably de-
crease the oxygen contamination. However, the thermal decompoéition 6f
crystals of ammonium bifluoride inside a grade AGOT graphite crucibié ap-
parently was successful 1n removing the oxygen contamination. Fuel.lBO
has been exposed for 1000 hr to this graphite at 1300°F (704°C) without
any uranium precipitation that could be detected by radiography.

INOR-8 was carburized to a depth of 14 mils while in direct contact
for 3400 hr with grade TSF graphite at 1000 psi in fuel 30 (NaF—ZrF4—UF4,
50-46-4 mole %) at 1300°F (704°C).

vi

Three nickel-base brazing alloys (Coast Metals No. 52 and 53, and

General Electric No. 8l), the 82% Au—-18% Ni alloy, and the pure copper
brazing material showed good corrosion resistance to fuel 130 (LiF-BeF,-
UF,, 62-37-1 mole %) at 1300°F (704°C) during a 10,000-hr thermal-con-
vection loop test of INOR-8 Jjoints. It was noted that the commercial
nickel-base alloys were corroded when used with Inconel as the base metal.
Results of initial corrosion studies on experimental Au-Ni-Ta alloys have
indicated that they have good resistance to both the NaF-ZrF,— and the
LiF-BeF,—base fuels. These alloys are being developed to investigate

possible graphite-to-graphite and graphite-to-metal joining methods.,

2.2 Chemistry and Radiation Effects

According to preliminary checks on phase behavior, the substitution
or addition of ZrF, in LiF-BeF, fuels containing UF, and ThF, appears to
be a feasible route to potential improvements in corrosion behavior and
perhaps oxide tolerance while at the same time retaining favorable in-
ventories and physical properties, Such quinary fuels can be regarded as
ARE NaF-ZrF, fuels modified as follows: Substitute 67 mole % Li7F for all
the NaF and substitute BeF, for half of the ZrF, to reduce both the melt-
ing point and the ZrF, volatility, giving LiF-ZrF,-BeF, in a mole ratio
of 6-1-1; either UF, or ThF, may then be interchanged with ZrF, as de-
sired, with only minor alterations of the over-all chemical and physical
properties (except density, which increases by 50% for complete inter-
change with the heavy elements),

Differences, as well as similarities, in the behavior of UF, and ThF,,
particularly when in combination with an alkali fluoride such as NaF, have
been explored, and some unusual solid solutions have been found.

Selective precipitation of oxides continues to appear promising as a
means of processing breeder fuels to remove protactinium and uranium from
fluoride melts.

Unintended precipitation of oxides continues to be a very trouble-
some problem; qualitative information in this connection has been obtained
by examining quenched samples from high-temperature equilibrations con-

taining small amounts of added oxide. The margin of tolerance for oxide

vii

without precipitation appears to be quite low, and in addition, most
sources of oxide contamination during operation and maintenafiée also
cause corrosion in direct proportion to the amountuof oxide added.

In order to evaluate the applicability of information about dif-
fusion rates of Cr® and their role in controlling the rate of corrosion
of INOR-8 by fluoride fuels, two anisothermal convection loop tests have
been studied under conditions such that the diffusion behavior could be
followed. The agreement between calculated and experimental behavior was
very satisfactory.

Much of a long-standing discrepancy in the behavior of NiF, has been
resolved through a redetermination of the free energy of formation of NiF,
by means of experiments in which HF corrodes Ni9 under conditions such
that the equilibrium constant for the reaction can be measured accurately.
Periodic sampling and analysis of the corrosion products found in long-
term forced-circulation corrosion test loops has been continued; some of
the results are uninterpretable in terms of the expected steady-state be-
havior, but most of such unpredicted deviations are sufficiently pronounced
to be regarded as reflecting experimental difficulties rather than inherent
behavior of the intended systen.

In exploring the conditions under which intercalation of graphite by
salts can occur, no effect was found with NaCl-FeCl; (70-30 mole %) at
750°C or with NaCl-FeCl; (46-54 mole %) at 200°C for 7-1/2 hr, although
pure FeCl; crumbles graphite at 300°C in 2 hr or less. Both high temper-
atures and dilution of the intercalatant are presumed to diminish the tend-
ency toward reaction.

Qut-of-pile studies with fuels containing tracer rare earths in con-
tact with graphite have been initiated to provide a background for an in-
vestigation of permeated samples which have been irradiated.

The use of purified fuels in engineering and corrosion tests has been
greater than had been anticipated, so that a stockpile of fuels of current

interest no longer exists.

2.3 Fuel Reprocessing

Solutions of NO, in liguid anhydrous hydrogen fluoride appear attrac-

tive as an alternative to EF-H,0 sclutions for processing MSR fuel salt

viii

containing little or no thorium. Under proper conditions the NO,-HF so-
lutions dissolve UF, along with the LiF and BeF,, and decontamination from
rare earths is satisfactory.

Hot saturated aqueous ammonium fluoride soluticns have a reascnable
solubility for thorium fluoride, but the solubility of LiF-BeF,-ThF,-UF,
salts 1s much lower. No adequate processing method has yet been found for
fuel salts containing ThF,.

In a single experiment no protactinium was removed from MSR blanket

salt by fluorination at 650°C.
1.1
ll2
) 2.1

PART 1. ENGINEERING AND COMPONENT DEVELOPMENT

COMPONENT DEVELOPMENT AND TESTING:cesosrsoceocsossoansss coae

Molten-Salt-Lubricated Bearings for Fuel PumpS.eseesee.. con
Hydrodynamic Journal BearingS..eeeceseesersscascsascnases
Test of Pump Equipped with One Molten-~Salt-Lubricated

Journal Bearing...eeseeecesesansescaasssossssasronnossns
Self-Welding of INOR—8..................................

Mechanical Seals for PumpS...... e teeseaesererarernsan s

.Loop for Study of MF-Pump PerfOormancCe....ceeiverenssccscsncss

Frozen-lead Pump Seal...eovenven ceasessesnse I
Remote~-Maintenance Development Facllityeeseeieeriooesenneaans

Design, Construction, and Operation of Materials Testing

LOOPS IIIII 5 & ¢ 5 & 9 & 0 0 & 2 4 b & P b e b s ae b b e ¢ b e s s 20 w0 * 8 8 F ® 5 0 8 88 . 98
Forced-Circulation Corrosion LOOPS..cesvecsosss ceanaas ‘e
Mechanical JOIntS.eceetetoseccatssscaconncas et tsssstetanuns

Design and Testing of a Frozen-Salt Metal«to-Graphite

Joint.......... Cresessuesnas treetssecseernassarasereesens
In—Pile LOOPS..-..--.noooocooluotoo.o.cooc-ooooc cccccc o0 a
ENGINEERING RESEARCH:+vsecesvcons certaetates e asranseveoanae
Physical-Property Measurements.......... tecsseccseanetaanns

Enthalpy and Heat Capacity.seesviesraneenicananns cesasene
Thermal Conductivity...... ceeceteccctsacaaraeas casesnnas
Surface Tension.eeessessesesnnees cesicesassssassacnsanne

HeateTransfer StUdlES e eeesssasnssesssossesvsssnssssssssscanss

PART 2. MATERIALS STUDIES

METATIURGY -« c.. .. Ceheereere et eeetaiastaa et .-
Dynamic Corrosion StuUdieS..csvssrssssssssasassnssssraassnse
Thermal-Convection LoOpsS...... Ceee st esetesescccaenans .o
Inconel Forced=-=Convectlion LOOPS:iseeeetscsns beseerssenas
INOR-8 Forced-Convection LOOPS..... crereennan Ceeserreans
Graphite Brazing StudieS..ceieerectseecscsannnens ssecssanacs
General Corrosion StudieS...eceeeeeercescanes cecsrseveranss

Permeation of Graphite by Molten Salts cersseersenes
Removal of Oxygen Contamination from Graphite....ccoeevs
Compatibility of INOR-8 and Graphite..ceseseeciecrocecnan
Corrosion Resistance of Brazing Alloys to

Fluoride Fuel..ceeeeeaaans cees

iii

W w

& O 2 2 vw

2.2 CHEMISTRY AND RADIATION EFFECTS........

2.3

xii

Phase Equilibrium Studies.....ccceveen
The System LiF-BeF,=ZrF,evevieceenns

The System NabF-Zr¥F,-ThF,

The System NaF-ThF,=UF ccccceatecnas

Phase Diagrams for Fluoride Systems....

Fission-Product Behavior and Reprocessing Chemistry....
Extraction of Uranium and Protactinium from

LiF-BeF,~-ThF, Mixtures........... “eo

Chemistry of Corrosion Processes..... ‘o

Effect of Oxide on LiF-BeF,-ThF, and LlF-Bng-ThF4-

UF,; Mixtures...

Radiotracer Studies of the Role of Cr

Diffusion as a

Controlling Factor in Corrosion by Molten Fluorides....

The Free Energy of Formation of NiF...

Anelysis of Corrosion Test LoopsS....
Corrosion Equilibria with Unalloyed 03 SR
Melting Points of Chromous Fluoride and Chromic

Fluoride...... ceceiecacaanananna .
Graphite Compatibility.ceececnceeesans .
Intercalation of Graphite....... saeen

Permeation of Graphites by Molten Fluoride Fuels....

Radiation Effects...

Preparation of Purified Materials.......
Purification, Handling, and Service Operations

FUEL REPROCESSING. ..

NO,-HF Solutions......
Aqueous NHiFeeveaoan..

Protactinium Removal

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

® & ¥ B 8 B 6 0 8 B 2 b & % e b8

Behavior of a Rare-Earth Fission Product in a
Graphite—Molten-Salt System.......

65

65
65
65
66
68

68

68
74

T4

75
79
80
81
82

82
82
83

84

84
86
86
87
87
87
88

PART 1

ENGINEERING AND COMPONENT DEVELOPMENT

o

1.1 COMPONENT DEVELOPMENT AND TESTING

Molten-Salt-Lubricated Bearings for Fuel Pumps

Hydrodynamic Journal Bearings

Investigations were continued with Journal bearings operating sub-
merged in molten salt 130 (LiF-BeF,-UF,, 62-37-1 mole %). Test 22, re-
ported previously,l was completed. Test 18 was repeated, and two addi-
tional endurance tests were begun. A summary of the test conditions is
presented in Table 1.1.1 (see also Fig. 1.1.1).

Test 22 was terminated on schedule after 494 hr. During the latter
- part of the test, 170 start-stop cycles were performed.

Test 23 was a repeat of test 18, with new parts, with approximately
the same results. Impending seizure was evident at loads above 25 lbyr;
so the test was terminated.

Test 24 was performed to investigate the endurance of the bearing
at steady-state operating conditions. After 80 hr of operation the drive
motor experienced a sudden power surge, and the test was suspended for
examination. Examinaticn revealed that the flow restriction at the dis-
charge end of the grooves was not sufficient to prevent the bearing from
becoming "starved."

Test 25 was a repeat of test 24, with new bearing parts. In this
case the bearing grooves had a radius of 1/8 in. compared with the 1/4
in. in test 24. The test was terminated after 497 hr of operation be-

- cause of selzure of the bearing. Examination revealed that the flow re-
striction at the discharge end of the grooves was not concentric with

" the bearing bore and therefore the bearing rubbed against the Journal,
causing the seizure. Steps are being taken to ensure concentricity with

future bearings.

Test of Pump Equipped with One Molten-Salt-Lubricated Journal Bearing

Three additional tests were completed on the pump with a molten-
salt-lubricated Jjournal bearing, and ancther test is in progress. A

summary of these tests is presented in Table 1.1.2.

'MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 3-5.

(#8]
Table 1.1.1. Summary of Test Conditions for Testing Hydrodynamic Journal Bearings

Test Material Groove Speed Operating Radial Opergtlng Start-

No. Confipurations* (rpm) Temperature Load Time Stop

Bearing Journal (°F) (1be) (hr) Cycles

22 Inconel Inconel Axial; 1200 1200 200 494 170
radius, 1/4 in.

23 INOR-8 INOR-8, Axial; 1200 1500 25 3 1
carburized radius, 1/4 in.

24 INOR-8 INOR-8, Helical; 1200 1200 200 80 1
carburized radius, 1/4 in.

25 INOR-8 INOR-8, Helical; 1200 1200 200 497 1
carburized radius, 1/8 in.

*A11l bearings contained three grooves. See Fig. 1l.1l.1

for groove details.
UNCLASSIFIED
ORNL-LR-DOWG 38968AR

Fig. 1.1.1. Groove Details of Hydrodynamic
Journal Bearing.

The Jjournal bearing in test 3 experienced a seizure after 11 hr of
operation, and subsequent examination revealed that foreign material
entered the bearing clearance space, resulting in the seizure.

In tests 4, 5, and 6, the three-helical-groove bearing configuration
mounted with gimbals was utilized. After 1000 hr of operation in test 4,
the bearing and Jjournal were fully suitable for further operation. At
disassembly, the Journal was damaged in removal from the shaft and there-~
fore cculd not be re-used. The bearing and Jjournal used in test 5 were
fully suitable for further use and are under test in test 6. This test
has operated for more than 850 hr, to date, and pump rotation has been

stopped and started 41 times.

Self-Welding of INOR-8

One INOR-8 self-welding experiment was performed at 1500°F, on loose-
fitting pins pressing against the sides of the holes, a condition encoun-
tered with gimbal-mounted bearings. The test was performed in a bath of
salt 107 for 373 hr, under a load of 200 lbp. Examination of the test

pieces revealed no evidence of self-welding.

Table 1.1.2. Tests of Pumps in Molten Salt 130
Test 1 Test 2 Test 3 Test 4 Test 5 Test 6
Bearing grooving 3, axial 3, helical 3, axial 3, helical 3, helical 3, helical
Bearing mount Rigid Gimbals Rigid Gimbals Gimbals Gimbals
Room-temperature 0.005 0.005 0.005 0.005 0.005 0.005
radial clearance,
in.
Temperature of molten 1200 1200 1200 1200 1200 1100-1350
salt, °F
Shaft speed, rpm 1200 1150 1200 1400 1200 800—1400
Salt flow, gpm 50 50 50 100 50 50-260
Estimated radial load 100 100 100 120 100 75—-150
of bearing, 1lbg
Operating time, hr 1/3 0 11 1000 105 850 %
Number of start-stops 1 1 1 2 100 41 %
Reason for termination Seizure Seizure Seizure Schedule Schedule

*Test still in progress.
Mechanical Seals for Pumps

The operation of the modified Fulton Sylphon bellows-mounted seal,2
undergoing test in a PKP type of centrifugal pump circulating NaK, was
suspended due to an increase of leakage of oil from the lower seal of the
shaft. The total operating time was 18,834 hr at 1200 to 1225°F. The
average 0il leakage from the upper and lower seals, respectively, was 0.6
and 0.8 cc/day. Unaccounted~for oil leakage averaged 35 cc/d&y; presum-
ably, 1t leaked into the process system. Examination of the pump rotary
assembly revealed that (1) the lower seal faces had worn considerably;

(2) the upper seal was in good condition; and (3) the impeller was in good
condition, but with an etched appearance in the flow passages. The normal
drainage passages to the catch basin for oil leakage past the lower seal
had become filled with a viscous, greaselike substance. This plugging may
well have diverted the unaccounted-for oil leakage downward through the
pump-shaft annulus and into the system. The plug material was a mixture
of carbon and the oxides of sodium and potassium. The pump rotary assem-
bly, with modifications to the shaft lower seal, is being prepared for

further test circulation of Nak.

Loop for Study of MF-Pump Performance

An MF type of centrifugal pump has continued in operation? and has
logged more than 25,000 hr of continuous operation. Since the previous
report period, the pump has continued operating in a region of cavitation
at 2700 rpm, 645 gpm, and 2.5-psig pump-tank cover-gas pressure. Molten
salt 30 at 1200°F is being pumped. The average seal-oil leakage collected
from the upper and lower seals was 26 and 16 cc/d&y, respectively. An un-
accounted-for loss of oil averaging 56 cc/day has been observed. Presum-
ably, it leaks into the process system. The pump was stopped nine times:
two times for 10 min each to replace brushes in the drive motor; two times
for 10 min each to replace brushes in the motor-generator set; three times
for 5 min each to replace the air filters in the motor-generator set; and

two times for 3 min each during a scheduled power outage.

“MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 9.

Frozen-Lead Pump Seal

The small frozen-lead-sealed pump,3 consisting of a centrifugal pump
mounted vertically over a fractional-horsepower motor drive, has operated
continuously during this period to accumulate 16,500 hr of continuous op-
eration. No seal leakage occurred, and no maintenance was required during

this period. The pump circulates molten salt at 1200°F.

Remote-Maintenance Development Facility

A remote-maintenance facility incorporating a high-temperature
(1200°F) mockup of a 20-Mw (thermal) molten-salt-fueled reactor has been
constructed (see Fig. 1.1.2). The purpose is to investigate the problems .
involved in remotely maintaining a reactor system which cannot be ap-
proached for direct maintenance because of radiocactive fission products
and induced activity in the reactor components. Tools, techniques, and
procedures have been developed for removing and replacing all maJjor com-
ponents, including heat exchangers, the primary fuel pump, the reactor
core vessel, pipe preheaters, and piping sections. All maintenance oOp-
erations are performed by a single operator from a remotely located con-
trol center, using closed-circuit television as the only means of viewing

(Fig. 1.1.3).

Prior to circulating molten salt in the system, techniques and pro-
cedures were checked out by repeated performance of the required mainte-
nance operations. Numerous modifications were incorporated in the loop,
as the need arose, either to improve visibility through the cameras or
to facilitate maintenance operations. A specially designed remotely op-
erated pneumatic bolt runner and torque tool was developed to loosen and »
tighten the bolts on the 3 1/2- and 6-in. freeze flange clamps (see Fig.

1.1.4).

After all tools were developed, and when all operations could be
performed by the operator, 120 gal of a molten-salt mixture consisting
of sodium fluoride, zirconium fluoride, and uranium fluoride was circu-

lated through the loop at 1200°F. The run was terminated and the molten

3MSR Quar. Prog. Rep. Oct. 31, 1958, ORNL-2626, p 23.

UNCLASSIFIED
ORNL-LR-DWG-46667-A

REMOTE MAINTENANCE DEVELOPMENT FACILITY
LEGEND

GENERAL MILLS MANIPULATOR
OVERHFEAD TRAVELING CRANE -5 TON
MOTOR LIFTING SLING
TOOL RACK
D.C. MOTOR - 30 H.P.
CENTRIFUGAL SUMP PUMP
RACK FOR HEATER AND THERMOCQUPLE DISCONNECTS
ELECTRIC HEATER AND THERMOCQUPLE DISCONNECTS
REACTOR VESSEL MOCK-UP
. HEATER - INSULATION UNITS
FREEZE FLANGE JOINTS FOR 3 ' in. PIPE (Total of 16}
. TELEVISION CAMERA WITH AUTO-ZOOM LENS
PIPE SUPPORT
14. FREEZE FLANGE JOINTS FOR 6 in. PIPE [ Total of 2)
15. FREEZE FLANGE JOINTS FOR % in PIPE (Total of 2)
16, SUMP TANK LIFTING SLING
17 SUMP TANK
18 AUXILIARY SCREW JACK FOR HEAT EXCHANGER
19. HEAT EXCHANGER MOCK-UP
20 LIFTING SLINGS AND DOLLIES FOR HEAT EXCHANGERS
21. STEREO TELEVISION CAMERAS
22. TRACK FOR TELEVISION CAMERAS
23 FILTER FOR FREEZE FLANGE AIR COOLING SUPPLY
24, BRIDGE FOR MANIPULATOR DOLLY
25 TRACK FOR MANIPULATOR BRIDGE
26, CELL LIGHTS
27 CONTROL ROOM
28 STEREO TELEVISION RECEIVERS
29 PRISMS OF STEREO VIEWER
30 TELEVISION CAMERA AND CAMERA DOLLY CONTROLS
31. MANIPULATOR CONSOLE
32. CONTROL VALVE FOR PNEUMATIC TOOLS
33 OVERHEAD CRANE CONTROLS
34. SOUND AMPLIFIER FOR CELL MICROPHONES

Gp=—QPENOORwN -~

Fig. 1.1.2. Remote-Maintenance Development Facility.

UNCLASSIFIED
PHOTO 35815

\

Fig. 1.1.3. Closed-Circuit-Television Stereo Viewer.

10
salt returned to the sump tank, as scheduled, after 3 hr of salt circu-
lation.

Following hot operations, maintenance operatlions were repeated in
order to determine the effect of salt contamination on maintenance oper-
ations to be performed on the circulating pump for the molten salt, the
pump motor, the reactor vessel, the heat exchanger, and the pipe pre-
heater. These components were successfully removed, and work is con-
tinuing on their replacement. (See Figs. 1.1.5 through 1.1.8.)

To remove and replace a component, it must be possible to remotely
break and remake pipe connections to leak-tight specifications. The
freeze flange developed at the Laboratory is used at 20 locations in this
system. Such a joint takes advantage of the ability of the molten salt
to form a frozen seal between flange faces (see Fig. 1.1.9).

Shooting of the color motion picture, "Remote Maintenance of Molten-

Salt Reactors," was completed.

Design, Construction, and Operation of Materials Testing Loops

Forced=-Circulation Corrosion Loops

The operation of long-term forced-circulation corrosion-testing loops
was continued. Including this and previous report periods, a total of 25
of these looPS,4 in various configurations and with various fluoride salts,

have operated in the MSR corrosion program. O0f this number, five each of

INOR-8 and of Inconel have been terminated for examination, after an av-
erage of 10,250 hr of operation. Three have been terminated for examina-
tion after less than a year of operation.

Loops presently operating and those terminated during this period
are shown in Table 1.1.3. Eight of the operating loops have exceeded a
year, while three more recently operated loops have less than a year of
operation at AT conditions.

Three INOR-8 loops, 9354-1, MSRP-13, and 9354-4, were terminated
during this period. INOR-8 loop 9354-1 was terminated after 14,378 hr

when a salt leak occurred in the second heater section. Loop MSRP-13 was

“MSR Quar. Prog. Reps. from Jan. 31, 1958, ORNL-2475, through Oct.
31, 1959, ORNL-2890.

12

UNCLASSIFIED
PHOTO 35816

Fig. 1.1.5. Heat Exchanger After Salt Run.

Fig. 1.1.6. Pump After Salt Run.

14

gy

4

Fig. 1.1.7. Reactor Vessel Suspended.

UNCLASSIFIED
PHOTO 35378

15

Fig. 1.1.8. Pipe Preheater After Salt Run.

UNCLASSIFIED
PHOTO 35817

Fig. 1.1.9. Freeze Flanges on Heat Exchanger After Salt Run, Showing Salt Seal.

Table 1.1.3. Molten-Salt Forced-Circulation Corrosion-Loop Operations Summary as of April 30, 1960

>
Loop Loop C;fiigzit;gn Hours of Operation Tg;gngilie o ent
Designation  Material . at Test Conditions PU u omments
Fluidx (°F)

9354-3 INOR-8 84 16,586 1200 Resumed operation after pump
change

9354-4 INOR-8& 130 15,140 1300 Terminated 5/3/60 for remov-
al of final sleeve insert
(see ref 5)

MSRP-7 INOR-8 133 14,713 1300 Normal operation

MSRP-6 INOR-8 134 14,636 1300 Normal operation

9354=1 INOR-8 126 14,378 1300 Terminated 11/27/59

MSRP-10 INOR-8 135 13,879 1300 Normal operation

MSRP-11 INOR-8 123 13,607 1300 Normal operation

MSRP-12 INOR-8 134 13,075 1300 Normal operation

9377-5 Inconel 134 12,353 1300 Normal operation

9377-6 Inconel 133 9,835 1300 Normal operation

MSRP-13 INOR-8 136 8,110 1300 Terminated 3/14/60

MSRP-15 INOR-8 BULT-14 4,493 1400 Contains double-walled in-
serts (see ref 6)

MSRP-14 INOR-8 BULT-14 4,255 1300 Contains double-walled in-
serts

MSRP-16 INOR-8 BULT-14 2,081 1500 Began operation 11/4/59; con-

tains double-~walled inserts

*Composition numbers, components, and mole percentages of components are shown below.

Salt No. Composition Mole Per Cent Salt No. Composition Mole Per Cent
84 NaF-LiF-BeF, 27-35-38 134 LiF-BeF,-ThF,-UF,  62-36.5-1-0.5
123 NaF-BeF,-UF, 53-46=-1 135 NaF-BeF;-Tht,-UF, 53-45.5-1-0.5
126 LiF-BeF,-UF, 53-46-1 136 LiF-BeF,-UF, 70-10-20
130 LiF-BeF,-UF, 62-37-1 BULT-14 LiF-BeF;-Th¥,-UF, 67-18.5-14-.0.5

133 LiF~BeF,~ThF, 71-16-13

terminated after 8110 hr when a pump failure indirectly caused rupture of
the cooler coil. ZLoop 9354-4,° which began operation with three double-
walled inserts, was terminated on schedule after 15,140 hr of operation.
An insert was removed after each 5000 hr of operation for weight-loss
determination. The results of the examination of these loops, when com-
pleted, will be reported by the Metallurgy section.

One new INOR-8 loop, MSRP-16, was placed in operation on Nov. 4,
1959, with salt mixture BULT-14. This loop is equipped with a sampling
device for molten salt and incorporates three double-walled inserts in

the second heater section.?®

The sections will be removed at different
operating intervals for weight-loss determination. This loop operates at

a high wall temperature of 1500°F and at a 200°F AT.

Mechanical Joints

A complete report on the development of mechanical joints will be
issued soon under the title '"Mechanical-Joint Testing Program for the

Molten-Salt Reactor Project."

Design and Testing of a Frozen-Salt Metal-to-Graphite Joint

During the past quarter, a frozen-salt graphite-to-metal joint was
tested in a circulating-salt system. The purpose of this test was to in-
dicate the feasibility of a frozen-salt seal between the core and blanket
salts of a graphite-moderated molten-salt reactor.

The test was conducted with a molten-salt pump and loop coupled to

the test piece shown in Fig. 1.1.10. The crucible was machined from a
molded block of type S-4 graphite. The frozen-salt seal was formed in
the annulus between the outside diameter of the graphite crucible and
the inside diameter of an Inconel cylinder which was water-cocoled. The
configuration simulates the inlet or outlet header of a molten-salt

containment vessel.

°MSR Quar. Prog. Rep. Jan. 31, 1958, ORNL-2474, p 32.
SMSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 16.

20

UNCLASSIFIED
ORNL-LR-DWG 45667TA

SALT INLET —-Q

FILL LINE—

SALT QUTLET
- e
|

[

l

J’I'l
7
AN

[

W HEATER
7,
| GRAPHITE TUBE (SLIP FIT)
/7
:;@
V] THERMOCOQUPLES
/] (TYPICAL)
COOLING COLS = ) | /
| #
O% | AN
_ !
/e —in. FROZEN SALT g
i6 A
SEAL ANNULLS ——__( g ’ e INCONEL
o " ’ D
A ¥ V!
GAS IRl / 0 /INSPECTION PORT
CONNECTION—" ERE g f
# |1
[ ’ D
b ¢
A [
e A
A g
2| /
a4 #
i g
# W
v L v
/|
|-
gs : N
0 N/
M SALT LEAKAGE
N CATCH BASIN
\\
\\
R
\§§1
\\\\i
-
N
\\“‘ GRAPHITE
Y] 5 SN
A O
N

HEATER

EJ,E r//
LAVA SUPPORT

INCHES E?;//THERMOCOUPLE

Fig. 1.1.10. Frozen-Salt Metal-to-Graphite Joint.

The salt flow entered the top of the test piece through the small,
center graphite tube and up the crucible past the seal area to the out-
let. The Inconel skirt extending past the seal area serves as a thermal
barrier.

The test joint was operated under two conditions of differential
pressure across the seal. The seal differential pressure, which was con-
trolled by pump head and helium backup pressure on the outside of the
crucible, was held at 0.5 and 18 psi for the two tests. There was no salt
leakage past the seal into the catch basin in either test. However, the
salt did extrude through the crucible walls during the second test. A
complete report of test conditions is in process.’

The test piece was dismounted after the second test, as shown in
Fig. 1.1.11. The upper flange, containing the graphite crucible and
Inconel joint, was removed from the pot and cut as shown at 1. The
frozen-salt seal annulus formed can be seen at 2. The salt frozen in the
thermal-barrier annulus formed by the Inconel skirt can be seen at 3. The
upper rim and lower portion of the graphite crucible are shown at 4 and

5, respectively.

In-Pile Loops

The first graphite-fuel capsule irradiation test, designated ORNL~-
MIR-47-1, has been completed. This test was for the purpose of studying
fuel and fission-product penetration into graphite under conditions sim-
ulating those of a graphite-core molten-salt reactor. The capsules have
been shipped back to ORNL and now are awaiting disassembly. The only
difficulty experienced in this test, which lasted 720 hr, was the loss
of the sodium-tank heaters. Operation continued without interruption.

The second test, designated ORNL-MIR-47-2, was inserted in the MIR
during the March 28 shutdown and is scheduled to be removed June 20. This
test has operated satisfactorily for 1000 hr, to date.

’J. L. Crowley and W. B. McDonald, A Metal to Graphite Joint for a
Molten Salt System (to be published).

w

UNCLASSIFIED
PHOTO 35567A

Fig. 1.1.11. Graphite-to-Metal Joint Dismantled, Showing Frozen-Salt Annuli.
1.2 ENGINEERING RESEARCH

Physical-Property Measurements

Enthalpy and Heat Capacity

Analysis of the previously reported datal on the enthalpy of the BULT-
14 mixture (LiF-BeF,-UF,;-ThF,, 67-18.5-0.5-14 mole %) has been completed.
The results can be expressed as follows:
1. For the solid (100-400°C),
H, — Hio (cal/g) = —5.0 + 0.185t + (7.8 x 1072)t2
where t is the temperature (°C).

2. For the liquid (500-800°C),
H, — Hso (cal/g) = —61.5 + 0.493t — (11.71 x 1077)t?
3. The heat of fusion at 450°C is 44.5 cal/g.

The heat capacities in the liquid state of the BeF,-containing salt

mixtures studied to date are summarized in Table 1.2.1.

Thermal Conductivity

A longitudinal-heat-flow comparison-type apparatus has been used to
measure the thermal conductivity of an INOR-8 specimen at temperatures up
to 1000°F. The essential features of the experimental device are illus-
trated in Fig. 1.2.1. The test specimen (a 6-in.-long, l-in.-dia rod) is
contained between cylindrical heat meters of Armco iron of like diameter.
Thermocouples (located at the axial center line in radial wells) are spaced
1 in. apart along the length of the apparatus. Radial guard heating is
provided at discrete intervals by a set of heated plates. An electrical
heat source at the top of the upper heat meter and a water sink below the
lower meter complete the apparatus. The entire assembly is contained with-
in an inert-atmosphere box to ensure against surface oxidation at the
higher operating temperatures.

A typical longitudinal temperature profile is shown in Fig. 1.2.2;

the temperature was observed to vary linearly with distance in both the

IMSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 21.

Table 1.2.1. Liquid-Phase Heat Capacities of BeF,-Contalning Salts

Salt Composition (mole %) Empirical Constants*
Mixture LiF NaF BeF, UF, ThF, a b
x 1077
1 76 12 12 0.325
3 25 60 15 0,315
123 53 46 0.312 21.74
126 53 46 0.530  —3.80 .
130 62 37 0.488 5.04
133 71 16 13 0.473 —23.80 .
134 62 36.5 0.5 1 0.545 —-8.06
136 70 10 20 0.289 —6.42
Bel.T-15 67 18 15 0.418 —14.00
BULT-14 67 18.5 0.5 14 0.493 —23.42

*Constants appear in the equation c_[cal-g *+(°C)™*] = a + bt, for
the temperature t expressed in °C. P

UNCLASSIFIED
ORNL-LR-DWG 43475

MAIN HEATER

THERMOCOUPLE
3RMCO IRON HEAT METER WELLS

TYPICAL GUARD HEATERS

SAMPLE

ARMCO IRON HEAT METER

COPPER HEAT SINK

Fig. 1.2.1. Thermal Conductivity Apparatus.

24

UNCLASSIFIED
ORNL-LR-DWG 49795

750 ‘
L]
“ |
700 [ - e
'\'
\.
™~ |
650 - -——] ____X.\___ I U
— ‘ \.
: \
W 550 : \ ' - —
=) 1‘
5 \
kagg=9.5 Btu/hr-ft-°F
'fi'_" 500 | —4 \ 498 R
: |
[I1]
: \
L
450 | f e \\ S AN p—
400 - = - \ [ |
N
330 | b . —
= HEAT METE R-mfms— SPECIMEN —=} HE/JT\-\
®,
| ’ | ’ ‘ METER | e
1

300

!

|
0 2 4 6 8 10 12 19 16 18
AXIAL POSITION (in.)

Fig. 1.2.2. Typical Longitudinal Temperature
Profile Obtained in Study of Thermal Conductivity
of an INOR-8 Specimen.

heat-meter and specimen regions. Although thin disks of a low-melting
solder were located between the INOR-8 bar and the Armco rods to provide

a low-thermal-resistance bond at operating temperatures, the small dis-
continuities (~ 5°F) at each end of the specimen show that this was not
completely effective, It is probable that the tin solder did not wet the
INOR-8 surface. This results in an increased radial heat flow, and a sub-
sequent loss in experimental precision. Thus, for the case illustrated

in Fig. 1.2.2, a 10% difference exists between the heat flows in the upper
and lower heat meters., Since the thermal conductivity of the specimen is
based on the heat flux indicated by the heat meters, a *5% variation is
introduced into the value for the thermal conductivity of the INOR-8 rod.
This is shown in Fig. 1.2.3, which summarizes the results of the measure-
ments with INOR-8 in the temperature range of 300 to 900°F. The corre-

lating line, which has been included to indicate the trend of the data

25

with temperature, can be represented to within +5% by the equation
k[Btushr 1-££71:(°F)71] = 7.2 + 0.0045%

for t in °F. These results are compared further in Fig. 1.2.4 with data
obtained at the Battelle Memorial Institute?® for Hastelloy B and Inconel,

using a comparable longitudinal-heat-flow apparatus.

Surface Tension

The surface tension of an NaF-BeF, (64-36 mole %) mixture has been
determined over the temperature range 600-800°C using the maximum-bubble-

3 The results are given in Fig. 1.2.5 and compared

pressure technique.
with earlier measurements® with mixtures LiF-BeF,-UF, (62-37-1 mole %)
and LiF-BeF,-ThF,-UF, (62-36.5-1-0.5 mole %).

Improvements in the experimental procedure and in the instrumenta-
tion have resulted in an increase in both the precision and the accuracy
of these determinations. The original pressure-sensing and -recording
arrangement (a differential-pressure cell and strip-chart recorder) was
subject to calibration shifts of the order of +3%, Replacement of these
units by a precision manometer resulted in an increase in reproducibility
of several orders of magnitude. At the same time, the greater sensitivity
of the pressure measurement allowed correction for two effects leading to
premature bubble collapse, namely, the rate of bubble formation and fluc-

tuations in the envirommental pressure. The total error in this deter-

mination is now estimated to be of the order of +3%, as compared with *8%

for the earlier studies.

°H. W. Deem, Battelle Memorial Institute, personal communication.
3W. D. Powers, MSR Quar. Prog. Rep. June 30, 1958, ORNL-2551, p 38.
“MSR Quar. Prog. Rep. Apr. 30, 1959, ORNL-2723, p 39.

UNCLASSIFIED UNCLASSIFIED

r ORNL-LR-DWG 45796 - ORNL-LR-DWG 49797
s ® i > 15 I l T i :
St =T T 4 S = | WNOR-8 | ="
ocx 10 _'—i———.- F z S 10 i t

~ o s e e} o — — .
g‘g I 25 +— HASTELLOY B
e~ ‘ E @ | | o
£ 5 ! z o) | |
& 200 300 400 500 600 700 800 900 1000 < 200 300 400 500 6CO 700 - 80O 900 1000

TEMPERATURE {°F) TEMPERATURE (°F)

Fig. 1.2.3. Thermal Conductivity of an INOR-8 Fig. 1.2.4. Comparison of Thermal Conductivities
Specimen. of INOR-8, Inconel, and Hastelloy B.

26

To correct for the effect of the depth of immersion of the capillary

tube on the bubble pressure, measurements were made at a number of levels

(up to 8 mm below the salt surface) for each temperature.

it 1s possible to make an estimate of the density of the salt.

From these data,

Preliminary

results are compared in Fig. 1.2.6 with the density estimated from the cor-

relation by Cohen and Jones;” the agreement 1s reasonable.

Heat-Transfer Studies

Operation of the system6

for determining heat-transfer coefficients

with the salt mixture BULT-14 (LiF-BeF,-UF;-ThF,;, 67-18.5-0.5-14 mole %)

flowing in heated Inconel and INOR-8 tubes was continued. Approximately

4500 hr of primarily isothermal (1150°F) flow have been accumulated. A

scheduled shutdown took place after 3750 hr for the purpose of replacing

the thermocouples and the oxidized copper power leads on both test sec-

tions.

On resumption of circulation, the turbine-type flowmeter was Ffound

to be Inoperable; and for the succeeding 700-hr period, flow rates were

determined by using the flow-rpm characteristic’ of the pump.

meter signal was spontaneocusly recovered at 4450 hr.

The flow-

Subsequent compari-

son of the flow rates as indicated by the turbine meter and the punp char-

acteristic shows a 3% discrepancy, with the flowmeter giving the higher

°S. I. Cohen and T. N. Jones, A Summary of Density Measurements on
Molten Fluoride Mixtures and a Correlation for Predicting Densities of

Fluoride Mixtures, ORNL-1702 (July 1954).
®MSR Quar. Prog. Rep. Oct. 31, 1958, ORNL-2626, p 46; MSR Quar. Prog.

Rep. July 31, 1959, ORNL-2799, p 39.

“MSR Quar. Prog. Rep. Apr. 30, 1959, ORNL-2723, p 41.

UNCLASSIFIED
ORNL-LR-DWG 49798

250

200 pe—cci . —

—
o
—

{62-37-1 mole %) “'fi%
— LiF-BeF,~Thk,~UF, -

{62-36.5-1-0.5mole %}

100 ' i
500 550 600 850 700 750 800 850
TEMPERATURE (°C)

o
)

@, SURFACE TENSION (dynes/cm)
C
7
@
@
o7
1
c
. &
[ ]
; L
[ J

Fig. 1.2.5.
(64-36 mole %).

Surface Tension of Nch-BeF2

UNCLASSIFIED
ORNL-LR-DWG 49799

2.2

T
od/

i - EXPERIMENTAL

\T\
\ b
--..__‘.

\ .
ESTIMATED —~
FROM CORRELATION \

—OF COHEN AND JONES

p. DENSITY (g fem)
M
(&}
¥

]

1.8

600 650 700 750 800 850
TEMPERATURE (°C)

Fig. 1.2.6. Density of NaF-BeF, (64-36 mole %).

27

result. Since the pump curve was obtained initially by calibration with
reference to the flowmeter (the performance characteristic for the turbine
meter having been established independently), a 3% inconsistency exists

in the flow rates for all At runs in the MB series. The designations MA
and MB have been used to differentiate between the data taken before and
after the shutdown, respectively. This flow uncertainty affects directly

both the calculated Reynolds modulus and the heat-transfer coefficient.
0.4)
Pr
are shown in Fig. 1.2.7 for the INOR-8 section and in Fig. 1.2.8 for the

The experimental results (as the heat-transfer parameter, NNu/N

Inconel section. All the data® have been re-evaluated, using the recently
determined values for the heat capacity of the BULT-14 mixture. (See the
previous section, "Physical-Property Measurements.") This resulted in an
increase in the magnitude of the heat-transfer parameter of about 10%. At
the same time, the heat balances (the ratio of the heat gained by the salt
in passing through the test section to the electrical energy input) were
adversely affected to the extent that all values are now greater than
unity (ranging as high as 1.6). A detailed re-examination of the experi-
mental instrumentation and a redetermination of the enthalpy of the salt
are planned in an effort to obtain an explanation for this situation.

The data given in Figs. 1.2.7 and 1.2.8 show a general scatter of the

order of *10%. However, the variability of the Inconel-tube data within

BMSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 23.

UNCLASSIFIED

UNCLASSIFIED -LR-
ORNL-LR-DWG 49800 » 100 I L [ L_+ } ORNL-LR-DWG 49801
100 — —_— - S
@ i S —— T _1 TT ‘,,},i_nl Jn 3 80 [ O A A B
. ! ! Ny /N
5 80+ — — ’ 04 : . a ,Nu/ Pr )
D R T - R b ;fl
B I ‘ 1 T @
x e ,J - Lol l | L LiF-NoF-KF-(46.5-11.5-42mole %), ; ’
L i ‘ 1 | 0 NoOH, AND NoNOz—-NaNOx—KNOx!
% I i , L IR E N A B S O z 40— 3~ KNO3,
= %0 — 4} | | }% < (40-7-53 wt %) Pz
o i fifl 4 ‘ . [ i
FI_ - - ( - A + +- T -+ '|_ — E— - ;‘,-‘ P
= ‘ 7 \&La———- LiF - NoF ~ KF (46.5-115-42 < \ \ | ‘ ARE
W e mole %), NoOH AND NaNO,,~ T NoF-2rF,~UF, (50-46-4mole %) .5 | |
- O // A NoNOz~ KNOz (40-7-53 wi %) <" 20[AND LiF-Befy-URy T T 77 O
S, A F —-— NoF-Zrf;-UF, {50-46-4 Sk (62-37-1male %)
g S0 mole %} AND LiF - BeF,-UF, z !
3 / ‘ : (62-37-1 male %) :
SN R NN 2
= L] | I B S |
3 4 5 : '
3X140 5 10 2 5 10 103 > 5
Ny, REYNOLDS MODULUS
Re Nge» REYNOLDS MODULUS
Fig. 1.2.7. Heat Transfer with BULT-14 (LiF- Fig. 1.2.8. Heat Transfer with BULT-14 (LiF-
Ber-UF4-ThF4, 67-18.5-0.5-14 mole %) Flowing Ber-UF4-ThF4, 67-18.5-0.5-14 mole %) Flowing
in a Heated INOR-8 Tube. in a Heated Inconel Tube.

28

this scatter band is greater than that observed for the INOR-8 tube. The
cause of this increased deviation is not yet understood. Since the salt
was drained from the loop during the thermocouple and electrical repair
operation, the data have been differentiated as to this aspect (series MA
and MB). There does not appear to have been a significant effect, although
(for the limited data available) a number of unexpectedly low values of
the heat-transfer parameter were calculated, No systematic temporal vari-
ation of the heat-transfer coefficient has been observed.

For comparison, the heat-transfer characteristics for a number of
9=12

previously studied salts are also given in Figs. 1.2.7 and 1.2.8. The

data for both the Inconel and INOR-8 tubes are seen to lie about 25% be-

low the general heat-transfer correlationt?
0.4 0.8
Ny, /Np,. = 0.023 N;

for normal fluids in turbulent flow within circular tubes under moderate
At conditions, as well as below the experimental results for other molten
salts. This discrepancy, 1f attributed entirely to the lack of experi-
mental data on the thermal conductivity of the BULT-14 salt, would require
a decrease in the thermal conductivity by almost a factor of 2 [from the
estimated value of 1.3 Bturhr *+ft 1+(°F)71]. The previously observed®
difference between the INOR-8 and Inconel data was found to arise from an
error in the value of the tube surface area used In the analysis. After
correction, the originally low results for the INOR-8 tube agreed closely
with the Inconel-tube data.

H. W. Hoffman, Turbulent Forced-Convection Heat Transfer in Circular
Tubes Containing Molten Sodium Hydroxide, ORNL-1370 (Oct. 3, 1952).

10y, W. Hoffman, Molten Salt Heat Transfer, ORNL CF-58-2-40 (Feb. 18,
1958).

11y, W. Hoffman and S. I. Cohen, Fused Salt Heat Transfer — Part III:
Forced-Convection Heat Transfer in Circular Tubes Containing the Salt Mix-
ture NaNO,-NaNO5;-KNO;, ORNL-2433 (Mar. 1, 1960).

123, C. Amos, R. E. MacPherson, and R. L. Senn, Preliminary Report on
Fused Salt Mixture No. 130 Heat Transfer Coefficient Test, ORNL CF-58-4-23
(Apr. 2, 1958).

13y, H. McAdams, Heat Transmission, 34 ed., p 219, McGraw-Hill, New
York, 1954,

PART 2

MATERIALS STUDIES

2.1 METALLURGY

Dynamic Corrosion Studies

Thermal-Convection Loops

Test results for 16 thermal-convection loops (nine of INOR-8 and seven
of Inconel) which were operated with fluoride mixtures are summarized in
Tables 2.1.1 and 2.1.2, These tests mark the completion of all phases of

1 dealing with thermal-convection ex-

the molten-fluoride corrosion program
periments.

As may be seen in Table 2.1.1, the attack in five of the INOR-8 ex-
periments, all of which operated for one year, was limited to surface
roughening and pitting to a maximum depth of 1/2 mil. Figure 2.1.1 is
representative of the attack found in these loops. One loop, 1224, which
circulated salt 123 (NeF-IiF-BeF,, 53-46-1 mole %) for 8760 hr at a maximum
temperature of 1350°F, showed slightly greater attack, in the form of sur-
face pitting to a maximum depth of 1 mil. Three loops, 1216, 1226, and
1244, exhibited attack in the form of shallow subsurface voids. The max-
imum depth of attack in loop 1216 was 1-1/2 mils, whereas in 1226 and 1244
the maximum depth of attack was limited to 1/2 mil. No attack, except for
light roughening of the surfaces, was found in the cold legs of the loops.
Also, examination of the cold regions revealed no deposits or other evi-
dence of mass transfer.

An extremely thin corrosion film was found on the surfaces of several
of these loops. As previoucsly discussed,2 filming has occurred in a ma-
Jority of the thermal-convection tests which have operated for more than
3000 hr. Attempts are being made to identify the film, and a discussion
of the results obtalned i1s included in the subsequent section, "Analysis
of Corrosion Film.,"

Intergranular and general subsurface voids formed in the hot-leg
regions of all seven Inconel loops (Table 2.1.2). The attack ranged from
depths of 4 to 7 mils in loops which operated at 1250°F, and from 7-1/2 to

1J. H. DeVan, J. R. DiStefano, and R. S. Crouse, MSR Quar. Prog. Rep.
Oct. 31, 1957, ORNL-2431, p 23.

°MSR Quar. Prog. Rep. July 31, 1959, ORNL-2799, p 50.

Table 2.1.1. Results of Metallographic Examination of INOR-8 Thermal-Convection Loops

(98}
FeN
Maximum . . .
Loop Pziizd Fluid-Metal Salt Metallographic Examination
No. (hr) Interface No.* Hot-Leg Appearance Cold-Leg
Temperature (°F) & APP Appearance
1224 8760 1350 123 Moderate surface roughening and No attack
pitting to 1 mil
1226 8760 1250 131 Moderate surface roughening, irregular  Light surface
corrosion film varying from 1/2 mil roughening, with
to 1 mil thick; there are scattered 1/2-mil corro-
voids within this zone sion film
1231 8760 1250 134 Light surface roughening and pitting No attack
to < 1/2 mil, corrosion film to
< 1/2 mi1
1238 8760 1350 134 Light surface roughening No attack, very
shallow corro-
sion film present
1244 g760 1250 135 Light surface roughening, moderate No attack
voids and pits to < 1 mil
1246 8760 1350 135 Light surface roughening No attack
1233 g760 1250 133 No attack; very shallow corrosion film  No attack
present
1240 g760 1350 133 Light surface roughening No attack
1216 8760 1350 127 Moderate surface roughening and pitting No attack

to < 1/2 mil, light scattered inter-
granular voids to < 2 mils

¥Compositions:
123 NaF-BeF,-UF, (53-46-1 mole %) 133 LiF-BeF,-ThF, (71-16-13 mole %)
127 LiF-BeF,-Thr,; (58-35-7 mole %) 134 LiF-BeF,-ThF,-UF, (62-36.5-1-0.5 mole %)
131 LiF-BeF,-UF, (60-36-4 mole %) 135 WNaF-BeF,-ThF,-UF, (53-45.5-1-0.5 mole %)

Table 2.1.2. Results of Metallographic Examination of Inconel Thermal-Convection Loops
Test Maximum Metallographic Examination
Loop Period Fluid-Metal Salt
No. ‘(3;;) Interface No.* Hob-les Avboarance Cold-Leg
Temperature (°F) & SPP Appearance
1225 g760 1250 123 Moderate surface roughening and pitting to Light surface
1 mil; moderate intergranular voids to roughening
maximum depth of 7 mils
1236 g760 1250 134 Light surface roughening and pitting to Light surface
1 mil; moderate intergranular voids to roughening and
maximum depth of 5 mils pitting to 1 mil
1237 g760 1350 134 Light surface roughening and pitting to Light surface
1 mil; heavy intergranular voids to roughening
maximum depth of 12 mils
1245 8760 1250 135 Moderate surface roughening and pitting to Light surface
1 mil; moderate intergranular voids to pits
maximum depth of 6 mils
1247 8760 1350 135 Moderate surface roughening; heavy general  Light surface
voids to maximum depth of 15 mils roughening
and pitting
to 1/2 mil
1235 7789 1250 133 Light surface roughening and pitting to A few pits to
1l mil; a few intergranular voids to 1 mil
maximum depth of 4 mils
1239 8760 1350 133 Moderate surface roughening; moderate Moderate sur-
intergranular voids to maximum depth face roughen-
of 7-1/2 mils ing and pit-
: ting to 1 mil
@ *Compositions:
123 NaF-BeF,-UF, (53-46-1 mole %) 134 LiF-BeF,-ThF,;-UF, (62-36.5-1-0.5 mole %)
133 LiF-BeF,-ThF,; (71-16-13 mole %) 135 NaF-BeF,-ThF,-UF, (53-45.5-1-0.5 mole %)

UNCL ASSIFIED,
J.18981

Fig. 2.1.1. Specimen Taken from Hot Leg of INOR-8 Thermal-Convection
Loop 1231 at Point of Maximum Loop Temperature (1250°F). Loop was
operated for one year with salt mixture LiF-BeF ,-ThF ,-UF , (62-36.5-1-0.5
mole %). Etchant: 3 parts HCI, 2 parts H,0, 1 part 10% chromic acid.

15 mils in loops which operated at 1350°F. In the cold-leg regions, sur-
face roughening and pitting occurred to a maximum depth of 1 mil. No evi-

dence of mass-transfer deposits was found in the cold-leg regions.

Inconel Forced-Convection Loops

The status of Inconel forced-convection loops presently in operation -
is discussed in Sec 1.1 of this report. A report discussing the results
for four Inconel loops (9377-1, -2, -3, and -4) was recently issued.® The -
results are summarized in Table 2.1.3.

In general, the attack in all four loops followed the characteristic
pattern for Inconel loops in contact with fluoride mixtures at higher tem-

peratures; namely, the attack showed a strong dependence on the inside wall

3R. C. Schulze and J. H. DeVan, Examination of Inconel Forced Con-
vection Loops 9377-1, 9377-2, 9377-3 and 9377-4, ORNL CF-60-4-54 (Apr. 6,
1960) .

36

Table 2.1.3. Results of Metallographic Examination of Inconel Forced-Convection Loops

Test Maxcimum Flow
Loop Period Salt Fluid-Metal ew Reynolds Rate Metallographic Examination
No. (hr) No.* Interface (°F) Number ( epm)
Temperature (°F) Ep
9377-1 3390 126 1300 200 1600 2.0 Intergranular voids to 7 mils
-2 3046 130 1300 200 3000 2 Intergranular voids to 8 mils
-3 8746 131 1300 200 3400 2.0 Intergranular and general sub-
surface voids to 14 mils
-4 9574 130 1300 200 2600 1.75 Intergranular and general sub-

surface voids to 15 mils

*Compositions:
126 LiF-BeF,-UF,; (53-46-1 mole %)
130 LiF-BeF,-UF, (62-37-1 mole %)
131 LiF-BeF,-UF, (60-36-4 mole %)

tem.perature.4 The attack in all the loops was limited to regions where the
loop wall temperature exceeded 1200°F, and the depths of attack at these
positions were quite consistent when viewed at similar time periods. As
can be seen in Table 2.1.3, the rates of attack found at the 3000-hr time

interval and the 2000-hr time interval were consistent within 1 mil,

INOR-8 Forced-Convection Loops

As discussed in Sec 1.1, eleven INOR-8 forced-convection loops have
completed operation for one year or longer with fused fluoride mixtures.
Operation of six of these loops has been terminated, and four of the six
loops have been examined (Table 2.1.4). Loop 9354-5, which contained spec-
imens of graphite in the hot-leg section, was discussed previously.® As
noted at that time, examination of this loop revealed no manifestations of
corrosive attack beyond the start of a thin surface film.

Similar results (Table 2.1.4) were obtained from the recent examina-
tions of loops MSRP-8 and -9, which circulated fluoride mixtures for 9633
and 9687 hr, respectively. Both loops appeared unattacked, except for the
development of the thin surface layer. The extent of this layer can be
seen in Figs. 2.1.2 and 2.1.3, which show the metallographic appearance of
the hot-leg sections from the latter loops. (Results of chemical analyses
of some of these layers are discussed in the next section.)

Considerably greater attack was noted, however, in loop 9354-1, which
was terminated after 14,563 hr because of a leak which developed in the
second heater leg. This loop encountered operational difficulties at two
earlier periods, each of which necessitated suspension of operation and the
replacement of loop com.ponents.6 As noted in Table 2.1.4, metallographic
examinations of specimens removed from the hot-leg regions revealed general -
surface roughening and pitting to a maximum depth of 1—1/2 mils (Fig.

2.1.4)., Attack elsewhere in the loop appeared negligible. In contrast,
two specimens removed from the hot legs after 3000 hr (at the time of the

first of the component failures) revealed no attack.

“W. D. Manly et al., "Metallurgical Problems in Molten Fluoride Sys-
tems," Second U.N. Intern. Conf. Peaceful Uses Atomic Energy, Geneva, 1958,
paper A/Conf 15/P/1990, p 12 (June 1958).

°MSR Quar. Prog. Rep. July 31, 1959, ORNL-2799, p 55.
6MSR Quar. Prog. Rep. Oct. 31, 1958, ORNL-2626, p 56.

38

Table 2.1.4. Results of Metallographic Examination of INOR-8 Forced-Convection Loops

Test Maxclmun | Flow
Loop Period Salt Fluid-Metal ANT Reynolds Rate Metallographic Examination
No. (1r) No . % Interface (°F) Number (eom) (Maximum Attack)
Temperature (°F) ep
9354-5 8,950 130 1300 200 3000 2.5 No attack¥*
MSRP-8 9,633 124 1300 200 4000 2 No attack¥*
MSRP-9 9,687 134 1300 200 2300 1.8 No attack¥**
9354-1 14,563 126 1300 200 2000 2.5 Heavy surface roughening and pit-

ting to 1-1/2 mils (nickel de-
posits were found in the cooler
coil)

*Compositions:
124 NaF-BeF,-ThF, (58-35-7 mole %)
126 LiF-BeF,-UF, (53-46-1 mole %)
130 LiF-BeF,-UF, (62-37-1 mole %)
134 LiF-BeF,-ThF,-UF, (62-36.5-1-0.5 mole %)

**¥Corrosion film found on interior surfaces of the tubing.

UNCLASSIFIED
T.19073 _

Fig. 2.1.2. Appearance of Specimen Removed from Point of Maximum
Salt-Metal Interface Temperature (1300°F) of INOR-8 Forced-Convection
Loop MSRP-8. Loop was operated for 9633 hr with salt mixture NaF-
BeF ,-ThF, (56-35-7 mole %). Etchant: 3 parts HCI, 2 parts H,0, 1 part 10%

chromic acid.

UNCL ASSIFIED
T.19080

Fig. 2.1.3. Specimen Removed from Point of Maximum Salt-Metal Interface
Temperature (1300°F) from INOR-8 Forced-Convection Loop MSRP-9. Loop
was operated for 9687 hr with salt mixture LiF-BeF ,-ThF ,-UF , (62-36.5-1-0.5
mole %). Etchant: 3 parts HCI, 2 parts H,0, 1 part 10% chromic acid.

40
UNCL ASSIFIED
T.18724

H

Fig. 2.1.4. Specimen Removed from INOR-8 Forced-Convection Loop
9354-1 at Point of Maximum Salt-Metal Interface Temperature (1300°F) Showing
Maximum Attack Found. Loop operated for 14,563 hr with salt mixture LiF-
BeF ,-UF , (53-46-1 mole %). Etchant: 3 parts HCI, 2 parts H,0, 1 part 10%
chromic acid.

Examinations of the cold-leg regions at the end of this test revealed
magnetic metal crystals loosely adherent to the cold-leg wall. This de-
posit was obtained for chemical analyses by sectioning the length of tubing
over which it appeared into approximately 6-in. segments and then scraping
out the deposited material. The lengths of tubing and the quantities and
compositions of material removed are listed in Table 2.1.5. The deposit
was composed predominantly of nickel and contained only minor amounts of
iron and chromium, with the greatest concentration of chromium being found
where the material was first precipitated from the salt. Molybdenum, which
comprised 17 wt % of the base metal, was present in only insignificant con-
centrations. Altogether the weight of the deposit totaled 22 g.

The abnormally high rate of attack and mass transport of nickel in
this loop necessarily implies that an extremely effective oxidant was made
available to the salt circulating in the loop. An indication of the nature

of this oxidant was afforded by after-test salt analyses, which indicated

41

Table 2.1.5. Chemical Analysis of Deposit*

Length of Weight of Weight % Mo
Specimen Tubing Deposit ( )
(in.) (g) Ni Cr Fe ppm
1 11-1/8 1.9000 gd.2  7.33 0.23 7035
2 42-1/8 4 4994 94.3 1.22 1.34 6380
3 18 b 4134 92.5 1.26 1.37 6215
A 32-7/8 4.0994 92.5 1.95 1.39 7850
5 R4-5/8 7.1898 9.3  0.68 0.8l 5690 :

¥Deposit contained small amounts of salt residue.

that a relatively large percentage of UF, had been converted to U0, during
the test. The production of U0, most probably was affected by H,O con-
tamination which entered the loop during one or more of the tubing fail-
ures. Reaction of this contaminant with UF, would also produce HF, which
is potentially capable of oxidizing nickel as well as other constituents
of INOR-8.

Analysis of Corrosion Film. — Chemical and metallurgical studies were

undertaken to investigate the nature and causes of the thin corrosion films
observed in long-term INCR-& corrosion loops, as discussed above. The film
appears as a continuous second phase which extends below the exposed sur-
faces to a thickness of 1/2 mil or less. No transition or diffusion zone
has been evident between the film and the base metal (Figs. 2.1.1, 2.1.2,
and 2,1.3), In thermal-convection loops the film has been observed both

in hot- and cold-leg specimens; in forced-convection loops it has been
found only along the tubing between the pump exit and the end of the second
heater leg.

Comparative hardnesses of the film and parent metal of one of the
specimens were recently determined by using a Bergman microhardness tester
with a 1-g load. The portion of the film examined was approximately 1/3
mil thick and was removed from the point of maximum salt-metal interface

temperature of INOR-8 forced-convection loop 9354-5.7 On the basis of the

'MSR Quar. Prog. Rep. July 31, 1959, ORNL-2799, p 55.

42

1-g load, the film was found to have a DPH (diamond pyramid hardness) of

518, and the DPH for the parent metal was 269. A check of the base metal
using the Tukon hardness tester with a 200-g load indicated a DPH of 218.
Although the numbers based on the Bergman tester appear to be slightly
high, the values obtained demonstrate that the film is approximately twice
as hard as the parent metal, as shown in Fig. 2.1.5.

Since none of the films have been more than 1/2 mil thick, it has been
extremely difficult to determine their chemical composition. However,
qualitative analyses of these films were recently carried out with the aid
of C. E. Zachary of the Metallographic Group. These analyses were made on
material "scrubbed" from the inside surfaces of the tubes with a corundum
slurry, the specimens being weighed before and after to determine the amount
of material removed. The resulting mixture of slurry and metal, along with
an unused sample of the slurry as a blank, was then submitted for spectro-
graphic analysis. i

Analyses obtained for several different specimens of this surface film

are listed in Table 2.1.6. The films were composed primarily of nickel and

UNCLASSIFIED
T.18104

INCH

Fig. 2.1.5. Hordness Measurements Made on Specimen Removed from
Point of Maximum Salt-Metal Interface Temperature (1300°F) of INOR-8
Forced-Convection Loop 9354-5.

43

Table 2.1.6. Spectrographic Analyses of Surface Films

Loop from Approximate . .
Which . Temperature Salt Spectrographic Analysis (wt )
i Location . .
Specimen of Specimen Circulated Ni Cr Fe Mo
Was Removed (°F)
1226 Cold leg 1078 131 62 4.3 29 4.7
9354-5 End of first 1269 130 80 1.5 14.5 A
heater leg
9354-5 End of second 1296 130 70.6% 29, 4%
heater leg
MSRP-"/ Fnd of second 1300 133 67.9% 32.1%
heater leg

*These analyses assume Ni + Mo = 100%.
contained smaller amounts of chromium, iron, and molybdenum. At present,
there is no apparent correlation of the film composition with any of the
test loop variables (temperature, time, salt, etc.). Interestingly, how-
ever, analyses of the specimen removed from the end of the second heater
leg of forced-convection loop 9354-5 showed it to have a composition ap-

proximately the same as the B-phase (Ni,Mo) of the Ni-Mo system.

Graphite Brazing Studies

The development of techniques for brazing graphite to itself and to
metals is of continuing interest for molten-salt power reactor systems
with a high-breeding-ratio potential., A study of these procedures has been
under way during the past year, and the progress has been reported in the
last few progress reports.‘g'9 Recent findings are summarized in this re-
port.

The promising brazing characteristics of the 4&% Ti—-48% Zr—4% Be
brazing alloy were discussed in the last report.? In order to demonstrate
the general feasibility of brazing assemblies of graphite tubes to metal
headers, a demonstration assembly was fabricated with this allgy. An ex-
ploded view of the assembly is shown in Fig. 2.1.6.

The first step was to braze the CT-158 fine-grained extruded graphite
tubing to a molybdenum ring using the Ti-Zr-Be alloy. The molybdenum ring
was used as a transition layer between the graphite and the INOR-8 base
plate. Because of its low coefficient of thermal expansion and its ability
to absorb shearing stresses, the molybdenum acts as a buffer and tends to
minimize the possibility of shearing in the graphite by stresses which re-
sult from the large difference in thermal expansion between graphite and
INOR-8.

The final assembly was made by brazing the molybdenum in the graphite-
molybdenum subassemblies to the INOR-8 plate with a commercially available
alloy, Coast Metals No. 53 (Ni-Cr-B-Si-Fe). The completed assembly is shown
in Fig. 2.1.7.

8MSR Quar. Prog. Rep. July 31, 1959, ORNL-2799, p 71-72.
MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 33-35,

UNCLASSIFIED
FINAL Y-32366

ASSEMBLY
SUBASSEMBLY

GRAPHITE
MOLYBDENUM
RiNG BRAZING
ALLOY

i

Fig. 2.1.6. Sequence for Assembly of Graphite Tube to INOR-8 Plate.

UNCLASSIFIED
Y-33370

ONE INCH

Fig. 2.1.7. Completed Assembly of Graphite Tube to INOR-8 Plate.

46
The utilization of the Ti-Zr-Be alloy for molten-salt reactor appli-
cations has been shown to be severely limited by virtue of its pcor cor-

rosion resistance to fluoride salts.?

A brazing-alloy development progran
was subsequently initiated for the express purpose of obtaining brazing
alloys which wet and flow on graphite at reasconable brazing temperatures
(< 1300°C) and which are corrosion resistant to the fused salts. The cor-
rosion-resistant 82% Au—18% Ni alloyt© was selected as a potential starting
material, and it may find use in certain selected applications. Corro-
sion-resistant carbide-forming elements such as tantalum or molybdenum
were added to promote wetting.

Promising ternary compositions were selected through an analysis of
the available binary phase diagrams of the Au-Ni-Ta and Au-Ni-Mo systems.
To date, 10-g buttons of approximately forty Au-Ni-Ta and thirty Au-Ni-Mo
alloys have been arc-cast and evaluated. Figures 2.1.8 and 2.1.9, respec-
tively, indicate the compositions of these alloys. The flow behavior, in
vacuum, of small fragments of alloys in the Au-~Ni-Mo system on graphite at

1300°C is shown in Fig. 2.1.10.

Flowability studies have been conducted with these alloys on both
graphite-to-graphite joints and graphite-to-molybdenum joints. The alloy
compositions in the two systems which flow well on graphite-to-graphite
joints are enclosed by dashed lines in Figs. 2.1.8 and 2.1.9. However,
alloys containing lower percentages of carbide-forming elements flow ad-
equately on graphite-to~-molybdenum Jjoints and are more ductile than those
required for graphite-to-graphite Jjoints.

The 35% Au—35% Ni—30% Mo alloy has been investigated in view of its
application for Joints of both types. Figure 2.1.11 shows joints contain-
ing low-permeability graphite tubes butt-brazed to a l/4—in.—thick molyb-
denum plate and to an AGOT graphite plate. The alloy was preplaced around
the outside diameter of the graphite tubes before induction brazing. Full
penetration of the braze through the Jjoints and subsequent filleting on
the inside diameters can be seen, A photomicrograph of a graphite-to-mo-

lybdenum joint brazed with this alloy is shown in Fig. 2.1.12.

1%, E. Hoffman et al., An Evaluation of the Corrosion and Oxidation
Resistance of High-Temperature Brazing Alloys, ORNL-1934 (October 1956).

47

UNCLASSIFIED
ORNL-LR-DWG 48802

T03Ni2?

COMPOSITIONS IN wt 7

EUTECTIC
~1400°C

ToNig

EUTECTIC
~1360°C

|
MINIMUM
~950°C

Fig. 2.1.8. Compositions of Experimental Au-Ni-Ta Alloys Prepared for Graphite Brazing Studies.

Preliminary thermal-cycling and leak-test studies, coupled with me-
tallographic examinations, indicate that those alloys which flow best on
graphite are brittle and tend to crack, as shown in Fig. 2.1.13. The
brittleness of these alloys is also evidenced by their relative ease of
crushing as arc-melted buttons. It should be possible, however, to obtain
adequate flow on graphite-to-molybdenum joints with alloys of improved
ductility by reducing the molybdenum or tantalum contents. This possi-

bility is being investigated.

48

UNCLASSIFIED
ORNL-LR-DWG 49803

COMPOSITIONS IN wt %

MoNi

EUTECTIC
~A4345°C

Au Ni

MINIMUM
~350°C

Fig. 2.1.9. Compositions of Experimental Au-Ni-Mo Alloys Prepared for Graphite Brazing Studies.

The porous graphites, such as type AGOT, have greater joint strengths,
as evidenced by the mechanical testing of T-joint specimens. Specimens of
AGOT graphite fail deep in the parent material, while low-permeability
graphite specimens faill at the graphite—braze-metal interface. The deep
penetration of brazing alloy into the pores of the AGOT graphite is shown
in Fig. 2.1.14; no evidence of this condition has been observed in the low-
permeability material. The small-sized pores in this type of graphite re-

duce the possibility of a mechanical "keying" action and reduce the braze-

graphite contact area.

UNCLASSIFIED
ORNL-LR—DWG 48425

Mo

COMPOSITION IN wt%

EUTECTIC
~1315°C

e AN
AVAYA - TavAVAS

AVaYAy, . VAVAVAN

WV - < \VAVAVAVAV

Ay T
MINIMUM
~950°C

Fig. 2.1.10. Flow Behavior of Au-Ni-Mo Alloys on Graphite at 1300°C.

Remedies for this problem seem to lie in improved joint designs a.nd/or
the development of new brazing alloys with stronger bonding characteristics

and lower coefficients of thermal expansion. Studies along these lines are

being made.

50
UNCLASSIFIED
Y-34872

010in/DIV
(a) (b)

Fig. 2.1.11. (a) Type RLM-24 Graphite Tubing Induction-Brazed to Molybdenum with
35% Au-35% Ni—30% Mo. (b) Type RLM-24 Graphite Tubing Induction-Brazed fo Type
AGOT Graphite with 35% Au—35% Ni~30% Mo.

UNCL ASSIFIED
Y-34987

0.02

Fig. 2.1.12. Photomicrograph of Type RLM-24 Graphite Tubing Induction-Brazed to Molybdenum with
35% Au-35% Ni-30% Mo.

51

UNCL ASSIFIED
Y-34988

Fig. 2.1.13. Photomicrograph of Type RLM-24 Graphite Tubing Induction-Brazed to Molybdenum with
35% Au—35% Ni-30% Mo and Thermally Cycled 16 Times from 700°C to Room Temperature.

UNCLASSIFIED
Y-34989

0.10in./DIV

L | | | I |

Fig. 2.1.14. Photomacrograph of Type AGOT Graphite-to-Graphite T-
Joint.

52
General Corrosion Studies

Permeation of Graphite by Molten Salts

The use of unclad graphite as a moderator in a molten-salt reactor
requires that not more than a small percentage of the graphite pore space
be filled with salt. To ensure this condition, a test program*! has been
in progress to study the permeation of various potential grades of graphite
by molten salt (BULT 14—.5U) at 1300°F and for various times and pressures.

Thirty-one grades of graphite have been tested under the conditions
described above. All grades were manufactured as low-permeability grades
except seven: CCN, CEY-B, R-0013, GT-123, SM-1l, TSF, and AGOT. Results
of the screening tests are summarized in Table 2.1.7. There are now four
grades, B-1, $S-4-1LB, GT-123-82, and CS-112-S, that meet the permeation
specification, under the screening-test conditions, that less than 0.5% of
the bulk volume of graphite be filled with salts. These results are from
specimens taken from single pieces of graphite that were relatively small.

In an attempt to determine the effect of pressure on permeation, 20
grades of graphite that were available at the time this test was assembled
were screened at a lower pressure. Three or more specimens of each of the
20 grades were exposed simultaneously for 100 hr to LiF-BeF,-Th¥,-UF, (67-
18.5-14-0.5 mole %, BULT 14—.5U) at 1300°F (704°C) and 95 psig, a potential
reactor fluoride-salt pressure. The results of this test are compared in
Table 2.1.8 with those obtained from previcus tests in which the pressure
was 150 psig. This test suggests that if the reactor pressure were 95
psig, four additional grades, CT-150, CEY-1350, CT-158, and CEY-G, might
be brought within the MSRP specification.

The grades of graphite that showed the greatest reductions in quantity
of salt permeation as a result of the lower pressure have the highest per-
centage of small pores comprising their total accessible voids. These
graphites also are those that were fabricated as relatively small shapes.

Under the assumption that other reactor conditions will not signifi-

cantly modify permeation, it appears that for some grades of graphite,

lly, H. Cook and D. H. Jansen, A Preliminary Summary of Studies of
INOR-8, Inconel, Graphite and Fluoride Systems for the MSRP for the Period
May 1, 1958 to Dec. 31, 1958, ORNL CF-59-1-4, p 1—20.

Table 2.1.7. Permeation of Various Grades of Graphite by LiF-BeF,-
ThF,-UF, (67-18.5-14-0.5 mole %, BULT 14-.5U) at 150 psig

Test conditions:
Temperature: 1300°F (704°C)
Exposure period: 100 hr

All densities and percentages are averages of three with the exception of
those with numerical superscripts, which denote the number of values av-
eraged

Specimen Dimensions,

Graphite  Bulk Volume Nominal (in.)

Graphite Apparent of Graphite

Shape

Grade D?2722§ Per?;?ted D;imgger D Length(s) .
B-1 1.91 0.0, Rod 0.90% 0.50 )
S-4-LB 1.90l4)  g.p(4) Block  0.50 0.75
or-123-82° 1.91(®) 0.36) Rod 0.50% 1.50
cs-112-8  1.89¢%) 0.5¢%) Rod 0.90% | 0.50
RH-1 1.89t6) 0.6 Rod 0.90% 0.50
CEY-1350  1.90 0.7 Pipe 1.24% 0.86%  0.25
CEY-G 1.88 0.9 Pipe 1.25% 0.86°  0.50
S-4-1A 1.89(4) 1.0(4) Block 0.50 0.75
CEY 1.91(6) 1,008 Pipe 1.24° 0.85% 0.50, 0.25
CT-150 1.80 1.8 Pipe 0.53% 0.26% 1.50
R-0009° 1.9240) 1.9¢6) Block  0.50 1.50 )
cEY-82° 1.87 1.9 Pipe 1.25% 0.86%  0.50 -
R-4C 1.02013) 2013 ok 0.50 1.50
R-0009 RG® 1.900%) 2.6(6) Block  0.50 1.50
cr-158° 1.76(%) 3.6(4) Pipe 0.40% 0.19%  1.50
186" 1.868)  3.g(6) Bar 0.50 1.50
cen® 1.92 b2 Block  0.50 1.50
CEY-B° 1.79 4.3 Pipe  1.25%  0.86> 0.50

54

Table 2.1.7 (continued)

Specimen Dimensions,

Graphite  Bulk Volume Nominal (in.)

Graphite Apparent of Graphite

. Shape .

Grade D?2722§ Per?;?ted D;imgBer T Length(s)
5-4° 1.85 4.8 Block  0.50 1.50
R-0025° 18703 5 (24) Block  0.50 1.50, 0.50
rim-24S 1.83(8) 5.7(13) Pipe  0.50 1.50, 0.50
ATJ-82 RG® 1.83 6.4 Block  0.50 1.50
r-0013°  1.87(8) 6.64%) Block  0.50 1.50
wiim-g2 181034 g p(24) Block  0.50 1.50, 0.50
Cs-82 RGb 1.80 8.1 Block 0.50 1.50
GT-123 1.79 8.9 Rod 0.50 1.50
ATT-82 1.80 10.8 Block  0.50 1.50
MHALM 1.76 11.4 Block  0.50 1.50
SM-1 1.72 13.9 Block 0.50 1.50
acor® 1.6817) 13,6(27) Bar 0.50 1.50, 0.50
TsFP 1.67 V4.4 Bar 0.50 1.50

a'I'hese are the original dimensions of the graphite as manufactured.

bThese were previously reported in MSR Quar. Prog. Oct. 31, 1959,
ORNL-2890, p 44.

cSpecimens for R-0025 were taken from a piece 39 in., in diameter and
11 in. long.

dSpecimens for RIM-24 were taken from the walls of a 5-in.-OD, 3-5/8-
in,-ID, 26-in.- long pipe.

Table 2.1.8, Permeation of Various Grades of Graphite by LiF-BeF,-
ThF,-UF, (67-18.5-14-0.5 mole %, BULT 14—.5U) at 95 and 150 psig

Test conditiocons:

Temperature: 1300°F (704°C)
‘ Exposure period: 100 hr
|

Bulk Volume of Graphite
Graphite Grade Permeated (%)a P95:Pl5OC
At 95 psig At 150 psigb
CT-158 0.04 3.6 O.Ol .
CT-150 0.3 1.8 0.2
CEY-13250 0.2 0.7 0.3
CEY-G 0.3 0.9 0.3
CEY-82 1.0 1.9 0.5
CsS-82 RG 5.3 g.1 0.6
CEY-B 3.0 4.3 0.7
GT-123 6.6 g.9 0.7
GT-123-82 O.25 0.3 0.8
MH4IM 8.9 11.4 0.8
R-0013 5.4 6.6 0.8
SM-1 12.2 13.9 0.9
S-4 A 4.8 0.9
ATI-82 10.6 10.8 1.0
ATJ-82 RG 6.3 v 1.0
R-0009 RG 2.6 .5 1.0 -
R-4 2.3 2 1.0
186 1 3.8 1.1 )
AGOT 16.1 14.6 1.1
R-0009 2.2 1.8 1.2

a'Each value 1s an average of three or more values.

b
These data have been reported previously; they are included here
for comparison purposes.

CThe ratio of the per cent of the bulk volume of the graphite per-
meated by BULT 14—.5U at 95 psig to that at 150 psig.

56

salt permeation can be decreased appreciably by proper pressure reductions.
However, a specific grade or grades of graphite chosen for a reactor should
be thoroughly investigated, since the low-permeability grades are still
relatively new.

Studies of fluoride permeation of graphite as a function of time have
been initiated with AGOT, S-4, and R-4 grades, which were chosen because
they are in plentiful supply. A partially completed series of tests indi-
cates that a l-hr exposure of these grades to BULT 1l4—.5U at 1300°F (704°C)
under a pressure of 150 psig produced the same degree of permeation as did
100-hr exposures under like conditions.

Grade R-4 graphite is one in which relatively large, accessible voids
were partially filled with carbon to decrease its permeabllity. The graph-
ite was impregnated with pitch, and the pitch was thermally converted to
amorphous carbon. Metallographic examinations indicated that the original
voids were approximately two-thirds filled with carbon. An exploratory
test was made to determine the relationship of permeability vs location in
this grade of graphite.

A standard permeation screening test as described above was made on
four specimens from the original surface of the graphite and three specl-
mens approximately 1/2 to 1 in. away from this surface. The speclimens were
nominally 1/2 in. in diameter and 1-1/2 in. long. The bulk volume of the
graphite permeated by BULT 14—.5U averaged 1.4% (0.7 to 2.7%) for specimens
that were located at the surface and 3.0% (2.3 to 4.2%) for those not at
the surface. More tests would be necessary to fully evaluate the varia-
tions in the fluoride-salt permeability of grade R-4. However, these data
indicate that for the impregnation technique used the lowest permeability
values extend over short distances, 1/2 to 1 in, The average percentage
for all seven specimens was 2.1, which compares well with a 1.9 average
value in previous tests for six similar specimens from random positicns.

Grade AGOT graphite was permeated with molten bismuth as a pretreat-
ment to a standard fuel permeation test.*? The purpose of the bismuth
permeation was twofold: (1) to determine if the bismuth would occupy the

pore spaces in the graphite and exclude fuel from these pores and (2) to

12MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 43.

determine if the bismuth which had permeated the graphite pores would be
retained within them during reactor temperature and pressure cycles. The
latter was important not only in maintaining resistance to salt permeation
but also in preventing contact of the bismuth with the nickel-base alloy
container, since nickel-base alloys have limited corrosion resistance to
bismuth at temperatures as low as 520°F (270°C).13

Results of the standard fuel-permeation test with the bismuth-im-
pregnated graphite showed that the bismuth suppressed the permeation by
fuel but did not eliminate it. Chemical analyses of six successive 1/16-
in.-thick layers machined from the graphite specimen showed a decreasing
uranium content as shown in Table 2.1.9.

Bismuth was not completely retained in the graphite pores during the
fuel-permeation test, and this resulted in some attack of the Inconel con-
tainer by the bismuth.

Bismuth impregnation will be tested to determine its effectiveness
in inhibiting fuel permeation of low-permeability small-pore-diameter

graphite, as the graphite becomes available.

131iquid Metals Handbook, p 176, June 1952,

Table 2.1.9. Effect of Impregnation with Bismuth in Inhibiting
Fuel Permeation of AGOT Graphite

Pretreatment conditions: molten bismuth at 1022°F (550°C),
200 hr, 125 psig
Fuel-permeation conditions: fuel 130 at 1300°F (704°C), 100
hr, 150 psig

Depth of Cut Uranium Analyses (mg of U per g of graphite)

(in.) Untreated Graphite Pretreated Graphite
0%—1/16 6.6 2.6
1/16-1/8 7.5 2.6
1/8-3/16 7.2 24
3/16~1 /4 7.6 1.7
1/4—5/16 7.4 L.3
5/16-3/8 7.3 C.9

*Surface of specimen.

58

\/ Removal of Oxygen Contamination from Graphite

Work has continued on methods for removing oxygen contamination from
graphite as a means of eliminating precipitation of UO, from LiF-BelF,-UF,
salt systems. As previously reported,l4 an expensive method of purging
graphite of oxygen contaminants has been found. Other methods investigated
include the use of preheated hydrogen and the thermal decomposition of am-
monium bifluoride, NH,F-.HF.

In a single test, flushing an AGOT-grade graphite crucible with pre-
heated hydrogen for 50 hr at 1300°F (704°C) prior to charging it with fuel
130 (LiF-BeF,-UF,, 62-37-1 mole %) did not detectably alter the quantity
of uranium that is normally precipitated from fuel 130 by contaminants of
this graphite.

The test system was evacuated for 15 hr at less than 2 u Hg pressure
at 1300°F (704°C) prior to the hydrogen flushing operation. The hydrogen
(~ 99.5% H,) was passed through a commercial deoxygenation unit and a
molisture trap, was heated to 1300°F (704°C), and was then passed through
the test system, which was also at 1300°F (704°C). The calculated flow
rate of the hydrogen was 2.8 ft3/hr. The charge of fuel 130 for the test
was stored in a chamber integral with the test unit. Therefore the system
was not opened to the atmosphere during the transfer of frozen fuel to the
graphite crucible at 1300°F (704°C). Following the fuel transfer, the test
unit was evacuated (< 2 u Hg pressure) and sealed and was then held at
temperature for 100 hr. Radiographic examinations were used to detect the
quantity of uranium that precipitated.

The value of hydrogen as a gettering agent for the contaminants of
graphite will be further investigated as a function of temperature and/or
time. It is also planned to determine whether a catalyst will make the
hydrogen an effective gettering agent.

A test has been completed in which oxygen contaminants were partially
or completely removed from a crucible of AGOT-grade graphite with a nominal
bulk volume of 190 cc by the thermal decomposition of 1 g of NH,F.HF. The
graphite crucible containing crystals of NH,F.HF was sealed in a vacuum

of 5 in. Hg in an Inconel container. The residual gas in the container

14MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 44.

59

was pure argon. The test unit was heated from room temperature to 1300°F
(704°C). The heatup of the argon plus the thermal decomposition of the
NH,F-HF to form NH, and HF raised the pressure of the system to 35 psig.
After 22 hr, the system was evacuated to less than 1 u, and a solid charge
of fuel 130 was placed in the graphite crucible. The unit had been de-
signed so that this could be accomplished with the crucible at 1300°F
(704°C) and without opening the system to the atmosphere. The ratio of
the bulk volume of graphite to the volume of fuel 130 at 1300°F (704°C)
was 27:1.

Uranium precipitation was not detected by radiogr@pp;c_examinations
after 90- and 1000-hr exposures. Normally, in an untreated graphite—fuel
130 system, the precipitation of uranium occurs in 5 hr or less. The ab- _ |
sence of the precipitate in the test suggests that the pretreatment with
the NH,F-HF was effective in removing the oxide contaminants. The test
is being continued to determine whether tenaciously held oxide contami-
nants might eventually cause precipitation. Past data concerning uranium
precipitation in graphite—fuel 130 systems indicate that this should not
happen.

Compatibility of INOR-8 and Graphite

The tendency for INOR-8 to carburize when under pressure and in direct
contact with graphite was investigated. A modified stress-rupture appa-
ratus was used to hold grade TSF graphite in contact with INOR-8& (heat
1665) at 1000 psi in fuel 30 (NaF-ZrF,-UF,, 50-46-4 mole %). The test
system was maintained at 1300°F (704°C) for 3400 hr. Metallographic ex-
amination (Fig. 2.1.15) indicated that carburization had occurred to a
depth of about 14 mils. Two successive sets of millings, each 7 mils
thick, taken from the INOR-8 platen at the metal-graphite contact surface
analyzed 0.38 and 0.12 wt % carbon, respectively. Nominal carbon content
of the INOR-8 is 0.067 wt %. The other side of the platen not in contact
with the graphite and the surfaces of the female threads in the platen
were examined metallographically; no carburization was detected on them,

A test is currently in progress to determine whether the rate of car-
burization of INOR-8 is dependent upon the stress on the material. An
INCR-8 specimen is being tested in a strongly carburizing medium while

being subjected to various stress concentrations,

60

UNCL ASSIFIED
50419

Fig. 2.1.15. Surface of INOR-8 (Heat No. 1665) Platen Which Was Held in Contact with
Grade TSF Graphite for 3400 hr at 1000 psi in Fuel 30.

61

Corrosion Resistance of Brazing Alloys to Fluoride Fuel

Since a barren-salt-to-fuel-salt heat exchanger may be used in the
Molten-Salt Reactor, brazing alloys which could be used in fabricating such
equipment have been examined for corrosion resistance to fuel 130 (LiF-
BeF,-UF,, 62-37-1 mole %), a potential reactor fuel.

Five different brazing materials, three nickel-base alloys, one gold-
base alloy, and pure copper, were chosen which had exhibited good resist-
ance to fuels 30 and 44 (the NaF-ZrF,—base fuels). The tests were con-
ducted by brazing small coupons together to form lap joints of Inconel to
Inconel and of INOR-8 to INOR-8. Ten of these Jjoints were suspended in an
enlarged section in the hot leg of each of three thermal-convection loops.
The loops were operated for 1000, 5000, and 10,000 hr at hot-leg tempera-
tures of 1300°F (704°C). The results from the 1000- and 5000-hr tests have
been reported previously.l?

The 10,000-hr test has been terminated and examined, and the results
are listed in Table 2.1.10. All the brazed Jjoints showed good corrosion
resistance when used to join INOR-8. However, Inconel Joints brazed with
Coast Metals No. 52 and 53 and General Electric No. 8l were attacked.

The results from the 1000-, 5000-, and 10,000-hr corrosion tests in-
dicate that several brazing materials suitable for joining INOR-8 or In-
conel have adequate corrosion resistance to the type of fluoride salt rep-
resented by fuel 130.

An alloy development program is being conducted by the Welding and
Brazing Group in an effort to develop brazing alloys which would be useful
in joining graphite to graphite and graphite to metal. As these alloys
are developed, they are corrosion-tested to determine their potential use
Tor Molten-Salt Reactor applications. Recent work has been done primarily
on gold-base alloys containing various amounts of tantalum, nickel, and
other minor constituents to provide proper flowability and melting point.
The results of corrosion tests on these alloys in different fuel salts are
listed in Table 2.1.11. A moderate concentration of tantalum was detected
spectrographically on the inside walls of the nickel container from the

60% Au—-30% Ta—10% Ni test. This dissimilar-metal mass transfer was not

15MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 46.

62

Table 2.1.10. Results of Corrosion Tests on Brazing Alloys
in LiF-BeF,-UF, (62-37-1 mole %, Fuel 130)
in a Thermal-Convection ILoop

Test conditions:
Time: 10,000 hr
Temperature: 1300°F (704°C)

Metallographic Results

Nominal Composition of

Brazing Material (wt %) Base Material

INOR-8 Inconel
Coast Metals No. 52 No attack observed; Diffusion voids to
(89 Ni—4 B-5 Si—2 Fe) no porosity 3 mils in depth
below surface
Coast Metals No. 53 No attack or poros- Severe porosity 15
(81 Ni—-8 Cr— B4 Si-3 Fe) ity mils deep
General Electric No. 81 No attack observed Complete attack;
(70 Ni—20 cr—10 si) severe porosity
through fillet
82 Au—18 Ni No attack No attack
Pure copper No attack or dif- No attack; small
fusion voids diffusion voids
are present in
fillet

intense enough to be detected on the nickel capsule by metallographic ex-
amination.
. The relatively large weight gain on the graphite-to-molybdenum T-joint
was due to pickup of fuel by the graphite. The two alloys listed last in
the table showed good resistance to fuel 30, which was used in this test

to simulate the components of a proposed pump-loop test being planned.

Table 2.1.11.

Test conditions:
Time: 100 hr
Temperature:

1300°F (704°C)

Results of Fuel Corrosion Tests on Newly Developed Brazing Alloys

_ Fuel Weight
Alloy(gzm5381tlon Test Container TygeegimEESt Used Change Metallographic Results-
’ P as Bath#* (mg)
60 Au—30 Ta—10 Ni Nickel capsule Alloy button 130 -3 Slight, scattered sub-
surface voids to 1
mil in depth
62 Au—26 Ta—12 Ni Nickel capsule Graphite-Mo 130 +36 Slight, scattered sub-
T-joint surface voids to 2
mils in depth
58 Au—27 Ni-8 Ta- Inconel capsule Alloy button 30 -0.9 No attack observed
3 Mo—2 Cr—-2 Fe
63 Au—29 Ni—-3.5 Mo— Inconel capsule Alloy button 30 N.A. %% No attack observed

2.5 Cr—2 Te

¥130: LiF-BeF,-UF, (62-37-1 mole %)
30: NaF-ZrF,-UF, (50-46-4 mole %)
**Not available.
2.2 CHEMISTRY AND RADIATION EFFECTS

Phase Equilibrium Studies

The System Li¥F-BeF,=-Zrf,

For a minimum temperature of 500°C the LiF~BeF,; binary system can
serve as a fuel solvent for a combined total of about 15 mole % UF, and
ThF,, together or singly, and there are indications that both the oxide
tolerance and the corrosivity of fuels are improved by the presence of a
relatively high proportion of the gquadrivalent cations. Thus, tc mini-
mize uranium inventories and yet retain the advantages of a more "acidic"
fuel, the addition of ZrF, is an attractive possibility; optimized fuels
in the future may well be chosen from the quinary system LiF-BeF,;-ZrF,-
Th¥F,-UF, .

In order to estimate whether such quinary mixtures — suitable as re-
actor fuels and containing 60 to 75 mole % LiF — are molten at 500°C,
preliminary studies of the phase relationships in the system LiF-BeF,-
Zr¥, were made. The equilibrium phases present in quenched samples of
seven LiF-BeF,~ZrF, mixtures show that at 60 to 75 mole % LiF, at least
15 mole % ZrF, can be contained in liquid solutions of LiF-BeF, at 500°C.
The composition sections 3LiF*ZrF,-2LiF*BeF, are not quasi-binary at any
temperature interval, because of the occurrences of the primary-phase
field of an unidentified phase on these composition sections. This phase
is probably a ternary compound of LiF-BeF,-ZrF, containing at least 75
mole % LiF. The compound is uniaxial negative and has such low bire-
fringence that only the average refractive index, 1.418, could be meas-
ured. At each of the four compositions where it was observed, the opti-
cal properties of the ternary compound were identical, from which fact
cne may infer that the compound forms little or no solid solution in the
ternary system. An x-ray diffraction standard pattern for the compound

has been derived.

The System NaF-ZrF,-ThF,

Renewed attention has been directed to the phase studies in the sys-
tem NaF-ZrF,-ThF,, because the system NaF-ZrF,-ThF,-UF, could provide al-
ternative converter-breeder reactor fuels and because of possible advan-

tages for fuel reprocessing cycles. Current efforts are concerned with

65
the determination of phase relationships in the composition area least
studied previously,l that of compositions containing less than 50 mole %
NaF. The determination of the refractive indices of the ZrF,-ThF, solid-
solution compositions furnishes a definitive method of establishing tie
lines in the Zr¥F,-ThF, solid-solution primary-phase area of the system
NaF-Zr¥,-ThF,. This information, once obtained, will be used to construct
fractionation paths in this primary-phase area. The combination of frac-
tionation paths and liquidus isotherms on the diagram will indicate the
composition of the liquid and the solid phase in equilibrium for each
compesition in the primary-phase area. The experimental determinations
of the refractive indices of the ZrF,~ThF, solid-solution series and of
the fractionation paths for the Zr¥F,-ThF, solid-solution primary-phase

area are in progress.

The System NaF-ThF,-UF,

The similarity of crystal properties of ThF, and UF, is conducive to
similarities in the compounds formed by these substances with other metal
fluorides. This is reflected by several similarities in the phase dia-
grams for the systems NaF-ThF, and NaF-UF,, but not all the compounds in
the two systems occur with identical NaF/XF4 molecular ratios. Each of
the equilibrium phases in the system NaF-ThF, takes significant amounts
of UF,; into solid solution, and also the solid phases in the system NaF-
UF, form sclid solutions by including ThF,.
Continuing studies of the phase equilibria in the system NaF-ThF,-
UF, provide additional evidence that the two isostructural crystal forms .
of the compounds 4NaF-:ThF, and 3NaF'UF, form a continuous solid sclution
in spite of the difference in stoichiometry of the end members. Phase .
data from some 15 thermal-gradient quenching experiments indicate that
the composition of the solid solution is confined to the line B-4NaF:ThF,—
B-3NaF-UF,;. On the other hand, at higher temperatures, where the alpha
modifications are stable, partial solubility of UF, occurs in Q;éNaF'ThF4,
and of the ThF, in &-3NaF:UF,;, so that the alpha-modification solid solu-

tions contain constant proportions of NaF, either 80 or 75 mole %. The

l1R. E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materials,
ORNL-2548, p 82.

66

-

TEMPERATURE (°C)

solubility of UF; in Q-4NaF:ThF, is limited to approximately 13 mole %,
and the solubility of ThF, in =3NaF-UF, is limited to approximately 4

mole %.

In Fig. 2.2.1 are shown the results obtained from thermal-gra-

dient quenching experiments on the composition line 4NaFTh¥,-3NaF - UF,.

Tt is unusual for isostructural crystal modifications to occur for a pair

of MF-X¥, compounds whose molecular ratios, ME‘/)CF;,, are not identical.

Considerable evidence exists which indicates that within the system NaF-

Th¥,-UF;, there is a similar continuous solid solution between the iso-

structural cubic pairs 3NaF:2ThF, and 5NaF-3UF,;, and limited solubility

725

700 -

675

650

625

600

575

525

500

475
4NaF-Thf, 5 10 15 20

UF, (male %)

3NaF-UF,

N = om = = a = m m = .
COBAPO LNV AND O SO

UNCLASSIFIED
ORNL-LR-DWG 48668A

LIQUID

NaoF + LIQUID

a-4Naf -Th{U)F, + NoF + LIQUID

a-4NaF -Th{U)F, + LIQUID

a-4NaF - Th(U)F + B,-2NaF -Th(U}f + LIQUID

a-4NoF Th(U)f, + B,-2NaF-Th(U)R,

a-4NaF Th(U)F, + B-4NaF-Thf — B-3NaF - Ufss + B,-2NaF -Th(U)F,
a-4NaF -Th(U)f, + B-4NaF-Thf, — 8-3NoF -Uf,ss + LIQUID
B-4NaF -ThE,— 3-3NaF - Ufzss + LIQUID

a-3NaF - U(Th)F, + 8- 4NaF Thf,— B-3NaF - Uf,ss + LIQUID
a-3NaF -U(Th)F, + LIQUID

a-3NaF -U{Th)F, + NaF +LIQUID

a-3NaF-U(Th)F, + NaF

@-3NaF-U(Th)F, + NoF +3-4NaF-Th —B-3NaF -Ufss
B-4NaF Thf; — B-3NaF -Ufss

NaF + B-4NoF-Thfy — B8-3NaF URss + B85~ 2NaF -Th(LIF,
NGF+,82~2N0F-Th(U)F4

NaF + B,-2NaF -ThU}f + 8-2NaF-U(Thif

NaF + B8-4NaF -Thf,— B8-3NaF-Ufss + 8-2NaF-U(Th)f
NaF +8-2NaF-U{Th)f,

Fig. 2.2.1. System NaF-ThF ;-UF ,; The Section 4NaF.ThF ;-3NaF-UF ,.

between the nonisostructural pairs 2NaF+:ThF,-2NaF'UF, and NaF+«2ThF,-
NaF +2UF, .

Phase Diagrams for Fluoride Systems

Detailed descriptions of the phase equilibria in systems of alkali
fluorides with zirconium, thorium, and uranium fluorides have recently
appeared in an annual report,2 and a comprehensive compllation of dia-
grams of such systems, along with others of significance to fluid-fuel
reactor development, all of which were investigated at ORNL, has also

been issued.l

Fission-Product Behavior and Reprocessing Chemistry

Extraction of Uranium and Protactinium from LiF-BeF,-ThF, Mixtures

The advantages of a molten-salt nuclear reactor as a thermal breeder

U232 and

could be greatly enhanced by improved processes for recovering
protactinium from a blanket salt. A recent.review of various oxide pre-
cipitation reactions® provides examples of the potential usefulness of
oxide precipitants in molten-salt mixtures containing uranium, thorium,
rare earths (represented as cerium), and beryllium to selectively remove
these elements from solution as oxides. The development of a suitable
chemical reprocessing scheme based on these reactions has been pursued
with favorable results.

In the proposed blanket material, LiF-BeF,-Th¥, (67-18-15 mole %),
the problem is to remove trace quantities of U233, produced by the breed-
ing process, in the presence of rather large thorium concentrations. The
difference in the behavior of uranium and thorium in reacting with calci-
um oxide, as illustrated in Fig. 2.2.2, provides the basis for oxide sep- "
aration schemes, although calcium oxide is not the preferred additive.

The production of y223 by thermal breeding is achieved by the decay
of Pa?33, However, Pa??? will also absorb neutrons by a similar reaction
to produce U??%, Hence, isolation of Pa??? from the neutron flux field

removes an important source of nuclear poisoning. An oxide precipitation

“Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960, ORNL-2931, p 3-27.
’Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960, CRNL-2931, p 84.

68

UNCLASSIFIED
ORNL-LR-DWG 43019
[

THEORETICAL END PQINTS
UF4 Tth CBF3
8 | aUBY CHEMICAL ANALYSIS
. ® Th BY ACTIVATION ANALYSIS
Iy - ®Ce BY RADIOCHEMICAL ANALYSIS
\‘

METAL FOUND (wt %)

e} 200 400 600 800 {000 1200
CaQ ADDED (meq)

Fig. 2.2.2. Reaction of CaO with UF, ThF ,
and CeF 5 in LiF-NaF (60-40 mole %) at 750°C.

scheme for protactinium removal appears to be useful for accomplishing
this objective.

Since the treatment to recover uranium and protactinium should not
cause serious alterations in the composition of the blanket material, the
choice of precipitants is probably limited to BeO, ThO,, UO,, or possibly
water vapor. The reaction of BeO with trace quantities of UF, dissolved
in LiF-BeF,-ThF,; (67-18-15 mole %) is illustrated in Fig. 2.2.3. In three
separate experiments, the addition of approximately 30 g of BeQ per kg of
melt effectively removed approximately 2000 ppm of dissolved uranium.
Roughly comparable results were obtained on using either ThO, or BeO to
precipitate uranium from the proposed blanket mixture in the experiments
summarized by Fig. 2.2.4.

The precipitation of prectactinium from solution in LiF-BeF,-ThF,
(67-18-15 mole %) was examined in a separate series of experiments. Ini-
tially, the removal of Pa?’? activity, obtained by the irradiation of
thorium, was followed as BeO and Ca0 were added, giving results typified
by Fig. 2.2.5. Following each precipitation experiment, the protactinium

was successfully redissolved by hydrofluorination in situ. The removal

of an estimated 50 ppm of Pa23l, labeled with Pa233, by reaction with BeO

is illustrated in Fig. 2.2.6. Similar results, shown in Fig. 2.2.7, were

UNCLASSIFIED UNCLASSIFIED

ORNL— LR— DWG 46377 ORNL-LR-DWG 4797%
2000 x104)
[ ]
=)
£
= 500 INITIAL WEIGHT OF SOLVENT: 496 g e |
& SURFACE AREA OF BeQ: 25.8m2/g L
— =
o = ® BeQ
q = o Ca0
@ J
- o]
= &
b z
z —
s s
= 5
E 1000 =
> 2
5 2
b [+
©
2 \
o
=
=
w
S \ <00
S 500 t ‘ R
o] ° ——
0 | 2 3 4 5
OXIDE PRECIPITANT ADDED (grams)
1S Fig. 2.2.5. Removal of Protactinium (Pa 233y
\ < 1 opm from Solution in LiF-BeF ,-ThF , (67-18-15 mole %)
[ ] . .
N~ by Reaction with BeO and CaO at 650°C.
0 e o—o—-o—o—.—lo
0 5 10 15 20 25
BeC ADDED (grams)
Fig. 2.2.3. Extraction of Uranium from LiF-
UNCLASSIFIED
BeF 2'ThF4 (67']8']5 mOIe %) at 650°Co ORNL-LR—-DWG 47976
(x 104}
UNCLASSIFIED
2000 ¢—e. ORNL-LR-DWG 472378
N, . "\
\ ~ 20
o
. ® ThO £
1600 \ ¢ &
[= O BeQ o
a
= 8 15
: . 5 A
5 1200 o}
J w
3 \ z
=
3 \ S 10
2 z \
S a0 =
© 800 v E ®
\ > AN
: \ £ s
o
o
400 . \ o
\\ 9
Y \v
\-0\ o} é
. — 0 0.5 1.0 15 2.0 2.5

0 100 200 300 400 5GO BeO ADDED (groms)

OXIDE ADDED (millieauivalents) : Y -
mieauatents Fig. 2.2.6. Removal of Protactinium (Pa?3!

Fig. 2.2.4. Reactions of ThO, and BeO with Labeled with Pa?*?) from Solution in LiF-BeF ,-
UF, in LiF-Ber-ThF4 (67-18-15 mole %) ot ThF, (67-18-15 mole %) by Reaction with BeO at
650°C. 650°C.

70

Al

UNCLASSIFIED :
ORNL—LR—DWG 48075

(x 10%)

20 o

PRCTACTINIUM IN SOLUTICON {counts/min - g}

\..__

0 1 2 3 4
ThO, ADDED (g)

o e

Fig. 2.2.7. Removal of Protactinium (Pa 231
Labeled with Pa233) from Solution in LiF-BeF ,-
ThF , (67-18-15 mole %) by Reaction with ThO, at
650°C.

obtained with ThO, as the precipitant. Such examples demonstrate that
realistic quantities of both uranium and protactinium can be precipitated
from solution in the proposed blanket melt for a molten-salt breeder re-
actor by using oxide reactions.

Current studies are directed toward the effect of added oxide on the
composition of the blanket material and toward solid-liquid separations
following precipitation reactions. Since BeO and ThO, are ceramic mate-
rials, they can be processed intc shapes having variable densities and
surface areas, and if adherent precipitates of uranium and protactinium
form on BeQ or ThO, ceramic bodies, the separation from the ligquid would
be facilitated. Preliminary results on the relative reaction rates with
BeQ, extruded into 1/16-in.-dia x 1/4-in.-long cylinders having average
surface areas of 8.5 m?/g, are shown in Fig. 2.2.8. In another experi-
ment, an apparent adherence of UQ, to extruded ceramic bodies of BeO was

demonstrated (Fig. 2.2.9) by the disappearance of uranium from molten

UNCLASSIFIED

UNGCLASSIFIED -LR-
ORNL-CR-DWG 46877 2000 QRNL-LR-DWG 48669
1500 , ‘ ]
[
4 TOTAL WEIGHT OF SOLUTION: 2.4kg
TOTAL WEIGHT OF BeOQ: 1209
EXTRUDED SHAPE OF BeQ: CYLINDERS
1/16—“1. DIA x 1/4-ifl. LENGTH
CONCENTRATION BeO: 4Cg /kg melt SURFACE AREA Be(:8.5 rnzlg
SURFACE AREA Be0:8.5m2/q 1500
EXTRUDED BeO: Y%g-in DIA x Y4 in E
1000 2
E
2 g
z =
2
<
5 ?
§ z
= g 1000 \
@ = -
¢ =z
z \ 3
= 500 2 \
-} 1 4
: : -
(3 2 ™
5 \\‘ z
o
-} \
500
\“ \\4
0
0 1 2 3 4 5
TIME {hr)
. . . . . o)
Fig. 2.2.8. Reaction of BeQO with UF, in LiF- o =0 100 150 200 250
BeF ,-ThF, (67-18-15 mole %) at 650°C. EXTRACTION TIME {min)

Fig. 2.2.9. Extraction of UF, in LiF-BeF,-ThF
(67-18-15 mole %) by Reaction with Extruded BeO
at 650°C.
LiF-BeF,-ThF, (67-18-15 mole %) collected from a solid bed by forcing
the liquid phase through a nickel screen, having approximately 900-u
openings, into a separate container. The analyses shown in Fig. 2.2.9
were obtained after hydrofluorination of the separated liquid.
The effect of an added oxide precipitant is determined by the equi-
librium for the reaction of the oxide with the fluoride components. For s

example, the equilibrium reaction
ZBeO + ThF, = 2BeF, + ThO,

influences the conditions which must be maintained to avoid precipita-
tion of thorium during the extraction of protactinium and uranium. The
relative stability of BeO and ThO, in LiF-NaF (60-40 mole %), as evi-
denced by the reaction of calcium oxide with dissolved BeF, and ThF., is
shown in Fig. 2.2.10. If BeO is added to ThF, dissolved (8 wt % Th) in

this same solvent, an equilibrium which apparently involves both solid

72

T

oxides is encountered (Fig. 2.2.11).

With more BeF,; present, as in the

case of the solvent LiF-BeF, (63-37 mole %), shown in Fig. 2.2.12, no

precipitation of ThO; occurs when BeO is added to a solution again con-

taining 8 wt % thorium.

Further study of this reaction is in progress.

UNCLASSIFIED
ORNL-LR-DWG 47977
T

& THORIUM :
o BERYLLIUM

200 .\Q \ - : : -]
‘ \.—: .

¢ 100 200 300 400 500 600 700
CaQ ADDED (milliequivalents)

&)
Q
(@]

400 ]

100 [

METAL FOUND IN SOLUTION {milliequivalents)

Fig. 2.2,10. Reaction of CaO with ThF, and
BeF, in LiF-NaF (60-40 mole %) at 750°C.

UNCLASSIFIED
ORNL-LR-DWG 47420

12 T T 1
® URANIUM
[
A THORIUM} BY ACTIVATION ANALYSIS
UNCLASSIFIED
ORNL-LR-DWG 47979
10 10
~ . 53
82 8 T 8 AT § A A
-— = 3 A ‘—-—-—-—.
% | g A
3 | 3 :
2 6 e——END POINT ThFy4 2 6
o] 4
w ‘ (]
=2 \ Z
S | 2
=2 | -
c 4 M ‘ T 4
ul
s | =
S
&
2 e ’ 2
END T
| POINT ENDTSC?'N
| U0 2
| 2
0 * 0 ‘ ; .‘-__e 'y Y
0 2 4 (53 8 10 0 250 500 750 1000 1250

Fig. 2.2.1L

Reaction of BeQ with ThF4 in
LLiF-NaF (60-40 mole %) at 750°C.

BeO ADDED (grams) BeO ADDED (meq)

Fig. 2.2.12. Reaction of BeO with UF, and
ThF , in LiF-BeF, (63-37 mole %) at 600°C.

73

The insolubility of uranium oxide in comparison with thorium oxide
was not anticipated from the relative stability of the crystals and
hence must depend on as-yet-unelucidated solution behavior, but the fea-
sibility of extracting uranium and protactinium from solution in LiF-
BeF,-ThF, (67-18-15 mole %) seems well established for the concentrations
of uranium and protactinium to be expected in the proposed blanket of a
two-region molten-salt breeder reactor. Of further interest in this con-
nection is the experimental observation that protactinium was precipi-
tated from sclution in the presence of approximately 1000 to 2000 ppm
uranium. This suggests that the precipitated oxide of protactinium is ‘
more steble than UO,. A study of the possible application of this reac-
tion to a one-region molten-salt breeder reactor is now in progress, &as
are studies to provide further experimental information for exploiting
the reactions of BeO and ThO, with uranium and protactinium in & chemi-

cal reprocessing scheme.

Chemistry of Corrosion Processes

Effect of Oxide on LiF-BeF,~ThF, and LiF-BeF,-ThF,~UF, Mixtures

The applicability of thermal-gradient quenching techniques for ob-
taining information on the effect of oxide in LiF-BeF,-ThF, and LiF-BeF,-
ThF,-UF, melts at high temperatures has been explored. Oxide was in-
cluded in the equilibrated samples by substituting beryllium oxide for a
fraction of the BeF, contained in the mixtures LiF-BeF,-ThF, (67-18-15
mole %) and LiF-BeF,-ThF,-UF,; (67-18-14.5-0.5 mole %). Samples with in-
crements of BeQ between 0.006 and 2.9 wt % were annealed in the tempera- )
ture ranges between 450 and 550°C for one week before quenching. Deduc-
tions concerning the reactions occurring at the high temperatures were
made by noting the phases present in the quenched samples as revealed by
petrographic and x-ray diffraction examinations.

In LiF-BeF,-ThF, the occurrence of ThO, was found only in samples
containing 0.3 wt % or more BeO. However, the samples containing 0.06,
0.03, and 0.006 wt % BeQ gave rise to an unidentified, highly bire-
fringent phase (refractive index: high N = 1.70, low N = 1.55), which
might well be a thorium oxyfluoride. In none of these samples were

traces of unreacted BeO observed.

74 -

In the quaternary mixtures, Be0O was added in increments upward from
0.3 wt %. At 0.3 wt % BeO, the only oxide phase encountered was UOp. At
0.8 wt % and upward, both UO, and ThO, were cbserved, mixed but uncom-
bined between 450 and 500°C and as a U0,;-ThO, sclid solution at tempera-
tures above 500°C. The results of these experiments appear to corrobo-
rate the sequential precipitation of ThO, and UO, previously noted by
Watson and Shaffer3'* when BeQ is used as =a precipitant in LiF-BeF,; melts,
and also indicate that U0,-ThO, solid solutions are readily formed at much
lower temperatures in fused-salt media than by solid-state reaction.

No quantitative conclusions regarding the solubility of BeO can be
made from these quenching experiments, partly because of the possibility
that small amounts of BeQ remain dissolved in the glassy portion of the
quenched samples without noticeably altering the refractive index of the
glass. However, the detection of a distinguishable oxygen-containing
phase at concentrations as low as 0.003 wt % BeQO in samples that would
have otherwise corresponded to complete liquefaction can be regarded as

evidence that the oxide tolerance of the sample was less than 30 ppm.

Radiotracer Studies of the Role of Cr® Diffusion as a Controlling Factor

in Corrosion by Molten Fluorides

A thorough understanding of the solid-state diffusion of chromium
in MSRE alloys is an important aspect of corrosion studies. The present
state of knowledge in this field has been summarized recently4 and will
appear in topical reports. -

A large number of the chromium diffusion coefficients available in
the literature® were applicable only at temperatures above 900°C. How-
ever, extrapolation of this information to temperatures of reactor in-

terest (600 to 900°C) could be misleading if the diffusion mechanism

“W. R. Grimes, G. M. Watson, et al., Radiotracer Techniques in the
Study of Corrosion by Molten Fluoffaégj for presentation at Copenhagen
Conference on the Uses of Radioisotopes, September 1960.

°R. B. Evans et al., Self-Diffusion of Chromium in Nickel-Based
Alloys, ORNL report (in press).

6P, L. Gruzin and G. B. Federov, Doklady Akad. Nauk 5.5.3.R. 105,
264-67 (1955). —

75

were to change within the extrapolated region. Accordingly, self-diffu-
sion coefficients of chromium in nickel-base alloys in the temperature
range 600 to 900°C were determined both by monitoring the depletion of
tracer from melts contalining Cr5lF2 and by successively electropolishing
the metal to obtain the concentration profile of the tracer in the alloy.

As reported previOusly,7’8 the diffusion coefficients change from
10716 cm?/sec at 600°C to about 10712 at 800°C but can vary from 10716
to 10715 em?/sec at 600°C depending on the annealing history of the
Inconel. Low diffusion coefficients are found in samples exhibiting
large grain growth. At temperatures above 800°C the diffusion coeffi-
cents in Inconel as measured by the different techniques were the same,
but at lower temperatures there were discrepancies in such a direction
as to suggest the emergence of a grain-boundary-path mechanism. With
INOR-8 no such differences were found. Also, the self-diffusion coeffi-
cients of chromium in INOR-8 and Inconel appear to be about the same if
the higher values found for Inconel at low temperatures are adopted as
the basis of comparison.

In order to obtain corrosion data directly amenable to analysis, two
anisothermal convection loops of INOR-8 were placed in operation under
identical conditions with tracer CrF,, but no corrodant, in the melt. The
first loop, No. 1248, was operated for 500 hr and then drained and exam-
ined to obtain the values for the self-diffusion coefficients of chromium
in INOR-8 shown in Fig. 2.2.13. A small difference was noticeable between
the values obtained by the various methods used; the coefficients obtained
on total specimen counts tended to be low at high temperatures and vice
versa. The reason for this trend is not immediately apparent and is pres-
ently ascribed to small systematic errors which lie within the over-all
experimental precision of #0.4 of a logarithmic cycle as normally encoun-
tered in such studies.

In order to determine the validity of the self-diffusion coefficients
determined in loop 1248 for the prediction of chromium diffusion rates

under corrosive conditions, the second loop, No. 1249, was operated under

“MSR Quar. Prog. Rep. June 30, 1958, ORNL-2551, p 95.
8MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 55.

76

log,o & (cm2/sec)

UNCLASSIFIED
ORNL-LR-DWG 46403

TEMPERATURE (°C) °
850 800 750 700

-1 I I ‘ ‘| ‘ ‘l

A COEFFICIENTS BASED ON TOTAL SPECIMEN COUNTS

® COEFFICIENTS BASED ON Cr°' CONCENTRATION —
GRADIENT MEASURED BY COUNTING ELECTROLYTE

O COEFFICIENTS BASED ON Cr®' CONCENTRATION
GRADIENT MEASURED BY COUNTING SPECIMEN

o
O
A\ ® ° .
00
A A \ .
SN 8
A
\:‘k
-13 T~
N A
0 \ A
A
OBTAINED UNDER NON-CORROSIVE CONDITIONS . © ~-
Bv ADDITION OF Cr*Fz TO NGF-ZrF4
-14
8.75 9.00 9.25 9.50 9.75 10.00 10.25 10.50

10,000/1 (oyc-1)
Fig. 2.2.13. Self-Diffusion Coefficients for Cr3Yin INOR-8; Loop 1248.

noncorrosive conditions for 792 hr and then a corrosive agent (Fng) was
added without otherwise disturbing the operation. After 264 hr of oper-
ation under corrosive conditions, the loop was drained and examined. As
expected, the presence of FeF, in the solution caused the chromium con-

centration at the surface of the metal toc be reduced to essentially zero.

The reaction®

cr® + FeF, = CrF, + FeC

has an equilibrium quotient of 1.5 x 10% at 600°C and greater at higher

°C. M. Blood et al., Activity Coefficients of Ferrous Fluoride and
of Nickel Fluoride in Molten Sodium-Fluoride—Zirconium-Fluoride Solu-
tions, Paper No. 75, Division of Physical and Inorganic Chemistry, 132nd
Meeting of the American Chemical Society, New York (Sept. 813, 1957).

temperatures, so that FeF,; is a drastic oxidant for cr® put has little
effect on the Ni-Mo matrix of the alloy. In Fig. 2.2.14, the results ob-
tained with a specimen from loop 1249 are compared with the results ob-
tained with the corresponding specimen from loop 1248, showing that the
experimental points agree well with the calculated curves.

Figure 2.2.15 shows the concentration profiles in the specimens as
calculated from the diffusion coefficients shown in Fig. 2.2.13. Curve 1
gives the expected concentration of tracer for 1056 hr of operation under
noncorrosive conditions. Curve 2 represents the expected tracer-concen-
tration profile in the specimen from loop 1249, altered from curve 1 by
adding FeF, after 792 hr of operation. Curve 3 shows the actual tracer-

concentration profile in the same specimen as calculated from the ratio

UNCLASSIFIED
ORNL—LR—DWG 46402

100 \u\./i"/o ] ‘ | l
AW NN o
N/ a SPECIMENS WERE LOCATED AT SIMILAR
0 X ¢ TEMPERATURE ZONES IN TRACER LOOPS
‘/Kf 1248 AND 1249
80 SIAN ,
\ ~ #'= TIME OF SPECIMEN EXPOSURE

1 > & TO LOOP SALT
| ~ 70 N N

a \\\ a, 4,= TIME AT WHICH CORRODENT WAS ADDED

(o]

=

o P \ LOOP 1249 (TRACER FOLLOWED BY FeF, ADDITION) |

=

z N . .
} <§I \ X f(x)=1/rr_oferfc2‘/07 dy \/r'r;_/erfc BT dy
| i \ 2vD VI = Wiy,
= 40 [— LOOP 1248—" \ — .

) (CONTAINED NO 5\\\ \\\\\\ 4 DATA POINTS BASED ON

@ 30 | — CORROSIVE AGENT = N NV COUNTING ELECTROLYTE — .

o DATA NORMALIZED TO { & DATA POINTS BASED ON

20 |- AS LOOP 1249)

X
10 f(x)=1/_1r_/‘erfc 4 d(2
o]

N . —
4 \)& A 2 A

VOt

0 ‘ | | |
0 5 10 15 20 25 30 35 40 45

X { microns below surface)

Fig. 2.2.14. Per Cent of Tracer Remaining in Alloy as a Function of Depth Removed from Specimen
Surface by Electropolishing.

78

UNCLASSIFIED
ORNL-I_.R-DWG 46401A

100 —
g
s {5)
80 4
N\ /1/ (1) Cr™F, ONLY FOR 7 + At
\/ {2) Cr™ F, FOR 7, THEN FeF, FOR
60 / At (PREDICTED) |
/ (1 {4) CALCULATED [Cr°] BASED ON(1)
40 /| \ {5) CALCULATED [Cr®] BASED ON(3)
7B NIRN i

(3) SAME AS (2) (ACTUAL)
/, - \\'\
Vs
20 / f’/ ~ \

INITIAL SURFACE CHROMIUM CONCENTRATION (%)

/ b Y ~
/ /, \\ {3)
i \\
0 a —
0] 5 10 15 20 25 30 35 40

PENETRATION (microns)

Fig. 2.2.15. Various Chromium-Concentration

Profiles at 813°C in INOR-8 Loop Specimens.

of the activity of the corroded specimen to that of the salt. Disagree-

ments at a depth of about 25 p or greater may arise from a lack of pre-

cision resulting from the fact that, at these penetrations, only about
25% of the tracer remains. However, in many other specimens the experi-
mentally determined tracer concentrations at the greater penetrations
were found to be greater than calculated. The deviations are within the
experimental uncertainty, but the trend is always in the same direction,
which suggests that the effect is real. 1In any case, a relatively minor
contribution to the over-all diffusion rates is involved. Curves 4 and
5 of Fig. 2.2.15 represent the total chromium-concentration profile in
INOR-8 after 264 hr of selectively leaching the chromium, showing that
considerable reliance can be placed on both the diffusion coefficients
and the mathematical analyses employed for treating metal diffusion in

corrosion tests with thermal-convection loops.

The Free Energy of Formation of NiF,

Determinations of the free energy of formation of NilF,, based on

measurements of the equilibrium constant for the reaction
NiF, + Hp = il + 2HF

in molten LiF-BeF, as a solvent, and including experiments on saturated

solutions in equilibrium with solid NiF,, have given less negative values

than those found in the litefature. This resolves much of a long-stand-
ing discrepancy previously noted in connection with the empirical corro-
sion results in nickel alloys, and also with a marked difference in the
solution behavior in fluoride melts of Nitt as compared with Fe'™™ and

Cr*tt ions. The details, as discussed in a recent report,lo lead to free
energies of formation of -123 and -126 kcal/mole, compared with a liter-
ature estimatel! of -128 and -131, at 600 and 500°C, and further suggest

that the entropy of fusion of NiF,; is nearer to 10 e.u. than to 5 e.u.

Analysis of Corrosion Test Loops

Periodic sampling of the melts in the forced-circulation corrosion
test loops has continued.'? There are now four loops in operation, and
two have been terminated recently.

The oldest loop, MSRP-12, an INOR-8 loop containing LiF-BeF,-ThF,-
UF, (62-36.5-1.0-0.5 mole %), continues to operate without sampling in

13 It has now operated

order to conserve the supply of salt in the loop.
for more than 13,500 hr.

Loop 9377-5, an Inconel loop charged with the same salt as MSRP-12,
has shown a continuing increase in chromium concentration for almost
12,000 hr and now contains 0.6 wt % Cr*t, which is far in excess of the
amount attributable to equilibrium with normal fuel. A further indica-
tion of extraneous oxidation is found in the decrease in uranium concen-
tration, as though UQO, were being precipitated.

Loop MSRP-13, an INOR-8 loop charged with LiF-BeF,-UF, (70-10-20
mole %), operated for 8200 hr before failing because of an inadvertent
freezing. As shown in Fig. 2.2.16, the chromium concentration climbed
at an increasing rate; although this type of curvature is atypical, the

terminal Crtt ion concentration is not exceptionally high for a fuel

containing 20 mole % UF,.

10c, M. Blood et al., Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960,
ORNL-2931, p 39.

111, Brewer et al., The Chemistry and Metallurgy of Miscellaneous
Materials: Thermodynamics (ed. by L. L. Quill), pp 65, 110 McGraw-
Hill, New York, 1950.

12MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 59.
13MSR Quar. Prog. Rep. Oct. 31, 1959, ORNL-2890, p 61.

UNCLASSIFIED
. ORNL-LR-DWG 49804
1800 . - R ‘
; ‘

i ; .
LOOP: MSRP-13 i
1500 - MATERIAL: INOR-8 !
SALT: LiF - BeF,~ UF, (70 - 10-20 mole %)

CORRCSION MEASURED BY CHROMIUM IN FUEL {ppm}

TIME thr)

Fig. 2.2,16. Corrosion of INOR-8 by a Fuel
Containing 20 mole % UF -

Loop MSRP-14, constructed of INOR-8 and charged with LiF-BeF,-ThF,-
UF, (67-18.5-14-0.5 mole %), has operated for 2300 hr. So far, the Crtt
ion concentration has not exceeded 300 ppm and has returned to near this
value after a dilution resulting from the addition of fresh melt to re-
store the liquid level; a steady-state concentration of Cr** at several
hundred ppm is characteristic of normal behavior.

Loop MSRP-15, a duplicate of MSRP-14, has operated for 4000 hr with-
out any difficulties and currently exhibits a Cr*t concentration of 500
ppm; the approach to the presumed steady-state level was made in less
than 500 hr, which is also a characteristic of normal behavior.

Loop MSRP-16, another duplicate of MSRP-14, exhibited low Cr**t ion

concentrations for 2750 hr before pump failure.

Corrosion Equilibria with Unalloyed Cr®

In order to better understand the corrosion behavior of Cro, LiF-
BeF,-ThF, (67-18-15 mole %) salt has been circulated over 1800 g of pure
chromium metal in an isothermal assembly constructed of Inconel. The

plan of the experiment was to determine the concentration of chromium in
the fluoride melt without UF, as a function of temperature, and then to
add UF, as an oxidizing agent. Two separate attempts, each involving
analyses of samples drawn over a period of months, gave unsatisfactory

results; the Cr** concentration did not follow a consistent pattern. The

first assembly was behaving so erratically after 2230 hr of operation

that it was drained for examination. A deposition of chromium metal had
taken place at the liquid level in the pump bowl, which was a cold spot
in spite of precautions to maintain isothermal conditions. The second

experiment is still under surveilllance.

Melting Points of Chromous Fluoride and Chromic Fluoride

Melting points were determined for both CrF, and CrFs;. Solutions of
these compounds result from the corrosion of chromium-containing alloys
by fluoride fuels. Both compounds were previously reported to melt near
1000°¢c. 14

Conventional thermal analysis procedures showed pure CrF, to have a
melting point of 900 % 10°C.

Chromic fluoride required special precautions because it dispropor-

tionates into CrF, and volatile CrFs at high temperatures.'?

To prevent
disproportionation, the samples were sealed in platinum capsules contain-
ing a helium atmosphere. Capsules were heated to various temperatures,
and the cooled product was checked visually for evidence of melting.
Melting was deemed to have occurred in a capsule heated to 1419°C, but
not in one heated to 1388°C; therefore the melting point is 1404 = 25°C.
Possibly the previously reported low value was depressed due to admixture

with CrF, resulting from disproportionation.

Graphite Compatibility

Intercalation of Graphite

Studies were continued relative to the conditions under which inter-
calation of graphite does or does not occur. Although certain molten
salts can damage graphite by the formation of intercalation compounds, no

evidence of such an effect has been found with breeder fuels.!® 17 Also,

141.. Brewer, "The Fusion and Vaporization Data of the Halides,"
p 201 in The Chemistry and Metallurgy of Miscellaneous Materials:
Thermodynamics (ed. by L. L. Quill), McGraw-Hill, New York, 1950.

l5Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960, CRNL-2931, p 186.
16MSR Quar. Prog. Rep. Oct. 31, 1956, ORNL-2890, p 66—67.
17Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960, ORNL-2931, p 70.

82

most of the fission products have been either melted as fluorides in
graphite crucibles or included as major constituents of fluoride melts

in graphite; again no evidence of intercalation has been noticed. It ap-
pears likely that reactor fuels melt at too high a temperature to favor
the formation of intercalation compounds. For example, it was found that
although pure FeCls at 300°C crumbles graphite tubes in about 2 hr to
form an intercalated powder, NaCl-FeCl; (70-30 mole %) shows no effect at
750°C. Dilution is also important, since intercalation was not found
with NaCl-FeCl; (46-54 mole %) at 200°C for 7-1/2 hr; the maximum concen-
tration of cations known to have [intercalating proclivities in the MSRE

> L,
rfario e T,
"

/ y
is expected to be much less than 0.0fibat. %.

Permeation of Graphites by Molten Fluoride Fuels

Continued tests on the permeation of graphite by LiF-BeF,-ThF,-UF,
(67-18.5-14-0.5 mole %) and LiF-BeF,-ThF, (67-18-15 mole %) are summa-
rized in Table 2.2.1. Most of the permeating salt in the CEY samples was
found, by chemical analyses, to be near the exposed ends, and there was a
selective depletion in the uranium concentration relative to other fuel

constituents with increasing distance from the ends.

Table 2.2.1. Permeation of Graphites Expressed as Weight Gains

Temperature: 1300°F
Pressure: 60-75 psig

Weight Gains (%)

Graphite
LiF-RBeF,-ThF,, LiF-BeF,~-ThF ,~UF,
AGOT 2023
A 3.4-6.0
GT-123 0.2 0.2
MH4TM-85 10.5 10.4
CEY* (RLM-24) 1.0 0.1-0.2
CEY ** 1.0 1.0

*This graphite is in the form of extruded tubing, 1-1/4-in. OD X
7/8-in. ID.
*¥¥This graphite is in the form of l/2—in.-OD extruded tubing.

83

Radiation Effects

Rehavior of a Rare-Earth Fission Product in a Graphite—Molten-Salt System

In a graphite-moderated fused-salt reactor, retention of fission
products in the graphite could appreciably reduce the breeding efficiency.
The effect of xenon has been evaluated in other reports.18 Rare earths
also comprise an important part of the high-cross-section fission prod-
ucts. FEuropium has been selected as a typical rare earth, with useful
tracer properties and a high cross section, for study in a molten-salt—
graphite system. To provide a background for future work with irradiated N
specimens, AGOT graphite was treated with LiF-BeF,-UF,-ThF, (67-18.5-0.5-

152,154 g5 described below.

14 mole %), containing a tracer quantity of Eu
A group of three AGOT graphite specimens in the form of rods, 0.75-
in. diameter and 3-in. length, were exposed to the tagged fuel for 1 hr
at 700°C under a helium pressure of 30 psia and under conditions which
should have given penetration of at least 15 wt %. Examination of these
rods revealed that the degree of impregnation, as evidenced by weight
gains, was only about 1%. Although the impregnation was not satisfactory,
and there was considerable difference between rods, contact radioauto-
graphs were made of the ends of O0.5-in. lengths sectioned from the three
rods. Figure 2.2.17 illustrates the variation found along a rod. Visual
observation of a sectioned face by sterecmicroscope revealed salt deposits
in large random pores or interconnecting faults extending inward from the
outer surface of the graphite rod.
Other AGOT graphite specimens have been treated with fuel salt con- -
taining europium by a different method. A hole 0.3 in. in diameter was
drilled 2.5 in. along the axis of a graphite rod 0.9 in. in dlameter and *
3 in. long, and this hole was filled with fuel containing Eu'’2'%5%. The
rod was then lowered into a molten-salt bath to a depth of approximately
2.75 in. This operation was carried out in a controlled-atmosphere vac-
uum-tight furnace. A thick Lucite window in the top of the furnace con-
tainer permitted observation during treatment of the graphite.
One AGOT rod was maintained in the molten-salt bath for 24 hr at a

temperature of 700°C under a helium pressure of 7.4 psia. Upon recovery

18Reactor Chem. Ann. Prog. Rep. Jan. 31, 1960, ORNL-2931, p 37.

84

UNCLASSIFIED
ORNL-LR-DWG 48552

152,154

Fig. 2.2.17. Radioautographs Showing Variation of Penetration of BULT-14 Salt Containing Eu
into AGOT Graphite Rod.

of the specimen and removal of the inner plug of salt, no penetration of
the graphite by the salt was noted, as might have been anticipated from
the fact that the rod was incompletely immersed, and hence no pressure
difference was available for pushing the salt into the graphite. The rod
was then cut into 0.5-in. sections and a 24-hr contact radioautograph
made using Kodak type M x-ray film. No blackening of the film was ob-

served.

85

A second graphite rod was treated for 24 hr, as described above,
except that the helium pressure was cycled twice between 7.4 and 35 psia
during the last 2 hr in an attempt to develop a pressure difference. The
attempt was again unsuccessful, and no blackening of the film was produced
by a 24-hr contact period.

The sections of both rods were then placed in contact with film for
six days, and a very slight darkening of the film occurred in the position
of the axial hole in each section. Gamma-ray analysis of the very small
amount of radioactivity in the graphite sections indicated that the film
darkening was due to the uranium and thorium content of the salt rather
than Eut32:+34 activity. These limited data indicate that europium re-
mains in the salt as expected and does not deposit on graphite surfaces.

A third AGOT specimen is being treated under a vacuum of 100 to 200
u at 550°C and then under a helium pressure of 41 psia at 700°C to induce
penetration of the graphite walls by tracer-bearing salt. Once permeation
of europium activity by this treatment has been accomplished, the arrange-
ment will be reversed so that the graphite specimen containing europium
will be exposed to clean salt, permitting the leaching or interchange of

the europium activity between the tagged and nontagged salt to be followed.

Preparation of Purified Materials

Purification, Handling, and Service Operations

The use of purified fuels has been somewhat greater than had been
anticipated, so that a stockpile of fuels of current interest no longer
exists, although four batches of beryllium-based fuels were processed re-
cently. Several capsules for in-pile testing were filled with a purified
fuel containing U%?° and Li”.

The large-scale production facility has recently been operated almos?t
continuously to treat 950 kg of fluoride melts for use in recovery proc-

esses under development by the Chemical Technology Division.

86

2.3 TFUEL REPROCESSING

NO,-HF Solutions

Further investigation of the dissolution of LiF-BeF,-ThF,-UF, (62-
36.5-1.0-0.5 mole %) in anhydrous hydrogen fluoride containing nitrogen
dioxide showed that under proper conditions the UF, 1s dissolved along
with the LiF and BeF,. Salt of this composition was contacted several
times successively with liquid HF containing 5 mole % NO,, each pass with
7 ml of solvent per gram of original salt. ZEssentially all the LiF and
BeF, and 73% of the UF, dissolved in the first two treatments, and the
rest of the uranium dissolved in the third. Thorium fluoride was not sol-
uble, and rare earths were sufficiently Insoluble that the dissolved salt
was decontaminated by a factor greater than 50.

A large power reactor would probably use a fuel salt contalning less
UF, than in this experiment, probably around 0.3 mole %, and perhaps no
ThF,. Such a salt could be dissolved with HF-NO, solutions in two con-
tacts. Since UF, is dissolved along with the LiF and BeF,, the section
of the fluoride volatility plant used for recovering uranium from the fuel
salt could be eliminated. The LiF and BeF, solubilities may be such that
the total fuel-salt solubility is lower than for agueous HF solutions,

thus requiring a somewhat larger plant for the same processing capacity.

Aqueous NH,F

In tests with blanket salt, LiF-BeF,-~ThF,;-UF, (67-18.5-14-0.5 mole %),
some component, probably LiF, interfered with ThF, solubility. A sample
of salt was added to 50% NH,F solution at 100°C. After 1 hr all the BeF,
and UF, dissolved, the respective concentrations being 13 and 2.15 mg/ml;
but only 12 to 15% of the LiF and ThF, dissolved, their respective solu-
bilities being 2.2 and 9.3 mg/ml. After one day, during which time the
PH decreased considerably due to loss of NH3 from the ammonium fluoride
solution, all the solubilities were lower except that of BeF,. Neutrali-
zation with NH,OH did not restore the original solubilities. In every

case all the BeF, present was in solution, in agreement with its very high

solubility in water. The solid phases containing ThF, obtained upon

freezing this salt were 3LiF:ThF, solid solution and a little LiF-2ThF,;
such compounds generally do not have properties similar to those of the
components, LiF and ThF,. In this case the solubility of ThF, in equilib-
rium with the compounds is much lower than that in equilibrium with pure
ThT,, .

No satisfactory process has been found for decontaminating salts con-
taining large amounts of Th¥,. The fluoride volatility process is entirely
satisfactory for processing these salts if they are used in the blanket of
a two-reglon breeder reactor, since decontamination is not required for
many years, if ever. However, decontamination would be required if these

salts are used for fuel, as they might be in a one-region converter reactor.

Protactinium Removal

A method for economically removing protactinium from the blanket of a
molten-salt breeder reactor at a rate competitive with its decay would be
advantageous. A single experiment was made to investigate the possibility
that a volatile fluoride might be formed. Natural protactinium (Pa231)
was added to LiF-BeF,-ThF,-UF, (67-18.5-14-0.5 mole %) to a concentration
of 0.012 wt %, and the salt was fluorinated for 2 hr at 650°C. No meas-
urable amount of protactinium was removed from the salt, but the uranium

was removed.

88

ORNL-2973
UC-81 — Reactors — Power
TID=-4500 (l5th ed.)

INTERNAL DISTRIBUTION

1. R. G. Affel 48. M. T. Kelley

2. L. G. Alexander 49. F. Kertesz

3. E. S. Bettis 50. B. W. Kinyon

4., D. 8. Billington 51. M. E. Lackey

5. F. F. Blankenship 52. J. A. Lane

6. E. P. Blizard 53. D. A. Lee

7. A. L. Boch 54. R. S. Livingston

8. C. J. Borkowski 55. H. G. MacPherson

9. W. F. Boudreau 56. W. D. Manly
10. G. E. Boyd 57. E. R. Mann
11. M. A. Bredig 58. L. A. Mann

12. E. J. Breeding 59. W. B. McDonald
13. R. B. Briggs 60. H. F. McDuffie
14. W. E. Browning 61l. J. R. McNally
15. D. 0. Campbell 62. H. J. Metz

16, W. H. Carr 63. R. P. Milford
17. G. I. Cathers 64. E. C. Miller
18. C. E. Center (X-25) 65. J. W. Miller
19. R. A. Charpie 66. K. Z. Morgan
20. R. S. Cockreham 67. J. P. Murray (Y-12)
2l. J. H. Coobs 68. M. L. Nelson
22, F. L. Culler 69. G. J. Nessle
23. J. H. DeVan 70. W. R. Osborn
24. D. A. Douglas 71. L. F. Parsly
25. L. B. Emlet (K-25) 72. P. Patriarca
26. W. K. Ergen 73. A. M. Perry
27. J. Y. BEstabrook 74. D. Phillips
28. D. E. Ferguson 75. W. D. Reel

29. A. P. Fraas 76. P. M. Reyling
30. E. A. Franco-Ferreira 77. J. T. Roberts
31. J. H. Frye, dJr. 78. M. T. Robinson
32. W. R. Gall 79. H. W. Savage
33. A. T. Gresky g0. J. L. Scott
34. J. L. Gregg €l. H. E. Seagren

35.37. W. R. Grimes 82. E. D. Shipley

38. E. Guth 3. M. J. Skinner
39. C. S. Harrill 84. G. M. Slaughter
40. M. R. Hill 85. A. H. Snell
41. E. E. Hoffman 86. E. Storto
42. H. W. Hoffman 87. C. D. Susano
43. A. Hollaender 88. J. A. Swartout
44. A. 8. Householder 89. A. Taboada
45. W. H. Jordan 90. E. H. Taylor
46. G. W. Keilholtz 91. R. E. Thoma
47. C. P. Keim 92. D. B. Trauger

93. F.
94. G.
95. A.
6. M.
97. J.
98. G.

G

149.
150.
151-727.

%0

GQuald=a= 20

Vonderlage
Watson
Weinberg
Whatley
White
Whitman

. Williams

D. H. Groelsema, AEC, Washington

100.
101.
102-105.

106-145.
146,
147-148.

C. E. Winters

J. Zasler

ORNL — Y-12 Technical Library,
Document Reference Section
Laboratory Records Department
Laboratory Records, ORNL R.C.
Central Research Library

EXTERNAT, DISTRIBUTION

Division of Research and Development, AEC, CRO
Given distribution as shown in TID-4500 (15th ed.) under
Reactors-Power category (75 copies — OTS)