ORNL-4616
UC-80 — Reactor Technology

Contract No. W-7405-Eng-26

REACTOR CHEMISTRY DIVISION

PREPARATION AND HANDLING OF SALT MIXTURES
FOR THE MOLTEN SALT REACTOR EXPERIMENT

James H. Shaffer

- LEGAL NOTICE

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

 

 

 

JANUARY 1971

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

CUMERNT LS UM BT

 
 

HETIRIBUTION UF THIS DO
 

 

 

 

 
CONTENTS

 

A IaCt . . o e 1

L It oduC i Orn L e e e e 1

2. Fuel, Coolant, and Flush Salt Requirements for the MSRE . ... ... ... ... . ... ... ... ......... 2

Quantities of Materials . ... .. ... . . e 3

Procurement of Starting Materials . . . ... ... . . 3

Lithium Fluoride . ... ... 3

Uranium Tetrafluoride . ... ... . . 3

Zirconium Tetrafluoride . . ... ..o 4

Beryllium Fluoride . . ... 5

General Chemical Specificalions ... ... ottt e 5

Production Methodology . .. ... .. o 5

3. Chemical Development of the Production Process . .. ... .. i i i 7

Oxide Removal . ... . e 7

Sulfur Removal . ... .o e 9

Removal of Structural-Metal Impurities .. ... .. ... . . . . ... . . . 11

Reduction of Structural Metals by Hydrogen . .. ... .. ... . . . 11

Reduction of Structural Metals by Beryllium and Zirconium . ......... ... ... .. ... ... .... 13

G 4. The Production Plant . .. ... 14
- Raw Materials Charge . ... . ... i e e 15
Meltdown and Pretreatment ... ... ... .. e 17

The Batch Process . ... 18

Auxiliary Systems . . .. e e 21

VUM SV I . . .. e e e 21

HE SUPPIY 21

Helium Supply . .o e e 21

Hydrogen Supply ..o oo e e e 22

S 0 ) 1~ 22

5. Production of Coolant, Flush, and Fuel Solvent . ...... ... ... ... .. ... ... . ... ... .. ... ..... 23

Process Operating Conditions .. ... ... ittt e e et e e e 23

HE-H, Treatment . .. ... e 23

Reduction of Structural Metals ... ... . . e 26

Process Control . ... 27

Lithium Fluoride Densification . .. ... ... ... .. ... e e e 28

7 6. Preparation of Enriched Fuel Concentrate . ... ... ... ... ... ittt 30
iy 7. Preparation of Fuel EnrichingCapsules . ....... .. .. .. .. ... ... . il 32
8. Reactor Loading Operations . .. ... .. .. e 36

9. Production Ecomomics . ... ... i e e e e 38

ACKROWIEdEIMEntS . . L .o . e e 3%
 

 

 

 

 

 
PREPARATION AND HANDLING OF SALT MIXTURES
FOR THE MOLTEN-SALT REACTOR EXPERIMENT

James H. Shaffer

 

ABSTRACT

A molten mixture of LiF, BeF,, ZrF,, and UF, served as the circulating fuel for the Molten-Salt
Reactor Experiment. Its secondary coolant for transferring heat to an air-cooled radiator was a motten
mixture of LiF and BeF,. A third mixture that was chemically identical te the coolant mixture was
used in place of the fuel for prenuclear operations and subsequently to flush the reactor core after a
fuel drain. Approximately 26,000 ib of these fused fluoride mixtures were prepared from component
fluoride salts and loaded into the reactor facility by CRNL’s Reactor Chemistry Division. Techniques
for handling molten fluorides and their production process for attaining high chemical purity were
developed and applied simuitaneously with the development of the molten-salt nuclear reactor
concept. The plans and operations which were a part of the fueling of the MSRE are described.

 

1. INTRODUCTION

The Molten-Salt Reactor Experiment (MSRE) was operated by the Oak Ridge National Laboratory
during the period June 1, 1965, to December 12, 1969, for experimental purposes and as a demonstration
of the molten-salt nuclear reactor concept. The MSRE was then placed on a standby operational status
pending further developments of the Molten-Salt Reactor Program (MSRP) in its pursuit of a thermal
breeder machine. Development efforts by ORNL which led to the design and construction of the MSRE
also included development of processes for the preparation of fused salt mixtures suitable for reactor use
and techniques for handling these materials in their liquid state at high temperatures. Thus the successful
demonstration of the molten-salt nuclear reactor concept also illustrated the relatively simple and
economical manner by which these reactors can be fueled.

Techniques for preparing and handling molten salts have been developed at ORNL over the past 18
years. During support of the Aircraft Nuclear Propulsion (ANP} program the application of these
procedures to nuclear technology was successfully demonstrated during the preparation of fluoride
mixtures and their loading into the Aircraft Reactor Experiment (ARE) beginning October 23, 1954.°
During the interim period and following a similar fueling operation of the Aircraft Reactor Test beginning
November 20¢, 1956, the molten-salt production facility was operated by the Reactor Chemistry Division as
an integral part of the Molten-Salt Reactor Program to provide fused fluoride mixtures for its chemical and
engineering tests and for other related projects of ORNL and the USAEC. Prior to the preparation of salt
mixtures for the MSRE, this facility had produced over 132,000 1b of fluoride mixtures of high chemical
purity. In addition to the operation of the production facility, handling techniques were further developed
by operations such as filling, sampling, and emptying engineering test loops. Similar operations with liquid
metals were also performed routinely.

The fluoride production facility was constructed as a batch process. Each of two processing units had a
capacity of about 2 ft* of fused salt per batch. During development of the MSRE concept, this production
plant was adequately sized for supplying materials for the engineering tests of the program and for the
repetitive preparation of relatively small quantities of fluoride mixtures having very diverse chemical
compositions. The requirements for fluoride mixtures of some 26,500 b for the operation of the MSRE

S represented the largest production effort undertaken by the program. Although this quantity exceeded the

 

- 'G. 1. Nessle and W. R. Grimes, Chem. Eng. Progr., Symp. Ser. 56(28), 51 (1960),
reasonable capacity of the preduction facility, its use with existing technology was the most feasible
approach both technically and economically available. Commercial sources of fused fluoride mixtures
which would meet specifications for the MSRE are as yet nonexistent.

In addition to the production of the various fluoride mixtures for the MSRE, this commitment also
included their loading into the fuel and secondary coolant systems of the reactor and the preparation of
incremental charges of 225U needed for sustained nuclear operations. This report is a description of plans
and operations followed in the fueling of the MSRE.

2. FUEL, COOLANT, AND FLUSH SALT REQUIREMENTS FOR THE MSRE

Specific fluoride mixtures for the MSRE were carefully selected on the basis of their nuclear, chemical,
and physical properties and of their potential application in a molten-salt breeder reactor.” As a result of
these considerations, mixtures based on the LiF-BeF, diluent system were used. The phase diagram shown
in Fig. 1 is a current interpretation of this system.® The reactor fuel mixture was to contain nominally (in
mole %) 65 LiF, 29.1 BeF,, 5 ZrF4, and 0.9 UF,; (liquidus temperature of 450°C). The actual fuel
composition was dependent upon the amount of uranium required to bring the system to the critical, and
then to the operating, condition. Fissionable 25U comprised about one-third of the uranium inventory;,
the balance, as nonfissionable 2°*U, was included for chemical purposes. Zirconium was a constituent of

 

2W. R, Grimes, MSR Program Semiann, Progr. Rept. July 31, 1964, ORNL-3708, p. 214,
3R, E. Thoma (ed.), Phase Diagrams of Nuclear Reactor Materigls, ORNL-2548, p. 33 (Nov. 2, 1959).

ORNL—LR—DWG 164 26R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

900 |
|
800 j\\ -
700 —- - RS
S ;
L 800 LiF + LIQUID ; _—
i : i
g | \
g : \
& j
W
$ 500 < X -
e : /
_: g \\ Befp + LIQUID
: |
400 ' .. e E 2L|F; BeFy \\// - ‘
LIQUID ) :.
LiF + 2LiF - BeF,
? & 2LiF - BeFp + BeFy (KIGH QUARTZ TYPE}
; a e e i L —— . —d
300 i @ ~1 I
| ~ 2LiF-BeFp & ! !
| ™ ' . ez @ LiF - BeFs + BeFp (HIGH QUARTZ TYPE)
, : | 1 |
200 LiF-BeF; 3 Lif - BeFa+ BeFz (LOW QUARTZ TYPE]

LiF {0 20 30 40 50 60 70 80 90 BeFp
BeF, (mole %)

Fig. 1. The System LiF-BeF,.

S
the fuel mixture to prevent the precipitation of UQ, and resultant criticality hazards in the event that

 

oxide contamination of the fuel occurred. At a concentration of 5 mole % ZrF, in the fuel, significant and
recognizable quantities of Zr(0, would be preferentially precipitated prior to loss of any uranium from the
fuel solution as U0, .

The secondary coolant was a simple binary mixture containing 66 mole % LiF in BeF,, selected to
avoid energetic reactions with the fuel (as with an alkali metal coolant) or contamination of the fuel in the
event of a failure in the heat exchanger. This mixture had a liquidus temperature (455°C) slightly higher
than that of the fuel mixture.

Prenuclear operation of the MSRE used a uranium-free LiF-BeF,; mixture of the same chemical
composition as the secondary coolant. This material, commonly referred to as the flush saft, also provided
for the removal of oxide and oxygen-bearing species from the system prior to fuel loading and subsequently
for flushing the circuit after a fuel drain.

Quantities of Materials

The volume of the fuel circuit of the MSRE was estimated, for fueling purposes, at about 73 ft°,
Production of the fuel was based on a 10% excess of this volume and a calculated salt density® of 142
Ib/ft®. Materials requirements for the MSRE fuel are listed in Table 1. The secondary coolant circuit had an
estimated fifl volume of about 42 ft*. Since the chemical composition of the coolant was identical with the
flush salt, production estimates were based on a 10% excess of their combined volumes and a calculated
density of about 120 Ib/ft®. Materials requirements for the secondary coolant and flush salt mixtures are
listed in Table 2.

Procurement of Starting Materials

Fluoride starting materials were purchased from commercial sources or otherwise obtained from the
USAEC on the basis of estimates in the preceding section. Table 3 lists a summary of these requirements
and actual quantities of materials ordered.

Lithium Fluoride. - For neutron-absorption cross-section consideration, all lithium fluoride used in the
MSRE production operation was almost isotopically pure ’Li. Its analyzed isotopic assay was at least
$9.99% 7Li. Since this material was available as the hydroxide, arrangements were made with the Y-12
Piant for its conversion to fluoride and for maintaining the isotopic purity of each production batch.
Because of this unique demand, sufficient ” LiF was prepared for the initial loading of the MSRE and for
the replacement of the fuel or coolant charge.

Uranium Tetrafluoride. — Although the 25U enrichment in the MSRE fuel mixture during nuclear
operation was about 32%, all of the U obtained for processing was highly enriched (93% in **°U).°
About 90 kg (198 1b) of 2*5U was obtained for the initial fueling of the MSRE and for its sustained
operation during scheduled tests of the MSRP. The balance of the uranium inventory in the fuel charge had
been depleted of 35U, These materials were available directly as their tetrafluoride salts from USAEC
sources.” |

 

4Reactor Chem, Div. Ann, Progr. Rept. Jan. 31, 1963, ORNL-3417, p. 38; C. F. Baes, Jr., J. H. Shaffer, and H. I'.
McDuffie, Trans. Am. Nucl Scec. 6(2), 393 (1963); Reactor Chem, Div. Ann. Progr. Rept. Jan. 31, 1964, ORNL-3591, p.
o 45,
%S, Cantor, Reactor Chem, Div. Ann. Progr. Rept. Jan. 31, 1962, ORNL-3262, p. 38.
®Tentative plans and approvals were based on 92% enrichment.
7 Authorization No. 2181, USAEC, Oak Ridge Operations, Sept. 15, 1964,

 
Table 1. Materials Reguirements for
MSRE Fuel Mixture

Estimated volume: 80.3ofit3 (10% excess)
Density of salt at 600°C: 142 b/ft3

 

 

Salt Compesition Weight for
Component Mole % Weight % 80.3 £t3 (1)
LiF 65 40.48 4,616
BeF, 29.1 32.76 3,735
Z:F, 5.0 20.02 2,283
238707, 8.61 4,59 523
235yp, 0.29 2.16 246

Total weight: 11,403

 

Avg molecular weight: 41.74

Table 2. Materials Requirements for
MSRE Flush and Secondary Coolant Mixtures

Estimated volume: 126.50ft3 (10% excess)
Density of salt at 600°C: 120 Ib/ft3

 

 

Salt Composition Weight for
Component Mole % Weight % 126.5 ft3 (Ib)
LiF 66 51.78 7,860
BeF, 34 48.22 7,320

Total weight: 15,180

 

Average molecular weight: 33.14

Table 3. Summary of Fluorides Procured

 

 

for MSRE
Fluoride Estimate Procured Source
Sait (Ib) (1b)
TLiF 12,476 22,000 USAEC
BelF, 11,0558 12,008 Commercial
Z1F, 2,283 2,300 Commercial
238y, 523 529 USAEC
235UF, 246 262 USAEC

 

Zirconium Tetrafiuoride. — Normal commercial grades of zirconium compounds may contain from 1 to
3% hafnium as an impurity and would invoke a severe penalty in neutron economy if used in the MSRE.
However, separations processes, based on an early development of the nuclear industry,® are well known.
Accordingly, zirconium tetrafluoride that was essentially “hafnium free” (<{50 ppm Hf) was available from
commercial sources on a competitive bid arrangement.

 

8¢, 1. Barton, Sr., L. G. Gverholser, and J. W. Ramsey, Separation of Hafnium from Zirconium, U.S. Pat. 2,938,769,
May 31, 1960; C. 1. Barton, Sr., et al., Separation of Hafnium and Zirconium by Extraction of Thiocyanate Complexes,
USAEC Report Y431 (June 1949); C. 1. Barton, Sr., L. G. Overholser, and W, R. Grimes, Separation of Hafrium and
Zirconium by Extraction of Thiocyanate Complexes, Chemical Studies Pare II, USAEC Report Y477 {September 1949);
C. J. Barton, Sr., L. G. Overholser, and W, R. Grimes, Preferential Extraction of Zirconium and Hafnium Thiocyanates -
Preparation of Pure Hafnium, USAEC Report Y-611 (June 1950); W. R. Grimes et al., Preparation of Pure Zirconium
Oxide ~ Laboratory Studies, USAEC Report Y-560 (February 1950),

 
 

 

Table 4. General Chemical Specifications
for MSRE Fluoride Mixtures

 

Allowable Concentration
Impurity (wt %)
(1 ppm = 0.0001 wt %)

 

Water 0.1

Cu 0.005
Fe 0.01
Ni 0.0025
S 0.025
Cr 0.0025
Al 0.015
Si 0,01

B 0.0005
Na 0.03
Ca 0.01
Mg 0.01

K 0.01
Li {natural) 0.005
Zr (natural) 0.025
Cd 0.001
Rare earths (total) 2.001

 

Beryllium Fluoride. - Since beryllium fluoride was normally available from commercial sources as a
manufacturer’s intermediate product, its chemical purity was not normally regulated to meet chemical
specifications of other users. However, two major producers of beryllium compounds undertock quality
improvement programs to meet the requirements of the MSRE. As a result of this cooperative effort,
beryllium fluoride was purchased by competitive bid at costs previously incurred for less-pure material.

General Chemical Specifications

Specifications regulating the maximum allowable impurities in fluorides obtained for the MSRE are
listed in Table 4. Those elements which would constitute nuclear poisons were given prime consideration.
However, aside from hafnium in zirconium and ®Li in LiF, none were major impurities in fluoride salts used
in the MSRE. Accordingly, restrictive specifications of nuclear poisons were established to prevent their
possible deliberate addition. The other chemical specifications were determined on a “best commercially
available BeF,” basis. Allowable impurity levels were based on chemical and specirochemical analyses of
numerous product samples from commercial vendors and from those materials obtained from the USAEC.
While all materials obtained for use in preparing MSRE fluoride mixtures were generally within these
specified limits, iron concentrations of 250 and 500 ppm were allowed in BeF, and LiF respectively. Some
carbonaceous impurities were also allowed since they could be readily removed as carbon by gas sparging
and were inherent to some manufacturing processes.

Production Methodology

The fluoride production method is generally independent of fluoride mixture composition provided
that the liquidus temperature is within the capability of the process equipment. The production of
multicomponent mixtures, however, is sometimes facilitated by the preparation and subsequent
combination of simpler binary or ternary mixtures. Thus the mode of production activities could be
directly oriented toward procedures by which the actual fuel loading would be accomplished.
 

 

 

Fig. 2. Facility for Reclaiming Sal¢t-Contaminated Equipment by Wet Sandblasting.

The operational schedule of the MSRE together with the limited storage capacity for prepared fluorides
necessitated the sequential preparation of the secondary coolant and flush salts followed by the fuei charge.
Since the two mixtures required in the prenuclear operation of the MSRE were of identical chemical
composition, their production was considered as a single operation. However, batches of "LiF used in
preparing the LiF-BeF, (66-34 mole %) mixture were selected on the basis of their isotopic purity so that
materials having the least concentrations of ® Li could be reserved for the fuel and flush salt mixtures.

To provide for conservation of fissionable material, for nuclear safety, and for the planned reactor
loading operation, the MSRE fuel salt was prepared as a fuel concentrate mixture and as a barren fuel
solvent mixture. The fuel concenirate was the binary eutectic mixture ' LiF-UF, (73-27 mole %) and was
the form in which all uranium was introduced into the MSRE. The fuel concentrate mixture was further
differentiated as the enriched fue! concentrate mixture, which contained ail ***U as highly enriched
235UF,, and as the depleted fuel concentrate mixture, which contained the balance of nonfissionable
uranium required for the fuel salt mixture, All BeF,, ZtF,, and remaining ” LiF needed for the fuel mixture
was combined as the barren fuel solvent. As caiculated from these requirements, the fuel solvent was
prepared as a ternary mixture containing {in mole %) 64.7 LiF, 30.1 BeF,, and 5.2 ZrF,.

As an economic measure the storage of prepared bulk mixtures and their transport to the reactor site
were accomplished by use of existing batch-sized containers. Costs for fabricating large heated vessels which
would accommodate a complete reactor charge were considered prohibitive for the single use foreseen for
the program. About 50 batch-sized containers which had becn previously used for nonberyllium salts were

 

 
 

cut open, cleaned by sandblasting, and lengthened by 12 in. upon reassembly. To make use of 20 additional
containers which had been used for beryllium salts, two wet-sandblasting cabinets were purchased and
installed according to beryllium handling requirements for less than $10,000. As shown in Fig. 2, these
units were installed in a tandem arrangement. One unit was used to remove salt deposits and scale by wet
sandblasting; the second unit was used for rinsing contaminants from the cleaned equipment. This facility
has since provided valuable service in reclaiming beryllium-contaminated equipment from continued
experimental programs on molten salts within the Reactor Chemistry Division.

3. CHEMICAL DEVELOPMENT OF THE PRODUCTION PROCESS

Aside from the physical mixing of salts to obtain lower liquidus temperatures, the primary purpose of
the production process is to achieve further purification of the resultant molten fluoride mixture. Although
starting materials of reasonably high purity are normally available from commercial sources, impurities
which contribute to chemical corrosion processes and to the deposition of solids can be very detrimental in
high-temperature molten-salt systems even at low concentrations. The removal of a limited number of these
impurity species during the production operation is achieved by treatment of the fluoride melt with
anhydrous hydrogen fluoride, hydrogen, and, in some instances, strong metallic reducing agents. Impurities
which can be volatilized are removed in the process gas effluent stream; those which can be rendered as
insoluble particles are removed by decantation and filtration.

Oxide Removal

Oxides in molten fluorides may arise from various oxygenated impurities in the starting mater:als.
However, the most abundant source results from the incomplete evaporation of absorbed water and
subsequent pyrohydrolysis during the initial melting of the fluoride mixture. Although oxide impurities in
themselves are probably not detrimental, their presence in the molten fluoride can result in the deposition
of solid particles or scale. In applications such as those of the MSRE, these heterogeneous systems may alter
heat transfer properties of the reactor components and, as an extreme case, might also create localized heat
sources in the core of a nuclear reactor by the deposition of uranium dioxide. Thus the chemistry of oxide
behavior in molten fluorides and of its effective removal by suitable processing methods has been of
continued interest in the MSRP.

Oxides are removed during the initial gas sparging of the molten fluoride melt with anhydrous hydrogen
fluoride. They react directly with HF by the reaction

0% +2HF = 2F  + H,0 (1)

and are conveniently removed from the process as water vapor. Extensive measurements of equilibrium
quotients for this reaction have been made.” They confirm prior production practices and show
quantitatively that the reaction is more favorable at lower temperatures and that oxide removal by this
reaction is highly effective. In fact, this equilibrium was further developed as an analytical method for
determining oxide concentrations at very low levels in the MSRE fluoride mixtures.'® Analytical methods
in use during preparation of MSRE materials were quite sensitive but were not sufficiently consistent for

 

’A. L. Mathews, B, F. Hitch, and C. F. Baes., Jt., Reactor Chem. Div. Ann. Progr. Rept. Jan, 31, 1965, ORNL-3789, p.
56,

10, F. Apple and J. M, Dale, “Determination of Oxides in MSRE Salts,”” Anal. Chem. Div. Ann. Progr. Rept, Oct. 31,
1967, ORNL4196, p. 15.
ORNL—LR—-DWG 56426R

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2500 ‘ ; , l
é s INITIAL WEIGHT OF MELT: 2.5 kg
HF: Ho =~
S 2000 %—L ——————————— 2 —d ]
o :
£ ¢ ;
[=X
L
=
g 1500 {- S—
o
o
! v
o
» :
z {000 o e
o)
=
-
3
i
LLi 500 . el . e}
o
5
e
\’“o--e-__fi @
O o
0 5 10 15 20 25

HF PASSED (g equivalent wt)

Fig. 3. Removal of Oxide from LiF-BeF, (63-37 mole %) at 760°C by Treatment with HF-H, Mixtures.

use as a process control method. Accordingly, the production process treatment with HF was continued
beyond practical reaction completion to ensure a suitable “oxide capacity” of those fluoride mixtures for
inadvertent contamination during reactor operations.

Although HF has heen used for oxide removal since the inception of the production process, procedures
foliowed prior to the MSRE production effort utilized an alternate HF-H, treatment.’ ! Hydrogen fluoride
will readily attack structural metals and alloys that are suitable as salt containers at the operating
temperatures of the production process by reactions of the type

M? + 2HF = MF, + H, . (2)

This reaction is arrested in the gas phase of the treatment vessel by the formation of a rather impervious
layer of the structural-metal fluorides on the metal surfaces. However, those surfaces which are in contact
with the fluoride mixture are continually renewed by the dissolution of the structural-metal fluorides into
the melt. Thus, by the alternate gas treatment method, removal times continually increased by the alternate
oxidation and reduction of structural metals untit faiture of the treatment vessel occurred.

Studies of the thermodynamics of the corrosion mechanism, noted by Eq. (2}, showed that chemical
equilibrium in fluorides of interest in the MSBR could be achieved by sparging with mixtures of HF and
hydrogen at controlied partial pressures.’” On the basis of this investigation, studies of oxide remova!
according to Eq. (1) were made with HF admixed with hydrogen at concentrations which were essentially
noncorrosive toward the salt container. Typical laboratory results of this study are shown in Fig. 3.

The application of HF-H; mixtures in the fluoride purification process was further demonstrated on a
larger scale by the in situ oxide cleanup of the simulated MSRE fuel salt used in the Engineering Test Loop

 

1y g, Eorgan et al, Reactor Chem, Div. Ann, Progr. Rept, Jan, 31, 1960, ORNL-2931, p, 64.

2¢ M. Blood, Solubility and Stability of Structural Metal Diftuorides in Molten Fluoride Mixtures, ORNL-CF-61-5-4
(Sept. 21, 1964); C, M. Blood er a¢f, ““Activitics of Some Transition Metal Fluorides in Molten Fluoride Mixtures,” in
Proceedings of the Imternational Conf. on Coordination Chemistry, 7th, Stockholm and Uppsala, June 25--29, 1962,
Butterworths, London, 1963,

 
S

 

 

 

 

 

 

 

 

 

 

‘-N

s 1 1T [
e ! |
= HF-H, RATIO:~1/5 (vol) /Lo/.
2 5o | He FLOW RATE: 25 liters/min .

b TOTAL WEIGHT OF FLUCRIDE SALT: o= |

> ' 137 kg ._/ ;

g 100 i el AT _fi___— |
a- ;
W » !
a ' / » i
L - ! ;
> 50 | ,/ ; ! ;
o : ! 1
R Pl L__i I

x G tQ 20 30 40 50 60 70

HF TREATMENT TiME (br}

Fig. 4. Removal of Oxides from the Engineering Test Loop by HF-H, Treatment at 1050°F.

Facility.! ? Since the salt container of the loop was fabricated of Inconel, this demonstration illustrated the
use of HF-H, mixtures to reprocess fluorides contained in materials which are rapidly corroded by HF
alone. The rate at which oxides were removed from the melt (Fig. 4) was determined by measurements of
water evolution in the gas effluent. The results of chemical analyses of salt samples withdrawn periodically
during the HF-H, treatment showed that the dissolved oxide concentration diminished from values of
500 ppm (apparent saturation with ZrQO,) to less than 200 ppm. The concentrations of structural metals
dissolved in the fluoride melt were virtually unaltered by the HF-H, treatment. However, metallographic
examinations of the Inconel dip tubes used for sparging the fluoride melt with HF-H, mixture showed that
mild corrosion had occurred. These findings were more consistent with the measured corrosion equilibrium
values at the HF concentration in H, used for this operation. It is perhaps reasonable to assume that
chromium and iron had been leached from the metal surfaces of the salt container whereby their rate of

corrosion was restricted by their relatively low rate of diffusion in the metal.’*

Sulfur Removal

Sulfur impurities must essentially be eliminated (<10 ppm) from molten-salt mixtures because of their
corrosive attack on nickel-base alloys at elevated temperatures. These impurities are found in the starting
materials primarily as sulfates and have been the most difficult impurity to remove. As currently
understood, sulfates must first be reduced to sulfide ion; removal can then be effected by its volatilization
as H, S by reaction with HF.

Farlier production procedures utilized the alternate HF and H, treatment for sulfur removal. Although
this method was reasonably effective, the discontinuity of sulfide removal by reaction with HF presented
some difficulties in ascertaining quality control of the production batch. For example, incomplete
reduction of sulfate prior to the last HF treatment would result in its reduction to sulfide during the final
treatment of the melt with hydrogen. Therefore the number of alternate HF-H, sparge treatments was
normally increased for those mixtures known to contain significant concentrations of sulfur in the starting
materials.

The development of the simultaneous HF-H, sparge treatment for oxide removal was also applicable for
sulfur removal. By continuous reduction of sulfate by hydrogen and volatilization as H, S by HF, effective
sulfur removal should be achieved with minimum treatment periods. The resuits of a typical laboratory test

 

3mSR Program Semiann, Progr. Rept. Jan. 31, 1963, ORNL-3419, p. 33,
14G. M. Watson et al,, Reacror Chem. Div. Ann, Progr. Rept. Jan. 31, 1960, ORNL-2931, p. 52.
10

of this procedure are shown in Fig. 5. However, these data indicate that the average remcval rate
corresponded to about 1% reaction of sulfide ion with HF. The rate-contrelling step was presumed as the
initial reduction of sulfate by hydrogen.

 

ORNL-LR-DWG 56425R

 

 

 

 

 

 

 

 

 

 

 

 

500 t ’ -
¢

g- \ , HF: H2 =~

§ 400 INITIAL WEIGHT OF MELT: 2.5kg —

<

=

>

_.l

3 300

z

O

=

=2

S 200

o

2

.

1

? 100 ]

\O : :

O 5 10 5 20 25

HF PASSED (g equivalent wt)

Fig. 5. Removal of Sulfur from LiF-BeF; (63-37 mole %) by Treatment with HF-H, Mixtures at 700°C.

ORNL-DWG 63-2T749R

 

 

g
4 [ - )’N‘XS}’

 

 

Prizs/Priy

 

 

 

 

 

 

 

 

 

 

TEMPERATURE (°C)

Fig. 6. Ratio of H;S to H, Pressures Reguired to Produce Sulfides of Nickel and Copper. The data points are from T.
Rosenquist [J. fron Siteel Inst. (Londonj 176, 37 (1954)]; the lower solid curve was calculated from free-energy data [J. F. o
Elliott and M. Gleiser, Thermochemistry for Steelmaking, vol. 1, Addison-Wesley, Reading, Mass., 1960]. o
11

S Separate studies of sulfate removal from molten LiF-BeF, (66-34 mole %) indicated that a principal
sulfur removal mechanism, in addition to the evolution of H,S, is corrosion of the nickel or copper salt

container.’

This was similarly true when sparging with helium or hydrogen zalone. The available
thermodynamic data indicated that, indeed, direct reaction of SO, 2, its thermal decomposition products
SO; and SO;, or H,;S with nickel or copper to form metal sulfides and oxides is to be expected at process
temperatures of 600 to 800°C. This investigation, therefore, pursued the rapid reduction of sulfate by an

active metal such as beryllium,

80,4 %" (soln) + 4Be? (cryst) = 4BeQ (cryst) + $* (soln) , (3)
followed by sulfide removal with HF,

2HF () + S* (soln) = H, S (g} + 2F ™ (soln) . (4)

According to published data shown in Fig. 6, a control of H; S to H, ratios should prevent the reaction of
H; S with nickel.

Removal of Structural-Metal Impurities

The fused salt sysiems of the MSRE were constructed of Hastelioy N, a nickel-base alloy which
contained 6 to 8% chromium as a constituent. In reactor fuel systems of this type some depletion of the
chromium activity in the surface layer was anticipated® ® until the following equilibrium was established:

Cr® + 2UF, = 2UF; + CiF, . (5)

If the molten fluoride fuel mixture additionally contained nonequilibrium concentrations of structural-
metal fluorides more easily reduced than UF,; (e.g., NiF,, FeF;, or FeF,), then excessive chemical
corrosion of the Hastelloy N container would occur. Similar corrosion mechanisms would also occur from
non-uraniwm-bearing fluoride mixtures such as the secondary coolant of the MSRE. Structural-metal
fluorides of this type might be present as impurities in the fluoride raw materials and may also be
introduced by corrosion of the process cquipment during production operations. Thus the control of their
concentrations in the purified fluoride mixtures has been an important process consideration.

Although there are several structural-metal fluorides which would contribute to the corrosion process,
production practices have generally been concerned with chromium, nickel, and iron as potentially
significant impurities in the fused salt mixtures. Commercially available fluoride salts normally contain only
iron as a major impurity; however, contamination by all three of these metals may result from corrosion of
the process equipment. Chemical development studies have pursued reduction both by hydrogen and by
active metals as methods for purifying fluoride mixtures.

Reduction of Structural Metals by Hydrogen. -- Since the inception of the flueride purification process,
structural-metal impurities have been reduced from solutjon in the molten fluoride by a final gas sparge
treatment with hydrogen. At the operating temperatures of the process, nickel fluoride is readily reduced,

 

Y5y, E. Corgan et al., Reactor Chem. Div. Ann, Progr. Rept. Jan, 31, 1964, ORNL-3591, p. 63,
165 A, Lane, H. G. MacPherson, and F. Maslan (eds.), Fiuid Fuel Reactors, p. 599, Addison-Wesley, Reading, Mass.,
1958.

 
12

ORNL-DWG 63-6489
130 - oo —-- ‘

 

 

 

 

 

 

 

 

 

 

AV
TEMPERATURE .
=C) i.FeE+:|O {ppm) l
125 b 8 800 600 L —_—

~ e TOO 550 |
% A 600 700 ;
o
a
+ 00 ¢
NCD :
L
Lo
o
E
5 75 b
|14
.4
<1
>
s
Z 50
o :

25

c ¥

0 0.5 1.0 1.5

HYDROGEN FLOW RATE (liters,/min)

Fig. 7. Reduction of Fe? in LiF-BeF, (66-34 mole %) at 600 to 800°C by Hydrogen Sparging.

iron fluoride is somewhat more difficult, and chromium fluoride is essentially inert to reduction for prac-
tical purposes. However, maximum chromium concentrations of 25 to 50 ppm that are present in most
starting materials are not prohibitive for reactor applications.

Further development of the hydrogen reduction method was primarily realized from the use of HF-H,
mixtures for the removal of oxide ion under reducing conditions. By using pure nicke! or copper for salt
containment within stainless steel vessels, the introduction of corrosion products during HF treatment was
effectively eliminated. As suggested by studies of high-temperature thermocouple research,'” the presence
of hydrogen would also reduce the corrosiveness of the HF-H, O effluent gas mixture which accompanies
the conversion of oxides to fluorides.

In adapting the production facility for the preparation of the relatively large quantities of fluorides
needed for the MSRE, some consideration was given to the rates at which structural metals could be
reduced from the molten solution. The results of a small-scale experiment which examined the reduction of
iron from approximately 2 kg of LiF-BeF, (66-34 mole %) are shown in Fig. 7. Although the gas-liquid
contact conditions of these experiments could not be scaled to larger equipment, the results clearly
indicated the effect of temperature on the reduction process. The initial instantaneous rates at realistic iron
concentrations in the fluoride melt were not prohibitively slow. In a similar but larger experiment
conducted in the production facility, values for the relative reaction velocity constant as a function of
hydrogen sparge rates were determined. These results, shown in Fig. 8, indicated that hydrogen sparging
rates of about 10 liters/min would accelerate the reduction rate by about 200% over that obtained with the
hydrogen flow rate of 3 liters/min customarily used. Sparge rates greater than 10 liters/min caused frequent
entrainment of salt in the gas effluent lines.

 

Y7G. W, Keilholtz et al., Reactor Chem, Div. Ann, Progr. Rept. Jan. 31, 1961, ORNL-3127, p. 133.

 
 

 

13

ORNL-DWG €3-6491

i J
QUANTITY OF SALT: 70 kg (ASSUMED {st ORDER
REACTION DEPENDENCE ON FeZ' CONCENTRATION)
0.08 | INITIAL Fe®* CONCENTRATION: 200-3C0 ppm
| ,

 

i
i
]
|

0.06 -

 

 

 

 

k., RELATIVE REACTION VELOCITY CONSTANT the' )

 

 

 

 

0.04 S / ; _—
/*
0.02 | / -
i
|
o | _
a 2.5 5 7.5 10 12.5

HYDROGEN FLOW RATE {liters/min)

Fig. 8. Effgct of Hydrogen Sparge Rates on the Relative Rate of Reduction of Fe™ from Solution in LiF-BeF, (66-34
mole %} at 700 C in the Fluoride Production Facility.

ORNL-DWG 63-6492

 

 

 

25 T T T
o TOTAL Cr, Ni AND Fe
e Fe
& Cr

20 5 & Ni

WEIGHT OF MELT:~2kg

 

 

 

 

 

METAL FOUND IN SOLUTION (millimoles/kg)

 

  

 

 

 

 

[ ———p AN A
0.4 0.2 0.3 0.4 0.5
BERYLLIUM METAL ADDED (g)

 

Fig. 9. Reduction of Structural-Metal Fluorides from Solution in LiF-BeF, (66-34 mole %) at 600°C by Beryiliwn
Metal.

Reduction of Structural Metals by Beryllium and Zirconium. — Although the process of hydrogen re-
duction was adequate for routine production, a capability for reclaiming materials contaminated with chro-
mium was needed. Such instances would arise from failure of the nickel or copper liner within the stainiess
steel preparation vessel, from detachment of metal oxide or fluoride scale formed on stainless steel surfaces
in the gas phase of the preparation vessel, and from raw materials containing impurities in excess of those
14

ORNL.-DWG 63-6493

 

 

 

 

 

 

 

 

 

 

 

 

 

 

a ioo r l T T I
S . o—ir— END POINT FOR €= /%"
_g !
£ AN
£ g0 ]
E .
=
Q &
|.‘..
3 60 b
3 |
[¥3)]
=
S
Z 40 L
O
U
= .
o
= I _
= 20
= _
& LEND POINT FOR
@ /
5 é crét — ¢.°
0 o1
0 20 40 60 80 100

ZIRCONIUM METAL ADDED (millimoles/kg)

Fig. 10. Removal of Chromium from Solution in LiF-BeF, (66-34 mole %) at 600°C by Additions of Zirconium
Metal,

given by representative analyses. The choice of reducing agents for use in production for the MSRE was
limited to those metals whose cations were a primary constituent of the molten fluoride mixture. However,
metallic lithium was not considered because of its low density, its low melting point, and its incompatibility
with nickel or copper at process temperatures. Thus beryllium was proposed as a reducing agent for the
MSRE coolant, flush, and fucl solvent salt mixtures; zirconium was an alternative for reducing impurities
from the fuel solvent mixture.

In one development test, beryllium metal turnings were added in weighed increments to a molten
mixture of LiF-BeF, (66-34 mole %) at 600°C. The melt initially contained about 12 millimoles of iron per
kilogram of melt (670 ppm by weight} and about 4 miflimoles of chromium per kilogram of melt {210 ppm
by weight}. The analytical results from filtered samples of the salt taken after each beryllium addition are
shown in Fig. 9. These data indicated that structural metals could be reduced to reasonably low
concentrations by berylliuim. The results of a similar experiment iflustrating the reduction of chromium
from the same fluoride solvent by zirconium are shown in Fig. 10. A third experiment examined the
comparative reducing power of bervllium and zirconium. The results of this test showed that beryllium
would reduce ZrF, from sofution in LiF-BeF, (66-34 mole %).

4. THE PRODUCTION PLANT

The fluoride production facility was constructed on a 40- by 40-ft area in the high bay of Building
9201-3 within the Y-12 Plant. To meet industrial hygiene requirements for handling beryllium and other
toxic fluoride saits, the facility was essentially totally enclosed and equipped with an zir ventilation and
filiration system that provided about three air changes per minute (14,000 cfm). This air flow also
maintained the atmosphere of the production facility at a negative pressure with respect to the cutside area
to control airborne contamination. Specific work areas within the production plant were compartmented
and provided with direct exhaust air ducts and filtered makeup air. All exhaust air from the facility was
15

ORNL ~BWG 64-6%98

 

 

 

 

   

AN
DOWN 1T \
SHOWER— | N\
PROTECTIVE . = - i
CLOTHING
STORAGE — ‘

 

 

i, SCALES

|

  

LOADING ROOM F—LDI\DiNG HOPPER
rd
,/ /

 

  
    

. SALT
TREATMENT -

 

MELT DOWN

L

SaLT

 

 

'\/_'

X
| e
\CONTROL

PANELS -

 

 

 

 

 

 

Fig. 11. Fluoride Production Facility — Layout of Operating Level.

passed through a bank of absolute filters before discharge outside the building. In addition, all liquid wastes
from the plant were piped to a large holdup tank for analysis before discharge into the Y-12 waste system.
On the basis of industrial hygiene reports of analyses of surface, waste, and airborne samples throughout
the production history, these provisions have adequately maintained a safe working environment within the
facility as welil as the surrounding area.

The production facility was constructed on three levels to facilitate maintenance and operating
procedures. The lower level housed change room facilities, auxiliary equipment, and accesses to process
furnaces for maintenance purposes. The second floor was the main operating level. The third floor provided
hatched accesses to the furnace compartments and fluoride loading room for service purposes. The roof of
the production facility also had removable sections to permit use of the overhead building crane for
inloading and outioading heavy equipment and supplies. The equipment and supply storage area was on the
lower building level and was also accessible to the building crane.

Fused fluoride mixtures were produced in two batch processing units. Prior to the MSRE production
commitment each batch unit was individually loaded with starting materials from a transfer hopper. The
normal batch cycle included a 48-hr cooling period before the furnace could be reloaded. Requirements for
large quantities of fluorides having identical chemical composition made feasible the use of a single furnace
assembly for the initial loading and melting of starting materials. This modification to the facility provided
a molten charge to each batch processing unit. By proper scheduling the two production units could be
operated on a semicontinuous schedule without loss of time for furnace cooling. In addition, the relocation
of the raw materials loading area adjacent to the meltdown furnace facilitated materials handling. Figure 11
shows the floor plan of the main operating level of the production facility and indicates the flow of
materials through the processing units.

Raw Materials Charge

Because of the toxicity of fluoride salts (beryllium fluoride in particular) the design of the loading room
and formulation of handling procedures were intended to minimize exposure of operating personnel to
16

hazardous conditions and to confine fluoride contaminants. The loading room was isolated from other areas
of the production plant by shower facilities and air locks. Two doors which opened to the main operating
level were sealed for emergency use only. To gain normal entrance to the loading room, the operator
donned only long underwear, shoes, and socks in the change room (lower level) and proceeded through a
shower stall, contaminated change room, and air lock to the stairs leading to the operating level. At this
station the operator dressed in a {ully protective plastic fresh-air suit, stepped through a shower facility, and
then entered the loading room. Fresh-air supply points for the suits were located in the loading room,
shower, and protective clothing area for convenience. Upon completion of loading room duties, the
operator departed by the entrance route. The shower adjacent to the loading room served for
decontaminating the plastic fresh-air suits. The operator showered himself before entering the main change
room. The loading room was manned by two operators and one outside observer.

Selected fluoride salts were individually loaded into a tared transfer container by hoist assist, moved, in
line, to scales for weighing, and then transferred to the meltdown furnace by means of a vibratory
conveyor. The outside observer verified all weighings and oversaw the loading of the meltdown furnace.
Following each furnace loading operation the loading room was vacuum cleaned. This operation, together
with room air exhaust near the loading hopper, prevented the accumulation of hazardous materials within
the facility. Figure 12 depicts operation of the loading room just prior to its initial use.

FHROTO 70837

 

 

Fig, 12. Fluoride Production Facility — Raw Materials Handling Area.

 

 
 

 

17

Meltdown and Pretreatment

The meltdown fumace assembly, shown in Fig. 13, adjoins the raw materials loading room and was
operated to provide a molten material charge to each of the two adjacent batch processing units. Materials
were fed through the extended 3-in.-diam tube of the vibratory conveyor and were dropped into the
furnace assembly. This vessel was constructed of 6 ft of 12-in.-ips sched 40 stainless steel 304L pipe lined
with %-in. nickel sheet and was surrounded by a 50-kVA furnace of commercial design.

In addition to the initial melting and mixing of raw materials, some pretreatment of the fluoride
mixture was achieved. The meltdown operation required approximately 4 to 6 hr, during which absorbed
water was volatilized and removed by gas purge. Metallic reducing agents were also incorporated in the
loading operation to reduce sulfates and structural-metal fluorides. Upon melting, gas sparge rates with
hydrogen and argon were increased to remove carbon by entrainment. At the conclusion of this

PHOTO 70838

 

 

 

 

 

 

 

 

 

Fig. 13, Fluoride Production Facility — Meltdown Furnace Assembly.
18

pretreatment period the molten fluoride was piped on demand to either batch processing unit. Some
separation of insoluble materials by decantation was achieved during this operation.

The Batch Process

Each of the two processing units was designed with two primary salt containers. The larger of the two,
the salt treatment vessel, was constructed from a 6-ft length of 12-in.-ips 304L stainiess steel pipe and was
sized to contain the solid bulk charge of salts that would correspond to about 2 ft* in the molten state. An
inner liner fabricated from %-in. grade A nickel sheet provided primary salt containment. Except for
conventional flanged access ports and threaded gas line connectors, the vessel was of welded construction.
In addition, the inner surfaces of the stainless steel vessel were plated with nickel to retard corrosion of the
vessel by HF and water vapor. The salt storage vessel (salt receiver) was incorporated as a pretreatment
for oxide removal and for minimizing contamination of the prepared salt mixture. This vessel was
constructed of grade A nickel, 12 in. in diameter and about 3 ft in length. Five access ports for
small-diameter (% -in.) tubing were welded into the top of the vessel.

The salt receiver vessel was connected by a smali-diameter tube to the dip line in the treatment vessel as
shown generally by Fig. 14. During the salt purification step, treatment gases were introduced into the

ORNL~LR-DWG 20157

 
 

TEFLON GASKETED FLANGE

— 4—in.~DIA CHARGING PORT

3/g~in—NIGKEL
TRANSFER LINE

FILTER

SAMPLER

 

 

 

T
7

 

 

W

REACTION VESSEL

 

 

 

RECEIVER VESSEL-

Fig. 14. Fluoride Production Facility — Salt Freatment and Receiver Vessels,

 
 

 

19

receiver vessel, passed through the connecting line, and flowed out of the dip tube for sparging the fluoride
mixture. At the conclusion of the production procedure, the differential gas pressure on the two vessels was
adjusted so that the salt charge would flow through the connecting transfer line and a filter disk of sintered
nickel (0.0015 in. pore diameter) before discharge into the salt receiver. A reference sample of the salt was
withdrawn after salt transfer rather than as shown in the sketch.

Temperature requirements for this batch process were achieved by electrical resistance heating. The salt
treatment vessel was surrounded by a 50-kVA furnace with a 15-in.-diam well of commercial design. A
stainless steel vessel with a closed bottom and open, but flanged, top served as a furnace liner and support
for the salt treatment vessel. The salt receiver was heated by a similar but smaller furnace rated at 23 kVA.
However, this vessel was suspended from a support frame. The furnace was mounted on a cable, pulley, and
counterweight assembly which allowed the furnace to be lowered from around the salt receiver for rapid
cooling. Small-diameter tubing which served as transfer lines for molten salts was heated directly by a
low-voltage alternating current. A photograph of one of the batch facilities is shown as Fig. 15. Furnace

¢ PHOTO 70834

 

 

 

Fig. 15. Fluoride Production Facility — Batch Processing Unit.
20

 

Fig. 16. Fluoride Production Facility — Process Control Unit.

controls and the gas influent manifold were located in a central control panel outside the process cubicle
(Fig. 16). Temperature control was achieved by two on-off controllers connected in series to protect against
an excess temperature condition. This control circuit would also activate a locat and general building alarm
sysiem,

A simplified schematic flow diagram of the unit process is shown in Fig. 17. The gas influent system
provided for the alternate use of hydrogen, helium, or vacuum and for mixing anhydrous HF with either
process gas. Flow rates for hydrogen and helium were determined by a conventional rotameter.

Concentrations of HF in the influent gas were determined (relative to the flow of carrier gas) by direct
titration of a side stream with standard caustic solution. The gas manifold of each production unit was
additionally arranged for control of salt transfer, both into the treatment vessel from the meltdown furnace
and into the sait storage vessel.

Process control was exercised primarily on the basis of results obtained from off-gas analyses. This
analytical station was located outside of but adjacent to the furnace cubicle. Direct titration of a side
stream for its HF content provided material balance information during the oxide removal step and denoted
the extent of reduction by hydrogen during the final phase of the purification process. The gas effluent was
passed through a cold trap, maintained at —12°C by refrigeration, to condense essentially ali of the water
vapor and most of the HF during the oxide removal process. The contents of this trap were drained
periodically and analyzed for HF. The inferred volume of water collected was also indicative of the rate of
oxide removal from the salt.

 

 
21

CRNL- DWG €4-6983

S

 
 
 
 

 

   
  
    
  
  

 

 

 

 

 

 

     

T HF SUPPLY AND
METERING SYSTEM . YGAS MIXING CHAMBER
~. i
s —~FILTER
. HELIUM ; R
i 5 o . P SALT
| HYDROGEN GAS GAS FoooaLl
- + — ¥ F
: VACUUM : MANIFOLD ANALYSES N iSESER‘ OIR
L o = FURNAGE
. CARBON TRAP-1
EXHAUST TO ;SEE\ ’ N
ATMOSPHERE \ / RS- SALT
) L PNTREATMENT
GAS ’ “‘N\]?i\?cw_/ a VESSEL
AR A i ,.\_,/’ ALY DR *:}“\
ANALYSES NoF s S FURNAGE
o 3 ~ N
PELLETS . CONDENSATE
HF REMOVAL DRAIN LINE
WATER
REMCVAL

Fig. 17. Fluoride Production Facility — Simplified Schemaiic Process Diagram.

Waste HF could be effectively removed from the effluent gas by absorption on sodium fluoride pellets.
Two columns, 3%, in. in diameter and 3 ft long, sufficed for this operation. The gas effluent downstream
from these columns was analyzed periodically to determine the frequency of regeneration of the sodium
fluoride beds. Waste gases were bubbled through a “seal pot” filled with a fluorocarbon oil and exhausted
to the atmosphere through a ' -in.-ips sched 40 iron pipe. Figure 18 is a photograph of the off-gas system
and shows the cold trap, seal pot, and exhaust gas line.

Auxiliary Systems

Although the gas handling systems of the production plant were quite simple, their design met specific
requirements of the purification process. Each auxiliary system supplied both batch production units and
the mettdown furnace.

Vacuum System. — Two Kinney pumps, each rated at 110 c¢fm and driven by 5-hp motors, were used
alternately as the supply and reserve source of vacuum. Although HF was not routinely pumped into the
vacuum system, soda-lime traps followed by columns packed with activated alumina were incorporated in
the manifold to permit limited exposure of the system to corrosive conditions of the process.

HF Supply. -~ Anhydrous hydrogen fluoride of high purity was obtained commerciaily in 200-1b
cylinders. Since HF boils at about 19°C, its pressure in the supply system was regulated by temperature
control. The cylinder was totally enclosed in a constant-temperature air bath and operated at 10 to 15 psig
in the system. Gas lines were traced with insulated Nichrome wire and maintained above 100°C by
electrical resistance heating. This feature prevented the condensation of HF in the lines and stabilized the
polymerization state of the gas.'® The HF flow into each production unit was controlled by a needle valve
and indicated by the differential gas pressure across a sintered nickel filter barrier. All gas lines were of
nickel; valves of commercial design were of Monel or stainless steel construction.

Helium Supply. — Helium was obtained from the Y-12 Plant manifold system. However, because of the
sensitivity of the molten fluorides to water vapor, it was passed through Linde molecular sieve 4A to ensure

 

18p L. Jarry and W, Davis, Jr., J. Phys. Chem. 57, 600 (1953),

 
22

PHOTC 7GE33

 

Fig. 18. Fluoride Production Facility — Off-Gas System.

dryness. Helium entering the process was pericdically monitored by a MEECO electrolytic moisture
analyzer to detect moisture breakthrough in the trap. Maximum water vapor concentrations were specified
at 10 ppm and normaily did not exceed 5 ppm.

Hydrogen Supply. — Hydrogen was supplied solely to the fluoride production facility from a 12-bottle
manifold. Standard-grade hydrogen (99.5%) was used for all operations involving hydrogen. However,
impurity oxygen was converted tc water by reaction on a Deoxo platinum catalyst unit having a 50-cfm
capability. Water was removed by absorption on molecular sieve. Because of the relatively large
concentration of water, four colummns, 3% in. in diameter by 5 ft long, filled with absorbent, were operated
in series to maintain water concentrations at less than 35 ppm. Three of the four columns were in
continuous use. A rotation schedule was followed so that the fourth column could be regenerated as
required. Water concentrations in the hydrogen effluent from the absorption columns were monitored
semicontinuously with the MEECO moisture analyzer.

Salt Storage. — After cooling the finished salt batch and removing it from the processing facility, the
salt storage vessel was wiped by wet sponging to remove possible salt contaminants. The container was then
moved to the salt storage facility. This facility consisted of a helium manifold system and was arranged to
allow access te each container by electric hoist and identification by a batch card system. Although the sait
containers could be sealed against atmospheric contamination, their connection to a live helium pressure
(~10C psig) ensured against possible leaks through the tube fittings or undetected cracks in the vessel walls.

 

 
23

5. PRODUCTION OF COOLANT, FLUSH, AND FUEL SOLVENT

Actual production of fluoride mixtures for the MSRE was started on March 1, 1964, in anticipation of
reactor loading during the second quarter of FY 1965. A three-shift seven-day work schedule was pursued
throughout production operations with one technician-operator and one maintenance mechanic assigned to
each rotating shift. Production of the coolant and flush salts, ? LiF-BeF, (66-34 mole %), was completed on
September 7, 1964, A total of 15,964 ib of the salt mixture was produced out of 16,104 ib of raw materials
charged to the facility in 61 batch operations. On this basis the overall production rate was about 4.7 ft>
(560 1b) per week, including time for plant shakedown and interruptions for maintenance. The average
batch cycle was about 150 hr.

Although this material was considered acceptable for use in the MSRE, a reprocessing schedule was
undertaken for the interim period before loading the coolant system of the reactor during the fatter part of
October 1964. This operation not only permitted upgrading those batches having higher impurity
concenirations but also permitted an evaluation of oxide removal. Twenty-one batches were recycled
through the purification process on the basis of their production history. Three of the first four reprocessed
batches showed further water removal of 125 to 235 ppm oxide equivalence in the salt, two batches of the
next six showed 130 and 187 ppm oxide, and one batch of the remaining eleven had 125 ppm of removable
oxide. All other batches had less than 30 ppm oxide equivalence in the salt.

Production of the fuel solvent mixture, ’ LiF-BeF,-ZrF, (64.7-30.1-5.2 moic %), began on Nov. 4,
1964, and was completed on March 10, 1965. A total of 10,545 Ib of this material was produced from
10,690 1b of raw materials charged to the production facility in 36 batch operations. The average batch
cycle time was 152 hr.

The depleted fuel concentrate, ' LiF-2*®UF, (73-27 mole %), was produced in a single batch at the
conclusion of the fuel solvent preparation. About 624 Ib of starting material was charged directly to a2 new
salt treatment vessel, heated above its melting point, and ireated for 122 hr. The batch was then divided
between two salt storage containers to facilitate its handling. A total of 620 Ib of this mixture was
recovered for delivery to the MSRE. The operation of the main production facility was concluded with this

preparation on March 12, 1565.

Process Operating Conditions

The operating conditions used in the preparation of these fluoride mixtures for the MSRE were based
primarily on prior production practices and existing equipment. However, some modifications were made
possible as a result of chemica! development studies. These changes were largely reflected in the
hydrofluorination step and process control procedures.

HF-H, Treatment. — Although the purification procedure was generally considered a two-step process,
the operating conditions chosen for the hydrofluorination step were also related to the subsequent
reduction of structural-metal difluorides. The corrosion of the nickel-lined process vessel according to the

reaction
2HF + Ni® = NjF, + H, (®)

was considered in establishing upper limits for HF concentrations in hydrogen. Discrete values for the
equilibrium quotients for the reactions

0" + 2HF = H,0 + 2F~ (7)
24

and
OH +HF=H,0 +F"~ (8)

were not available when production operations began; however, experimental results had shown that
relatively low HF concentrations would suffice for effective oxide removal from the fluoride mixtures.
Process temperature was a third factor in that lower temperatures favored oxide removal and also corrosion
of the process equipment. Operating conditions established on these process variables were further modified
by limitations from off-gas analytical and handling procedures.

Although liquidus temperatures for the MSRE fluoride mixtures were in the range of 450 to 500°C, a
process temperature of 600°C was chosen for the oxide removal step on consideration of the polythermal
conditions which existed in the treatment vessel under static conditions and the effects of temperature on
the process reactions. According to equilibrium data by Blood,'? the concentrations of HF in hydrogen
which would maintzin the NiF, content of the melt at an arbitrary value of 25 ppm would be about 38%
by volume at 600°C and decrease to about 23% at 500°C. Production operations were begun with HF
concentrations of 10% in hydrogen at I atm pressure. Minimum conditions for discontinuing the HF
treatment were set at a water evolution rate of less than 2 g/hr on the basis of the sensitivity of the off-gas
analytical method. The results of early production runs showed that water evolution rates initially
approached stoichiometric limits and then rapidly diminished below the arbitrary control level at the
conclusion of the process step. Although this control level for water removal corresponded to an average
oxide removal rate of 15 ppm/hr, subsequent calculations of oxide and hydroxide ion concentrations in the
melt which would be in equilibrium with 1 atm of HF and a water vapor pressure of 0.0046 atm,
corresponding to the control limit, yielded very low values. According to equilibrium data by Mathews and
Bues,!® the oxide content of the melt would vary from 0.44 ppm at 600°C to 0.02 ppm at 500°C, and
correspoending hydroxide concentrations would vary from 11 to 4 ppm.

The planned operation of the production facility called for an evaluation of oxide removal rates at
higher HF concentrations up to limiting values. However, prolonged use of higher concentrations
introduced complications in the off-gas system. As an attempt to reduce possible contact of the fluorides
with extraneous water vapor sources, HF was stripped from the waste gas stream by adscrption on scdium
fluoride pellets rather than by conventicnal caustic scrubbers. This trapping system was sufficiently
effective at the 10% HF concentration level to permit direct discard of the hydrogen effluent to the
atmosphere. At significantly higher HF concentrations in the process influent stream, the partial pressure of
HF in the waste gas became prohibitively high for direct discard. As shown by Fig. 19, typical oxide
removal times were about 35 hr for processing the flush and coolant mixture. This time requirement was
considered compatible with the degree of operator control available for the process. Consequently,
increasing the concentration of HF would require modifications of the off-gas system and increased
operator attention.

The removal of sulfur from the fluoride mixtures was also accomplished during the hydrofiuorination
treatment step. However, its effective removal proved more difficult than was anticipated from earlier
development studies. The H,S in the gas effluent was collected periodically in an ammoniacal cadmium
chloride solution and titrated with a standard iodine solution. Material balances on sulfur evolved, as H, S,
were compared with the quantity of sulfur believed to be present in the raw materials charge. These values

 

19 A, L.. Mathews and C. F. Baes, I1., Oxide Chemistry and Thermodynamics of Molten Lithium Fluoride- Beryllium
Flyoride by Equilibration with Gaseous Water—Hydrogen Fluoride Mixtures, ORNL-TM-1129 (May 7, 1965).
25

ORNL-DWG 64-6294

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

A
100 % ] TR . T
o
s ol
= 80 o C Y
5 A C—133 a
- B C-132 A
0 4
e 60 & C30 - -
~ A C-128
o
s
g 40 TN
LL_!
o
bopob _
&
=
0
0 5 10 15 20 25 30 35

PROCESS TIME (hr)

Fig. 19. Fluoride Production for MSRE — Removal of Oxides from TLiF-BeF, (66-34 mole %) During HF-H,
Treatment at 600°C.

ORNL-DWG 64 -6995

 

0.7 T |
BATCH NO. C-4&
WEIGHT: 420 kg

0.6 |-— HF CONCENTRATION: 4 meg/liter Hy ——nu ]
INITIAL SULFUR: 8%ppm=1{0.2 ¢

 

 

 

 

 

 

£ ol |

; 0.5 p— | ¥ =
pe : l

Y

oal—d— ,__L_ R RN B
&

g

Y 0.3 _L__ ]
7

b

o

LL] S -t
= 0.2

2 |

 

AREA = 6.5g SULFUR
o f -

N

o : ‘
Q 5 {0 15 20 25 30
PROCESS TIME (hr)

 

 

 

 

 

 

Fig. 20. Fluoride Prgduction for MSRE — Removal of Sulfur Impurities from 7LiF-=Ber (66-34 mole %} During
HF-H, Treatment at 600 C.

were often in disagreement and probably resulted from the deposition of nickel sulfide in the meltdown
furnace. However, the results of these off-gas analyses could be empirically related to the concentration of
sulfur remaining in the melt. Following each addition of beryliium metal, the concentration of H, S in the
off-gas was found to increase (Fig. 20) when sulfur, presumably as sulfate, remained in the melt.
26

Consequently, sulfur removal was considered complete when an addition of beryllium metal failed to result
in an increase in the H, S evolution rate. Repeated additions of beryllium metal followed by prolonged HF
treatment periods (bevond that required for oxide removal) were frequently necessary to satisfy this
control condition. Subsequent analyses of salt samples from these preparations consistently showed sulfur
concentrations of less than 5 ppm in the finished product.

Although an efficient sulfur removal process was not attained prior to or by this production effort,
sulfur impurities were present in the BeF, raw material of only one-third of the production batches. Its
removal was not unduly restrictive on the overall production effort. The development of sulfur-free
beryllium fluoride by the commercial manufacturers has precluded further interest in the development of
this aspect of the fluoride production process.

Reduction of Structural Metals. - The reduction of structural metals from solution in the molten
fluorides was generally considered as the final phase of the production process. However, the use of a
separate meltdown furnace for this production commitment was adaptable to pretreatment of the fluoride
mixtures with beryliium metal prior to their transfer into the salt treatment vessels. The amount of
bervllium metal added with each raw materials charge was based on the concentrations of reactive
impurities (sulfur, chromium, nickel, iron, and water) in the starting salts. As conducted, the cffects of this
reduction step were incomplete, but iron concentrations of materials discharged to the salt treatment vessel
were reduced to about 200 ppm. The effectiveness of this reductive measure was further limited by the
efficiency of the liquid-solid separation process. Filtration of salt into the treatment vessel was not feasible
because suspended oxide particles in the salt mixtures rapidly plugged the sintered nickel filter. Therefore,
separation of reduced metal particles from the molten fluorides depended upon decantation.

Beryllium and zirconium metals were found to be effective reductants for chromium, nickel, and iron
fluorides during development tests. However, their use in the final phase of the purification procedure was
curtailed because of process control limitations. The reactions of these active metals in reducing
structural-metal fluoride impurities were found to be nonstoichicmetric during development tests, probably
because of surface coating of the active metals with reduced materials. In addition, these particles were
believed to be more extensively suspended in the fluoride melt, on the basis of filtration tests, because of
reduced particle size or lower density of the coated particle. Thus, one consideration of the use of active
metals as reducing agents was the dependency of the actual separations process on the integrity of the filter
used during transfer of finished batch to the storage container. The effect of these metals on the MSRE fuei
system had not been evaluated at this time. Since the addition of large excesses of active metals to the
fluoride mixture was also not advised, process control measures would have been required to ensure the
satisfactory reduction of structural-metal impurities from the salt mixture. Thus, process time saved by
rapid reaction of strong reducing agents was diminished by time reguired for chemical analyses or other,
more direct, analytical methods. The routine use of these active metals would have also required the
development of apparatus for their addition to the melt without exposure to air or other source of oxide
contamination.

The removal of structural metals from solution in the fluoride mixtures depended primarily on
reduction by hydrogen. Since reduction rates were known to increase with increased temperature, the melt
temperature was raised to 700°C. Higher temperatures were avoided to prolong the service life of the
furnaces. By the hydrogen sparge technique, the reduction of structural-metal fluorides, together with the
correspending evolution of HF, was a steady-state process. Therefore the concentration of HF in the gas
effluent stream could be empirically related to the concentration of iron fluoride which remained in the
melt, This relation, obtained for production of the MSRE fuel solvent mixture, is shown in Fig. 21. The
hydrogen sparge treatment was terminated when the HF concentration in the gas effluent dropped below

 
ORNL-DWG 65-2555

 

 

 

   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L 600 ‘ 1 e |
R . O BATCH F-178
. X BATCH F—177 7
£ 500 — SALT TEMPERATURE: 700°C ~
& Hs FLOW: 10 liters/min
—
I
w400
z
=
o
B 300
o
}_
=
2 200
C
o
=
O
x
= 400t
ob
0 002 004 006 008 040 042

HF CONCENTRATION IN GAS EFFLUENT (meq /titer Hy)
Fig. 21. Reduction of Iron Fluoride from MSRE Barren Fuel Solvent at 700°C by Hydrogen Sparging.

0.02 milliequivalent per liter of hydrogen. A similar determination was made for the binary coolant and
flush salt mixture. This control limit was set at 0.01 milliequivalent of HF per liter of H,. The difference in
these control points was not explored but was assumed as a solvent effect.

Process Control

The preparation of acceptable fluoride mixtures for the MSRE depended on process control measures
for each phase of the production operation. Records were provided that showed the source, quality, and
quantity of raw material used in each production batch. Each storage container had numbers impressed into
the top and side for positive identification.

Although operational control of the process depended upon analyses of the process gas streams, the
quality of each salt batch was determined from chemical analyses of the i)roduct. Samples of the melt were
withdrawn at the conciusion of each treatment period and at more frequent intervals when needed for
further evaluation of the process. These samples were obtained in copper filter tubes which had been
previously fired in an atmosphere of flowing hydrogen. The holding device was designed to protect the
sampler from atmospheric contamination following the hydrogen firing operation and throughout the
sampling procedure. This assembly also permitted sampling of the melt without disturbing process
conditions. In addition, a larger sample of the salt was withdrawn from the finished batch after its transfer
to the storage container. This sample was retained for reference purposes.

Chemical analyses of the filtered salt sample taken just prior to salt transfer into its storage container
were considered to be representative of the production batch. Averages of concentrations of pertinent
impurities found in the various production batches are shown in Table 5. Values shown for “oxide
removed”’ were determined from quantities of water collected in the effluent gas cold trap during the
hydrofluorination treatment.

Since the oxide removed column reflects water removal after melting and during hydrofluorination, the
quantities removed indicate the initial oxide and hydroxide content of the melt. The reported value for the

s oxide solubility of the flush and coolant mixture, LiF-BeF, (66-34 mole %), is about 200 ppm at 600°C;
28

Table §. Fluoride Production for the MSRE —
Average of Chemical Analyses of Salt Batches

 

 

 

) Chemicat Composition Average Concentration of Impurities (ppm}
Salt Mixture o
(mole %) Cr Ni Fe S Oxide Removed

Coolant ?LiF-Beli, 19 26 166 <5 1460
(66-34)

Flush TLiF-BeF, 16 39 123 <5 1650
(66-34)

Fuel solvent TLiF-BeF,-ZiF, 21 15 77 <5 728
(64.7-30.1-5.2)

Depleted fuel TLiIF-2?3UTF, 9 15 50 <5 386

concentrate (7327

 

the corresponding value for the fuel solvent mixture, LiF-BeF,-ZrF, (64.7-30.1-5.2 mole %), is about 380
ppm.2® Thus the oxide removed from the fluoride mixtures was clearly in excess of solubility limitations.
The lower quantities of water removed from the fuel solvent resulted from an alteration in the production
procedure. During the first two batch preparations of the fuel solvent mixture, hydrofluorination periods in
excess of 100 hr were necessary for oxide removal. These requirements indicated that the rate-controlling
step was either the reaction of HF with solid ZrO, or the dissolution of ZrQ, in the melt. During
subsequent preparations the molten charge in the meltdown furnace was mazintained quiescent before
transfer to the treatment vessel, and the length of transfer tube in the meitdown furnace was also shortened
by 2 to 4 in. This technique permitted effective decantation of the fluoride mixture from solid ZrQ, . The

quantity of zirconium lost from each batch was negligibie, and the HF treatment times were reduced to
about 30 hr.

Lithium Fluoride Densification

With the exception of 7LiF, fluoride salt starting materials, as received, could be charged directly to the
meitdown furnace. The lithium fluoride was found to have a very low bulk density and to contain excessive
quantities of water. Direct use of this material would have resulted in loss of production capacity, frequent
changeout of filters in the main ventilation system, uncertainties in material balances, and perhaps excessive
oxide contamination in the molten charge to the batch processing units. Consequently, pretreatment of the
"LiF to improve its bulk properties was desired. Although this material was found economicaily unsuitable
for pelletizing, the results of @ study of its sintering characteristics,”’ shown in Fig. 22, indicated that an
acceptable material could be obtained by a relatively inexpensive operation.

Intermediate-scale tests were made to develop a processing procedure and to examine the feasibility of
pretreating the entire amount (22,000 Ib) of LiF that was on hand. These tesis showed that periodic
agitation of the LiF while heating to 650°C was necessary to produce a free-flowing granular product.
Anhydrous HF was admixed with the helium sweep gas while the charge was heated to about 400°C to
convert LiOH, either initially present or formed by pyrohydrolysis, to LiF. Otherwise, heating to 650°C
permitted the LiOH to fuse with the LiF and form an intractable mass.

 

298, F. Hitch and C. F. Baes, Jr., Reactor Chem. Div. Ann, Progr. Rept. Dec, 31, 1966, ORNL-4076, p. 19,

213, 1. Sturm, “A Method for Densifying Lithium Fiuoride Powder,” MSR-62-94, Nov. 20, 1962, internal
memaoranéum,

..............
29

ORNL—LR— DWG 77866

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2.5 — ‘ ,
| |
I ‘.r | i e =
« -
1
/
2.0+ - —
> Sl Tt
=
o
&
Woas— — S
¥
3
35
o - L e —
&
1.0 —--- - ) o
—— |
—_— . d- { —_ ‘
0.5 i i L

 

 

150 260 350 450 550 650 50 850
TEMPERATURE (°C)

Fig. 22. Densification of Precipitated LiF by Annealing at Various Temperatures.

 

 

Fig. 23. Fluoride Production Facility — Horizontal Densification Kiln.
30

The production-scale equipment was a horizontal Monel reaction vessel (17 in. in diameter by 8 ft long}
equipped with a full-length agitator and heating jacket. This apparatus had been fabricated and used during
the ANP program for the conversion of solid ZrCl; to ZrF, by treatment with anhydrous HF at elevated
temperatures. As shown by Fig. 23, the apparatus was loaded directly from materials storage drums in
amounts up to 300 1b per batch. At the conclusion of the densification, the témperature was lowered to
about 200°C. The reaction vessel was then tilted by 2 hydraulic mechanism and the material dumped into a
stainless steel drum. After cooling to room temperature, the material was sampled and transferred to fiber
drums for storage. The average bulk density of the 7 LiF was increased from about 0.6 g/cc to about 1.1
g/cc by this operation.

This production operation was started on February 10, 1964, and was concleded on November 6, 1964,
Operation of this facility was achieved by personnel assigned to the large production facility. During this
period all 22,000 1b of 7 LiF was processed.

6. PREPARATION OF ENRICHED FUEL CONCENTRATE

The scheduled operations of the MSRE estzblished a requirement for 90 kg of *?° U contained in the
binary salt mixture " LiF-UF, (73-27 mole %p}. This guantity provided for bulk additions of fissile uranium
to the fuel drain tank during the initial fueling of the reactor and for incremental additions to the fuel
mixture in the pump bowl for sustained nuclear operations. Production of this mixture commenced on
December 17, 1964, and was concluded on January 29, 1965, Full shift coverage was provided during this
period except for a two-week vacation break in operations.

Approval of this operation by the ORNL Criticality Committee?
of the proposed production procedures and relevant chemical and physical properties of the salt mixture.?”>
The stipulations of this approval, the relatively small quantity of materials, and the monetary value of the
enriched uranium (~$12,000 per kilogram of 22%U) favored smail batch sizes. Therefore, this preparation
was conducted in the intermediate-scale production facility. Except for certain restrictions and added
control measures, the production procedure was essentially the same as that followed for the preparation of

? was based on a detailed description

other MSRE fluoride mixtures in the large-scale facility.

As a result of a nuclear criticality review, a batch size containing 15 kg of *?* U was allowed. Therefore,
six batch preparations were required to fulfill obligations for the MSRE. Each batch contained 26.16 kg of
the eutectic salt mixture and had a total uranium content of 61.65 wt %. The salt treatment vessel, also
subjected to this review, was constructed from a 36-in. length of 6-in.-ips sched 40 pipe (stainless steel
304L) with an inner liner of Y5-in. nickel. The design of this vessel, shown in Fig. 24, was essentially the
same as that used for large-scale production. Salt storage containers for the finished batches of enriched fuel
concentrate mixture were constructed from 36-in. lengths of 4-in.-ips sched 40 grade A nickel pipe. The salt
batch had a liquid depth of about 29 in. in the salt storage container and a dry-mix depth of about 26 in. in
the salt treatment vessel. The sali treatment vessel was heated by a 23kVA furnace and the salt storage
container by a 7500-W furnace (Fig. 25) in an arrangement similar to that of the main production plant.

 

22RNL Criticality Review Report No. 4, Aug. 11, 1964, and Report No., 11, Oct. 15, 1964,
231 m. Shaffer, “Preparation of Enriched Fuel Concentrate Mixture for the MSRE,” MSR-64-42, July 7, 1964,
internal memorandum,

 
 

31

PHOTO 74923

 

Fig. 24. Fiuoride Production for MSRE -- Salt Treatment Vessel for Enriched Fuel Concentrate Mixture.

Batches of raw material were blended and loaded into the salt treatment vessels by the Special
Processing groups of the Y-12 Plant and transported by truck to the flueride production area. After
processing and transfer to its storage container, the salt batch was placed in a nuclear-safe transport
container (Fig. 26) and shipped to a security warehouse within the Y-12 Plant. To further minimize the
possibility of a nuclear incident, the quantity of >°° U permitted in the production facility was limited to
one batch. However, by using two salt treatment vessels and scheduling materials transfer operations, the
production facility was operated aimost continually during preparation of the six batches of concentrate

mixture.
32

PHQTO 7186%

 

 

 

Fig, 25, Fluoride Production for MSRE — Process Equipment for Enriched Fuel Concentrate Mixture.

7. PREPARATION OF FUEL ENRICHING CAPSULES

One of the unique features of the MSRE was the method by which supplementary quantities of 2*° U
were added to the reactor fuel. These incremental additions were made by dissolving a fused eutectic
mixture of "LiF-2?SUF, (73-27 mole %) from small capsules that were lowered by a windlass into the
bowl of the fuel pump. The filling of these capsules with the fused fluoride mixture was a part of the
fluoride production effort and was accomplished during March 22—26, 1965. Approval of this operation by
the ORNL Criticality Committee’* was again based upon a detailed description of the handling
procedures.?®

The fuel enriching capsules were constructed from a 6-in. length of nickel tubing (% in. OD by 0.035
in. wall) with a hemispherical bottom. The top plug was penetrated by two “%-in.-0OD, 0.025-in.-wall nickel
fill tubes. Each filled capsule contained about 85 g of **3U, or about 148 g of salt mixture. Sufficient
capsules (161) were filled to fulfill scheduled tests of the MSRE. All capsules were filled from a single
production batch of the fuel concenirate mixture by means of a salt transfer tube (*; in. OD by 0.065 in.
wall) that extended almost to the bottom of the salt storage container.

 

24ORNL Criticality Review Report No, 25, Feb. 1, 1965.
255, H. Shaffer, “Preparation of Fuel Enriching Capsules for the MSRE,” MSR-65-4, Jan, 14, 1965, internal
memorandum,.

 

 
 

 

33

PHOTO 74924

 

 

 

 

 

 

 

 

Fig. 26. Fluoride Production for MSRE — Storage Container and Holder for Enriched Fuel Concentrate Mixture.

For filling purposes, seven capsules were connected in series by their Ys-in. fill tubes and clustered
within a 4-in.-diam heating chamber. The inlet and outlet fill tubes to the ciuster were connected by tube
fittings to the salt transfer line and to the overflow reservoir. The capsule cluster was held in place by an
adjustable support (Fig. 27) which also served as a distributor for the helium purge stream about the
capsules. The surrounding heating chamber had a flanged top, a body constructed from a 12-in. length of
4-in.-0D, 0.065-in.-wall stainless steel tubing, and a welded bottom of Y-in. stainless steel plate. Heat was
supplied by Calrods wrapped about the outer periphery of the chamber.

Since 23 fill operations were required, provisions were made for rapid assembly and disassembly of the
capsule clusters in the filling apparatus. The top cover flange of the assembly was rigidly fixed to a support
stand. The salt transfer line and the helium supply and exhaust lines penetrated the cover flange through
gastight connections. The heating chamber was suspended from a counterbalanced cable system so that it
34

- PHOTO 72084

 

 

Fig. 27. Fiuoride Production for MSRE — Fuel Enriching Capsules Prepared for Filling with “LiP->>°UF, (73-27
mole %).

could be raised or lowered as required during each fill operation. The assembly of capsules in this apparatus
is shown in Fig. 28.

The salt storage container, the salt transfer line, and the clustered capsules were heated to 600°C, and
sufficient helium pressure was applied to the salt storage container to cause the molten fluoride mixture to
flow into the clustered capsules. Displaced gases were vented through the top of the overflow reservoir. The
liguid levels in the capsules were visually observed by radiography using a Norelco 160-kV, 6-mA portable
x-ray unit and a TVX camera. A photograph of the equipment during a typical filling operation is shown in
Fig. 29. When the last capsule in the cluster was filled, helium pressure was vented from the salt storage
container and applied to the top of the overflow reservoir. Thus any excess sait in the capsule fill assembly
was returned to the salt storage container. The capsule cluster and salt transfer line were cooled to near

 
 

 

35

 

PHOTO 72085

 

Fig. 28. Fluoride Production for MSRE — Assembly of Fuel Enriching Capsules in Filling A pparatus.

room temperature while backflowing helium through the system. The filled capsule cluster was
disconnected from the assembly, and its exposed tubes were capped. After the net weight of salt mixture in
the cluster was determined, it was sealed in a watertight can. The individually canned capsule clusters were
placed in a holder within a nuclear-safe carrier for storage. All capsules were filled at a rate of about five
clusters per day during a two-shift five-day period.

Although the accountability of 2*°U was maintained for each capsule cluster, each capsule was
photographed by x ray before and after filling with the enriched fue! concentrate mixture. Thus, minor
variations in weight could be calculated from measurements on the contact prints. This examination
revealed no defects in the capsule clusters nor any variation in uranium density in the frozen salt mixture.

The fuel enriching capsules, as delivered to the MSRE, were suitable for indefinite storage and needed
only minor mechanical preparation before use. Each capsule was removed from its cluster by clipping the
 

 

 

Fig. 29, Fluoride Production for MSRE - Fuel Enriching Capsule Fill Operation Using Portable X-Ray Unit and TVX
Camera for Control,

"g-in. nickel fill tubes. near the top plug of the capsule. The cuter surface of the capsule was buffed as
required to remove any oxide, and the nickel jacket was slotted on a milling machine located in a glove box.
A wire bail was inserted through a predrilled hole in the top plug for fastening the capsule to the windlass
cable.

8. REACTOR LOADING OPERATIONS

When construction of the MSRE was substantially complete, salt batches selected for the secondary
coolant were transported from storage in the production facility to the reactor site. During the latter part
of October 1964, approximately 5755 Ib of coolant mixture, ’ LiF-BeF, (66-34 mole %), was transferred
from 22 salt storage containers to the coolant drain tank. Salt batches for the flush salt mixture were then
transported to the reactor site and loaded into one of the fuel drain tanks during November 1964,
Approximately 9200 b of this mixture from 36 storage containers was required. With the unloading of
these materials, the MSRE began its prenuclear test operations.

Reactor fueling operations began on April 20, 1965, with the loading of the barren fuel solvent and the
depleted fuel concentrate mixture. Approximately 10,050 1b of the solvent, ”LiF-BeF,-ZrF, (64.7-
30.1-5.2 mole %), from 35 batch containers and 520 Ib of depleted fuel concentrate, " LiF-233 UF, (73-27
mole %}, from 2 batch containers were added directly into a fuel drain tank. The addition of the enriched

 

 
 

 

37

PHOTCQ 67865

 

Fig. 30. Fluoride Production for MSRE — Transfer of Fluoride Mixtures into the MSRE Fuel System.

fuel concentrate mixture, * LiF->?%UF, (73-27 mole %), to the fuel solvent was accomplished during the
latter part of May 1965 and was coordinated with preparations for the zero-power experiments of the
reactor system.?®

Although the loading of fluorides into the reactor system was a unique operation, the techniques
involved were very similar to those used for filling the various engineering test assemblies during the
development of the MSR concept. Two portable furnace units were used for unloading the large production
batches. Each unit used a 25-kVA furnace identical to those used to heat the salt storage containers in the
main production facility. A simple gas manifold provided the necessary connections for controlling the gas
pressure over the salt mixture in the storage container. This apparatus, during the loading of the flush salt
into the reactor fuel system, is shown in Fig. 30. For this operation, the fill station was located some 15 ft
above the fuel drain tank. The flanged access point of the tank was fitted with a heated reentrant tube
which extended down to the tank cavity. A heated pipe was connected to the flanged port and was
terminated about 5 ft above the floor of the fill station. A filter of sintered nickel was connected on top of
the fill pipe and provided with a separate heater circuit. This arrangement facilitated frequent replacement
of the filter during fill operations. When a salt batch in either furnace became molten, a dip tube was
inserted in the salt container to within %, in. of the bottom. This line also connected to the top of the filter

 

26 1SR Program Semiann. Progr. Rept. Aug. 31, 1965, ORNL-3872, p. 7.
38

and was heated by a low-voltage alternating current. Transfer of the salt to the reactor was coordinated
with reactor operations personnel. The amount of salt delivered to the reactor system from each salt batch
was determined by the weight difference of the storage container before and after each transfer operation.
Although the flush and fuel salt mixtures were loaded from the same station, the same apparatus was used
in a different location for the initial loading of the secondary coolant.

Bulk additions of the enriched fuel concentrate mixture were made from the station used for the fuel
solvent. However, a smaller furnace unit was used toc accommodate the smaller baich containers. The first
major addition consisted of the transfer of about 44.17 kg of **°U from three containers. Three
subsequent additions of 2**U to the fuel solution increased its >** U inventory to 59.35, 64.42, and finaily
68.76 kg. Transfers of less than batch-size quantities of fuel concentrate mixture were made by inserting
the salt transfer line to a predetermined depth in the batch container. As planned, these bulk additions of
fuel concentrate mixture to the fuel solution increased its 223U content to within | kg of the critical
loading. The balance of 235U nceded to reach nuclear criticality and subsequent 2*° U additions were made
with the fuel enriching capsules by the MSRE operating staff.

The loading of all fluoride mixtures — cooclant, flush, fuel solvent, and fuel concentrate - into the
MSRE required 101 separate transfer operations. These were accomptlished by the fluoride production staff,
with assistance from reactor operating and maintenance personnel, in a routine manner and without
detectable beryilium contamination to the reactor facility.

9. PRODUCTION ECONOMICS

During the course of fused fluoride production at ORNL, modifications and revisions have been made
on the facility to meet specific safety requirements and to incorporate procedural changes which resulted
from process development programs. When the use of beryllium fluoride in fused fluoride mixtures became
attractive to the Molten-Salt Reactor Program, extensive modifications in the physical plant were made to
cope with health hazards which accompany the handling of beryllium compounds. The required
preparation of large quantities of identical fluoride mixtures for the MSRE made the use of a single
meltdown furnace assembly both effective and economically feasible. The installation of this unit and the
relocation of the materials handling area were the only major revisions in the production plant after iis
modification for handling beryllium salts. Six plastic fresh-air suits at about $120 each sufficed for the
entire production program and provided more positive protection of loading room operators from toxic
fluorides than was normaily attained from safety devices previously used. Other production features which
were developed for the process required only minor revision of the facility. Although the as-developed
investment in the production facility was probably excessive, its replacement value was estimated in the
range of $300,000 to $500,000.

The operating budget for the fluoride production commitment averaged about $20,000 per month and
included maintenance and expendable equipment costs. The large production facility was operated on a
seven-day three-shift schedule. Each of the four shifts required a technical operator and an assistant for
routine maintenance. Other supporting maintenance crafts and analytical services were employed as
required. Two supervisory personnel were assigned full time for the entire production operation.

Fluoride starting materials that were acquired for the production of MSRE materials had values as
shown in Table 6. The raw materials cost for the coolant and flush salt mixture, ’ LiF-BeF, (66-34 mole %),
was $11.29 per pound and that of the fuel salt, exciuding 25U costs, was $10.13 per pound. Combining
estimated operating costs, the delivered value of the coolant and flush salt mixture was calculated at $19.71
per pound and that of the fuel salt, excluding 22° U, was $17.33 per pound. Thus the total vatue of fluoride
salt mixtures required for operation of the MSRE was about $484,000, exclusive of plant amortization and
 

39

Table 6, Costs of Raw Materials Used
in MSRE Fluoride Production

 

 

. . Unit Cost Toatal Cost
Material Quantity (doliars) (dollars)
L 12,919 1b 16.504 213,164
BeF, 11,472 v 5.70 65,390
ZrF, 2,265 Ib 8.00 18,120
UF, 90 kg 12,000.00 1,080,000
(2 35y basis}

Total 1,376,674

%ncludes $1.82 per pound for preparation as fluoride salt.

Table 7, MSRE Fuel Enriching
Capsule Costs

 

ltem Cost per Cluster Cost per Capsule

 

(doliars} {dollars)
Fabrication 125 17.85
Tilling operation 77 11.60
Salt production 33 4.72
Materials costs
235F, 7320 1,045.00
TLiF 7 1.00
Total cost 7562 1,079.57
Total cost less 2>°U 2432 34,57

 

uranium costs, The estimated values associated with the preparation of the fuel enriching capsules are given
in Table 7. Since each capsule contained about 85 g of 2?°U, the preparation charges for sustaining the
fuel during nuclear operation can be assessed at $0.41 per gram of 2351,

ACKNOWLEDGMENTS

One of the primary responsibilities of the Reactor Chemistry Division, and its parent section of the
Chemistry Division, to the Molten-Salt Reactor Program has been the development and operation of a
production facility to provide various fluoride mixtures for the chemical and engineering tests of the
program. Under the directorship of W. R. Grimes, the scope of this program was enlarged and directly
associated with the general chemical development of molten fluoride systems. The successful completion of
the commitment for fueling the MSRE has been largely due to this intense personal interest in the
application of molten salts in the field of nuclear technology.

The ability to prepare molten fluoride mixtures of sufficient purity for high-temperature systems and
the techniques for the practical handling of these materials in their liquid state have been under continual
development by the fluoride production group since the inception of the MSRP at ORNL. These practices
and procedures resulted primarily from the efforts of G. J. Nessle, former production group leader, and his
associates.

The preparation of materials for the MSRE represented the largest single production effort of the
program. Although the production process had been essentially developed to its present state, the operation
of the plant at this production level and the degree of quality control demanded for the reactor experiment
40

required the individual dedication of the production group members to be successful. F. A. Doss, who had
been a member of the group since its beginning, served as the operations chief. Those who operated the
facilities were W. K. R. Finnell, W. Jennings, Jr., W. P. Teichert, and C. E. Roberts and were assisted
by R. W. Ray, R. G, Ross, and B. F. Hitch — all of the Reactor Chemistry Division. Maintenance
operations were the responsibility of the Y-12 Plant and were discharged by C. W. Mason. Members
of his staff who were 2 part of the shift operations included F. H. Deford, G. D. DeLozier, D. L.
Craig, H. V. Tripp, P. E. Barker, B. F. Butler, and P. R. Schuliz of the Y-12 Plant Maintenance
Division.
41

ORNL-4616
UC-80 — Reactor Technology

INTERNAL DISTRIBUTION

1. J. L. Anderson 22. J.H. Frye 43. A.M. Perry

2. C.F.Baes 23. C.H. Gabbard 44—68. M. W. Rosenthal

3. S.E. Beall 24. W. R. Grimes 69. Dunlap Scott

4. M. Bender 25. A.G. Grindell 70-74. }.H. Shaffer

5. E.S. Bettis 26. P.N. Haubenreich 75. M. J. Skinner

6. D.S. Billington 27. R.F. Hibbs 76. D. A. Sundberg

7. F. F. Blankenship 28. W. R. Huntley 77. R.E. Thoma

8. E. G. Bohlmann 29. W.H. Jordan 78. D. B. Trauger

9. C.J. Borkowski 30. P.R. Kasten 79. G. M. Watson

10. G.E. Boyd 31, M. T. Kelley 80. A. M. Weinberg

11. R. B. Briggs 32. J.J. Keyes 81. J. R. Weir

12. S. Cantor 33. R.B. Lindauer 82. M. E. Whatley

13. J. L. Crowley 34. M.I. Lundin 83. I.C. White

i4. F. L. Culler 35. H. G. MacPherson 84. Gale Young

15. I.R. Distenfano 36. R. E. MacPherson 85—88. Central Research Lih:ary

16. S. J. Bitto 37. H. E. McCoy 89. ORNL — Y-12 Technica! Library
17. F.A. Dos 38. L. E. McNeese Document Reference Section
18. W.P.Ea therly 39. A.S.Meyer 90--127. Laboratory Records Department
19. J.R. Engel 40. R. L. Moore 128. Laboratory Records, ORNL R.C.
20. D. E. Ferguson 41. E. L. Nichoison

21. L. M. Ferris 42. R. B.Parker

EXTERNAL DISTRIBUTION

129. M. Shaw, Atomic Energy Commission, Washington
130. T. W. MclIntosh, Atomic Energy Commission, Washington
131. D.F.Cope, Atomic Energy Commission, Oak Ridge
132. A.R.DeGrazia, Atomic Energy Commission, Washington
133. David Elias, Atomic Energy Commission, Washington
134. Norton Habermann, Atomic Energy Commission, Washington
135. Herbert H. Kellogg, Columbia University, New York 10027
136. John D. Corbett, Iowa State University, Ames, lowa 50010
137. Lloyd Zumwalt, North Carolina State University, Raleigh, N.C. 27607
138. Paul Cohen, Westinghouse Electric Corporation, P.O. Box 158, Madison, Pa. 15663
139. J. A. Swartout, Union Carbide Corporation, New York, N.Y. 10017
140. Laboratory and University Division, AEC, ORO
141. Patent Branch, AEC, ORO
142—-361. Given distribution as shown in TID4500 under Reactor Technology category (25 copies —NTIS)