OAK RIDGE NATIONAL LABORATORY
operated by

UNION CARBIDE CORPORATION
for the

U.S5. ATOMIC ENERGY COMMISSION

 

ORNL- TM-1853

COPY NO. - 282
DATE — 6-6-6T
CESTI BRRICES

H.C. J'oog MN b8

CHEMICAL RESEARCH AND DEVELOPMENT FOR MOLTEN-

SALT BREEDER REACTORS

W. R. Grimes

ABSTRACT

% Results of the 15-year program of chemical research and develop-
2 ment for molten salt reactors are summarized in this document. These
. results indicate that 7LiF—Bng-UF)_L mixtures are feasible fuels for
thermal breeder reactors. ©Such mixtures show satisfactory phase be-
havior, they are compatible with Hastelloy N and moderator graphite,
and they appear to resist radiation and tolerate fission product ac-
cumulation. Mixtures of TLiF-BeFQ-ThFu similarly appear suitable as
blankets for such machines. Several possible secondary coolant mix-
tures are available; NaF-NaBF3 systems seem, at present, to be the

most likely possibility.

Gaps in the technology are presented along with the accomplish-
ments, and an attempt is made to define the information (and the
research and development program) needed before a Molten Salt Thermal
Breeder can be operated with confidence.

NOTICE

o This document contains information of a preliminary nature and was prepared
primarily for internal use at the Qak Ridge National Laboratory. It is subject
to revision or correction ond therefore does not represent a final report, The
information is not to be abstracted, reprinted or otherwise given public dis-

- semination without the approval of the ORNL patent branch, Legal and Infor-

T ~ mation Control Department.

IfifimanHCEIHSEcufiflfiflfllmyggflfl
;j -‘A
e T

 

 

 

 

 

— LEGAL- NOTICE

‘This report was prepored as an .uccount of Govarnment sponsored work. Neither the United Statas,‘
‘nor the Commission, nor any person acting on behalf of the Commission: o L
- A. Mokes any warranty or representation, sxpressed or implied, with respect to the accuracy,

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

ony information, cpparotus, gneihod,r..or..process disclosed in this report may net infrings ..

privately owned rights; or

_B. Assumes ony liabilities with usp.ci to the use of or for domages nsulfmg from fhe use of,

any information, apparatus, method, or process disclosed in this report.

~ As used in the above, “‘person flchnq on behalf of the Commission® includes ony omployee or
contractor of the Commission, or smployee of such contractor, to the extent that such empioyce
or contractor of the Commission, or employes of such contractor prepares, disseminates, or -

provides access fo, any information pursuunt 1o his amployment or contract wnh the Commrssion,
or his employment with such contractor,” -

 

 

 

 

 
 

 

 

CONTENTS

Abstract . . ¢ & i v i e e e e e e e e e e e e e
Selection of MSBR Salt Mixtures . . . . . . . .

General Requirements for the Fluids . . . . .
Choice of Fuel and Blanket Composition . . .
Oxide-Fluoride Equilibria . . « « o« « o « . .
Fuel and Blanket Compositions . . . . . . . .
Choice of Coolant . . . . . ., . . . . . . ..

Physical Properties of MSBR Liquids . . . . .

Chemical Compatibility of MSRE Materials . . . . .

Stability of UF3 and UFy, . . . . . . . . . .
Oxidation (Corrosion) of Hastelloy N . . . .
Compatibility of Graphite with Fluorides . .
Radiation Effects e e e e e e . e e
Behavior of Fission Products in Molten Salts . . .
Physical Chemistry of Fission Products . . .
Net Oxidizing Potential of Fission Process .
Chemical Behavior in MSRE e e e s e e e e e
General . . . .« ¢ ¢ i v b 4 h bt e e e e e
Corrosion in MSRE . . . . . . . . . . + . . .
Behavior of Fission Products . . . . . . . .
Molten-5alt Production Technology . « . . . . . .
Production Process . . . .. ¢ v ¢ v v v v . o

MSRE Salt-Production Economies . . . . . . .

Separations Processes in MSBR Fuels and Blankets .

~Possible Separation of Rare Earths from Fuel

--MSBR'In-Line'Analysis Program . . . . ¢« « « « o« .

Proposed Program of Chemical Development . . . . .

Refel‘ences * e - - . . e . - - . - . . - - - .

e
— " LEGAL NOTICE
- eport was prepared as an lccounht of Govai-nment _iponsored work, Nt;lth.er the Unltqd :
This T p"r,m Commission, nor sny person acting on behalf of the Commission: o e aoen-
suw.:."n;akes any warranty or representation, expressed or implied, with “:tpez; o the hocu-
c this report,
y fulness of the information contained in ’
r:c:;yw&:::\t:‘::o:l" :pl;::nem method, or process disclosed in this report may not infringe 7
o . ) , 1 c .

) wned sor - - B o
Pfl“:lizsumosfl lf;'uuu‘;umes with respect to the use of, Tr g:; i?la‘x::.iagets pl;eriu.lfl.ng from the
‘ 'x ppAr: , ot process disclo s re .
. mtormau?n, . “. m;o:e::tfig:np;ehflf of the Commission® includes any em-
As used In e e Com: mployee of such contractor, to the oxtent that

of the Commission, or & 7
p“::!:z:l;l:;::r:iw:;n&mwr of the Commission, or employee of such contractor prepares,
su

disseminates, or provides accesd §0, any information pursuant to his employment or eont;-act
' -
with the Com'miuion. or his employma_l_:__t wlf.'f: snch contracmr n s R

ASTRIBU

21
29
30
3k
37
37
Lo
k5
L6
5T
o7
65
68
68
69
T1
82
82
86
89
89
102
120
133

TION OF THIS. DOCUMENT g UE[M‘@

 
 

 

CHEMI CAL. RESEARCH AND DEVELOPMENT FOR MOLTEN-

SALT BREEDER REACTORS

Use of molten fluorides as fuels, blankets, and coolants offers a.
promising and versatile route to thorium breeder reactors. Mixtures con-
taining fissile and/or fertile materials have been studied in consider-
able detail, and sflown to possess liquidus temperatures, phase stability,
and physical properties which are suitable for the purpose. These fluo-
ride mixtures éppear to be compatible with structural metals and with
graphite suitable for use in a Molten Salt Breeder Reactor; such compati-
bility seems assured under irradiation at MSBR conditions. Cheap, low-
melting fluoride coolants for MSBR have not yet been demonstrated but
promising leads are available; the relative simplicity of the coolant
problem lends assurance that a reasonable solution can be found.

A reference design for a 1000 MW(e) Molten Salt Breeder Reactor has
recently been,published.l The state of knowledge of molten salts as
materials for use in that reactor and in attractive alternatives or
improfiements is described in some detail in the following pages. An at-
tempt is made to define thbse areaswwhereadditional knowledge is neces-
sary or fery_desirableand_té_estimafié the effort_required'to obtain this

knowledge for a molten'saltfbreeder reactor and a breeder reactor experi-

‘ment.

SELECTION OF MSBR SALT MIXTURES®™”

General RéQfiirEments for the Fluids

 

A molten salt'reaétor'makes the following stringént minimum demands

upon its fluid fuel. The fuel must consist of elements of low (and prefer-

 
 

 

 

ably very low) capture cross section for neutrons typical of the energy
spéctrum of the chosen design. The fuel must dissolve more than the
critical concentration of fissionable material at temperatures safely be-
low the temperature at which the fuel leaves the heat exchanger. The mix-
ture must be thermally stable and its vapor pressure must be low over the
qperatihg fiemperature range. The fuel mixture must possess heat transfer
and hydrodynemic properties adequate for its service as a heat-exchange
fluid. It must be relatively non-sggressive toward some otherwise suit-
able materisl--presumably & metal--of construction and_toward some . suit-
eble modérator material. The fuel must be stable toward reactor radia-

- tion, must be able to survive fission of the uranium--or other fissionable
material--and must tolerate fission product accumulation without serious
deterioration of its useful properties.

If such reactors are to produce econamical power we must add to this.
list the need for reactor temperatures sufficiéntly high to achieve
genuinely high quality steam, and we must provide a suitable link (a
secondary coolant) between the fuel circuit and the steam system. We
mustlalso be assured of a genuinely low fuel cycle cost; this presupposes
a cheap fuel and en effective turn-around of. the unburned fissionable
material or (more reasonably) an effective and economical decontamination
and reprocessing scheme for the fuel.

If the reactor is to be a breeder we must impose even more. stringent
limits on permissible parasitic neutron captures b& the reactor materials
and provide sufficient fertile matefial either in a breeder blankét or in

the fuel (or in both). If & blanket is used it must be separated from
 

 

the fuel by some material.of very low neutron cross section.

The demands imposed upon the coolant and blanket fluids differ in
obvious ways from those imposed upon the fuel system. Radiation intensity
will be considerably less in the blanket--and markedly less in the cool-
ant--than in the fuel. Efficiency of the blanket mixture as a heat trans-
fer agent may be felatively unimportant, but a high concentration of
fertile material is essential and an effective recovery of bred material
is likely to bervital.

Choice of Ffiel and Blanket Composition
General Considerations

The compounds which are permissible major constituents of fuels or
blankets for thermal breeders are those tha# can be prepared from beryl-
ljum, bismuth, boron-11, carbon, deuterium, fluorine, lithium-T, nitro-
gen-15, oxygen, and the fissionable and fertile isotopes. As minor
constituents one can probably tolerate compounds containing the elements
listed in Table 1.

Of the known compounds containing useful concentrations of hydrogen

(or deuterium) only the hydroxides of the alkali metals, the saline hy-

‘drides of lithium and calcium, and certain interstitial hydrides (zir-

conium hydride, for example).shofi adequate thermal stability in the 1000°F
to 1300°F temperature range. [Acid fluorides (NsHFp, for example] might
be'pérmissible'in IGW“conéentrations at lower temperatures.] The hy-

drides are veryzstrong:réQucing agents and are most unlikely to be useful

components of any uraniferous iiQuid'fuel-system. Alkali hydroxides dis-

solve extremely'small Quantitieé of uranium compounds at useful reactor

temperatures and are very corrosive to virtually all useful metals at

 
 

 

such temperatures. One concludes, therefore, that hydrogen-rich com-
pounds, which might provide self-moderation to molten fuels,-are-not use-
ful in practical fuel or blanket mixtures.

The non-metals carbon, nitrogen, silicon, sulfur, phosphorus, and
oxygen each form only high melting and generally unsuitable binary com~
pounds with the metals of Table l. From these non-metals, however, a
wide variety of oxygenated anions are available. Nitrates, nitrites,
sulfates, and sulfites can be dismissed as lacking adequate thermal sta-
bility; silicates can be dismissed because of undesirably high viscosi~
ties. Phosphates, borates, and carbonates are not so easy to eliminate
without study, and phosphates have, in fact, received some attention.

The several problems of thermal stability, corrosion, solubility of urani-

un and thorium compounds, and, especially, radiation stability would seem

b

to make the use of any such compounds very doubtful. - L
When the oxygenated anions are eliminated only fluorides and

chlorides remain. Chlorides offer mixtures that are, in general, lower

melting than fluorides; in addition UCl3 is probably more stable than UF3

with respect to the analogous tetravalent compounds. For thermal reactors,

fluorides appear much more suitable for reasons which include (1) useful-

ness of the element without isotope separation, (2) better neutron

economy, (3) higher chemical stability, (l4) lower vapor pressure, and

(5) higher heat capacity per unit weight or volume. Fluoride mixtures

are, accordingly, preferred as fuel and blanket mixtures for thermal

reactors. The fluoride ion is capable of some moderation of neutrons ;

this moderation is insufficient for thermal reactors withrcores'of reason-— —

able size. An additional moderator material is, accordingly, required.

*
 

ey

2

Table 1.

 

Elements or Isotopes Which

May be Tolerable in.High Temperature Reactor Fuels

 

Absorption Cross Section

 

Rubidium

Material
- (Barns)
Nitrogen-15 0.000024
Oxygen 0.0002
Deuterium 0.00057
- Carbon 0.0033
Fluorine 0.009
Beryllium 0.010
Bismuth 0.032
Lithium-7 0.033
Boron—li 0.05
Magnesium 0.063
Silicon 0.13
Lead 0.17
Zirconium 0.18
Phosphorus 0.21
Aluminum 0.23
Hydrogen 0.33
Calcium 0.43
Sulfur 0.49
‘Sodium 0.53
Chlorine-37 0.56
Tin 0.6
Cerium 0.7
0.7

 

 
 

7
G

Choice of Active Material

Uranium Fluoride. -~ Uranium hexafluoride is a highly volatile com-

 

pound clearLy unsuited as a.componept of high-temperature liquids.'_UOQFg,
though relatively nonvolatile, is a strong oxidant which should prove
very difficult to contain. Fluérides of pentavalent uranium (UF5, U2F9,
etc.) are not thermally stable and should prbve prohibitively corrosive
if they could be stabilized in solution. |

Uranium tetrafluoride (UFh) is a relatively stable, non-volatile,
non-~-hygroscopic material, which is readily prepared in high purity. It
melts at 1035°C, but this freezing point is markedly depressed by several
useful diluent fluorides. Uranium.trifluoride (UF3) is stable, under
inert atmospheres, to temperatures above 1000°C, but it disproportionates

at higher temperatures by the reaction <
o
hUF3 S 3UF, + U

Uranium trifluoride is appreciably less stable in molten fluoride solu-

4,5

tions. It is tolerable in reactor fuels only insofar as the equilibri-

un activity of uranium metal is sufficiently low to avoid reaction with

2

the moderator graphite or alloying with the container metal. Apprecisable

concentrations of UF3 (see below) are tolerable in LiF—BeF2 mixtures such
as those used in MSRE and proposed for MSBR. In general, however, urenium

tetrafluoride must be the major uraniferous compound in the fuel.

Thorium Fluoride. - All the normal compounds of thorium are quadri-

 

valent; ThF, (melting at 1115°C) must be used in any thorium-bearing
fluoride melt. Fortunately, the marked freezing point depression by use-

ful diluents noted ebove for uranium tetrafluoride applies also to thorium \
 

 

)

tetrafluoride.
Choice of Fluoride Diluents

The fuel systems for thermal reactors of the MSRE and MSBR types
require low concentrations (0.2 to 1 mole %) of uranium, and the proper-
ties (especially the melting temperature) of such fuels will be essential-
ly those of the diluent mixture. Blanket mixtures (and perhaps fuel
systems for one-region breeders) will require considerable concentrations
of high-melting ThFh. The fuels must, if they are to be compatible with
large steam turbines, be completely molten at 9T5°F (525°C).

Simple consideration of the nuclear properties leads one to prefer

T

as diluents the fluorides of Be, Bi, 'Li, Mg, Pb, Zr, Ca, Na, and Sn (in

that order). Equally simple considerations (Table 2) of the stability of
diluent fluorides toward reduction by common structural metals,6’7 how-
ever, serve to eliminate BiFB, PbF2, and probably SnFé from consideration.

No single fluoride can serve as a useful diluent for the active
fluorides. BeF2 is the only stable compound listed whose melting point
is close tqthe required level; this compound is too viscous for use in
the purerstate.

The:very sfiable flubridés.éf the alkaline earths and of yftrium and
cerium do not seem to be useful major constltuents of low melting fluids.
Mlxtures contalnlng about 10 mole % of alkallne earth fluorlde with BeF2
melt below 500°C but the v180051ty of such melts is certalnly t.oo hlgh
for serlous con31deration.,_,__;:- 

Same of thepossible_pofibifiations'of'aikali fluorides have suifable;
fféézing poinfis.8> Eqfiimdlér'miitures of LiF and'KF méit-at:h90°c, and

mixtures with 40 mole % LiF and 60 mole % RbF melt at 470°C. The ternary

 
 

 

 

10

Table 2. Relative Stabilitya of Fluorides

For Use in High Temperature Reactors .

 

 

Free Energy . Absorption Cross .
Compound of Formation Melting Section®? for
| at 1000°K Point -~ Thermsal Neutrons
(kcal/F atom) (°c) (barns)
Structural Metal
Fluorides
CrFo -Th 1100 3.1
FeF»> -66.5 930 2.5
NiF5 -58 ' 1330 4.6
Diluent Fluorides
CaFo -125 1330 0.43
LiF ~125 848 | 0.033°
BaFo -124 1280 | 1.17
SrFo -123 1400 1.16
CeF3 © -118 1430 0.7
YF3 -113 11hk l1.27 .
MgFo -113 1270 0.063
RbF =112 192 0.70
NaF -112 : 995 0.53
KF =109 856 1.97
BeFo -104 548 0.010
ZrF), -94 903 0.180
AlF3 -90 1404 0.23
SnFo -62 213 0.6
PbFo -62 850 0.17
BiF3 =50 127 0.032
Active Fluorides
ThF), -101 1111 -
UFy, -95.3 1035 -
UF3 -100.4 1495 -

 

BReference state is the pure crystalline solid; these values are, accord-
ingly, only very approximately those for solutions in molten mixtures.
bof Metallic ion. 7

CCross section for 'Li.

L L)

. O
 

&Y

0

 

")

(.a! P

L)

A

4/

7,-eutect1c (52 mole % BeF

11

[T

~

systems LiF-NaF-KF andeiF-NaF—RbF have lower melting regions than do
these binaries. .All these systems will dissolve UFh at concentrations up

to several mole % at temfiefatures below 525°C. They might well prove use-

ful as reactor fuels if no:mixtures-with more attractive properties were

available. .

MixtureS‘with-useful melting points over relatively wide ranges of

camposition are avallable if ZrF, is a major component of the system..
L .

- Phase relatlonshlps NaF—Zth system show low melting p01nts over the

interval yo to 55 mole %_ZrFu. A mixture of UFh with NaF and Zth served

- as fule for the Aircraftheaetor Experiment.

The lowest melting binary.mixtures of the usable diluent fluorides

" are those containing BeF, with NaF or LiF.8 (The ternary system

L1F—NaF—BeF2 has been examlned in some detail, but 1t seems to have no

1mportant advantage over e;ther blnary.) Slnce Be offers the best cross

section of the diluents (and'TLi_ranks very hlgh), fuels based on the

o
The binary System.LiF-Ber has melting points below 500°C over the

LiF-BeF, diluent system weme‘chesen,fqr'MSRE and are prop0sed for MSBR.

concentration range from 33 to'80 mole'% BeF2 8 The presently accepted

LiF--BeF2 system dlagram, presented in Fig. 1, is characterlzed by a 31ngle

,_meltlng at 360°C) between BeF and 2LiF-BeF

2 2 2°
- The cempound 2L1F BeF2 melts 1ncongruently to LiF and llquld at h58°C.
-7L1F BeF2 is formed by the reactlon of solld BeF2 and solld 2L1F BeF2 be-
low 280°C. | |

 
 

 

ORNL-DWG 66-T7632

 

900

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig.

BeF, (mole %)

l. The System LiF-BeF

2

848
800
700 N
LIF+LIQUID
S -
o
* 600
s \ 555 <_ -
x
2 ™
2 500 ' e A
- 458 . - \
: ~ fier-(HIGH QUARTZ TYPE)
\ / +LIQUID
400
\\/ 360
LiF+2LiF-BeF, ‘ - '
2 2LiF-BeF, +BeF, (HIGH QUARTZ TYPE)
| |
300 - = [ 1 280
\ B|  2LiF-BeF, & . LIF-BeF, + BeF, (HIGH QUARTZ TYPE)
w -+ _° @] LiF-BeF, + BeF, (LOW QUARTZ TYPE)\_ 220
»o0 = LiF-Befp & i i , -[ <
LiF 10 - 20 30 40 50 60 70 80 90

A

.,

J )
(h

¥

0

[ =
 

")

L

"}

&/

 

13

" LiF-BeF, Systems with Active Fluorides

2

The phase diagram of thetBeF2-UFh system (Fig. 2) shows & single
eutectic containing very little UFh..9 That of the LiF—UF:h system (Fig. 3)
shows.three-oompounds, mOne of which melts congruently andone of whioh
shows a low temperature_limittof st&bility.lo The emteetic mixture of
hLiF-UFh'end LiF-UF) occurs at 27 mole'% UF), and melts at L90°C. ‘The ter-
nary system_LiF—BereUFh,lof primary importance in_reactor fuels, is shown
4.2 |

as Fig. The system shows two eutectics. These are at 1 mole % UF),

; they melt at

and 52 mole % BeF 3

o and’ at 8 mole % UF) and 26 mole % BeF

350 and L435°C, respectively. Moreover, the system shows a very wide range’

-

of compositions melting'below 525°C.

The system BeF -ThF is very similar to the analogous UF system.ll
L b

‘The L1F-ThFh system (Flg. 5) contalns four compounds.12 - The compound

3LiF- ThFh melts congruently at 580°C and forms eutectlcs at 570°C and 22
mole % ThFh and 560°C:and 29 mole.% ThFh with LlF and Wlth LlF'ThFh,

respectively. The compoundsLiF'ThFh,-LiFfQTth,_and.LiF“hThFh melt in-

congruently at 595°C and—890°C; The ternary system LiF-BeF —ThFh (see

F1g._6) shows only a 51ng1e eutectlc with the comp051t10n hT 0 mole % LiF
11

etand 1. 5 mole % ThFh meltlng at 356°C._ In splte of small dlfferences due

to the phase flelds of L1F 2ThFh, 3L1F ThFh, and hLlF UFh’ the systems

represented by Flgures h and 6 are very similar.

 
 

TEMPERATURE (°C)

14

ORNL-LR-DWG 28598A

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

URg (mole %)

Fig. 2 The Sysf.em UFh-Ber

1100
- e
1000 LIQUID ]
- ./. 1.
900 , ]
| —*" |
.‘.-—"
; ‘.,Cf'.
800 oo
leet*® (
- |UFy +L1QUID
700 &
| .
J
600 H
é((.t.-(((.‘l.*.(.—.".—.—#—.——O——.——i——.-—? ' ? -
500 | QpygHBeFo +LIQUID
- e QnigHBeF2 +UFg
200 L— L
~ BeF, 40 20 30 40 50 60 70O - 80 90 UF,

18y

0

-8

B
o)

%)

"

15

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

L ORNL-DOWG 17457A
1100 —
1000 //

900 | ,/
o
<
& 800 :
o T /| .
= 700 \ —
71
F. -
. 600 N—— /
. L - . ¢
. ) w <
500 v/ > 5
w w
| 4LiF-__UF4/ ~ - -
LiF 10 20 30... 40 50 60 . 70 80 . 80 UF,
: . UFg (mole %) L
_ Fig. 3. The System LiF-UF)

 
 

e

ALL TEMPERATURES ARE IN C

E
P

EUTECTIC
PERITECTIC

H

@ = PRIMARY PHASE FIELD

LiF - 4UF,

 

LiF -~

848

T

16

 

ORNL—DWG 66-7634

 
     

 

 

 

14)

G

L

L
 

¥)

*f‘

" as blanket.cempbsitidns_ ':T'-“

__' L 17

ThFh‘and UF), form & continuous series of solid solutions withrneither

-maximum nor minimum. The L__iF.—ThFh-UFh system (see Fig. T) shows no ter-

nary compounds and a sifigie eutectic13 (which contains 1.5 mole % ThF),

with 26.5 mole % UF), and freezes at 4L88°C). Most of the ares on the dia-

gram is occupied by primarytphase'fields of the solid solutions UFh‘ThFu3

LiF-hUFh-LiF;hThFh, and LiFfUFL—LiF'ThFh. Liquidus temperatures decresse,

generally, to the LiFeUFh edge”Of'the diagrsm;;

' It is clear from exemination of the diagrams shown that fuel systems

- .melting below 500°C are available over a wide range of compositions in the

LiFTBéF2—UFh system. Since (see Fig. 6) up'ts 28 mole % of ThF) can be
melted‘at-?emperatures below-1100°F,.blanket systems with very large ThFh
concentrations can be obtaiped.. Moredver, the very-greet similerity in
behavior of ThF) and UF) permits fractional replaeement of ThF) by UF) with
little.effect‘on-freezing'temperature overlthe cofipositiofi range of .

interest as fuel. Fuels for 51ngle reglon reactors should accordlngly,

;be availeble in the L1F—BeF —ThFh-UFh quaternary system.

Phase behav1or in the ternary systems LlF»BeF _UFh and LlF—BeF -ThFh

has, as_a consequence,of stud1es c1tedrabove, been examlned'ln_con51derable

'detail'and‘the phase diegrams_sre well'defined, If,‘asis'likely, fuels

and'blankets_for two~regien_breeders can be choseh'frem_these ternaries |

'--lthes theonly.fieeessary ad@itienal‘stu&y ef phase behavieris a more de;
_ffieiled exafiination ef.liéuidusiand'espeeially of~solidusrelatiopships\ahd.

crystellization.peth-behavier:in the regions near those chosen as fuel and

e e i e an  mam b v e et e e

 
 

 

 

 

 

 

 

 

 

TEMPERATURE (°C)

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ORNL-LR-DWG-265358B
1150
1050 // |
950 7
850 e / -
750 AN ‘ L/
650 o
\ / <
-
<
550 , -
3LiF- ThF, — LiF-ThF, —s={ LIF-2ThF, —
w0l | |1
LiF 10 20 30 40 50 60 70 80. 90  ThF,
- ThE, (mole %) - L

Fig. 5. The System LiF-ThF)

 

)]

i b
et ket

 

"

d“f"

 

7_19_

' . ORNL-LR-DWG 37420AR4
L o Th, 4444 |
‘ 1o TEMPERATURE IN °C
, COMPOSITION IN mole %o
1
[ 1050
. LiF-4Th, ’

1000

950

 

LiFthg A I\ T 900~

P 762 850

P 597
£568
3LIF-ThE
% »
£ 565 "%
700

650
~600 _

WX

0 ,
- , __,\’z%g‘:?oo‘%\ T _
| 80, 2 mo ) 550—— , 526
LiF /N BeF,
848 . 2UiFBeR;500450  400] 400 450 - . 500 . sgs
' ' , P4a58 £360 : -

 

~ Fig. 6. The System LiF-BeF,-ThF,

g

 
 

-y

LiF

. 848

N
P 597 A

£ 568

3LiF'ThF4

E 565 Z\

LWQ 06(‘%;"
% AN

20

©uy

   
   
  
 
   

PRIMARY-PHASE AREAS
{a) UF,~ThE, (ss) |
(b) LiF-4UR,~LiF-4ThF, {ss)
{¢) LiF-2Th(U)F, (ss)

~ TEMPERATURE IN °C,
COMPOSITION IN mole 7%

| LiF-4ThE, (d) 7LiF-6UF,-7LiF-6ThF (ss) -
(e) 3LIF-Th(UIF, (ss) :
(f) LiIF -
LlF'ZThF4 —
P BIT A
. LiF-ThE
PT62 A

, o
ALf) \, 3

 

h \/ .
4LiF-UF;” P5O0" '£490 P 610 { pis . LiF-4uF,
| ,  \LiF-UR, |
Fig. 7. The System LiF-ThF)-UF)

- - k

ORNL-LR-DWG 28215AR4 -

v,
1035,

)

3

R
 

21

Oxide-Fluoride Equilibria

The phase behavior of pure fluoride systems is such that adequate
fuels and blankets seem assured, but the behavior of such systems is mark-
edly altered by appreciable concentrations of oxide. Since all commercial
fluoride prepafations confiain some oxide (and water which reacts with the
fluorides at high temperature to produce oxide) methods must be devised
to remove this contaminant to safe levels before use of the fluoride mix-
ture in the reactor. Avoiding contamination by oxide of the molten mix-
tures during reactor operation and maintenance was possible in principle
but, before the excellent operating experience with MSRE, was not at all
certain in practice. Accordingly, careful studies of oxide-fluoride equi-
libria in fluoride melts have been made to establish (1) the effect of
contaminant oxide on MSRE fuel, and (2) the ease of removal of oxide to
tolerable levels prior to reactor usage of the melts.

Measurements of reaction equilibria between water vapor in hydrogen

carrier gas with LiF-BeF_ melts over a wide composition interval have been

2
. . . 14,15 e s .
examined in detail by Mathews and Baes. Equilibrium quotients for
the reaction
H,O + BeF,.,,y ¥ BeO + 2H 1
2%(g) T PFo(e) T BeO(s) T () (1)

(where %, g, and 4 refer to_liquid, gaseous, and dissolved states and s
indicates that Be0 was present as a saturating solid phase) were measured
- from 500 to TO0°C over the composition range,xBeF = 0.3 to 0.8. The
results are summarized by |
) =a+b x2 + ¢ xh

5 |
1og (PHF/PHEO *BeF,, LiF rip (2

wherein a, b, and ¢ all were linear functions of 1/T°K,

 
 

 

 

22

3.900 - h.h18(1o3/T),

o
i

b

fl

7.819 - 5.hh0(1o3/T),
¢ = -12.66 + 5.262(10°/T).
In the same investigation, measurements were made upon melts not saturated

with Be0. In addition to the reaction

- o
H.O + 2F Z0
(g) (

5 )+ 2HF(g) - (3)

d) (a
for the formation of oxide ion, it becsme evident, both from these measure-
ments and from those upon Be0 saturated melts, that hydroxide ion also was
formed
0 + F * OH™
1% T ( (

a) + HF( (L)

a)
Because of limitations inherent in the transpiration method used, the
equilibrium quotients for these two reactions were less aécurately deter-
mined than was the previous one for Be(Q saturated melts (ca. i_;O%,
respectively, compared to *+ 5%). They were sufficient to show, however,
that both oxide and hydroxide increase in stability with increasing tem-
perature. The stability of hydroxide with respect to oxide, however,
decreases with increasing temperature. Hydroxide can, accordingly, be
readily decomposed in these fluoride melts by sparging with an inert gas
(e.g., hydrogen).

OH + F Z‘HF(g) + 02‘, log K = 5.23 - 6.56(103/T) (5)
16

Similar measurements have also been made by Baes and Hitch ~ in

which the 2LiF-<BeF, contained added Zth. With x__ >~3 x lO-h, Zr0

is the stable saturating oxide aolid, and hence the following equilibrium

may be written

o

»
 

"

1y ¢

‘ff

'mlxtureS'of H .and HF. The measured equlllbrlum quotlents in 2LiF- BeF

 with apprec1able quantltles of a reactlve oxlde (such as H 0 (oo}

 

23

B0 * ey * U0x(e) * MF(g) - (6)

It was also found that’the ‘equilibria (3 and 4) for the formation of oxide

and hydrox1de ions were shlfted to the right with 1ncrea51ng x y 1.e.,

ir Fh

in the direction of greater stablllty of these ions.

- These results are con51stent with preV1ous dbservatlons that L1F-BeF2 -
melts are readlly freed of ox1de contam1natlon by treatment with gaseous

2 2

were used to calculate the eff1c1ency of HF ut1llzat1on in such a treat-

ment as a functlon,of temperature and HF partial pressure w1th the assump- |

tion that equlllbrlum is malntalned between the gas stream the molten

salt, and any BeO SOlld present. _This calculatlon (Flg. 8) shows that the

eff1c1ency in the removal of oxide to a flnal value of 16 Ppm

(XO24‘— 3.3 x 10 5) is qulte hlgh over a W1de range of condltlons.

By combination of reactlons (l) and (6), 1t is poss1ble to calculate

that both BeO and ZrO2 w1ll coex1st at equlllbrlum w1th 2L1F BeF2 contain-

ing oxlde ion o S . :
‘ o oml2E g M opag
Zr02(s)_+ QBe (d)f+ Zr gyt ZBeO(.) - _(T)

,when Zth is present at concentratlon of approxlmately 3 x 10 =k mole

*

_”fractlon. With 1arger amounts of added Zth, Zr0 becomes the less soluble

2

('s_ta‘ole) ox1de._, '

When'a molten mixture centaining only IiF, BeF2, and UFh is treated
3 FeO)

'”,preclpltat1on of U0, resul‘l‘.s.l‘L-’5 The uo, so produced 1s st01ch10metr1c,

.. and 1f-1t is ma;ntalned,ln;ccntact with the melt for.suff1c1ent time 1t

Such precipitation has been

forms'transparent ruby crystals_of uo,, 00"

 

 

 

 
 

2

 

S
o

. ORNL-DWG 65-4187
“INFLUENT HF PREs;SURE; (atm) |
00 1 —0.2-g 11+
200.C %1 1505

 

 

 

 

 

(o_
o

 

 

e |
//’.-—‘ . e

 

A\

 

o
O

 

T T = |
/ GOOOC //// | 0.2/______..—--—-— .

 

 

 

-~
O

 

 

y T P RS
. | T ’/////”/’””"‘_Sll?—‘“"—__-_—

 

o
o

 

 

 

  

AL~ .00 -

700°C -

 

HF GONVERTED TO Hp0 (%) -
o ‘
o

/

 

 

H
O

 

W
o

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

-t
ST
(VR

2 -5 101 2 5 100
INITIAL OXIDE GONTENT ( mole /kg )

Fig. 8. Efficiency of Removal of Oxide -
from 2LiF-BeF, by Treatment with HF
 

to publlshed UO -ZrO

 

. =

assumed to present\a danger;for the MSRE since slow precipitation of UOé
followed-by'a sudden'entranee'offthe material into the core could permit

. ' o S
uncontrolled increases in reactivity. Precautions were taken with the

MSRE to assure cieanlineSS"of:the'system, the fuel mixture, and the cover

gas, but it was anticipated,that“some inadvertent contamination of the

_system might occur. Accordlngly, 1t waes dec1ded to 1nclude Zth in the

MSRE fuel compos1tlon since measurements of the metathesis reactlon

Zroz(s) + UFh( ) z Zth(d) + UO o(s) (8)

have shown that theimole'ratio of.Zth to UF, at equilibrium with}both

UO2 and Zr02, while varylng somewhat with temperature and melt composition,

remains very far below that chosen for the fuel salt._ As atconsequence a

+ _ o
cons1derable-amount—of Zr#_-—an amount eas1ly detected by chemical analysis

of the fuel salt-—would be prec1p1tated by oxide contamlnatlon before an

apprec1able quantlty of UO should prec:Lpltate.l{’5 ;
In connectlon w1th these studles, it was ascertalned that, contrary

5 phase dlagrams;rronly very dllute SOlld solutions

are formed in the temperature range 500—700°C, Because of the obvious
'ilmportance of thls to the MSRE experlments have been carrled out in which

_ both UO »ZrO mlxtures and (U Zr)O solld solutions prepared by fus1on

2 2

r'were equlllbrated w1th L1F—BeF melts. .The resultlng phase dlagra'ml8 for

2

the UOE-_--ZrO.2 system over thejtemperature interval of-realaconcern.is shown

~in Fig. 9.

. The oxide concentratlon 1n 2L1F BeF2 saturated w1th BeO was estimated

by comblnlng the equlllbrlum results for reactions (l) and (3) to be

: 108_X02;'=§0.0h5»2,95.x 103/T. . . | ‘(9)

 
 

TEMPERATURE (°C)

99.5

26

MOLE PERCENT UO3

 

 

        
   

  

 

    

 

 

 

 

 

 

 

 

ORNL-DWG 66-963

0 05 0O

 

 

 

 

 

 

100 99.0 985 10090 80 70 60 50 40 30 20 10 O 15
, T T 1 T 1 TF T ¥ T 71 1 T T
ugo |- CUBICs.s. —{\es00 LIQUID 2800/~ TETRAGONAL s.s.  —11180
: © TETRAG.ss.
140 - 00 cue1i: s.s. —2400 2400 [= N MONOCLINIC s.s. 140
o TETRAG. s.5. \ CUBIC s.s. / o
400 |- 2000 |- : 42000 |- : — #00
CUBIC s.s. , TETR‘:_G- s.s. 0 .
! \ TETRAG. / CUBIC s.s. &
1060 |- — 1600 w00 - —1060 W
: CUBIC s.s. \ . CUBIC 5.5, N | A cuscss. 2
+ ’ . : + -
] MONOCLINIC s.s. l' TETRAGONAL s.s. || MONOCLINIC s.5. =
1020 — 200§ ) : A 1200 }- © wald1020 ¥
o > X . =
{ i “%q| £
= CUBIC s.s. ' | S \
980 -+ st i . cueic I ,/, goo__| 1 980
| I MONOCLINIC s.s. ‘ | II' | MONOCLINIC s.5.—-|
940 4 400l | b/ [Ha00 : H 940
il |'£ i '
I _
900 L i Ca a7 l'! 1 i 900
o0 0.5 .0 5 O 10 20 30 40 50 60 70 BO 90 100 98.5 99.0 100

S

Fig. 9. Phase Behavior in System U02'--Zr02

MOLE PERCENT Zr0»

LEGEND:
LIQUID-LAMBERTSON AND MUELLER {23)
- {ABOVE 1600°C)} COHEN AND SCHANE
——— e {BELOW 1600°C} ] (16} -
-+— MUMPTON AND ROY {22) .
{BELOW 1600°C ) ORNL PROPOSED

 

 

99.5

O "WET" CHEMICAL ANALYSIS
X ACTIVATION ANALYSIS FOR U

o

C

M
 

C

f

ty product of Zr0, could be-estimated. Wlth 1ncreas1ng X

r-comp031tlon on the act1v1ty coefflclents of the spec1es Zrh+ and O

-'fuel composition (x

 

_ o7
. The solubility 1ncreased vith temperature but no strong dependence on
KB F ‘was found In these measurements the mole fractlon of oxide at BeO
saturation probably was less than O. 002.

From the  similar measurements in Zth-containing meltsl‘ the solubili-

s the concen;

2 Zr Fh

Jtration of oxlde at ZrO saturatlon at first fells as would be exPected

from the equilibrium

Zr02(s) TIr Ty v RO g (10)

However, it then levels off and subsequently rises with,further increases

4 1

. in Xy @ (Fig.- 10). -This‘could be caused, at least in part, by:the forma-
, L 24
tion of a complex ion, Zr02_; S
B Zru+5-: + O z ZrO2+ (ll)
S B PO (g FEO (g |

or it could be caused entlrely by the influence of the changlng melt
2-

- The plot. in Fig 10 indlcates approx1mately the ' ox1de tolerance of MSRE .
fuel salt-flush salt mixtures; i.e., the_amount,of dissolved oxide these
‘mixtures can contain without oxide‘precipitation. It is seenrthat the

'ox1de tolerance 1ncreases rapldly w1th temperature, espec1ally near the

70 F 0 05), 1nd1cat1ng that any excess ox1de present
Lo

,fmight be removed by collecting ZrO on a relatlvely cool surface in the

VMSRE system.

These studies have deflned relatlvely well the 51tuat10n 1n L1F-BeF

| l;and in L1F-BeF -Zth melts." They have been of real value 1n assess1ng

fthe 1n1t1a1 purlflcatlon process (see ‘below) and in assurlng the 1nad—

vertent precipitation of-UOzfshould prove no problem 1n'MSRE.

 

 

 
 

OXIDE CONCENTRATION {mole/kg)

N

N

28

ORNL-DWG 65-2542R

 

600°C _
| -BeO SATURATION

l -
/ | ZrO, SATURATION
l . .

001 002 005 04 0.2 05 {4 - 2
ZrF, CONCENTRATION (mole/kg)

Fig. 10. Solubility of ZrO, in LiF-BeF,-ZrF)
Melts as Function of ZrF), Concentration -

-
 

29

Fuel and Blanket Compositions

T

LiF, BeF Zth and UFh

- The fuel chosen for MSRE was a mixture of o
consisting of 65-29.1-5-0.9 mole %, respectively, of these materials. The
Zth we:: added, as indicated above, to eliminate the possibility of pre-
cipitation of UO2 through inadvertent contamination of the system with
reactive oxide. [The general precautions regarding cleanliness in MSRE
and the apparent success of the fuel preparation and handling procedures
for that‘operation_have gone far to remove apprehension from this source.
No samples removed from MSRE have contained more than 100 ppm of oxide,
" and no precipitated oxides have been observed on examination by optical
.microscopy,] Since chemical reprocessing techniques (probably distilla-
tion)‘will certainly be applied to the MSBR fuel system and since such a
reprocessing scheme can be éxpected to remove oxides, it seems very likely
tha£ Zth,need fiot be a constituent of MSBR fuel.

On the basis of information presented above the reference fuel select-
ed for use in the MSBR is a ternary mixture of TLiF-BeF2-233UFh (68.3-
7'31;5—0.2.mole %) (see-Fig. k) which exhibits a liquidus temperature of ap-
proximately‘h50°c., Equilibrium crystallization of this fuel mixture
proceeds according to the follofiing sequence: On cooling in the tempera-
ture interval 450 to h38°C,.éLiF-BeF2 is deposited from the fuel. At
438°C, the salt mixtfireSolidifiés éhd produces a mixture of the two
éfYStalline phaées, 2LiF-BeF2 and LiFfUFh, comprised of approximately 89
, and épproximétely 11 wt % LiF-UF) .

wt % 2LiF:BeF
" The blanket salt selected for the MSBR is the TLiF--BeFe—Th_Fh-ternary

mixture (71-2-2T7 mole %) (see Fig. 6), which exhibits a liquidus tempera-

ture of approximately 560°C. Equilibrium crystallization of this blanket

 
 

 

30

mixture is as uncomplicated as that of the fuel. Only the two solid
phases, LiF-ThFh and a solid solution of 3LiF-ThFh which incorporates Be2+
in both interstitial and substitutional sites, are formed during solidifi-
cation, and these solids are coprecipitated throughout the crystallization
of the salt.ll
Choice of Coolant

The secondary coolant is required to remove heat from the fuel in the
primary heat exchanger and to transport this heat to the power generating
system, Ifi the MSBR the coolant must transport heat to supércritical
steam st minimum temperatures only modestly above TOO°F; in MSRE the heat
was rejected to an air cooled radiastor at markedly higher temperatures.

The coolant must be possessed of adequate heat transfer properties
and must be compatible with Hastelloy N structures. It should not react
energetically with fuel or with steam, it should consist of materials
whose leskage into the fuel would not necessitate expensive separations
procedures, and it should be relatively inexpensive. To assure easy com-
patibility with the steam generation circuit the melting temperature of
the coolant should be below (and preferably considerably below) TOO®F.
Other demands (especially in the neutron economy and in radiation stabili-
ty areas) are clearly less stringent than those upon fuel and blanket
mixtures.

The coolant mixture chosen for MSRE and apparently shown to be satis-

T

factory in that application is BeF, with 66 mole % of 'LiF. Use of this

2
mixture would require some changes in design of equipment for the MSBR
since its liquidus temperature is 851°F; moreover, it is an expensive

material. The eutectic mixture of LiF with BeFs, (48 mole % LiF) melts at
 

‘duced and, accordingly, corrosive or chlorides [AlCl

31

near TOO°F (see Fig. 1) but it is both viscous and expensive. The alkali

- metals, excellent coolants with real promise in other systems, are un-

desirable here since they react vigorously with both fuel and steam. Less
noble metal coolants such as Pb° or Bi° might be tolerated, but they may

not prove compatible with Hastelloy N.
19

Several binary chloride systems are known ™~ to have eutectics melting

below (in some cases much below) TOO°F. These binary systems do not, how-
ever, appear especially attractive since they contain high concentrations

of chlorldes [T1cl, ZnCl BiCl CdClE, or SnClE], which are easily re-

2? 2°

3» ZrClh, HfClh, or
BeCIQ];Which'are very volatile. The only binary systems of stable, non-
volatile chlorides are those.containing LiCl; LiCl-CsCl (330°C at 45 mole
% CsCl), LiCl-Kcl (355°C at 42 mole % KCl), LiC1l-RbCl (312°C at 45 mole %
RbCl). Such systems wouldrbe relatively expensive if made from TLiCl, and
they could lead to sérious contamination of the fuel if normal LiCl were
used.

~ Very few‘fluorides or mixtures of fluorides are known to melt at tem-

peratures below 370°C. Stannous fluoride (SnF ) melts at 212°C. This

material is probably not stable durlng long term service in Hastelloy N;

moreover, its phase diagrams W1th stable fluorldes (such as NaF or KF)

| probably show high meltlng p01nts at relatlvely low alkali fluorlde concen-—

tratlons.

Coolant mlxtures of most . 1nterest at present are those based on
fluoborates of the alkall metals. The blnary system NaF-NaBFh is describ-
eat? 2 ‘as having a eutectlc (at 60 mole % NaBFh) meltlng at 580°F. Pre-

llmlnary unpublished studles at this Laboratory suggest strongly that this

 
 

 

 

 

32

published disgram is in error, and that the NaF—NaBFh.eutectic melts at
near T16°F. There is same evidence to suggésf that boric oxide sub-
stantially lowers the freezing point of NaF—NaBFh mixtures and we believe
that the Russian workers may have used quite impure materials. It is like-
ly,'howéver, that the material (pérhaps even with a moderate amount of
1_3203) may be useful. It should prove sufficiently stable to radiation

for service as coolant, and the equilibrium pressure due to

NaBFhl z BF3 + Nan

shofild prove satisfactorily low. Estimates of the heat transfer.énd fluid
properties of this material appear attractive. The extraordinarily high
cross section of boron should permit small lesks in the heat exchanger to
be recognized immediately, and removal of traces of BF3 from the fuel by
continued treatment with HF should be possible. Compatiblity of the
NaF—NaBFh mixture with Hastelloy N will probably be satisfactory (see sub-
sequent sections), but such compatibility remains to be demonstrated.

If the NaF—NaBFh eutectic system proves unsuitable by virtue of its
freezing point, preliminary data (see Fig. 11) suggésts that freezing
points below TOO°F can be obtained in the ternary systemNaF-KF—BFB.

Should experience prove the NaF—NaBFh mixture (or its closerelatives)

unsuitable, coolant compositions which will meet the low liquidfis tempera~
8,19 or

9

ture specification may be chosen in the NaF—BeF2 .NaF-LiF—Ber,
KF—Zth—AlFSQ:L systems. These materials are almost certainly compatible
with Hastelloy N, and they possess adequate specific heats andrlow vapor
pressures (see section below). They (especially those including LiF) are

moderately expensive, and their viscosities at low temperature are certain-

O
 

a5

  

 

33
“ i
’
{
ORNL-DWG 66-7633
- BFy—127
\ TEMPERATURE IN °C
COMPOSITION IN MOLE %
. +H-+H+ DENOTES SOLID SOLUTION
}
) : ’ S ¥, '
"NoF - \/ \F \/ - \// \/ - KF
995 : R L FT0 . . 856

Fig. _ll,.':k The’-é-y_stém'NaF—KF-BF3 (Preliminary) O

4 ‘;"

 
 

 

Physical Pfdpprties of-MSBR Liquids

3k

Estimates of some of the physical properties of the'proposed MSBR

possible secondary coolants are given in Table kL.

‘blanket and fuel salts are listed in Teble 3. Estimated values for four

Table 3. Composition and Properties of Fuel and Blanket Salts .

 

 

 

 

Composition - _
(mole %) - Fuel Blanket
i - LiF  65.9 LiF .= Ti
BeFé 33.9 ThFh 2T
| UF,, 0.2 BeF,, 2
Liguidus Temperature:
°c L5T 560
oF 855 1040
Physical Properties: At 600°C At 600°C
1112°F 1112°F
Density, lb/ft3 125 280
Heat Capacity, Btu 1w~L(oF)~t 0.55 0.22
Viscosity, centipoise 8.6 21 |
Vapor Pressure, mm Negligible Negligible
Thermal Conductivity, . _ g
0.011 0.077

vatts/(°C-cm)
 

 

»

35

ly higher than are desirable. It is possible that substitution of Zth

or even AlF_ for some of the BeF, will provide liquids of lower viscosity

3 a 2

at no real expense in liquidus temperature.

Table k4. Ccmposition and Properties of Four.

Possible Secondery Coolants

 

Cdmposition :
(mole %) A B C D E

 

NeF 4  NaF 7.7 ILiF 5 LiF 23
NaBF) 96 NaBF, 83.65 NeF 53 NaF 41 NaF 57
KBFh 8.65 BeF2 k2 BeF2 36 BeF. 43

Liquidus Tem-

perature: |
°c 380 370 318 328 3L0
oF 716 700 604 622 634

Physical Proper-
ties at 850°F

 

4shog)e
Density, 1b/ft3 130 130 138 136 139
Heat Capacity
Btu. 1b-1l(°F)-1 0.» 0.4 0.45 0.h7 0.4
Viscosity, centi- | . | . | B
poise 15 (436°C) 25 50 35 55
Vapor Pressure at IR c'”‘ | _ o
1125°F (607°C)P, mm 310c - 2537 Negligible Negligible Negligible

Thermal Conductiv-

ity (watts/°C-cm) , 0;008"4"0;0075"~ 0.01 - 0.01 0.01

 

'aMean temperature of- coolant g01ng to the primary heat exchanger.

leghest normal operatlng temperature of coolant.

CRepresents decomposition pressure due to MBF), + BF, + MF.

 
 

 

36

The densities were calculasted from the molar volumes of the pure com-
fionents by assuming the volumes to be additive.. The heat caéécities were
estimated by assuming that each gram atom in the mixture contributesVB
calories.per degree centigrade. The value of 8 is the fipproximate aver-
age from & set of similar fluoyide mélts.22
The viscosity of the fuel andrcoolants C, D, and E were estimated

23,2h,25 the viscosity

from other measured LiF—BeF2 and NaF—BeF2 miitures;
of the blanket salt was estimatedrfrom megsurements2hof mixtures which
contained UFh instead of ThFh. rThé viscosity of coolant A could not be
reliably estimated because of the sbsence of measurements on this compo-
sition. However, the viscosity of the mador-components, NaBFu, is about
14 cp at 1#36°C.26

The vapor pressures of the fuel, blanket, and coolants C, D, and E
are considered negligible; extrapolation of measurements on similar mix-
tures yielded pressures less than 0.1 millimeter. The partial.pressure
of BF3 above the fluoroborate coolant mixture was calculated from measure-
ments on pure NaBFf7 by assuming that NaF, NaBFh, and KBFh form an idesal
(in the sense of Raoult's Law) solution.

The values given are unlikely to be in error to an extent sufficiént
to remove the fluid from consideration. It is clear from the’' fact that
estimates, rather than experimentally determined values, are used in these

tables that a program must be devoted to measurement of physical proper-

ties for the pertinent materials.
 

 

C

37

CHEMICAL COMPATIBILITY OF MSRE MATERTALS

Successful operation of the MSRE requires compatibility of the molten
fuel mixture with unclad graphite and Hastelloy N during years of rapid
circulation of the fuel through an sppreciable temperature gradient. Such
compatibility must, moreover, be assured while the fission process produces
its intense radiation field and the buildup of fission product species.

To evaluate these implied problems has required a large research and de-
velopment program in which many tests have‘been conducted over a period of
several years.

Details and specific findings of the large program of corrosion test-
ing are presented és e separate paper in this series.28 In brief, com-
patibility of the MSBR materials is assured by choosing as melt constitu-
ents only fluorides that are thermodynamically stable toward the moderator
graphite and toward thé structural metal, Hastelloy N, a nickel alloy con-
taining about 16% Mo, 7% Cr, and 5% Fe. The fuel and blanket components

(LiF, BeF,., UF, , and ThF, ) are much more stable than the structural metal
L L

2’
fluorides (NiFg, FeF2,

a minimal tendency_to corrcde the metal. ©Such selection, combined with

and CrF2); accordingly, the fuel and blanket have

proper purification procedures,'prqvides liquids whose corrosivity is
within tolerable"limits.  The chemical properties of the materials and the
nature of their several interactions, both with and without radiation and

fission, are described briefly in the following.

 

Stability -of UF3 and UF),
" Pure, crystalline uranium trifluoride is stable, under an inert at-

mosphere, to temperatures in excess of 1000°C, but it disproportionates at

 
 

 

38

sufficiently high temperatures by
2 o
Long, who studied the reaction

UFh+]§‘f122 UF3+-HF

obtained data28 wvhich when combined with other accepted values indicate
that the free energies (in.kcal/mole) for the pure crystalline materials

can be represented by

AFf = «351 + 52.8 x-10-3 T°K
UF3 7

and

f f : -3
UF3 _ AFUFh = +497.0 - 15.6 x 10_.

However, uranium trifluoride is appreciably less stable in molten fluoride

AF T°K.

solutions than in the crystalline state. Long's data for the reaction in

29

2LiF-BeF, solution yield the following equations™ for activity coef-

ficients of the materials in this solution

log yyp = -1.62 + 3.7 x 107> T°K

3
and

Uranium trifluoride is permissible in reactor fuels only insofar as
the equilibrium activity of U° which results is sufficiently low to avoid
reaction with the moderator graphite or appreciable alloy formation with
- the Hastelloy N. Use of the aétivity coefficients shown above to predict
at 1000°K (T727°C) the activity of uranium in equilibrium with melts con-

4

taining various U*3/U* ratios leads to the data of Table 6. It is obvi-

ous that large quantities of UFh must be reduced if apprecisble uranium
 

 

 

39

activities are to be obtained. UC, would form, for example, if 68% of the

UFh'were reduced to UF3;

Table 6. Calculated Values of the Fraction of the Total Uranium
in Solution Present in the Trivalent State (UF3/Total U)
in Equilibrium at 1000°K with Varlous Phases

(Total uranium in solution = 1 mole %)

 

 

Phase U° Activity - UF,/Total U(%)
U Metal 1.0 | > 99
UcC 3 x 107° 89
uc, . 5x 1077 68
Ni alloy 1078 Lg
Ni alloy 2 x 10720 20
Ni alloy = 1x 107 - 1

 

In fuel proeessing, hydrogen reduction ef the fuelrmixtures (as
described in the section efi Production Technology below) should lead to
reductiofi of no more than.abeutr2% pflthe UFh' Corrosien reactions such
as 2U?h + Cr ICrF2‘+ 2UF3_wpele:ificreese the UF, concentration to a
negligible extent;ebove-this'value.~-Thus, under reactor conditions, it
seems clear that the reductlon of the UFh normally encountered would intro-

duce no problems, only through drastic and V1rtually unlmaglnable reduc-~

tion could serious consequences arise._;

 
 

 

Lo

Oxidation (Corrosion) of Hastelloy N3_5 S o '

30

Blood™ has made & careful study of the reaction

MFo(a) * Baa) ¥ Me) * Hl(g)

where7M represenis Cr, Fe, or'Ni5 c, £, and d indicate that the species is

crystalline solid, gaseous, or dissolved in molten»LiF—Ber mixture. His

15

date (Teble 7) when combined ” with accepted values for HF, yield free

Table 7. Experimentelly Determined Equilibrium Constents
_ for the Reaction ‘

+ M, | +
Maa) * Ha(g) * M(e) * 2 (q)
in LiF-BeF, Mixture Containing 62 mole % LiF

 

 

Temperature KN for CrF2 KN for FeF2 ' KN for N1F2
1000°K 4.4 x 10~ 1.9 - | :
800°¢C 1.3 x 10"h 0.80
700°C 7.5 x 1077 0.53 7 x 10°
600°C 1.2 x 10~ 0.13 1.5 x 10%
P
where T ——
KN NMF X PH

 

energies of formation (along with those of Long for UFh and UF3) in

Table 8.
 

 

”

o~ 1
’

L1

Table 8. Free Energiesa for Solutes in
Molten 2LiF-BeF, (773-1000°K)

 

 

Solute AGT AGT (1000°K)
(kcal/mole) (kecal/F-)
U™+ WFT BAL.6 - 58.1 x 1073 7oK 96.6
vt 4 3F 336.7 - 40.5 x 1075 T°K 98.7
Ni%* 4+ 2FT  146.9 - 36.3 x 1073 ™k 55.3
Fe2t + 2F  15L.7 - 21.8 x 1073 T%K 61.5

cret 4 o 171.8

21.4 x 10’3 T°K 75.2

 

& The reference state is that hypothetical solution
with the solute at unit mole fraction and with the
activity coefficient it would have at infinite

- dilution.

These data reveal clearly that chromium is much more readily oxidized
than iron or nickel. Accordingly, any oxidative attack upon Hastelloy N
'should be expected to showzselective attack on the chromium. Such oxida-
tion and selective attack follows from reactions such as the following:

1. _Impuiities in the melt

5 -+ CrF2+‘Hi, or.

Cr + 2HF -+ CrF

Cr + NiF

2ty

2. Oxide films on the metal
. mio + BeF, + NiF, + Beoi

followed by reaction of NiF, with Cr
3. Reduction of UF, td’UFg o

Cr + 2UF, 112UF3

4'CrF2 )

 
 

 

 

42

Reactions implied under (1) and. (2) ebove will proceed essentially-
to completion at all temperatuies within the MSBR circuit. Accordingly,
such reactions can lead (if the system is poorly cleaned) to a fioticeable
rapid initial corrosion rate. thever, these reactions do not give a sus-
tained corrosive attack. | |

The feaction of‘UFh'with Cr, on the other hand, has en equilibrium
constant with a small temperature dependence; hence, when the salt is
forced to circulate through e tempereture gfadient, & possible mechanism
exists for mass transfer and continued attack.

If_nickel, iron, and.molybdenum'are assumed to be completely inert
diluents fof chromium (as is approximetely true), and if the circulation
rate in the MSBR is very rapid, the corrosion process can be simply de-
scribed. At high flow rates, uniform concentrations of UF3 and CrF2 are
maintained throughout the fluid cireuit; these concentrations satisfy (at
some intermediate temperature) the equilibrium constanf for the reaction.
Under these steady-state conditions, there existe some temperature-inter-
mediate between the maximum and minimum temperatures of the circuit, at
which the initial surface composition of the structural metal is at equi-
livrium with the fused salt. Since the equilibrium constant for the chemi-
cal reaction increases with increasing temperature,_the chromium concené
tration in the alloy surface tends to dec:eese at temperatures higher than
T and tends to increase at temperatures lower than T. [In some melts (NaF-
LiF-KF-UFh, for example) AG for the mass transfer reaction is qfiite large,
and the equilibrium constant changes sufficiently as a function of tem-
perature to cause formation of dendritic chromium crystals in the cold

zone.] For MSBR fuel and other LiF-BeF,-UF) mixtures, the temperature

V
 

 

"

43

dependence of the fiass—transfer reaction is small, and the equilibrium is
satisfied at reactor temperature conditions without the formation of
crystalline chromium. |

Thus, in the MSBR, the rate of chromium removal from the salt stream
by deposition at cold—fluid regions is controlled by the rate at which
chromium diffuses into the cold-fluid wéll; the chromium concentration

gradient tends to be small, and the reSulting corrosion is well within

- tolerable limits. In the hot-fluid region, the alloy surface becomes

’

depleted in chromium, and chromium from the interior of the wall diffuses
toward the surface. This rate of diffusion is dependent on the chromium
concentration gradient. Since diffusion occurs by a vacancy process and
in this particular situation is essentially monodirectional, an excess of
vacancies can accumulate in the depleted region. These vacancies precipi- .
tate in areas of disregistry, principally at graih boundaries and impuri-
ties, to form voids. The voids in turn asgglomerate and grow in size with
increasing time and temperature. The resulting subsurface voids are not

interconnected with each other or with the surface.
The mechanisms described above leed to such observations as (a) the
complete independence_of COrrosion'rate from flow rate for a given system

*

and (b) the increase in corrosion with increase in temperature drop as well

as with increase in mean'temperaturerwithin a system,

The results 6f;nfimérou§longftermteéts have shown that Hastelloy N
has'excellent cofrosipnlresiSténcé fiolmolten_fluoride mixtures at tempera—
tures well sbove those__ anticipated in MSBR. The attack from mixtures
similar to'the MSBfi fuel ét'témperatures as high as 1300°F is barely ob-

servable in tests of as long as 12,000 hr. A figure of 0.5 mil/yr might

 
 

 

Ly

be expected.31 Even less corrosion occurs in the blanket where the UFh

concentration is very low. Further, the mechanical properties of
Hastelloy N are virtually unaffected by long-time exposure to the molten
~fluoride fuel and blanket mixtures. Corrosion of the container metal 5y
the reactor fuel and blanket does not seem to be an important problem in
the MSER.

Thié encouraging status fof metal-salt compatibility certainly applies
to the coolant mixture if & reasonable NaF-BeF2 or NaF-LiF—BeF2 mixture is
chosen. It is likely that the'NaF—NaBFh coolant mixture will also prove

compatible with INOR-8, but no detailed experimental proof of this is

available. The free energy change for the chemical reaction

BF3(g) + 3/20r(s) > 3/2CrF2(E) + B(é)

is about +30 kcal at 800°K.32

The reaction is, therefore, quite unlikely
to occur, and similaer reactions with Fe, Mo, and Ni are much less so. In
addition, the above reaction becomes even less likely (perhaps by 10 kcal

or so) when one considers the energetics of formation of the compound

NaBFh and dilution of the NaBFh by NaF. However, the following reaction

BF () + (X + 3/2)M(s) > MkB(S) + 3/2MF

3(g 2(%)

is almost certainly the one to be expected. Thermbchemical-data for the
borides of Cr, Ni, Mo, and Fe do not seem to have been established. Very

stable borides such as TiB, and ZrB2 show free energies of formation of

2
-67 and -68 kcal/mole (or about -34 kcal/boron atom) at 800°K.33 The

borides of Mg (M’gB2 and MgBh) show free energies of formation of less than

33

=10 kcal per boron atom. Unless the borides of the Hastelloy N constitu-

ents are very stable, it would appear that the alloy will prove resistant

o
 

 

L

‘able solutions.

L5

to this coolant. However, such compatibility must be demonstrated by ex-
periments.
Compatibility of Graphife With Fluorides
Graphite_does not reaét chemicdlly with molten fluoride mixtures of
the type to be used in fhe-MSBR. Available thermodynamic data6 suggest

that the most likely reactiqn:

hUFh + CZ CFh + hUF3

should come to equilibrium at CFh pressures below 10-8 atm. CF) concen-
trations over graphite-salt systems maintained for long periods at elevat-

3k to be below the limit of detection

ed temperatures have been shown
(> 1 ppm) of this compound by mass spectrometry. Moreover, graphite has
been used as a container material for many NaF—Zth-UFh, LiF-BeF2—UFh, and
other salt mixtures with no evidence of chemical instability.

The MSBR will contain perhaps 20 tons of graphite. Several potential
problems in addition to that of chemical stability have been considered.
These include (1) hazardous incresse in uranium content of core through
permeations of the graphite by fuel, (2) reaction of fuel material with
oxygenated gaseous species'desorbed from the graphite, and (3) carburiza-
tion 6f the HasteiloyN sfrucfiufegby'éarbon dissolved, suspended, or other-

wise carried in the circulsting sait.r These possibilities have been

studied experimentally and found to be inconsequential or to have practic-

b5
_Graphite is not wetted by MSR fuel mixtures (or by other similar mix-
tfires)_at elevated temperaturéé;  The;éxteht tp which.graphite is permeat-

ed by the fuel is, accordingly,'defined by well-known reiationships among

 
 

 

L6

applied pressure, surface tension of the nonwetting liquid (about 130
dynes/cm), and the pore size spectrum of the graphite specimen. Hdwéver,

since the void volume of the graphite may be about 16% of the core fuel

volume, detailed testing of pérmeation behavior has been necessary. Typi

cal‘tests35

with MSRE graphite_have exposed enguated specimens to MSRE
fuel mixtures at 1300°F;»applied pressures were set at 150 1b, a value of
three times the reactor design pressure. The observed permeation did not
change with tifie after a few hours. In these tests 0.18% of the graphite
bulk volume was permeated by the salt; such permeation is well within that
considered tolerable during MSRE operation. Specimens permeated_to this
extent have been given 100 cycles between 390 and 1300°F without detect-

able change in properties or appearance.

Radiation Effectsl’S"S’Bh’36

A considerable body of information about the stability and compati-
bility of MSBR materials under irradiation from fissioning fuel has been
obtained. These studies were motivated by the concern that neutrons, beta
and gamma rays, and fission fragments might cause radiation damage to fuel,
metal, and graphite structural components. Fission fragments, which should
produce localized regions of dense ionization and radiolysis in the molten
salt, might affect fuel stability and corrosion behavior.

Early In-Pile Tests on NaF—Zth—UFh Fuels

 

The earliest studies of radiation effects on molten fluoride systems
were done in the molten-salt ANP program. These tests used NaF-Zth-UFh
mixtures and Inconel containers. Such ifradiations, with melts and metal

chemically similar to those propésed'for the MSBR, were performed over &

»
 

 

 

0

L7

wider range of power density and temperature thah were used in more recent
irradiation work in support of the MSRE. More than 100 static capsule
tests were carried out in thermal neutron fluxes from 10ll to 10lh neutrons
cm sec_l, with fission power densitiés from 80 to 8000 w/c, at tempera-
tures from 1500 to 1600°F, and for irradiation times from 300 to 800 hr.
Chemical, physical, ahd metallographic tests indicated no major changes in
the fuel or the Inconel which could be attribfited to the irradiation
conditions. Corrosion of Inconel was comparable to that found in unir-
radiated controls. Three types of Inconel forced-circulation in-pile
loops were operated with NaF—Zth-UFh melts at fission power densities of
400 to 800 w/cc, maximum temperatures of 1500 to 1600°C, and for 235 to
475 hr at full power; corrosive attack on the Inconel was no greater than

in corresponding out-of-pile tests (wall penetrations less than 3 mils).

Early Tests on LiF—BeF2-UFh Fuels

 

The first irradiation test on an'LiF—BeFLbased fuel was a graphite~
fuel compatibility test in the MTR. Two Inconel capsules containing
graephite liners filled with LiF-BeF,-UF) (62-37-1 mole %) were irradiated

at 1250°F for 1610 and 1492 hr, and at average power densities of 954 and

- 920 w/ece, respectively. Thé-eprsure reéulted in no apparent damage to

the graphite, and negligib;éféorrosion to the Incofiél which was exposed
fd,the salt thrOUgh:smaliuthés ih the graphite liner.

In the next test, two small Hastelloy N capsules fiere‘filled with the

same’ LiF-BeF,~UF, mixture and irradisted for 5275 hr in a flux of 1 to

2 x lO:VI',4 heutr¢nscm-2“sec-latan initial power density of 1170 w/cc and

a temperature of 1250°F to an estimated T5% burnup. The failure of one

 
 

 

48

capsule at this time forced términation of the experimént. The results of
later tests suggest that fuel radioiysis at ambient reactor temperature
during shutdowns may have contributed importantly to the capsule failure.

Two forced-circulation Hasteiloer loops containing the‘above-LiF-
BeFe—UFh mixture were also installed and operated in the MTR. These were
designed to operate at 1300°F maximum temperature, 190 w/cc power density,
and a linear flow velocity of 2.5 ft/sec. Pump failure terminated the
first test after 860 hr and the second after 1000 hr. Métallographic eX-
emination of the metal from the first loop revesled a moderately eroded
region (approximately 2 mils deep) in one of the sharp bends in the high-
flux region. Metal specimens from the second loop showed a negligible
degree of corrosive attack. Since later in-pile tests confirmed the good
corrosion resistance of Hastelloy N, it is suspected that the first loop
was fabricated from substandard alloy.
Testing of MSRE Fuels

The ORNL~-MTR-U4T7- series of capsule irradiation experiments was design-
ed to test the stability and compatibility of actual MSRE materials (graph-
ite, Hastelloy N, and fuel salt) under conditions approximately those of
the MSRE, with emphasis on the interfacial behavior of molten salt and
graphite. The capsules were relatively large to provide adequate speci-
mens of graphite, fuel, and Hastelloy N for thorough postirradiation exsmi-
nation.

In the U7-3 test, four Hastelloy N capsules (see Fig. 12) containing
graphite boats holding a pool of fuel salt (BeF2-UFh-LiF—ThFh-Zth, 23.2-
1.4-69.0-1.2-5.2 mole %) were irradiated for 1594 hr at meximum tempera- -

tures of 800°C and maximum power densities of 200 w/cc to a burnup of sbout

o

O
 

 

 

W

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

  

 

  

49
ORNL-LR-DWG 56754R
. wamRALS
MATERIALS _ , "SPECIMENS ~_GRAPHITE BLADE
1 [\\ - 1
. V : ™, # g / 4"' // E %
sl i F "
_ %= | % / o -
- 0 —— /a
’ h N 5
|
. £-MOLTEN- I
SALT FUEL GRAPHITE BOAT-" - . = - . ~THERMOCOUPLE WELL
L 0 . Y 1/2 314 q
Lo o e 0 1
_ _ 7 INCH .
Fig. 12. MSRE Graphite-Fuel Capsule Test - )
| ) . ORNL-MTR-LT-3
4

 
 

 

-

50

10%. Each capéuie also contained specimens of Hastelloy N, molybdenum,

and a pyrolytic graphite attached to a graphite blade dipping into the-

shallow pool of fuel. Two of the graphite boats were‘initially-ifipregnated‘

with the fuel to provide a more'extreme test of graphiteéffiel cdmpatibility
at high tempefature. When the capsules were dlsmantled the frozen fuel
exhibited nonwettlng contact angles with the graphite boats, the graphlte
blades, and the pyrolytic graphlte»spec1mens. The graphite structurekap-
peared undamaged visually and metallographlcally Howe&er thére wéfe
definite observatlons that fuel radloly51s had taken place (generatlon of
F2 and CFh)' Because of these observatlops, subsequent testg studied the
radiolytic instability of the fuel in detail: it wéé found that only when
the irradiated fuel was aliowed to freezé-an& cool below 100°C did radio-
lytic decomposition take placé. .

The 47-b irradiation assembly comprised of four large Hastelloy N cap-
sules (seérFig.'13), each containing a 0.5-in.-diam graphite specimen sur-
rounded by & 0.2-in. annulus of fuel, and two smai_l Hasfelldy N capsules

containing a 0.5-in.-0OD graphite cup nearly filled with fuel. The largé

‘capsules contained about 25 g of fuel (BeF -UFh-LlF-ThFh—Zth, 22.6-0. -

T1.0-1.0-4.7 mole %). One of the smaller capsules contained 10 g of the

same fuel; the other 10 g of & similar fuel of higher UFh concentration

(1.k mole_%). The capsules were irradiated for 1553 hr at temperatures up
to 800°c (900°C for the small 1.4 mole % UF), capsule), at average power

' densities from 40 to 260 w/cc, and to burnups from S to\lO%. There was a-

gain evidence that fuel radiolysis occurred at low temperatures during
reactor shutdown; however, metallographic examination of the Hastelloy N

capsule walls showed no discernible corrosion, and the graphite appeared

9
 

w

. -

 

 

 

51

ORNL—LR-DWG 67T744R

7

~ Cr-A THERMOCOUPLE\fl /NICKEL THERMOCOUPLE
| o WELL
NICKEL. POSITIONING LUGS (2) | o

NICKEL FILL LENE\?‘\ 1" /'(NICKEL VENT LINE

INOR-8 CAP
N

HELIUM COVER GAS
(3.5 ecm3)

  

 
 

 

 

 

 

 

 

 

~—|NOR-8 CAN

PUNCTURE AREA ,
FOR GAS SAMPLING 7

 

 

r

~Y 7 7 7T 77 7 77 777

 

 

 

-7 T 77777

——MOLTEN SALT FUEL

gl ~

 

 

 

o . - (25¢q)
. t Cr-Al
o { THERMOCOUPLE

 

 

 

 

, CGB GRAPHITE
e 12.4 cm? INTERFACE

_ | o N INOR-8 CENTERING PIN
‘o Vo 1~ S

 

 

ANCH O = NICKEL POSITIONING [UG

Fig..13. Large Fuel Capsule from ORVL-MTR 47-h

 
 

 

.;!

,52'

ot

undamaged'except for-the vapor-exposed region_of the small high power

den51ty capsule. - | |
To investigate the fuel rad101ys1s further two capsules in the hT-S

assembly, of design 51mllar to the 1arge hT-h capsules were equlpped with

.gas lines which permltted measurement of pressure w1th1n the capsule and | ~

withdrawal of cover gas sampleslwhilefhe irradiation was proceeding. Two |

large sealed capsules with w1dely dlfferent areas of graphlte and metal ex-

.posed to the fuel and two small capsules conta1n1ng fuel-lmpregnated

graphlte rods suspended in hellum completed the‘assemhly. Four of the

vcapsules contained salt having the compos1t1on L1F-BeF2-Zth-UFh with mole

ratios of 67.36-2T.73-4.26-0.66. Salt with lower uranium concentration

(LiF—BeF2—Zth-UFh, 67.19427.96-h.51—0.3h) was used in one of the gas—

swept capsules and in the'low-flux, impregnated-rod capsule._rThe hT-S

capsules were irradiatea for h-l/Q\uonths at average fluxes'between“

2 x 1072 and 3 x 10%3 neui:rons-c'.m-esec_l to burnups between T and 15%.

a "’

Gas samples were taken from the purged capsules under a varlety of operat—
ing conditioms, with fuel temperatures varylng from 190 to 1500°F and power
densities from 3 to 80 w/cc. During reactor shutdowns, when the assembly
cooled to about 35°C, pressure rises were,observed in the capsules equip-
ped with gas lines, and gas samples indicated tue presence of'flucrine.

. With the'reactor operating and the fuel molteu, the isolated capsules _;
sshowed no. fluorine. In a few of the 60 gas samples, barely detectable
traces of CF) (approkifiately 5 ppm) were found;'these were probably due to
iincomplete flushing of the system'since the'lastlreactorrshutdcwn., In any
case, the observed minute rates of CFh generation represented negligible

reduction of UFh to UF

3 und, accordingly, an inconsequential practical B 2
 

 

o}

lsl

i

53

- problem. In later.hot—cell'studies'of frozen irradiated fuel, it was es-

tablished that the gas evolved was pure fluorine and that the G value at

- 35°C was 0.02 molecules of fluorine per 100 ev of fission product decay
‘energy ebsorbed. The rate of radiolysis was greatest in the temperature

~range of'35 to 50°C;vit dropped to low values at -T0°C and to zero at tem-

peratures ebove 80°C

The k7-6 test vas de31gned to allay any llngerlng doubts that fuel
radiolysis and its consequences‘could be e11m1nated by maintaining the
fuel'moltenieven_dufing reacfior Shufdown;- Four cylindrical, Hastelloy N
eapsules.were.used (ldih. OD?x 2;615 in. long). Heatefa were provided for
all capsules, and,these'turged'on_automatdcally when the fuel temperature

approached the liquidus temperature, maintaining the fuel salt in a molten

condition even when the-feactof,was,shut‘down. The capsules contained
‘cylindrical graphite cores which were 0.5 in. in diameter and 1.35 in.

- long; the cores werelsurrounded by'0.2 in. of fuel salt and pierced by a

central Hastelloy N thermeeouple well. Two of the capsules (see Fig. 14)

. were.equipped-with gae 1ines.and differed from each other'only in that half
~the graph1te area in one was replaced by a Hastelloy N exten51on of ther

._ thermowell. These capsules were charged w1th an L1F—BeF -Zth-UFh fuel

- similar to that in the_hT-S_test_bup contalnlng 0.9 mole % UF,. The two
fsealed capsules containedifullasize éraphite cores and similar fuels with

'o 5 mole % and L. Omole % UFh

| The h7~6 assembly was 1rrad1ated in the MIR for 1500 hr to burnups

_fram l% (1n salt contalnlng 0. 5% UFh) to 5% (salt with 4.0% UFh) Gas
samples were taken at steady operatlon with the purged capsules at tempera-

‘tures from 850 to 1300°F and .at power densities from 20 w/ce to 75 w/ce.

 
 

 

 

 

 

 

~ THERMOCOUPLE —a

 

7

COVER GAS
SPACE (~ 3 cm3)

 

 

WATER JACKET ]

-

"HEATER -

 

. ANNULAR GAS GAP —|

 

*

 

7

THERMOCOUPLE»\\

5 € < i

//- PURGE TUBE

 

 

 

 

 

 

Ll Ll

 

 

 

 

  

 

 

 

N N N N N N N S NN N N SN NSNS AN NSSINCEN SN SN SN NS E NS SN AN NN N A
1
I

Ll

 

 

 

 

 

 

HEATER LEADS —7

-

 

 

 

 

SIS SN SNSNEIANNNNEN NSNS NN NN N NN AN N NN NN NN N NN N NS

N

il

 

\\, COOLANT WATER

SUPPLY TUBE

»
CGB GRAPHITE COREV_‘.
MOLTEN SALT E;UEL |
INdR-B CAPSULE B&;Y

. { |

WATER FLOW CHANNEL

 

 

" GAS CAP PURGE SUPPLY TUéE‘//

- Fig. 14. Fuel Capsule from ORNL-MTR-47-6

3 W

<’

1w

 
 

 

 

o}

02

The gas analyses detected no CFh or other fluorine-containing gases in any
of the 36 gas samples; CF), deliberately added to the capsules during ir-
radiation was radiolytically decomposed at a rate which decreased with tem-
perature and seldom exceeded L4%/hr. .
Particular care was given to the postirradiation examination of the
graphite specimens. No uranium deposits were found by chemical analysis,
by delayed neutron counting of neutron activated specimens, or by x-radio-

graphy of thin sections. It is therefore clear that uranium deposition on

graphite is associated only with fuel radiolysis at low temperatures, and

that the reaction does not take place between graphite and molten fission-

ing fuel. In addition, visual, metallographic, and x-ray diffraction ex-
aminations of the 47-6 graphite specimens failed to reveal any differences
between the irradiated graphite and unirradiated controls. Also, the
Hastelloy N capsule wall specimens from run 47-6 appeared unaffected by
the exposure based on visual and low-power magnification examination.
Metallographic examination of unetched specimens revealed no change in
wall thickness (less than 1 mil change). |
Conclusions from In-Pile Testing of Molten Salts

In summary, the 47- series'of_irrédiation studies has been generally
reéssuiing as to the radiatibfi stafiility and ccmpatibility of Hastelloy N,

graphite, and fuels based on lithium and beryllium fluorides. The cor-

- rosion that is known to occur, i.e., that due to mass transfer, does not

seem'to;be influenced by pcwer density. It has been shown that the prin-
cipal_disturbing effects_are‘conseQuences of low-temperature fuel radioly-
sis which is easily suppressed by maintaining the irradiated fuel at a

temperature above (conservatively) 200°C. On the basis of the L7- series

 
 

 

56

tests, the limits of this'reassufance}in'regard‘to radiatioh effects on
MSBR materiels extend to temperatures of about lhOObF and power densities
of about 100'w/cc.

The two previous loop tests onrsimilar LiF;BeF2 fuels, described a-
bove, extend the limits of dssurance tO‘power densities of 200 w/cc at
1300°F with regard to'corroéion of Hastelloy N in the absence of graphite.
There has been no indication of the U7- series experiments that graphite
introduces problems in addition to the expected one of Hastelloy N carburi-
zation (when the two are in close contact).

The previous capsule tests with LiF-Ber-UFh fuels, carried out with
no provisions to circumyent the low-temperature fuel radiolyéis effect,
suggested that salt power densities of at least 1 kw/cc may be permissible.
Also, the numerous tests with NaF-Zth-UFh fuels in Inconel at temperatures
up to 1600°F and power densities up to many kilowatts per cubic centimeter
exhibited tolerable compatibility characteristics. With respect to radia-
tion effects, there is no obvious chemical reason to suppose that a gross-

ly different salt behavior would be observed using MSBR materials.

C
 

 

4

o7

BEHAVIOR OF FISSION PRODUCTS IN MOLTEN SALTS3“S

Fission products will be produced in & 2225-Mw(th) MSBR &t a rate of
about 2.3 kg/day. In the reference MSBR'design, the fuel salt volume is

about T00 ft3

and the-fissilé inventory about 700 kg; with these values
the fission product concentration after 50 days accumulation would be about
15% of the fissile concentration. Thus, it is clear that fission product
concentrations can be significant even with high processing rates, and
that fission product behavior needs to be considefed in specifying reactor
operating condifions.
Physical Chemistry of Fission Products

Fission and its immediate aftermath must be a violent process; the
very energetic major fragments are probably deficient in electrons at their
origin, and, as they lose energy by cbllisions, they undoubtedly produce
additional ionization within the medium. It seems certain, however, that
electrical charge is conserved in this,précess; electrons and protons are
heither created.nor déstroyed by . the fission event. It follows, therefore,
that when fission of UF) occurs in an inert environment [as in a (hypo-

thetical).completely inert container] the reaction
| o o :7 - + -
- UFh-+ n ¥ 2FP +2n +,hF

must in a statlstlcal sense, satlsfy the conditions that (l) the salt be

_electrlcally neutral and (2) redox equllibrlum be established among the

numerous ionic spec1es.r In an 1nert contalner such catlon—anlon equlva-
lence (and redox equlllbrlum) mlght be satlsfled w1th uranlum valence

states above h and with p051t1ve 1on formatlon by Nb, Mo, Te, or Ru. The

MSBR contalner metal (Hastelloy N) is not completely inert and the fuel

 
 

 

58

contains a small concentration of UF3, so additional possibilities exist

- for this system. Should the fission product cations.firove inadequate for
the fluoride ions plus the fiésion product anions (notably I ), or should
they prove adequate only by assuming element valencerstatesltdo highrto be
- thermodynamically compatible with Hastelloy N, the container metal would
be constrained to supply the cation deficiency.

Thermochemical data from whiéh,the st&bilify of‘fissidnlproduct
fluorides in complex dilute solutions can be predicted‘are lacking in many
cases. BSuch information which appears definite is briefly described in
the following sections.

Rare Gases
The fission products krypton and xenon are volatilized from high-tem-

34,37

perature melts as elements. The solubilities of these gases in mol-

ten fluoride mixtures38’39’ho

obey Henry's law, increase with increasing
temperature, decrease with increasing atomic weight of the gas, and vary
somewhat with composition of the solvent. Henry's law constants and heats
of solution for the rare gases in LiF--BeF2 mixtures are shown in Table 9.

The positive heat of solution ensures that blanketing or sparging of the

fuel with helium or argon in a low-temperature region of the reactor can-

not lead to difficulty due to decreased solubility and bubble formation in

higher temperature regions of the system. [There is no evidence of
trouble from such source in MSRE where the He is appiied in the pump bowl
at the highest temperature in the circuit.]

The very low solubilities of these gases suggest tha£ they shofild be

readily removed from reactor systems. Only a small fraction of the calcu-

lated xenon poisoning was observed during operation of the Aircraft Reactor
 

U

»

59

Experimenthl where the only mechanism for xenon removal was the helium

"purge of the pump bowl.

Table 9. Solubilities and Heats of Solution for

Noble Gases in Molten LiF—BeF2 Mixtures at 600°C

 

 

LiF-BeF2 (6k-36 mole %)

 

Heat of
ga Solution
Gas K x 10 (kcal/mole)
Helium ' 11.55 j_O.h 5.2
Neon L.63 + 0.2 5.9
Argon - 0.98 + 0.02 8.6
Xenon 0.233 + 0.01 12.1

 

& = moles gas/(cm3 solvent)(atmosphere).

A sbmewhat,more ambitious scheme for insuring a low poison fraction
fér xenon (and krypfon) isotopes is to remove thevhalqgen precursors iodine
and bromine on & tifie cyclé short éompared to their halftimes for decay in-
to‘the hoble gases. Sincel35Xe is by far fhe worst poison of this class,
removal of its iodine precursor would be_most‘important; ifs decay half-
time is‘sfich that”itsresiaéfice.timg_in the reactor should be kept at 1
hour or.less. In.principle.if"(énd Er-) can.be removed by the reaction

T+ gy > Fg * HI(gy
where d'afid g-indicate that'tfieSpeéies is dissolved in thé melt or exists
ifi the gaseous'stéte;'Molfefi'fiuorides similar to MSBRVfuel and sPiked
withI--have been shown to yield the contained iodine readily on contact

with gaseousHF.ha These smell-scale (and preliminary) studies suggest

 
 

 

60

that the removal step is chemically feasible.
Elements in Periodic Groups I-A, II-A, II-B, and IB-B

Rubidium, cesium, étrontium, barium, zirconium, yttrium, and the
lanthanides form very stable fluorides."TheSe fission préducts should, ac-
cordingly, exist in the molten fuel in their ordinary valence states. A
variety of studies of many typésvéhows'that iarge amounts of Zth, the
alkali fluorides, and the alkaline earth fluorides can bédiséolved in MSBR
fuel mixtures at operating températures. Since the tfifluorides are less
soluble, the solubility behavior of the fluorides of yt£rium and the rare

43,4y L5

_earths, and of plutonium _haé been examined in some detail. The
saturating phase from solutions in LiF—BeF2 and related mixtures is the
simple trifluoride; when more than one rare earth is present, the saturat-
ing phase is & (nearly ideal) solid solution of the trifluorides. Such
solid solutions are known to accommodate UF3 and it is very likely that

they would include PuF. as well. The solubilities of these solid solutions

3
depend strongly on composition of the melt; the solubilities may be near
the minimum value for MSBR fuel compositions. Even then, howéver, the
solubility (near 0.5 mole % at MSBR operating temperatures) is such that
many months would be required for the reactor to saturate its fuel with
these fission products. In any case, reprocessing to remove the rare
earths, and particularly neédymium, is required in the interest of neutron
economy .

The above statements regarding rubidium and cesium do not apply to
that fraction of these elements originating in the graphite as daughters

of the rare gases which have permeated the moderator. These alkali metals

form compounds with graphite at high temperature but the absolute amounts
 

H

[ ]

61

are so small that difficulties from this source are unlikely. Damage to
the gfaphite.by this'mechanism will, as a matter of.course, be looked for
in all future radiation and fission sfudies.
Other Fission Products

These products include molybdenum, ruthenium, technétium, niocbium, and
telluriumproduced in relatively high yields with rhodium, palladium,
silver, cadmium; tin, and antimbny in yields ranging from small to trivial.

6,7,32,33,46 _ poests that the fluorides

The available thermochemical data
of these elemen£s-wou1d be (if they were present in the pure state) reduced
to the métal by chromium at its activity in Hastelloy N or by UF3 at
reasonable concentyations in the fuel salt.

The high-yield noble metals (Mo, Nb, Ru, Tc, and Te) have polyvalent
fluorides which are generélly quite volgtile and moderately unstable. The

formation free energies for NbF MoF6, and UF6 may be calculated with

5’
relatively good accuracy because of recent measurements at Argonne of the

heats of formation of these éompounds by fluorine bomb calorimetry.hT-hg

The entropies and heat capacity data also are available.so While the
people at Argonne have measured RuFE,Sl no entropy or heat capacity data

seem to be available:

- fi;‘_g_a f_s_g_gé | Reference
MoFgle)  3ra.3540.22 1213 . u8
UF(g) .-5'10.—77-'_4_»_ 0.h45 ~-67.01 49
- WF(g) -_h33;5 + 0.5  -91.56 AT
RuFg(s) 21301 + 0,35 51

From these values and the availsble heat capacity data the following ex-

pressions for AGf were derived. In the Case of RuF5, Glassner's6 earlier

 
 

 

62

estimate was corrected to be consistent with the above AHf measurement :

- 6T (R ,g) = -416.70 + 54.40(T/1000)
AcfkRuFS,g) = 200 + 25(T/1000),
AG* (MoF ,g) = -370.99 + 69.7(T/1000),
AGT(UF,8) = -509.9% + 65.15(T/1000) .

The following wvalues of Aaf have been reported previously for UF3and UFh
in 2LiF—BeF2:

~336.73 + 40.54(T/1000),

AEf(UF3,a)

i

Aaf(UFh,d) _hhh;61 + 58.13(T/1000).

From these free-energy values the following equilibrium conétanté have
been calculated for the formation of the volatile fluorides‘bylreaction'
with UFh(d) in the MSRE from the equation

log K = a + b(103/T):

Reaction K : | a b

No(sy * SUFy(a) ¥ MF5(g) * SF3(a) Fior XUF3/XUFh 7.33  -26.82
Ru(g) * 2Pu(a) ¥ BF5(g) * 5UF3(q) Frur XUF /XUFh 13.76  -Th.17

SUFh(d) z UF6(g) + 2UF3(d) PUF6XUF /XUF -32.88

In Fig. 15, calculated equilibrium partial pressures of the gases are
plotted vs the UF3/UFh ratio in the melt. As the oxidizing power of the

melt is increased,NbF5 is expected to appear first, followed by MoF6, and

then RuF_. Uranium hexafluoride has a lower dependence on oxidizing power

5
because its reduction product is UFh rather than the metal. It was as-
 

 

L))

"

63

.sumed in the case of NbFs, MoF6, and RuF5 that the reductlon product was '

the metal. The UF6 should not be formed in s1gn1f1cant amounts until the

melt is ox1d1z1ng enough to produce RuF5 If any stable intermediate

fluorldes of Nb Mo, and Ru are formed in the melt the result would be

correspondlngly lowered equlllbrlum gas pressures and lowered power

-dependences on the UFh/UF ratio.

- Tellurlum hexafluorlde has not been 1ncluded in thls listing, but this

.'compound seems certaln to be less stable than any shown here. No data

which would permlt,lnclus;on of the fluorides of technetlum seem to be

'aira.ilable. -

If the UF /UF ratlo in MSRE falls 51gn1f1cantly below 10 3 NbF .
L _5-

would be expected to volatlllze if the n1ob1um metal in equlllbrlum with

'the fused salt were at a near unlt act1v1ty. Apprec1eb1e pressures of

!

.MoF6, RuF » UFg, (and almost certalnly of TcFg or TeF6) would require much

- more ox1d1z1ng conditions 1n,the melt.

The actual state of these f1551on products is of moderate importance

_to”theleffectlveness of molten salt reactors as breeders;. If the molybden-
gum, nloblum, technetlum, and ruthenlum ex1st as metals (or perhaps as
- ;1ntermetalllc compounds) and plate the Hastelloy N portions of the reactor
.they Wlll be of 11tt1e consequence as p01sons, although they may prove a -
 serious nuisance or worse to heat exchanger malntenance. If they ex1st as
'._soluble fluorides_then they#csuse llttle trouble end are, 1n prlnclple,

iremov&ble in-the»processingdcycle.’ They'can cause‘moSt trouble by:forming

: carbldes or by adherlng in some other way to the graphlte moderator.
l, Molybdenum can form M02C and.MoC'ln the MSRE and MSBR temperature range,

the AG values for these compounds become negative at about 450°C and the

 

 
 

 

PRESSURE (atm)

64

ORNL-DWG 67-773

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

e - | : '
L | ‘ |
X 1
& s ‘
Qfi)/ ' B . :g N ,// _
& _ ,
) . S 0?6 )(\)Fsl -
R L - C
| & | — | |
g O .
4 - & . 4000°K
7 N Xy =0.01 _
q

 

 

|

 

- 108
X UF4/X_ UF,

10%0 | 10'2 0'4

Fig. 15. Equilibrium Pressures of Volatile

Fluorides as Function of UF&/UFB.

Ratio in MSRE Fuel.

 
 

o

¥

65

o2

compounds become more stable at increasing temperatures. Niobium carbide
(essentially NbC) has a large (33 kcal) negative heat of formation at 298°K
and is certainly stable under reactor conditions. Nothing appears to be

known concerning carbides of technetium, but it seems certain that no

‘carbide formation is expected from the platinum metals, silver, tellurium,

cadmium, antimony or tin.

Net Oxidizing Potential of Fission Process53

235

The fuel exposure tests have used U as fissile fuel, with thermal

13

flux exposures of about 3 x 107~ neutrons cm,-2 sec Table 10 shows the

relative yields of the several most important fission products3O resulting

235 3 peutrons —

from fission of U in a steady thermal flux of 3 x 10l
.‘s.ec-l for three selected time intervals. The listed fission products com-
prise at least 97% of all those produced (total yield is 2.0) at listed

times, with no fission product removal.

 

 

 

Table 10. Fission Yields from Thermal Fission of 235U
8, =3 x 1013 neutrons cm=2 s,ec'l
: Time Since Startup
Element 11.6 days 116 days 3.2 years
Br - 0.00030 0.00021 0.00021
I 0.0359 0.0145 0.0125
Kr + Xe 0.297 0.301 0.301
Rb 0.0387 0.0390 0.0393
Cs 0.0971 0.131 0.132
Sr 0.1kl 0.121 0.0980
Ba | . 0.105 0.068k4 0.0626
Rare Earths + Y - 0.528 0.560 0.559
Zr - , 0,318 . 0.318 0.317
Subtotal 1.564 1.553 1.522
Nb | 0.0040 0.0139 0.0028
Mo 0.201 _ 0.201 0.2k2
Te 0.0410 0.0586 0.0592
Ru - 0.140 0.126 0.11h
Total 1.950 1.953 1.940

 

 
 

 

 

66

If the chemically active fission products shown in Table 10 occur as

I~ ,Br ', R, Cs , sret

. Bée+, 13+, 3% (rare earths), Zrh+, and Mt
and if krypton, xenon, molybdenum, technetium, and ruthenium occur as
elements, and if no fission product species are removed from the réactor,'
then the total fission product yield mfiltipliéd by the valencé-(iXiZi) will
be 3.475 and 3.560 at 11.6 and 116 days, respectively. If all krypton and
xenon nuclides of half-life greater than 5 fiinutes are removed from the
system before they decay, the comparsable ZXiZi valueg become 3.21 and 3.26.
If all krypton and xenon nuclides with half;lives greater then 1 minute are
removed before decay, the IX.Z; values aré 3.06 and 3.09 at 11.6 and 116
days, respectively. The above EXiZi values (which seem inadequate to
satisfy the fluoride ions released by the fissioned uranium)suggeSf that
thé fission process is per se oxidizing to UF3 and ultimately to Hastelloy
N. Results of many in-pile tests of compatibility of the materials, how-
ever, suggest that fission does not lead to corrosion of this container
metal. |

If, on the other hand, all the molybdenum formed MoF6 and the techne-.
tium formed ‘I'cF5 then the fission process would require more than 4 fluo-
ride ions per fission event and the fission process would per se be reduc-
ing to UFh' Even for the rather unrealistic case where all xenon and
krypton species with half-lives in excess of 1 minute were removed the
X 2, values would be near 4.5. This would require reduction of one mole
of UFh to UF3 for each 2 moles of uranium fissioned.

Both extremes (that is a strongly oxidizing or a strongly reducing

action of the fission process) seem unlikely. It seems likely that a

fraction of the molybdenum, niobium, and technetium exist as fluorides
 

 

 

o

67

(of valence lower than this maximum) and that, accordingly, the net effect

of fission is neither markedly oxidizing nor markedly reducing to the

~Hastelloy N-UFh system.

Should. subsequent long-term tests at high burnup prove the fission
process to be oxidizing the cure would seem to be relatively simple; if the
burned uranium were made up by addition of UF3 (or UF3.+ UFh) the problem
would be solved. Similarly, if the fission processes were (unexpectedly)
reducing toward UFh the makeup of burned uranium could be as a mixture of

UF5 (or UF6) with UF).

 
 

 

68

C

CHEMICAL, BEHAVIOR IN MSRE
General

The Molten-Salt Resactor Experiment operated during six separate
periods in 1966; virtually all of the operating time accumulated aftéf mid-
May_was at the maxifium possible power of about T.5 M#. The refictor accumu-
lated approximately 11,200 Mwhr during the year. Additional operation in
1967 (essentially ali at maximum.powér) led to:accumulatioh of an addition-
al 21,000 Mwhr as of the scheduled shutdown on May 10,:1967;

During periods of reactor operation, samples 6f thé reactor salts were
remofed routinely and were ansglyzed for major constituents, corrosion
pfoducts and (less frequently) oxide co#tamifiation. Standard semples of
fuel are drawn three times per week; the LiF—BeF2 coolant salt is sampled
every two weeks. ' -

Current chemical analyses suggest no perceptible composition changes.
for the salts since they were first introduced inté the reactor some 20
months ago.

While analyses for Zth and for UFh agree quite well with the material

T

balance on quantities charged to the reactor tanks, the values for 'LiF and

BeF2 have never done so; analyses for LiF have shown lower and fOrw'BeF2

have shown highervalues than the book value since startup. Teble 11 shows
a comparison of current analysis with the original inventory value. While
the discrepancy in LiF and BeF2
nothing in the analysis (or in the behavior of the reactor) to suggest that

concentration remains a puzzle, there is

any changes have occurred.

Routine determinations of oxide (by study of salt-H,.0-HF equilibria)

e

2

continue to show low values (&bout 50 ppm) for 02-. There is no reason to
 

 

 

«0

69

believe that contemination of the fuel has been significant in operations

to the present.

Table 11. Current and Original Composition

of MSRE Fuel Mixture

 

 

" Constituent | Original Value Current Analysis
[ (mole %) (mole %)
TLiF ~ 63.40 + 0.49 64.35
BeF,, 30.63 + 0.55 | 29.83
ZrF), 5.14 + 0.12 | 5.02
U, . 0.821 + 0.008 0.803

 

MSRE maintenance operations have necessitated flushing the interior of
the drained reactor circuit on four occasions. The salt used for this oper-

7

ation consisted originally of an 'LiF-BeF, (66.0-34.0 mole %) mixture.
Analysis of this salt before and after each use shows that 215 ppm of
uranium is added to the flush salt in each flushing operation, correspond-

ing to the removal of 22.7 kg of'fuel-salt residue (about 0.5% of the

charge) from the reactor circuit.

 

Corrosion in MSRE
The chromium concentration in MSRE fuel is 64 ppm at present; the en-

tire operation seems to have increased the chromium concentration only 26

ppm. Thié increase cérreéfibndsfto,remdval of about 130 g of chromium from

the fietal of the fuel circfiith If;this were removed uniformly it would
repréSent removal of chrqmium;to é depth_of gbout 0.1 mil. Analyses for
i:oh and nickel in the system are relatively high (120 and 50 ppm respec-

+
tively) and do not seem to represent dissolved Fe2+ and N12 . While there

 
 

 

 

 

T0

is considerable scatter in these enalyses, there seems to be no indication
of corrosion of the Hastelloy N by the salt.

rThe absence of corrosion--though in general accord with resuits from
a wide variety of out-of-pile corrosion tests--seems somewhat surprising
for the following reasons. The UF3 concentration of the fuei'fidded to MSRE

was markedly less than intended. Careful reexsmination of the production

records and study of the reaction

-
%1{2+UFh+UF3+HF
‘on samples of surplus fuel concentrate show that the fuel salt had only a-

3+
bout 0.16% of its uranium as U3‘. Nearly 10 fold more than this was intend-

ed.
If--as seems virtually certein--the chromium content of the salt was
due to
lor + uF, 2 3crF, + UF
2 L €27 2 3

. _

an additional 1100 grams of U3' should have resulted. With that originally
3+

added the U3’ should have totaled sbout 1500 grams and as much as 0.65% of

the uranium in the system could have been trivalent. An attempt, however,

3+
to determine the U3 (by the H _-HF reaction above) after 11,000 Mwhr of

2
MSRE operation indicated that less than 0.1% of the uranium was trivalent.
Fission of the 550 grams of uranium (corresponding 11,000 Mwhr) could
certainly not have oxidized more than 40% of the 1350 grams of vt vhien
had apparently been oxidized. The remaining 800 grafis (approximately)
could have been oxidized by inadvertent contamination (as by 60 grams of

HEO desorbed from the moderator stack). However, the réte of corrosion

even by this relatively oxidizing'fuel melt remained imperceptibly slow.
 

 

.

T1

Addition of beryllium metal (as 3" rods of 3/8" diameter in a perfo-
rated basket of nickel) through the samplihg system in the pump bowl served
as a convenient means of reducing Uh+ to g3t during reactor operation. 1In
this form beryllium appeafs to react at about 1.25 grams per hour so that
some 600 grams of U3+ are produced by an 8 hour treatment. Some 30 grams
of Be have been added in this way to create an additional 1.6 kg of U3+.
During the subsequent 20,000 Mwhr of operation (which burned 1 kg of urani-
um) this 1.6 kg of U3 seems to have been oxidized. Again, it seems likely
that the fission process was responsible for okidizing a substantial frac-
tion, but ndt all of, this material.

Additidnal‘beryllium will be added to MSRE fuel as soon s power oper-
ation is resumed; it is tentatively planned to reduce at least 1% of the
U* to U3* at that time.

The lack of corrosion in MSRE by melts which appear to be more oxidiz-
ing than those intended can be rationalized by the assumption (1) that the
Hastelloy N has been depleted in Cr (and:Fe) at the surface so that only

Mo and Ni are eXposed to attack,'with Cr (and Fe) reacting only at the slow

rate at which it is furnished to the surface by diffusion, or (2) that the

noble-metal fission,produhfsf(see gections following) are forming an ad-

herent and protéctivé platé on the feactor_meta1.

Behavior of Fission Produétssy?5$°'

Helium is inffddficedfihtq the‘pufip-bOVl.of MSRE at & rate of about U4

liters per minute; this helium serves to strip Kr and Xe from the fuel in

the pump bowl’and to'swéep,these gases to the charcoal-filled traps far
downstream in the exit gas system. Since a relatively'small'fraétion (less

than 10%) of the fuel mixture is bypassed through the pump bowl, the ef-

 
 

 

 

 

T2

ficiency of removal of the fission product gases should not be very high.
However, the Xe poisoning of MSEE st T Mw is onlj about 0.3% in AK/K, &
value considerably less than was anticipated. This low poison level is
probably due to stripping of Xe within the fuel system into hélium bubbles
which are known to circulate (at perhaps 0.2% by volume) in the fuel salt.

- Samples of MSRE fuel, drawn in 10 to 50 cc metal samples at the sampl- -
ing station in the pump bowl, have been routinely analyzed, by radiochemi-
cal techniques, for 14 fission product isotopes and, in some cases, for
239Np and 239Pu produced in the fuel. In general, the fission product
species which are known to possess stable fluorides are present in the cir-
culating fuel at approximstely the éxpected concentration levels. Tfie best

143

monitors (9lsr and Ce) with convenient half-lives, stable non-volatile

fluorides, and no precursors of consequence typicaliy and consistently show
concentration levels some 15% lower than those calculated from power levels
based upon heat balances for the reactor.

Those elements whose fluorides are known to be relatively unstable
(molybdenum, niobium, ruthenium, tellurium, and silver) are found in the
salt at considersbly less than the expected concefitration. If calculations
of amounts expected are based upon concentrations of 9lsr, about 60% of the

103 13

99Mo, 30% of the °re appears in the melt. It

Ru, and about 30% of the
is not yet possible to state with certainty whether these materials.are
present in the salt as colloidal metal (or alloy) particles or as soluble
chemical species, though present evidence suggests that the former is the
more likely.

Iodine has been found (presumably as I")iat nearly the expected concen-

tration in the samples of fuel. Examinatibn of graphite and metal samples
 

 

@

w

73
and, especially, of specimens from the vapor phase as described below do
show several surprises. |

An assembly of MSRE graphite and Hastelloy N specimens was exposed on
the central stringerrwithin the MSRE core dufing its initial operation.
This assefibly was removed during the July 17 shutdown after 7800 Mwhr of
reactor operation, and many specimens have been carefully examined.

No evidence of alteration of the graphite was found under examinsation
by visual, x-radiographic, and metallograephic examination. Autoradiographs
showed that penetration of radioactive materials into the graphite was not
uniform andrdisclosed a thin (pefhaps 1- to 2-mil) lasyer of highly radio;

active materials on or near the exposed graphite surfaces. Examination of

the metal specimens showed no evidence of corrosion or other danger.

Rectangular bars of graphite frcfi'the top (outlet), middle, and bottom
(inlet) region of this central stringer‘were milled in the hot cell to re-
move six succeésive leyers from each surface. The removed lsyers were then
analyzed for several fission'product isotopes.

The results of analysis of the outer layer from the graphite specimens
are shown in Table 12. It iscleaf that, with the assumption of uniform
deposition on or in all fge*mddergtérjgréphite,‘appreciable.fractions of
Mo, Te, and Ru and a largé fracfiion.of.fihe&Nb are assbciated with the
grephite. No analyses for Tc have been obtained,

1ko_ 89 141  1hh 137

The behavicerf - Ba, Sr;' Ce, Ce, and Cs; all of which have

xenon or krypton precursors, can be accounted for in terms of laws of dif-

~ fusion and half+liVes'of‘tbe'firecurSOrs..'Figfire 16 shows the change in

concentration of the fission'prbduct isotope with depth in the graphite.

140

Those isotopes (such as Ba) which penetrated the graphite as noble gases

 
Table 12. Fission Product Deposition on Surface® of MSRE Graphité

 

 

Graphite Location

 

 

 

~ Isotope 5 Top Mzddlé Bgttom
dpm/cm Percent dpm/cm Percent dpm/cm Percent
of Totall of Totalb - of Totall
(x 107) (x 107) (x 107)
Pno 39.7 13.4 51.4 17.2 3h.2 11.5
132me 32,2 13.8 32.6 13.6 27.8 12.0
103y 8.3 1.4 7.5 10.3 1.8 6.3
Pon 4.6 12 22.8 59.2 2k.0 62.4
131, 0.21 0.16 | 0.42 0.33 0.33 0.25
99y 0.38 0.33 0.31 0.27 0.17 .15
kb, 0.016 0.052 0.083 0.27 0.044 0.14
8sr 3.52 3.2h. 3.58 B 3.30 '2.99 2.Th
1hog,, 3.56 1.38 4.76 | 1.85 2.93 1.1k
1o 0.32 0.19 1.03 0.63  0.58 0.36
3T 6.6 x 107 0.07 2.3x1073  0.25 2.0 x 1073 0.212

 

@Average of values in 7- to 10-mil cuts from each of three exposed .graphite faces.

6

2

Percent of total in reactor deposited on graphite if each cm2 of the 2 x'10" em of
moderator had the same concentration as the specimen.

L

 
 

 

L

_ 4013

DISINTEGRATIONS PER MINUTE PER GRAM OF GRAPHITE

75

" ORNL-DWG 67-774

- 103g, —

 

 

13—

 

o 1 . 20 30 4 50

~_ DISTANCE FROM SURFACE OF GRAPHITE (mils)

 Pig. 16. Concentration Profile of Fission Products
in MSRE Core Graphite After 8000 Mwhr

 

 
 

.

show stfaight lines on the logarithmic plot; they seem to have remaiped at

the point where the noble gas decayed. As expected, the gradient for luOBa

89

with a 16-sec lhoXe precufsor is much steeper than that for “Sr, which hes

~
L

a 3.2-min 89Kr firecursor. A1l the others shown show a much steeper concen-

tration dependence. Generallyrthe coneentration drops‘a faetor of,lOO from
’ the top 6 to 10 mils to the second layer. |
| It is p0951b1e that carblde formatlon is respon51b1e for the depo-
sition of Nb and possibly for that of Mo but it seems quite unllkely for
Ru and Te; the 1od1ne probably got in as its tellurlum precursor.i Since
these materials have been shown to appear.ln the exlt gas as volatlle,
species, it seems likely that they entered the graphite by the same mechan—
ism. The possibility that the strongly ox1d1z1ng fluorides such as M0F6
were present ralsed the questlon as to whether UF6 was accumulating in the
- graphite. An average of 0.23 ug/cm? was found in the surface of the graph-
ite; much less was present in interior samples. This amount of uranium,
| equivelent to less than 1 g in the core, wes'eonsiaered to be negligible.

j T&ble 13 ehows'the extent te whicfi varioué-fissioy product isotopes
are deposited on the Hastelloj N specimefis in the,cofe. A large ffactipn
of the molyfidenum ana tellurium and a substantial frection of the ruthenium

13;1 was carried into

a5

the specimen as its tellurium precursor. The values for °

prisingly high, since those for the lthe ahd 1the with-nobie-gas pre-

seem to be so deposited. It seems'possible that the

Zr seem sur-

cursors probably reflect the amount expected by direct reccil at the moment
of fission.
If the Nb and Tc are assumed: to behave like the Mo, Te; and Ru, it may

be-noted that the MSRE could have been uniformly'plaied during its‘operap

 
 

 

 

- Table 13. ‘Deposition of Fission Products on Hastelloy N in MSRE Core

 

Hastelloy Location

 

 

 

 

| Isotope  Top Middle Bottom
o, 2 2 2
dpm/cm”™ Percent dpm/cm Percent dpm/cm Percent
~ of Total?® of Total® | of Total®
SR (x 10?) . - (x 107) (x 107)
o '7212. k2.8 276 - 55.6 20k hi.2
1320 508 131 | 3k41 88 Lot 110 s
: ' -l
103pu 35.5 29.3  25.5 21 23.2 19.1
131, 8.2 3.8 4.0 1.8 5,2 2.k
g 1.8 1.0 1.8 1.0 2.6 1.3
}MICe 0.05 0.02 0.22 0.07 0.15 0.06
l1_”‘Ce 0.01 10.02 0.09 0.18 0.35 0.07

 

fpercent of total present in reactor which would deposit on the 1.2 x lO6 cm2 of Hastelloy N
if deposition on all surfaces was the same as on the specimen.

 
 

 

T8

tion with several hundred angstroms of relatively noble metals.

The only gas-liquid interface-in the MSRE (except_for~the contact be-
tween liquid and the gas-filled pores of the moderatof graphite) exists in
the pumplfiowl. There a salt flow of about 60 gpml(B% of the total system
flow) contacts a helium cover gas which flofisthrough #he bowl at U4 liters/
min. Provisions for direct seampling of this exit gas are planned.but have
not yet been installed in the MSRE. |

Samples of the liquid fuel are obtained by lowering arsampler, on a
stainless steel cable, through this cover gas and into the liquid. It has
been possible, accordingly, to detect chemically active fiésionproduct
species in this cover gas by radiochemical analysis of fihe.stainless steel
cable and its accessories which contact only the gas phase and by analysis
of special getter materials which are attached to the cable. Coils of
silver wire and specimens of Hastelloy N have generally been used as get-
ters for this purpose. No quantitative measure of the isotopes present in
the gas phase is possible, since no good estimate can be made of the gas
volume sampled. The quantity of material deposited on the wire specimen.
does not correlaste well with contact time (in the range 1 to 10 min) or
with the getter materials studied.

The quantity of material deposited, however, is relatively large.
Table 14 indicates relative amounts found in typical tests. There'is no
doubt that Mo, Ru, Te, (and from subseqfient tests, Nb) are sppearing in
the helium gas of the pump bowl. The quantities are, moreover, surprising-
ly large; if the materials are presumed to be vapors the partial pressures
would be above 10-6 atmosphere. The i¢dine isotopes show perceptibly dif-

ferent behavior. Iodine-135, whose tellurium precursor has a short half-
 

 

a)

79

 

 

 

Table 1. Qualitative Indication of Fission Product
- in MSRE Exit Gas |
Amount ®

Isotope onNi  On Ag  On Hastelloy Fra
Liquid

Mo 8 2 1 4

132Te 1k 6 T 9

105g, 10 3 3 5

106Ru | 6 2 1 1

1351 0 0 0 0

133, 2 1 2 2
131y 1.5 0.5 0.8

0.9

 

8The unit of quantity is that amount of the isotope in 1 g

of salt.

bOn stainless steel csable.

 
 

 

80

131 133

life, does not appear, while ~~ I and ~~-I, both of which have tellurium

precursors of appreciable half;life,'are foun&. These findings--along with
the fact that these iodine isotopes argvpresgnt in the salt at near their
expegted concentration--suggest that any iodine in the vapor phase comes

as a result of volatilization of the tellurium'precufsors.

Early attempts to find uranium on the wires (as from evolution of UF6)
were finsuccessful. More recent attempts--perhaps with the oxidation po-
tential of the salt at a higher level--have shown significantluranium depo-
sition corresponding to several parts pér million in the gas phase. It is
possible, But it seems unlikely, that "salt spray" could account for this
observed uranium. Salt spray certainly does not account for the observed
noble metal species carried in the gas.

The behavior of these fission product species in the gas phaSe'seems
to correlate poorly--if at all--with the UF3/UFh ratio in the fuel melt.
Concentrations of Mo, Nb, Ru, and Te in the gas phase seem to‘increase (or
decrease) together but were unaffected--within the considerable scatter of
the data--by the deliberate addition of beryllium to the MSRE melt. The
concentrations of these elements in the fuel decrease (after correction for
radiocactive decay) during reactor shutdowns; such behavior would be expect-
ed if they plate out upon metallic or other surfaces. Concentrations in
the gas phase decrease somewhat more than>those in the salt but the dif-
ferences seem much smaller than should be attributsble to (for example)
some radiation chemistry oxidation process to produce MoF6, etc.

It seems most unlikely thet these data can be reconciled as equilibri-
un behavior of the volatile fluorideé. It is possible that the MSRE metal

is plated with a noble-metal alloy whose thickness is several hundred
 

 

1}

81
L

angstroms, and it is conceivable that the UF, /UF_ ratio is near 10 . The
, | y/ U3

compound NbF_ could show an appreciable pressure under these circumstances.

>
The other possibilities such as MoF6,.TeF6, and RuF5 would require much
higher UFL/UF3 ratios, and it seems.most uplikely that any single redox po-
tential can yield the relative abundance observed for these isotopes.

The recent findings of‘silver and palladium isotopes in relatively
high concenfiration in the gas phase seem (since these elements certainly
lack volatile fluorides) to case additional dofibt on species such as MoF6,
Rqu, etc. as the gas-borne species.

One possible explanation of the available data is the following: The
noble metal species (Mo, Nb, Ru, Te, Ag, Pd, and probably Tc) occur--as
thermodynemics predicts--in the elemental state. They originate as (or
very rapidly become) individual metal atoms. They aggregate at some finite
rate, probably alloying.witfi one another in the process and become insolu-
ole as very_minute colloidal particles which then grow at a slower rate.
These colloidal particles are not wetted by the fuel, tend to collect at
gas-liquid interfsces, and oan readily be swept into the gas stream of the

helium purge of the pump bowl .They tend to plate upon the metal surfaces

of the system, to form carbides (Nb and Mo only) with the graphite, and

(as extraordlnarily flne smoke") to penetrate the outer layers of the

moderator. Whlle there are difflculties with thlS interpretation it seems

| more plauslble than others suggested to date.

It is clear that further study and addltlonal data from MSRE and from

sophlstlcated in-pile loop tests will be required before'the-detalls of

fission product behavior can be uoderstood.

 
 

 

82

MOLTEN-SALT PRODUCTION TECHNOLOGY
The fuel and blanket salts of a molten-salt breedér reactor céfi Be
prepaied by fechniques similarrto thése develdped for the broductiofi-of
fluoride mixtures for the MSRE. Commercially aveailasble fluoride salts,
which were used as starting materifils for the fluoride prbdfiétion prdcess,
required further purification bnly.to.femove a limited numher of—impurity
species. Chemiéal reactions used to effect éalf purificatiofi énd méthbds
by which process cofiditions were controlled are both edapteble to the
1arger;scale production capabilities that will be required to supply large-
scale MSBR;s. | .
Production Prdcgss
Fluoride mixtures required for the MSRE wefe prepared by-a batch

process in a facility initially designed to support the various:chemicél
and engineering tests of the program. A layout of the produétion process
is shown in Fig. 17. Starting materials were weighed into appropriate
batch sizes and simultaneously transferred.by vibratory conveyor to a melt-
down furnace assembly. In addition to providing a molten charge to each
of two adjacent processing units, the meltdown facilify was utilized for
preliminary purification of the fluoride mixtures. Beryllium%metal turn-
ings were added to reduce structural-metal impufities to their ihsoluble
metallic states. The molten mixture was also sparged with helium and hy-
drogen at relativeiy high flow rates to remove insoluble carbon by entrain-
ment.

 Primary salt purification was achieved in each of two batch prbcess-
ing units. The melts were inifially‘sparged with é géseous mixture of an-

hydrous HF in hydrogen (1:10 vol ratio). Oxides, either initially present
 

0

"

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

e f&s\\‘CONTRbLH

 

    
 
 

S g3
S ORNL —DWG 64-6998
DOWN
| 'SHOWER
" PROTECTIVE '
CLOTHING
STORAGE | | |
: -/ ~SCALES
= LOADING ROOM “<—LOADING HOPPER
SALT
TREATMENT—
o | MELT DOWN |
 SALT T\
RECEIVER -1 | \(/(%m".

 

 

 

PANELS

 

 

o Flg. 17

Operailng Level

' Fluoride Productlon Fac111ty, Layout ofV-

 

 

 
 

 

o 8k
or formed by reaction of the fluoride salts with their adsorbed water on
heating,'wére removed as water by the reaction - i\.

0% + 2HF Z 2F + HO.

As shown by Fig.’18,.the effiéiency of this reaction is QUité high. \Oxide '
removai rates fiere defermified by.conQensing wgter vapo; from‘the gas ef-
fluent‘in a cold trép; | |

| Sulfides were also re@ovgd (as HQS) by réaétion with'fiF.‘ waever,
any'remaining_sulfates/mfist be reduced b&_hydrogén or added berjilium_
metal bef@re sulfur removal by HF treatment ié efféctive. -Althqugh this
‘ impurity_ié difficult to remove, commercial vendors'of'fluoride salts used
in the MS.E were_successful, through process development éfforts,-in‘sub~
stantially reducing this impurity from their.products. CdnseQuéntly, sul- B
fur removal from fluorideKSalts should not be an impfirtfifit consideration
for future prdduction of fused fluoride mixtures for MSBR's.

Nonequilibrium concentrations of structuralqmetal fluoride impuritig§—

»that are more easiiy reduced than'UFh (e.g., fiiFa Qf FeF2) wofild‘result in
the depletion of chromium activity in the}Hastelloy N alioy uséd es the
sfiructural material in the MSRE. Since theée impurities are ffesent in
fluoride raw materials end may also.be introduced by corrosion of the.«_
process equipmefif, their concéntrations in fhe.purified fluqride mixtures
were an important procéss considerétion.- Following‘HF tréatment,.the fluo-

ride mixtures,were_spargéd with' H_ alone ati700°C to effect. the conversion

2

of impurities to insoluble metals by the reaction - - o
_ - o :
| K MF, + H, T M + 2HF.

Hydrogen was also introduced during HF treatment to reduce corrosion of the

hickel salt-containment vessel. Measurement of the HF concentration
 

 

0

o .

HF CONCENTRATION (megq/liter Hp)

5

W

N

-l

g5

ORNL-DWG 64-6996

 

_HF INPUT CONCENTRATION

 

 

 

 BATCH NO. C-#6 | - //0-
| WEIGHT:120 kg L
& FROM ANALYSIS O
| GAS EFFLUENT

 

 

 

 

10 45 20 25 30
.. .. PROCESS TIME (hr) '*

 

 

 

 

 

 

 

 

 

Fig. 18.  Fluoride Production for MSRE; Utilization
‘of HF During Purification of 'LiF-BeF, -

. (66-34 mole %) at 600°C.

 
 

 

86

~

ig the éas éffluént'during hydrogenlsparginé provided a_Convenient.process
control. As showfl by'Fig. 19 the concentration of HFlin ;he_éés‘effluent
‘was indicative of the iron concentration femaining in the salt_mixtures.

* At ‘the conclusion of the fi2 treatment, residual quantifiies of.HF were
" removed by sparging the melt with dry helium, The purified fluoride mix-
ture was then.franéferred to its storaée cofi?ainer. A sintered nickel
filter, inserted in the transfer line, removed entrained solids frqm the
melt. | - | |
| Thué.firifiafy contrél of the production process was exercised by analy-
sis of process gas‘sfireams. Filtered Sampleé of the salt mixtufes were ob-
fiained periodicaliy during the précess'for chemicél'andlyses. This second-
gzy'control measure provided the basis for acceptgnce of‘the_salt batch fbr
use in the MSRE.

235

All the U required for critical operatiofi of the reactor could be

prepared as a concentrate mixture,'TLiF-UFh (73-27 mole %), with UF), that

235

was highly enriched in “°7U. This facilitated compliance with nuclear

safety requirements and permitted an orderly approach to criticality during |

fueling operations through incremental additions of 235

235

UF), to the fuel -
';systemvof the reactdr. Since the density of U in the‘concentrate mix—
ture is relativély high (2.5 g/cc), the fueling methéd.employed‘for the
MSRE should suffice for all pragtical reactor systems.
MSRE Salt-Production Economics
The operatiofi of the production facility for the preéarationvof MSRE
materials waé conducted on a'seVen—day, three~shift schedule at a‘budgeted

~

cost of about $20,000 per month. The raw meterials cost for the 15,300 1b

T

of LiF—BeF2(66-3h mole %) used as the coolant and flush salt was $71.29

 

»
 

 

 

 

 

N

¥

IRON CONCENTRATION IN SALT (ppm)

87

ORNL-DWG 65-2555

 

- 600

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

O BATCH F—178 3 o
X BATCH F=177 7 |
500 —— SALT TEMPERATURE: 700°C -
Ho FLOW: {0 liters/min o
400 | ?
300 |
X
200 |—
100
0 002 004 006 008 040 042

Fig. 19. HF Concentration in Gas Effluent

(mynmrn)

 
 

At o e b et

88 .

per pound and that of the fuel salt (11,260 1b) excluding 235y costs -was
$10.13 per pound. As calculated from operating and raw materials costs

U costs) the coolant and flush salt cost

(but not plant amortization or
$19.71 per pound and the fuel salt cost was $17.33 per pound. Operating
costs, as well as faw_materials costs, should be substantially-reduced for

larger-scale production operations.
 

 

 

o

"

"

89

SEPARATIONS PROCESSES IN MSBR FUELS AND BLANKETS

Use of molten salt reactors as thermal.breeders will obviously require
effective schefies-for,decontamination of the fuel and for recovery of bred
uranium from the blanket. No provision, however, was made for on-streanm
removal of fission products from the MSRE, and fuel reprocessing has re-
ceived less attention to date than have more immediate materials problems
of this and éimilar machines. No detailed rgprocessing scheme has, accord-
ingly, been demonstréted.

Recovery of uranium from molten fluorides by volatilization as urani-

um hexafluoride and the subsequent purification of this UF6 by rectifica-

tion or by sorption-desorption on NaF beds is well demonstrated. Recovery

of bred urenium from blankets or removal of uranium (where necessary to
facilitate other processing operations) from the fuel, therefore, is clear-
ly feasible. Such volatility processing is described in some detail else-

56

where in this series.

57

More recent studies”’ have shown that the LiF, BeF2 (and Zth, if
present) can be recovered quantitatively, along with much of the uranium,
by vacuum distillation at temperatures near 1000°C; very encouraging de-
cohtamingtion-faétors frqm fare earth fluorides (which are left behind in
the stillbottoms).have beeh demonst:ated. As is described elsewhere in
this series’C this distillation procedure combined with recovery of UFy by

fluorination shows real promise as a ffielrprocessing technique.

‘Several other techniques have,shown promise, at least'in preliminary

testing.‘-A brief Summary;ofrthese is presented in the following.

Possible Separation70f Rafe Earths from Fuel

The rare earth fission products, which are the most important nuclear

 
 

 

20

poisons in a reactor fram which xenon is effectively removed, form vefy
stable trifluorides with a portion of that F_; released as fiséion of uran-
~ium as UFh' There is no doubf, therefore, fhat these fission products are
dissolved in the molten fuel and are availsble for reprocessing.
By Solid-Liquid=Equilibria | |

The limited solubility of these trifluorides (though sufficient to
prevent their precipitatibn under fiormal MSBR conditions) suggested &ears
8go a possible recovery scheme. When a LiF;Ber-UFh melt (in the MSBR
coficentration fange) that is saturated withld single rare‘earth fluoride

(LaF_,, for example) is cooled slowly the precipitate is the pure simple

3’

trifluoride. When the melt contains more than one rare earth fluoride the
precipitate is a (nearly ideal) solid solution of the trifluorides. Ac-
cordingly, addition of an excess of CaF3 or LaF3 to the melt followed by
heating to effect dissolution of the added trifluoride and cooling to ef-
fect crystallization effectively removes the fission product rare earths
from solu.tion.hh It is likely that effective removal of the rare earths

and yttrium (along with UF_ and PuF3) can be obtained by passage of the

3

fuel through a heated bed of solid CeF., or LaF_. The price, which is al-

3 3

most certainly too high, is that the resulting fuel solution is saturated

with the scavenger fluoride (LaF, or CeF3, whose cross section is far from

3
negligible) at the temperature of contact.

Since the rare earth fluorides seem to form with uranium trifiuoride
solid solutions similar to those described above it is possible to con-
sider UF3 as the scavenger material. It should be possible to reduce the
fuel UFh to UF3 and then by passage of the solution through a bed of UF3

to remove the contaminant rare earths; in principle, by careful control of

™
 

 

0

91

the columnn tempersture (and, thereby, the solubility of UF3) one could ob-
tain from the column a fuel of-the correct uranium concentration which
could be returned to the reactor after oxidation (by HF or HF--H2 mixture)
of UF, to UFh' While the process deserves further study, the great insta-

3

bility of UF, in solutions of high UF3/UFh ratios and the great ease with ‘

3
vhich the metallic uranium alloys with structural metals will probably make
the process-unattfactive in practice.

Removal of rare earth ions, and other ionic fission product species,
by use of cation exchangers also seems an appealing possibility. The ion
exchanger would, of course, need (1) to be quite insoluble, (2) to be ex-
tremely unreactive (in a gross sense) with the melt, and (3 ) to take up
rare earth cations in exchange ions of low neutron cross section. For
rare earth separations it would pfobably suffice if the material exchanged
normal Ce3+ or La3* for the fission product rare earths; other separation

: ' +
schemes (such as distillstion) would be required to remove the Ce3‘ or

La3

* but they could operate on a much longer time cycle. [The bed of CeF3
described above functions in an ion exchanger; it fails to.be truly bene-
ficial because it is t00flsqlfib1e'in the melt. ]
UnfortUnately,'thérefarefinbt many materials known to be truly stable
to the.fuel_mixture.Zirédhium oxideis stable (in its low temperature

- ke R
form) to melts whose-Zr +/U'IH‘_ ratio is in excess of. about 3. It is con-

~ ceivable that sufficiently dilute solid solutions of Ce,0, in Zr0, would

2°3

be stable and would exchange CeS* for other rare earth species. Inter-

: meta.l;ic' compounds of Ijare.-earths with'moderrately,noblg metals (or rare

earths in very'dilute alloyS:With,suCh metals) seem finlikely to be of use

because they are unlikely to be stable toward oxidation by UF). Compounds

 
 

 

92
with oxygenated anions (such as silicates ahd molybdates) .are decomposed
by the fluoride melt; they, and'simpleoxides(ZrO2 excepted) precipitate

2

as carbides or nitrides) of the rare earths either alone or in'solid di-

U0, from the fuel mixture. It is possible that refractory compounds (such

lute solution with analogous uranium compounds, mey prove uséful.A A con-
siderable amount of_exploratory'research will be required (and many of the
obvious possibilities have alreadyrbeen rejected) before such_a-technique
can be given consideration. |
Bi Reduction

The rare earth fluorides are very stable toward reduction to the |
metal. For example, at 1000°K the reaction

gLaF (¢c) + Be(c) -> %La(c) + BeF

377 3(c 2(2)

where ¢ and £ indicate crystalline solid and liquid, respectively, shows

+ 32.4 kcal for the free energy of reaction. With the LaF. in dilute solu-

3
tion and BeF2 in concentrated solution in LiF—BeF2 nmixture the free energy
change is, of course, even more_unfavorable. However, the rare earth
58

metals form extremely stable solutions”  in molten metals such as bismuth.
Beryllium is virtually insoluble in bismuth and forms no intermetallic
compounds with this metal. Accordingly, the reaction

2 2
3LeF3(q) + Be(o) < Flaqg;) + BeFyrqys

where d indicates that the species is dissolved in 2LiF-BeF2, ¢ indicates
crystalline solid, and Bi indicates a dilute alloy in bismuth, can be made
to proceed essentially to completion. Accordingly, LaFé can be reduced

and extracted into molten Bi from LiF—BeF2 mixtures.
 

 

 

 

 

© ey

"

]

~to

.93
10 IR ool 58,59
Since Li~ also forms stable solutions in molten bismuth, the

process of reducing the rare earths with beryllium cause some reduction of

"LiF and extraction of lithium'by the bismuth. In practice, it is more con-

venient to use 'Li in bismuth (at or just below the concentration which

yields crystalline Beo_atlequilibriUm)'as the reductant. Figufe 20 shows

the behavior-ofrseveral’rare esrths'when extracted from very dilute solu-

~ tioms in 2L1F BeF w1th L1-bear1ng B1 in s1mple equlpment.60 It is still’

too early to be sure that the separstlons avallable are suff1c1ently com-
plete, espec1ally_for heav;en rsre_earths, for the method'to be competltlve
with the distillation process. In addition, it ‘is uncertain whether re-

T

covery of the 'Li will be.neEeSssrj7and if so, how mnchtrecovery would be -

accomplished' However, the process seems at thls prellmlnary stage to be
worthy of further study.

It is clear that reductlon processes of th1s type can, at least in

: princ1ple be accompl1shed~electrochem1cally w1th the molten b1smuth as

-

the cathode and w1th some 1nert anode at which fluorlne gas can be generat-

ed. The concentratlons of rare earth metals, llthlum, and beryllium ob-

~ tained in the molten blsmuth w1ll be 1dent1cal to those Obtalned by chemi-
' 'f_ cal equlllbrlum as descrlbed above.,,Whether one'prefersrtheflelectrolyt1c
| method or direct chemlcal-eQulllbratlon fiill,'aocordingly;-depend upon the

economlcs of the competlng processes.

{
Recovery of Protactlnlum from Blanket

233

While removal of bred U from the blanket by fluorlnatlonss appears

fea51ble, the prlor removal of 33Pa from the blanket to permlt its decay

233U out31de the neutron fleld would be a most valuable contrlbutlon to

the breeding econOMy.r,Such a separative process must be simple, since it -

 

 

 
 

94

ORNL—-DWG 66—2425

10

ANTHANUM

O

CERIUM

-

LITHIUM FOUND IN METAL PHASE
(mole fraction % 402)

0.0 A_
0.04 0 { 10 400

MOLE FRACTION OF RARE EARTH IN METAL
MOLE FRACTION OF RARE EARTH iIN SALT

Fig. 20. Effect of Lithium Concentration in
: . Metal Phase on the Distribution of
. Rare Earths Between LiF-BeF, (66-3k4
mole %) and Bismuth at 600°C.
\

 

 

1000

"

U
 

 

e

‘be diminished appreciably by treatment with HF and then H

95

must be capable of handling the entire blanket in a time short compared

233

with the 31.5 day half-time for decay of Pa.

By Oxide-Fluoride Equilibria | - .
Removel of Pa by deliberate addition to LiF—BeFQ-ThFh blanket mix-

61,62

tures of BeO, Th02, or UO, has been demonstrated. Precipitation of

2
an oxide of protactinium, adsorption of protactinium on the added oxide,
or (more likely) formation of a solid solution of protactinium oxide with
the best oxide, has been shown to be essentially complete, The process
has been-defionstrated to be reversible; treatment of the oxide-fluoride
mixture with asnhydrous HF dissolves the added (or precipitated) oxide and
returns the protactinium to solution from which it can readily be repre-
cipitated. It seems likely that protactinium might be removed from the

blanket by passage of a side stream through a tower packed with ThO (or

2
possibly Be0O); the protactinium, in some unidentified form, would remain

on the bed and would there decay to 233U outside the neutron field. In

its passage through the packed bed of oxide the blanket melt becomes satu-
rated with oxide ion. This oxide ion concentration would probably have to
> before the melt
could be returned to the blankét sfiream.

. EE Reduction

The possibility ofrecovery'of pfotactinium from realistic LiF-BeF,-
ThFh blanket mixturesby_redgction has been‘examined experimentally with
su;prising'and‘encouraging_results.63' No real information exists as to
the free énergy formetion of thefluorides of protactinium. Accordingly,
' 233

experiments were performed in which traces of Pa were added to LiF-BeFe-

ThFh melts, the melts were carefully treated with HF and H2 to insure

 
 

 

 

96

conversion of protactinium to fluoride and its dissolution in the melt,

and the solution subsequently treated with a strong reducing agent. Some

experiments used ThPb, in lead or ThBi3 ifi bismuth as the reductant; other

2

tests have used metallic thorium. Ih each case, the protactinium remained
in molten fluoride solution (as judged by radiochemical analysis of filter-
ed samples) until the reducing agent was added and was removed, upon ad-
dition of reductant, to very lofi concentration levels. Figure 21 shows\
the data for a typical case. The removal has been shown to be nearly
quantitative at both traces (less than part per billion) levels and at
realistic concentrations (50 ppm) of 231p, traced with 233pg. The process
has also been shown to be reversible; sparging of the system with HF or
HF-H2 mixtures returns the protactinium quantitatively to the molten fluo-
ride solution.

lRecovery of the precipitated protactinium has proved to be more dif-
ficult. Attempts to obtain the deposited protactinium in molten Bi or Pb
have been generally unsuccessful in equipment of irdn, copper, niobium,
or steel; the deposited protactinium was only fleetingly (if ever) dis-
solved in the molten metal. When thorium was used as the reductant no ap-
preciable concentration of protactinium was found in the excess thorium.
Careful examination of sectioned apparatus shows some protactinium on the
vessel walls, and some appears to remain suspended (in easily filterable
form) in the salt. The mechenism of removal of protactinium from the salt
mixture remains far from certain. It appears likely that the thorium (or
slightly weaker reducing sgent) reduces protactinium to form a moderately

stable intermetallic compound (perhaps with Cr or Fe) which is filterable,

is not dissolved by the molten lead or bismuch, and is readily decomposed

O

4
"

"

 

 

233pg IN SALT PHASE (%)

60

40

- 20

 WITHOUT ADDED || WITH Odwt% |
- THORIUM ol Th IN Pb ADDED
L Eg 1 e
< | | |
) £LL_ |
o _20 40 - O 20 40
Bt 3 EXTRACTION TIME (hr)
- Fig. 21. Effect of Thorlum Metal on the

97

 

ORNL-DWG 66-970

 

 

 

 

'WEIGHT OF SALT: 6004g
WEIGHT OF LEAD: 9009

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Extractlon of

233Pa from L1F—BeF2-ThFh

(73—2—25 mole %) in Salt-Lead System at 600°Cr

 
 

98

by aahydroue HF./’

Attempts to recover the protactlnlum by reductlon with metalllc thor—
ium in steel equipment in the presence of added iron surface (steel WOOl)
have shown some promlse.sh In a typical experlment, some 320 grams of
LiF-ThF) (27 mole % ThEh) contaiaing'Blppm (26 mg) of‘protactiniafi was
reduced with thorium«in the presence of I grams of steel weol. The.
LiF-ThFh, previously”purified,_was placed in.awelded nickel~reaction

233 231

vessel, irradiated ThFh containiné & known amount of . Pa and Pa was .

added to the mixture, and it was'treated firstrwith & mixture of HF and H,

‘and then with H, alone. Four grams of steel wool (grade 00, 0.068 m /g

2

surface area) was placed in a low—carbon-steel llner inside another nickel

vessel. The contents of th;s vessel were then treated with purlfled hydro-

- gen at,800°0 for several hours torremove}as much as possible of the oxide
s ‘ ' : :

surfaee contaainatioh of the steel weol and liner. The two vessela fiere
then'connected together at room temperature and heated to about 650°C, and
the.salt was transferred to the steel—liaed vessel. After two separate‘
exposures of the salt to a.solld thorium surface, as 1nd1cated in Teble 15,

-

the salt was transferred back to its original conta1ner and allowed to
cool in helium. The steel—lined vessel was cut up, and samples were sub-
mittedkfor‘analysis. |

The data in Table 15 show that 99% of the protactiniuéiwas;precipi-‘
tated in a form that would not pass through a sintered copper filter after

a fairly short exposure to solid thorium, but nearly 7% was in the un-

filtered salt that was transferred back to the nickel vessel after ex- .

posure to thorium. About 69 g of salt was associated with the steel wool

in the steel liner in the form of a hard ball. ' Partial separation of the
 

 

”

0

99

salt from steel wool was effected by use of a magnet after crushing the
ball, and the ifen;rieh fraction hed‘the higher protectinium concentration.
The small amount ofprotactiniufi found on tfie‘veesel wall is especially
notable. “Coprecipitatioh of metallic protactinium and iron (and possibly
nickel) wouid help to account for the manner in which protactinium settled
out on, and adhered tp, the steel wool surface.

On the basis of presently available information, thorium reduction of

protactinium from molten breeder blanket mixtures in the presence of steel

- wool is believed to be a promising recovery method warranting further in-

vestigation.
Recent experiments have shown, in addition, that the handling of pro-
tactinium is simplified somewhat if graphite serves as the container.

233Pa) is dissolved in molten

When irradiated thorium metal (containing
bismuth in metal conteiners the protactinium disappears_fram the liquid
metal solution rapidly. Simiiar experiments using grephite vessels show
very slow negligible decreeses in protactinium concentration (after cor-
rection for radioactive decay) with time.

Accordingly, recent studies of reduction of protactinium from molten
fluoride solution have been eqnductedrin_vessels of graphite. An inter-
esting assembly whieh.has_been Stgdied in a preiiminary way uses & cylin-
drical grephite crucidble (as aiiinef inside a stainless steel or nickel

vessel) containing a pool of molten bismuth and a central cylindrical

chimney of. graphite with'ifis_ldwer.end immersed in the bismuth pool. The

chimney is cohnectedfto the 1lid of'the metal Jjacket vessel in a manner

~ such that the central chamber and the annular outer chember can be main-

tained under separate and different atmospheres. A LiF—ThFh blanket mix-

 
 

100

Table 15. Precipitation of Protactlnlum frqm Molten L1F-THFh '
(73-27 Mole %) by Thorium Reduction in the Presence of Steel Wool

 

 

231, . Total
Semple Concentration ~'231Pa“"
(mg/g) (mg)
Salt after HF-H, treatment ' .0.0634 | | 20.3
Salt just before transfer - 0.081 - . 26.1
Selt 35 min after transfer ©0.079 24.9
Salt after 50 min thorium exposure 0.0026 0.69
Selt after 45 min thorium exposure ' 0.0009 0.27
Nonmegnetic fraction of materiel 10.20 11.5
in steel liner
Magnetic fraction of material in 0.628 ' 10.2
steel liner -
Unfiltered salt after transfer to 0.0076 - 1.75
nickel vessel ' |
Steel liner wall 0.0006
Stainless steel dip leg 0.53.
Filings from thorium rod 0.29
All salt samples 1.35

Total protactinium recovered 25.5

 

 
 

»

"

101

ture containing protactinium‘fluoride is placed in the annular chamber and
& LiF-NaF~-KF mixture is_plaéed.in the inner chamber. An atmosphere of HF
is used to sparge»the LiF-NaF-KF mixture and a reducing metal and (beryl-
lium or'thofium) is added to the salt in the outer chamber. The protac-
tinium fluoride in the outer chamber ié reduced, dissolved in and transfer-
red thfough the bismuth and is oxidizedAby HF and dissolved in the LiF-

NaF-KF mixture in the inner cylinder. Additional study is necessary to

‘establish (1) the rate at which such a system can be made to work, (2) the

quentity of reducing metals transferred to the recovery salt, and the

| completeness to which the reaction can be easily driven. The system--

~which seems to have several useful variations--does, however, look promis-

ing.

It is also clear that, as in the rare earth reduction process,
electrochemical reduétion 6f the protactinium fluorides should be success-
ful. In fihis case, it might seem especially promising if (as now seems

likely) the protactinium is being reduced in the presence of metal to a

‘stable intermetallic compound. Attempts to reduce protactinium electro-

chemically with a variety of metallic electrodes to ascertain (1) the

type end composition of the intermetallic compound, and (2) whether a

- simple'recovery process with;a'solid'electrode'can be achieved are

scheduled for study.

 
 

 

 

 

102

MSBR IN-LINE ANALYSIS PROGRAM

The rapid acquisition of data concerning the compositions of the fuel,
coolant, and cover gas is highly desiréble in the operation of a fluid fuel
reactor. To be of most fialue the data should be representative of the re-
actor at zero time preferably with as little time-delay as possible in
order to evaluate changes in composition'ffom normal conditions and to
take requisite action. This state can only be attained by in-line analy-
sis. Investigastions are under way to‘developinstrumenfation capable of
providing instantaneous data. "It is proposed to devote considereble ef-
fort in this direction as part of the MSBR program. The alternative is to
sample the fuel and coolant.at periodic intervals and remove the sample .
for enalysis at an appropriate analytical laboratory. This procedure is
time-consuming and thereby suffers obviously from a definite time lag in
providing information so that unknown events and information concerning
these events are out of phase.

Although in-line instrumentation is a well-established technique, its
application to molten salt reactors is essentially in its infancy - parti-
cularly in regard to radiation and its effect on maintenance of operating
equipment. The objective is thus to apply the successful in~line techni=-
ques that hawe been used to control many nonradiocactive chemical processes
to control the reactor fuel, coolant, and cover gas.

Helium Cover Gas

Ip addition to the anticipated impurities (atmospheric contaminants,
CFh’ Kr, and Xe) have been found to represent significant contaminants in
the MSRE off-gas system. While it has not yet been possible to measure

hydrocarbons in the MSRE blanket gas, ofganic deposits have seriously
 

 

L4

103

interfered with the operation of the MSRE off-gas system and hydrocarbons

‘in concentrations of severaluhundred.parts per million have been found in

the off-gas from the MSRE pump test loop and in simulated pump lesk ex-
periments. (These experiments indicate that most, if not all, of the
hydrocarbon enters the pump bowl through a mechanical joint which can be
welded ifi future models.)  In these tests the total hydrocarbon concen-
tration was measured continuously by a flame ionization detector and the
individual_hjdrocarbons -~ principally light unsaturates -- were identi-
fied by ges chromatography.

Gas chrqmatography is a near perfect technique for automated analysis.
This technifiue is now highly devéloped ahd refined, and considerable ex-
perience hasrbeen gained from research in other reactor programs on the

analysis of helium by gas chromatographic techniques. The determination

of perménént gas impurities in molten salt reactot'blanket gases will re-

quire an instrumentrof imprQVed sensitivity thét~is compaiible with
intense radiatibn.. A simple ohrdmatograph has been used to measure ppm
and lower concentratlons of H 2, 2, CHh,_Kr, Xe, and CFh in the off-
gas frcm an MSRE in—p1le test. Thesé contaminants were resolved on a 10X
molecular sieve column and measured w1th a hellum dlscharge detector,

which has the follow1ng llmlts of detectlon.

 
 

 

104

- Table 16. Sensitivity for Detection of Contaminants in
Helium by Gas Chromatography

 

 

Component Parts per Billion

H20 : | 1 |
H2 100
02 >10

. Kr : >10
N, -
CHh | o >10.
CFh 20
co 20
Xe | - 10

 

The defermination of H20 and 002 will require a morelcomplex instru—
ment with multiple columns; probably a three-cplumn instrument will be
required for all fihe above camponents. Also it will be-necessary to eli-
minate all organic materials of construction completely if.extendéd depend-
able operstion is to be obtained with highly radioactive samples. An all-
metal pneumatically actuated sampling vaive is being developed for this
application. This valve will also be operable at high temperaturgs to
minimize the adsorption of traces of méisture.' The effects of hydro-
carbons on the chromatograph has not been tested bu@ will probably require
some modification of the proposed instrument.

Gas chromatography is the most highly developed method for the suto-
matic analysis of hydrocarbon mixtures; however, the resolution of the
complex mixtures anticipated in thg blanket gas requires columns packed

rwith organic substrates, which are not compatible with the highly radio-

active samples. Also, the experience with the pump test loop has indicat-
 

0"

_saits haxe'been_obtained.

105

ed that the continuous_measurement,of,the total concentration of hydro-
carbons would provide adequate information for. reactor operations. These
measureménts, made with a\flame ionization. detector, provided data to dif-
.ferentiaxe between possible locations of lesks; conversely, the complete
analyses were of value only in development studies for the selection of
means of removing the hydrocarbons. The flame ionization detector would

probably nof_be suitable for in-line analysis of the reactor blanket gases

because its operation would inject substantial quantities of air inmto the

off-gas system. An alternate method which will not introduce contaminants
is'being'develofied. VIn this tgchnique the hydrocarbons are oxidized to
carbon dioxide and water with copper oxide, and the thermal conductivity
of the combusted stream is compared with that of the same gas after the

002'and H20are removed by ascarite and magnesium perchlorate. This
method has been tested with a bench top apparatus and found to give a sig-
nal proportional to hydrocarbon concentration over the range of interest
with a limit of detection below 10 ppm. If possible, a similar spparatus
will be tested on the off-gas of the MSRE. |
Spectrqphotcmet:y of Molten Salts

Absoyption spectfophbtpmetfy'and'eléctrochemical analytical techni-

ques are potentiallyiapplicable,for'in-line analysis. The absorption

spectra of separate solutibfis'df:U(IV):andrU(III)_in fluoride~base molten

65&66_ Baged on a consideration of these spectra,
U(iII),péuld be detérminedfat a;wavéléhgfih”of,360'mu to a concentration

level of ca. 300 ppm in the presence of up to 1 mole % of U(IV) in molten

LiF-BeF,-ZrF) . ‘Such a spectrophotometric method, which is based on a

characteristic spectrum, would be a specific and direct method. Performed

 
 

 

 

106

"in-line," this determination would provide a direct, specific, ‘and con-
tinuous monitor of the U(III) concentration in the molten fuel sélt.-' 
Similarly & relatively weak peak at 1000 my in thé absorption spectra of
tetravalent uranium could be used to'monitof U(IV), provided the concen-
tration of U(III) does not exceed about 1000 ppm. Any corrosion products -
in the molten,sait, even if present at Several t;mesTthe concefitr&tion
level that is expected, will not interfere with the proposed determinations.
The effect of the spectra of the various fissidn.firoducts is-not known
primarily becéuse their equilibrium oxidatién-sfiates'are.not*known with
¢ertaifity. It seems reasonable to assume, however, that little if any ef-
fect fiill be 6bserved. Perhaps the most interference will pefrom the
rare earths, probably as soluble fluorides. 'On the basis of experimental
evidence the rare-earth spectra in molten fluoridq salts should present
sharp but insensitive absorption peaks.

Recently a very intense absorption peak at 235 mpu has been found for
U(IV) in LiF-BeF,, melts. Preliminary estimates indicate that this peak
could be used for the in-line measurement of uranium concentrations as
low as 5 to 10 ppm. If no interfering ions are present,‘the-fieak'could be
applied as a sensitive detector of leaks into coolant salt streams and to
meaéure residual uranium in depleted reprocessing streams.

The design of a spectrophotometer to be used in these proposed appli-
cations is rather well defined. Modification of an existing commercial
spectrophotometer, a Cary Model 1L4-H manufactured by Applied Physics
Company , wil; adequately meet the design criteria. In order to eliminate

most of the radiation which is present in the salt sample the optical path-

length of the spectrophotometer will be extended ca. three feet; at the
 

 

 

2»

[

"

»

107

same time the imaging‘of~the optical system will be-modified-tofprovide

more intense illumination of the sample area.

It appears that'the'piping which will deliver'the_molten saltvto the

 sample cell can be extended”a_convenient'distence from the reactor core

so that environmentalrrafiiétion mey be no problem to the servicing of the

electronic components of the spectrophotometer. If the radiation is a&bove

tolerance, separation of electronic and optical components can best. be
handled by building oneminstrument_housing the optical components and an-

other instrument containing'the electronics. Schematic diagrams of the

-cell de51gn and spac1ng are. shown in Flgs. 22 and 23..

A

If the sPectrophotometer is to monitor the spectrum of U(III) con-
tlnuously and monltor the. spectrum of U(IV) occa31onally, this type of
repetltlve analy51s is readlly adaptable to an automatlc cycllc operation

with the data recorded by dlgltlzlng equlpment

,.Electrochemical Studies

 

In pr1nc1ple electrochenicalranalyses_of molten salts are attractive

for in-line analysis sincefffie'technique-lends itself somwéli ‘to remote

.operations. In- addltlon, any spec1es 1n solutlons that can be ox1dlzed or
:',redficed:iS'determln&ble byielectrochemlstry. The chem1ca1 behaV1or of the
‘.;_solutlon and the reactlons 1nvolved must be known- however.r Ideally, one
-[fcould establlsh the normal potentlal of . the fuel and observe fluctuatlons

:'iand ‘deviations from thls norm._ In thls manner the normal operatlng be-

hav1or of the fuel 1s known and presumably changes 1n thls behav1or wouldr

~be-correlated wlth.observeditranslt;ons._ To accomp11sh thls task a reli-

.able reference electrodefliscneeded.; To}this,end, it is planned to in-

\

'vestigate rarious metal—metsl_ion-couples (nickel-nickel fluoride, nickel-

 

 

 
 

 

V/a-in- diam CELLl\\

Al,03 WINDOW—_

108

ORNL-DWG 65-3975A

T

FREEZE VALVE

 

   

 

 

 

 

 

A N N AN AN

 

N e e N

 

 

 

 

_FOCUSED LIGHT BEAM

 

r——

 

 

s
W

~—__Y/g-in-diam APERTURE

 

 

 

He —

MATERIAL : INOR-8"

'Fig. 22. Molten Salt Reactor In-Line Spectro-
 photometric Facility

"

L]
 

 

 

e

@

1}

*

109

" ORNL- DWG. 65-3976

'MIRROR

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Fig. 23. Cell Space Optical Design

 
 

...‘-_-,,_,,,

bridge" between the Ni-NiF

felectrode,68 and an isolated couhter‘electrode,

110

nickel oxide, berYllium—beryllium fluoride, for,exampie)'as possible refer-

ence electrodes thatrare compatible fiith flaoride melts. -In praetice, the
metal-metal ion reference appears to be rfie best choice from_the standi
point of investigating and setting up of electroanalytical methods for an-
alysis ef molten fluoridee. One model of tfie Ni;Ninelectrode fias been

tested and found to be reversible and reproducible‘but of limited service

life. The useful 11fet1me of this electrode is limjited to a few weeks by

the dlssolutlon of a thin membrane of boron n1tr1de Wthh serves as § salt :

2

be mechanically feasible to replace electrodes in reactor process streams .

half cell and,the molten sample.. While it mey

periodically, a much more dependsble system could be constructed if an in- _

sulating material that is compatible with molten fluorides eould be dis;
covered.

A three-electrode system, an indicator electrode,67 quasi-reference

: 69

has been applied to

molten salts successfully. Approximate potentials for observed electrode

reactions for several electroactive species are shown in Fig. 24. Theo-~

retically, it is pessible to measure the concentration of the meial ions
at their decamposition poteatials indefiendent\of the presence of other
metal ions as long as the potential difference is at leasf 0.3 v. The
presence of gross auantities of one metal and traee quantitiesrof another
often results in swamping of the decamp051t10n potentlal.

This technique has already been applled to samples from the MSRE to
determlne the oxidation state of iron and: nlckel which appeared to be

\

present' in the fuel in concentrations above that predlcted to exist in

equilibrium with INOR-8 at the_obServed concentrations of chromium. Con-

"

“
 

 

n

111

ORNL-DWG. 66- 6277

- CATHODIC

 

 

 

~ ANODIC |
L I R L T

+20 #O N ¢ -0 -2.0

POTENTIAL ,
7 VOLTS VERSUS PLATINUM QUASI-REFERENCE ELECTRODE

Flg. 2h. 'Approximate Poten’cials for Electrode
Reactlons in Molten L:LF-BeF -Zth ,
(65 6-29 h-s 0 mole %, at Temp 5oo°c).

 
 

12
centrations of ionic iron and nickel éf only about 10 and 1 ppm were défer—
mined by véltammetrié scans of remeltgd samples that had been.withdrawn
framthe_flSRE before it was operated at power. These_valueé compare with
tqtal concentratiohs.(determined cfiemically) of 125 and L5 ppfi, respective—‘
lf,land\indicate that the major fractions of these cbntéminants are prob;
jablyrpfeseht és fineiy divided metals. Thus thg'concentrations of thesé
corrosion producfs.in true ionic solution are more consistent with thermb—
&ynamlc predlctlons. | | |

The three-electrode system also offers significant potentlal as a
technique for in-line monitorihg_qf'uranium in reactor fuels. In MSRE
. type melts at 500°C U(IV) to U(III) reduction waves have beén'found to be
reproducible to befiter than 1% in measurements during e twouhour period,
and to about 2 to 3% for intermittent measurements taken over a one-month
period. If the reproduc1b111ty could be 1mproved the technlque could also
be used to_measure trivalent uranium. The\ratlo_of reverse to forward
scan currents is-unity vhen only U(IV) is present, but the ratio}increases
as UF, is added to the melt. One limit to the-reproducibility.o;‘the

3

voltammetric measurement is the precision of definitiofi of the area:of the
indicating electrode. With present instrumentation itiis neceésary to -
limit the electrode area by inserting a 20-gauge platinfim wire only 5rmm
into the meltsrto limit the_currénts to measurable valfies. It is appérent
.that only a small change in melt level will produce a significant error in
electrodeareé. / | |

A new voltammeter is being built that will fieasure twentyfold-higherk'

currents so that an electrode with more repfoducible area can be used.

This instrument also permits faster sweep rates which will minimize theV

1" .
 

113

effects of stirring in flow cells which will be necessary for process an-
alysis. With these refineménts it is possibie that uranium can be continu-
ously monitored with accufacy_that is comparable to that of hot cell analy-
ses. An alternate method for défining the electrode area is to use an
inéulating sheath. Boron nitride sheaths have been used with some success

but are slowly attacked by the salts. The teéhnique would be greatly

‘simplified if & really compatible insulator were available, and a materials

development program would appear to be merited.

A new phenomena which may offer a combihed electrolytic and gas analy-
sis technique for oxide determination has recently been observed. When
LiF--BeF2 melts are electrolyzed in vacuo at the potential (+ 1.0 v) of an
anodic_wave which has been attributed to the oxidation of o#ide ion, gas
evolution is noted at the indicator electrode. The gas was found to be
predominantly 002 (resulting from the reaction of electrolytic oxygen with

the pyrolytic graphite electrode or the graphite container) with lesser

quantities of CO and 02. If 100% current efficiency can be achieved, a

coulometric method would result. Alternately the evolved gases could be

purged from the electrolytic cell and analyzed ges chromatographically.

Determination of Oxide by Hydrofluorination

The duantitative evolution:of oxide as water by hydrofluorination of

“molten fluéride salt:mixtfires has been successfully applied to the deter-

mination of oxide in the highly radioactive MSRE fuel samples. The sampl-

ing ladle, contéining about 50 g;Of salt, is sealed in = nickel hydro-
fluorinator with a delivery tube spring-loaded sgainst the surface of the
salt."After7the'system is pfirged at 300°C with a hydrofluorinating gas

mixture of anhydrous HF in hydrogen, the salt is melted, the delivery tube

 
 

 

 

 

11k

is driven beneath the surface of the salt, and the melt is pfirged with |
hydrofluorinating gas mixture, the oxide being-evolved as water. ' The ef-
fluent from the hydrofluorinator is paséed through a sodium‘fiuoride
column at 70°C to remove the HF, and the water in a fractioh of thié‘gas
stream is measured with the cell of an electrolytic moisture monitor. The
integrated signal from the moisture monitorlcell is proportional to the
concentfation of oxide in the sample.‘-The water is efolved éuite rapidly
with anaelyses essentially complete within aboutl30 minutes after the salt
is melted. Most of this time is consumed in purging the water ffam the
sodium fluoride trap and in "drying down" thé cell. |

The components required to carry out this determination in the hot
cell are shown in Figs. 25 and 26. Figure 25 shows from left to right:
the samplifig ladle; a nickel liner, which protects the hydrofluorinator
bottom; the hydrofluorinator top, with its replaceable delivery tip and
baffles to retain the salt in the liner; the hydrofluorinator bottom and
a clamping yoke to seal the hydrofluorinator via a Teflon O-ring. Figure
26 shows the assembled hydrofluorinator in the furnace on the right con-
nected with a pneumstically actuated coupler to the compértmentrwhich_
contains the sodium fluoride column, the moisture cell, a capillary gas
stream splitter and the necessary valving and connections. |

At the reactor startup samples of flush salt and fuel were analyzed
by both the hydrofluorination and KBth methods with satisfactory agree-
ment. The KBth results were positively biased by abéut 20 ppm fihich is
readily explained by atmospheric contamination of the pulverized salt.
Since the reactor has been operating at power the results of the éamples

analyzed have fallen in the renge of 50 + 5 ppm.

i»
 
 
 

 

 

 

 

 

 

 

 

 

 

 

 

 

o ek « .
=
@ _
T 55 ;
o b | |
x
) Q. | | |
| m oy , o
o 3 .
o |
oy . ,
b : |
3 o
| e 4 ‘ ‘
. 5 ,
o : M
4 _ W
, T |
i = | | :
o - |
- . | ,
et .. , ) o
R A R .2 _ S
PR Te Nz rel= han bl i (oINS MVO g - _,
i . B
_ 8 o
v . L
, 3 o
. A NS ,
[ . . g . ! i ; : Lo ! : - g .. | ‘
4 ‘ " : . .

 
 
 

 
 
 

 

 

 

 

 
 

 

 

 

 

o  Fig. 26. Hot-Cell Apparatus for Oxide in
' MSRE Salts.

 

 

PHOTO 82177

 

 

EY

«

 

 
s s b e

 

117

The hydroflueridatdon method should be equally applicsble to the
analysis of MSBR samples, as no interference is anticipated from thorium.
Because the reaction is rapdd and quantitative it offers promise for ap-
plication to process analysis and might also be combined with a determi-
nation of reducing power. The reactions involved in the process determi-

nation are as follows:

2

0~ + oHF 3 H,0t + 2F (1)

and '
UFy + HF 2 UF) + 1/2 Bt (2)
or M° + nHF 2 MF_+ n/2 Ht , | (3)

with evolved water and hydrogen measured.

Application of reaction (1) could be carried out by either of two
techniques. In the simplest approach the molten salt would be subjected
to a single-stage equilibration with HF in.a hydrogen or helium carrier
and the oxide computed from known equilibrium constants. This approach is
subject to several problems; the most serious of which is.that activities
of oxide rather than concentrations are measured. Thus precipitated

oxides are not determined. Also, it would be necessary to maintain accu-

rate;temperature control because equilibrium constants of the reaction

are relatifely dependent on_temperature} An slternate approach which

would circumvent the sbove problem but would requlre a more complex ap-

rparatus is to equlllbrate a8 constant stream of the fuel with a counter-

" current flow of - hydrofluorinatlng gas. By proper selection of parameters

(HF concentratlon temperature contractor des1gn and flow rates) it is
theoretlcally poss1ble to approach quantltatlve removal of oxide from the

effluent salt so that a steady state is reached in which the water evolved

 
 

118 b

is equivalent to the oxide introduced in the salt stream. Rate constants
for hydrofluorination are not available.
Thorium and Protactinium

All of the experimental work on the proposed methods has been carried
out on MSRE type salts but should also provide adequate analyses for the
MSBR fuel. In the analysis of the MSBR blanket the presence of thorium |
and protactinium must also be considered. A§ this time the in-line analy-
sis of.fhorium does not appear essential to the operation of the reactor --
&8 possible exception is the monitoring of thorium in the core to detect
leaks between the blanket and thg éore; Also, on the basis of its spectro-
photometric, electrical and therfiédynamic properties, thorium is #ot ex—
pected to interfere significantly fiith any of the proposed methods. The
in-line analysis of protactinium must be cofisidered as a priority determi-
nation because the concentration of protactinium must be maintained at a
low level in the blanket for efficient breeding. In the sbsence of.ei- | ‘
perimental data, the spectrophotometric method appears to offer the most
profitable avenue of investigation. ,;
Reprocessing System |

Monitoring of the continuous fuel reprocessing system will probably
be of even more 1mportance than the monitoring of the main reactor ;ystem,
because the compositions of the reprocessing streams are more subject to
rapid operational control. Moreover, the reprocessing system offers
several avenues for the temporary or permanent loss of fissipngble
material. Salt streams which will require continuous measurement of trace
concentrations of uranium include the effluents from fluorinators of the

fuel and blanket reprocessing streams. Part of the residual uranjum in \EJ
 

of

119

either of these streams is subject to permanent loss either in the still
bottoms of the fuel Sysfiem or in the fiaste of the blanket fission product
disposal.. The in—line analysis of major concenfrations of uranium in the
meke-up stream frqm.the recombiner will also be required for inventory
control. With the possible exception of a change in the concentration and/
or nature of the corrosion products the techniqueé that are developed for
the analysis of reactor salts should be edually applicable to the reproces-
sing systém;

| Gaseous effluents streams from thé UF cold traps and the recombiner
systém could introduce temporary losses via transfer of uranium to the off-
gas.syStem and will reqfiire in-line analysis for trace concentrations.

Gas streams that contain major concentrations of UF6 (e.g., effluents from
the fluorinators) can probably be adequately monitdred by ultraviolet
spectrophotometers, bfit no cofipletely satisfactofy methfids,have yet been
found for the in-line analysis of trace concentrations of uranium in gas
streams. OSeveral techniques are being considered to monitor the Fluid

Bed Volatilify Pilot Plant, and any mefhods developed should be ideally

suited for the MSBR reprocessing system.

 
 

 

 

120

PROPOSED PROGRAM OF CHEMICAL DEVELOPMENT

The chemical status of molten fluorides as reactor materials, present-
ed in some detail in preceding sections, indicates strongly that thermal
breeders based upon these materials are feasible. The discussion<afiove,
however, points out several problem areas that remain and numerous speci-
fic details that require examination and experimental-invesfigation. A
brief summéxy'of these areas and specific plans for the necessary studies
islpresented under the several headings below. An estimate of the manpower
and money required, over the next 87years, to accomplish these research
and development activities is presented as Table 1T.

It is axiomatic that the course of researcfi and development activities
is seldom smooth and is difficult to predict in detail for a long period.
Researches lead to valuable findings that can be exploited, and unsuspect-
ed problems arise and require additional efforts for their solution. It
is very unlikely, therefore, that this budget breakdown will pfove accurate
in detail, but the overall sums and manpower, year by year, should be suf-
ficient for the purpose.

Phase Equilibrium Studies

Equilibrium phase behavior of the proposed fuel and blanket systems
is relatively well established. A careful and detailed examination, using
all the most advanced techniques, should be made of the region close to
the proposed compositions in the fuel and, especially, in the blanket
system. In addition, the join from the LiF—UFh eutectic, through the fuei
composition, to the barren fuel solvent will require some examination.
None of these studies is urgently needed in the next year or fwo. Be-

havior of the LiF—Ber-UFh‘system with moderate fractions of the UFh
P YU P

 

 

 

»

for MSER Program

 

Table 17. Projected Breskdown of Chemical Development

 

 

 

 

 

 

 

 

 

 

 

Development Area 1968 - 1969 1970 1971 1972 1973 197y . 1075
- M % MY $ ‘MY $ MY @ $ M - $ MY $ M. $ MY $

Phase Relationships o . A ' : ‘ : : 4 -

Fuels and Blankets . 0.5 20 0.5 20 1 ko 1 40 0.25 10 0.25 10  0.25 .10 . 0,25 10

Coolants w27 B0 1 10 1 ko 0.25 10 0.25 10 ° 0.25. 10 0.25 10 0.25 10
Oxide and Oxyfludridé‘thfivior 2 : 9d -‘ 2.1' - 90 2 90 a1 | ko 1 ko 1 : s I ‘ 1 ko .f 1 ko
Solution Therfiod&nahiééf‘  : 2. 80 2 80 2 8 2 . 8 2 8 2. .8 2 80 2 80
Physical Properties’ = | 3. 10 . & 10 b 150 4 150 2 8 2 & 1 k.
Rediation Effects = 5. 35 8 s00 10 80 10 80 _ 10 80 . 10 80 8 500 . 6 koo
Fission Product Behavior : 2 100 3 150 -3 150 3 150 3 150 3. 150 -2 100 - 100
Protactinium Chemistry 2. 80 2 80 3. 120 3 120 2 8 1 50 1 50 1 50
Fission Product Separationms 2 80 2 80 3 120 - & 160 L 160 2 80 2 \ 80 1 bo
Development of Continuous 0 0 0 o 2 100 2 100 2 100 0.5 25 0.5 25 O : 0

Production - : e ' ' : E ‘ '
Chemical Services 2 .70 2 _To 2 10 .3 _ _105 3 205 4 10 - b 1o 6 200

Subtotal | 22.5 1050  26.5 1260 33.0 1760 33.25 1755  29.5 1615  26.0 1465 22,0 1075 = 19.5 930
Analytical Developmgnt.“' 2 80k 160 b 115 S 225 5 225 4 170 - 170- . & 170
Analytical Services 3 100 -4 130 5 160 6 180 6 180 6 180 6 180 6 180

Total 27.5 1330 3h.5 1550 42,0 2095 © Wh.25 2160  40.5 2020  36.0 1815  32.0 1425 29,5 1280

 

 ®poes not include a considerable éapital_expenditure:for in~pile facility.

Estimate for thi

-~

s might be as high as $600,000.
 

122

reduced to UF3,,and more deflnltive information as to solubility of rare-
earth fluorides, alkaline earth fluorides, plutohium fluoride, etec. in the
mixtures near to the fuel mixture are of somewhat more urgency. At the

modest research level shown, theée,studies should be largely cpncluded in.

s four-year period, and very minor efforts are projected beyond that inter-

- val.

3

Phase Behavisr in the fluoroborate systems (or the altérnatives,pre-
sented above) proposed as-coolsnts is much less well understood. The items
of first priority are to define the phase behavior (ificludiné'equilibrium

BF3 pressure)‘in the NaF+BFé system, such studies should’ include a syste-

matic examlnatlon of the effect of B2q3 on the phase equlllbrlum Once

the binary system is established, the NaF-KF-BF. ternary system shou;d be

3

3 System appear unattractive (for ex-

ample} by reason of incampatibility-with Hasfielloy N) these phase studies

examined. However, should the BF

should immediately be shifted to examinatiofi of the most promising alter- _ v

-native. In addition, a small explbratory study of systems based upon SnF2
should be attempted, so that the interesting properties of this material

_san be expléited if its compatibility with Hastelloy Njcafi be démonstrsted;

- Oxide and Oxyfluoride Behavior

Behavior of oxides, and of oxide and hydroxide ions over fiertinent

regions of the LiF-Ber, LiF—BeFe-UFh, and LiF—BeFE-UFh-Zth systems is

-reassufing and is now reasonably well understood. Some hdditional effort

on such behav1or in the proposed MSBR fuel system (and its close comp051-'

tional relatlves) is still necessary. " Little is known of oxléé behavlor

in systems with moderate to high concentrations of thorlumr' Accordlngly,

a high priority in thess studies must be given to examination of oxides -\EJ

o

|
 

 

123

end hydroxides in LiFfBeFQ-ThFh systems &t and near the blanket compo-
sition. A study of distribution of uranium and thorium between the antici-
pated (U'--Th)O2 solid solution and the molten fluoride phase as a function
of temperature and melt composition will follow. Extension of this study
to include equilibrium distribution of protactinium will also be done if
oxide processes for this element still appear attractive. It is expected
that & 2-man effort can answer, during the next three years, the urgent
questions that affect fuel and blanket production techniques and system
cleanliness requirements. Minor questions and careful refinement of some
of the data are expected to justify a continuing effort at a slower pace
thereafter; |
Solution Thermodynamics

We feel a distinct need for more information about thermodynamics of
many possible fission product or corrosion product species in dilute solu~
tion in fluoride melts. We need to augment‘the program now under way
which attempts to obtain this data by EMF measurements and by direct
measurements of chemical équilibria. The following several items would be
acéorded’nearly_equal priority:

EMF Study of M/MF_Electrodes

 

Experiments using metal/metal fluoride electrodes with reference
, | - + :
electrodes such as Be/Be2 or,H2, HF/PA will attempt to determine the
identity, solubility, and thermodynamic stability of lower valence fluo-

rides of such elements as niobium; molybdenfim, ruthenium, technetium, and

copper in LiF-BeF2 melts. The priority order for these elements is prob-

ably the order listed. The method should also be used, as opportunity

 
 

 

Lttt 1§ e a1 82 007 0 Bt e AR SR s S e i

124

permits, to firm our present values for the fluorides of iron, chromium,
and nickel. Such studies should prove of real value in (1) decisions as

to suitability of improved container metals (such as niobium) and (2)

-evaluation of possible fission prodfict species and behavior. These EMF

studies would, at least in the initial stageé, be coupled with measure-
ments of equilibria such as
Nb + XHF 2 Nbe + X/2 H2.
Thermodynamics of Rare-Earth Fluorides
We will use the cell
LiF 4 | By, BF | Pa

M| MFy Gy | Mgy

for which the expected cell reaction is
+ = + .
M + 3HF MFB(S) 3[2H2

We hope to measure AGf and AHf for crystalline rare-earth trifluoridesi

These data are badly needed in order to calculate AGf values for the dis-
solved salts from solubility data.

Electrochemical Deposition Studies

Liquid bismuth solutions of niobium, molybdenum, ruthehium, and

technetium will be studied (1) to investigate the feasibility of electro-
chemical deposition and (2) to determine activity coefficients of metal
solutes in the liquid bismuth. These studies also could support present
studies of chemical reduction of rare earths (and Pa) into Bi, by_demon-

strating electrochemical reduction of the same ions.

Activities of LiF and BeF2

 

Measurements with the cell
Be | BeFp(q) | HFp, Hop | Pa

"
 

 

 

C

125

should be extended over a wider temperature interval and composition range
to confirm, improve, and extend our knowledge of thé’thermodynamics of the

LiF-BeF, system. [If a suitsble Th® | ThF) electrode can be demonstrated,

2
a similar study in the LiF-ThF) and LiF;-Ber;-ThFh systems will be made. ]

Other Studies

As time permits and the need requires we would attempt (1) study of
reaction equilibria involving BF3 (especiglly with structural metal ele-
ments and alloys) in mixtures of interest as coolants, and (2) continued
study of the reaction

oHF + 5°~ 2 H,S + H,.
Physical Properties

We believe it unlikely that the physical property values listed in
this_docUment for fhe fuel, the blanket, and the.several-coolants are in
error sufficient to cause rejection of the fluid. Howe;er, the state of
knowledgé of physical and heat transfer properties of these fluids is un-
satisfactory, and a considerable effort will be required to establish the
values with precision. - |

Vapor preééures need—to be evaluated for the several mixtures
(throughdecompositiop'pressurés of the BFbeaSed coolants will be estab-
lished with their'phase.behévior);Z'Since_the Valfies are known to be low,
these meaéurements d6 not pgguég deserve a high priority, but studies with
the'ffiel'should be included ear1y sihce they will essist with the distil-
iatibfi sffidies._- |

Density, ¢oefficiént'ofithérmal-exPansion, and viscosity data will be

required to confirm the prééént"eétimates. .Specific heat and heat of

 
 

 

 

126

fusion values are also needed. All these values will, if sufficientlyr
high accuracy can be achieved and if & sufficient concentration range is
covered, be helpful in checking'presentmethéds for estimating the proper-
ties and will be usefu; in attempts to f@rmulate a consistent theory of
high température liquids.

Surface tension measurements of fhe several salts, and theirrélose
compositional relatives, should be eétablished during the next two years
as time and resources permit. | |

fhermal conductivity is the property that is most difficult to
measure for molten salts (as for other.liquids), and the one for which‘a-
vailable information is most insufficient. A program for measurement of
this property for fuel, blanket, and any of‘the materials likely to be
chosen as coolant is urgently required. Thermal conductivity of liquids
can be estimated with some precisiqn if the velocity of sound in the liquid
is known; measurements of sonic velocity in the molten salt mixtu:es
should, therefore, be undertaken as a reasonable backup effort.

Radiation Effects and Fission Product Behavior

These two items promise to be the most demanding, and the most ex-
pensive, in the list of necessary chemical fievelopment activities. No ad-
verse effects of radiation upon the fuels, the‘moderator, or the compati-
bility of the fuel-graphite-metal system have beep observed. Howevef, no
realistic tests of these combinations have been made at power densities
so high as those proposed for MSBR. Studies presently under wey, and radi-
ation facilities presently available, should_by early FY 1968 permit long-
term tests to high fuel burnup at power densities in the 300_kw/lit¢r

range. Such studies are done in in-pile thermal convection loops. which
 

 

he

 

127

expose & very high fraction of the totai fuel to the highest flux; the as-
sembly is equipped so-thafi samples of gas can be taken at will, samples of
fuel can be withdrawn, and samples.of enriched fuel can be added as desir-
ed. These tests will be valuable both in assessment of possible radiation
damage problems and in evaluation of fission product behavior.

We are convinced that'exposures at.even higher power densities (up at
least to 1000 Kw/liter) are necessary in this program end we will attempt
to design and operate suchrfacilities. Success in this venture would per-
mit not only an accelerated tést program for the nufierous possible problem
areas but would also safely assess such reactor accident possibilities as
blocked flow channels, pUmp'stoppages, etc. We hope that such tests can

be conducted at the Oak Ridge Research Reactor in loops cooled by thermal

convection, and the considerable sums budgeted for the effort are predi-

cated on that hope. If the studies must be done elsewhere, or if forced
convectionfloops (with.the,attendant punp development problem) must be
used, then fhe estimatés of stgff and funding required are‘certainly too
low. |

With careful analysis of off-gases fram such systems and rapid radio-

,chemical analysis cf.fuelsafiplés,ve'éhould'get definitive data on

, fission’broduct'behaviorat'trulyrealistic~concentrations and production
crates; Careful checks of graphite and metal frcm such tests immediately

rafter termination of the run should afford realistic data on distribution

of flssion products in these materlals.

Radlation levels,'fram gamma;rays;and-frOm.the delayed neutrons, in
the coolant mixturecare;~clearly, much lower than those for the fuel, but’

radiation damage to the BFB-based'coolants.is not necessarily a trivial

 
 

 

 

 

128

m&tter. When such a coolant mixture is eSt&blished,as to phase behavior,
heat transfer capability, and compatibility, it shogld be givén.a long-
term test at higher-than-realistic radiatién levels to‘see whether such
damage is a possibility.

Fission Product Sgpérations

The distillation process is, at present, the expécted technique for
reprocessing of the fuel mixture, and development activities associated
with that process are described in another report in this series. A small
effort on vapor-liquid equilibria in direct suppoft of that development |
“will continue, as needed, as a portion of the present program.

Highest priority will, for the present, bé devoted to the study of
reduction of rare-earth fluorides from L:'LF--Be.F2 mixtures intq dilute salloys
of the rare-earth metals in bismuth, or into stable intermetallic compounds
of other types. These studies can be carried by mid-1968 to a point where
a sound evaluation of their potential can be made. If the process shows
promise, it will be useful to examine electrochemical vs chemical tech-
niques for its prosecution, and it msy prove necessary to investigate means
of recovery of lithium from the rare earth-bismuth alloy.

The search for insolfible compounds which are stable toward the molten
fuel and which are capsable of ion exchange reactions with rare-earth cat;
ions in the fuel mixture will be continuéd. Attempts will be made to use
Zr02 doped with traces of rare-earth oxides, uranium carbide dopéd with
rare-earch carbide, and other refractory compounds, as well as any promis-
ing rare-earth intermetallic compounds.

Exploratory studies of reduction of the more noble fission products

will be instituted as definite information on the nature of the species in

il

4
bt b et

 

 

C

129

solution becomes aveilsble. Should nicbium or moiybdenum, for example, be
shown to exist in the fuel mixture as‘-a fluoride, their removal by chemical
or electrochemical means will be attempted.
Study of the,equilibrium
HF + I -+ HI + F_

will be continued and extended to'include effects of melt composition and

temperature. This technlque may prove valuable in removing a major frac-

tion of xenon precursors on a short time cycle end may minimize require-
ments for impermeability_of the graphite'moderator and core structure. If
these studies continue to:aPPEar'prdmising,‘they should be extended to ine
clude possibie removal'of'tellurium7by'volatilization of H2Te.
Protactinium'Chemistry |

The.surprising fact tnat protactinium as fluoride is removed from
very dilute solutions in LiF-BeF,-ThF) by reduction with thorium metal (or
with moderately stable intermetallic.compounds of thorium) represents a
breakthrough which must be exploited. Accordingly, first priority will be
given to continued and increased study of this reaction. Primary attention
must be paid to determination of the ultlmate state of the protactlnlum, it

is- presently believed to be a stable 1ntermetallic compound. Success in

'this venture should permlt systematic study of means for recovery of the
”'element. Technlques which Wlll be applied at both tracer (ppb) and real-
:istlc (ppm) concentrations will 1nclude electrochemlcal reductlons with a

";varlety of metal electrodes, and chemical reductions in the presence of

: selected metallic constltuents._'

The process by whlch protactinium'is precipitated'by an excess of

 
 

 

 

 

 

130
> will continue to be examined. Attempts will be made to estab~
lish that passage of the melt through beds of oxide (ZrO2 will be included)

Be0 or ThO

will remove the protactinium without reaction with other constituents. If
this is true, as previous tests have strongly suggested, a:careful study
~of the effect of extraneous ions, of the behavior of uranium, and of the
extent of contamination of thé melt by oxide and hydroxide ion will be made.

233U pfoduct from

Mefhods for recovery of the protactiniumxor of the
whichever of these processes.séems promising will be undertaken as soon as
an understanding of the removal mechanism permits. |

Development of Continuocus Production Methods

As the discussion of production téchnology.above makes clear, the
present production methods have been adequate for materials for MSRE; fihe
fuel, coolapt, and flush salt were furnished in a high and completely
satisfactory state of purity. It seems very likely that the present unit
processes will serve to prepare MSBR fuel and (perhaps with minor modifi-
cations) blanket. However, the 25,000 1b of material for MSRE required
nearly a year to prepare in the existing batch processing facilities, and
provision of a considerably larger quantity for an MSBR would be gquite un-
economical if this equipment were used.

The purification process is quite a simple one. It seems ceitain
that it can be engineered into a continuous process with the throughput per
unit of time and manpower much greater thafi that of the present batchroper—
ation. The relatively smallldevelopment effort adjudged necesséry for this

conversion is scheduled so that the finished plant could be available for

run-in on large quentities of salt needed in the engineering-scale tests.

W
131

Chemical Services

Under this heading are lumped the many and diverse ways in which the
molten salt chemists perform services in direct support of other portions
of the development effort. These ways range from (1) examinstion and
identification (as by the optical microscope) of deposits found in engi-
neeriné.test loops, (2) determination of permesbility of graphite specimens,
(3) in-place hydrofluorination of batches of salt before reuse in test
equipment, (4) manufacture of small batches of special salt compositions
for corrosion or physical property tests, and (5’ liaison emong the engi-
neers, reactor.operators, hot-cell operators, and enalytical chemists so
that the many specisal samples receive proper handling and data from them
are reasonsably interpreted..

It is difficult to specify, long in advance, the details of such
services, but many years' experience encourages us in the belief that the
suggested level will be needed.

Analyticel Development

In order to apply in-line analytical techniques to the MSBR, con-
siderable preliminary.information and_data.mnst,be gatheredtso that a
sound evaluation of possible successful reactor,applicationé can be made.
This approach will.permit g'quifiumjshift of effort to those concepts that
appear to be most~fffiififfil._Fof}eiafifile;,fihe.expériencegained in the
analyéis of hydrocarbons inthefMSRE'offégaé,is being used now in the de-
Sign.of-a gaschrqmatograph_fipdetermine automatically the vafiouS'con-
stituentsfin the éover,gas.a-work on_this pibject.iS'currently finder way
end will be directed towards the MSER. |

The long term in-line analysis progrem is planned in this tentative

 
 

 

 

 

 

132

order of priority.

I. 8. Construct a lsboratory facility which will piovide s flowing
salt stream, probably driven by & gas 1lift. Provision will be made for the
addition of contaminants to the salt inclfiding oxide, sampling, capability
for hydrofluorination and electréiytic treatment of the salt. This facili-
ty will be used to provide tests of eléctrdchemical methods for uranium and
corrosion products and for measuring'the electrochemical potential of the
salt vs a standard reference cell. | |

b. Initiate investigation of a countercurrent equilibration
methpd for the determination of oxide by hydrofluorination.

c. Accurately determine reproduc{bility of operation of spectro-
photometric cell for future application to determination of uranium and

protactinium.

Ny}

ITI. Continue basic inves?igations of electrode-précesses to observe
if chromium, oxide, and trivalent uranium can be determined in this manner.
III. Investigate materials as insulators for reference electrode.

IV. Conduct in-pile testing of any in-line techniques which prove
successful in Section I. '

V. Develop gas chromatographic analyses compatible with high activi-
ty. Includes radiation testing of packing materials, testing solid ad-
sorbents for hydrocarbons. Development of all metal valving. Testing ef-
fect of radiation on detectors.

VI. Investigation of alternate continuous methods of in-iine gas

‘analysis, e.g., thermal conductivity, referencing gas after chemical sepa-

"ration to original gas stream.
 

 

 

5.

10,

11,

12.

133

REFERENCES
P. R. Kasten et al., De81gn Studles of 1000-Mw(E) Molten—Salt Breeder

Reactors , 0RNL-3996 (July 1966)

W. R. Grimes et al., "Chemical Aspects of Molten Fluoride Reactors,"

in Progress in Nuclear Energy, Series IV, Vol. 2, Technology, Engi-
neering, and Safety, Pergamon Press, London, _1969.3

W. R. Grimes et al., "Chemical Aspects of Molten Fluoride Salt
Reactor Fuels,”" in Fluid Fuel Reactors, ed. by J. A. Lane, H. G.
MacPherson,_and Frank Mgslan, Addison-Wesley Publishing Co., Inc.,
Mass., 1958. K | | |

W. R. Grimes, "Materials Problems in Molten Salt Reactors," in

Materials and Fuels for High Temperature Nuclear Energy Applications,

' ed, by M. T. Simnad and L. R. Zumwalt, the M.I.T, Press, Mass., 196k.

MSRP Semiann. Progr, Rept. July 31, 1964, ORNL-3708, p. 21k.

Alvin Glassner, The .'Therinochemi'cal Prop_ertiea of the Oxides, Fluo-
rides and Chlorides to 2500°K, ANL-5750. |

L. Brewer et al., MDDC-1553 (1945) and L. Brewver in The Chemistry and
Metallurgy of Miscellaneous Materials, Thermodynamics, ed. by ‘L. L.

Qulll McGraw-Hlll New York 1950, pp 76-192._

R. E. Thoma (ed ), Phase Diagrams of Nuclear Reactor Materials,

5]03NL-25h8 (Nov. 6 1959)

L. V. Jones et al "Phase Equalibria in the Ternary Fused—Salt

T“System LiF-BeF -UFh,{ J Am Ceram. Soc. h§(2) 79-83 (1962)
. J Barton et al., J. Am. Ceram. Soc. h1(2) 63-69 (1958)

R. E. Thcma et al., 7. Phxs. Chem. 6l, 865 (1960)

R. E. Thama et al., J. Phxs. Chem. 63 1266-7h (1959)

 
 

 

 

13.
1k,

15.

16.

717.

18.

19.

20.

21.

22.

23.

2k,

13h4

C. F. Weaver et al., J. Am. Cerem. Soc. 43(4), 213-18 (1960).

A. L. Mathews and C. F. Baes, USAEC Report ORNL-TM-1129 (1965).

C. F. Baes, "The Chemistry and Thermodynamics of Molten Sait
Reactor Fluoride Solutioné," in Proceedings of IAEA Symposium on
Thermodynemics with Emphasis on Nuclear Materials and Atomic Trans-
port in Solids, Vienna, Austria, July, 1965.

C. F. Baes and B. F. Hitch, Reactor Chemistry Div. Ann. Progr. Rept.
Jan. 31, 1965, ORNL-3789, pp. 61-5.

I. Cohen and B. E. Schaner, A Metallographic and X-Ray Study of the

U02-Zr02 System, USAEC Report WAPD-253, June 1962.

K. A. Romberger et al., "Phase Equilibrium Studies in the Uranium(IV)

T

Oxide Zirconium Oxide System," presented at 151st National Meeting
of the American Chemical Society, Pittsburgh, Penn., March 21-31,
1966. |

E. M. Levin, C. R. Robbins, and H. F. McMurdie, f’ha_se Dia:gréms for
Ceramists, The American Ceramic Society, Inc., Ohio, 196k4.

V. G. Selivanov and V. V. Stender, Zhur. Neorg. Khim. 3(2), 4k8
(1958).

R. E. Thoma and B. J. Sturm, Reactor Chem. Div. Ann. Progr. Rept.

Jan. 31, 1965, ORNI-3789, p. 306.

W. D. Powers and G. C. Blalock, Enthalpies and Heat Capécities of
Solid and Molten Fluoride Mixtures, ORNL-1956 (Feb. 1, 1956).
MSRP Semienn. Progr. Rept. Feb. 28, 1965, ORNL-3812, p. 1k5.

B. C. Blanke et al., Density and Viscosity of Fused Mixtures of «
 

25.
26.

27.

28.
29.

30.

- 31,

32.

- _33.

3k

135

Lithium, Beryllium, end Uranium Fluorides, MLM-1086 (Dec. 1956).

B. C. Blanke et al., Viscosity of Fused Mixtures of Sodium Fluoride,
Beryllium Fluoride, and Uranium Fluoride, MIM-10T79 (April 1956).

S. Cantor and W. T. Ward, Reactor Chem. Div. Ann. Progr. Rept. Dec.
31, 1965, ORNL-3913, p. 2T.

L. J. Klinkenberg, Rec. Trav. Chem. 56, 36 (1937).

G. M. Adamson, Materials Development for Molten-Salt Breeder Reac~
tors, ORNL TM-1854 (Aug. 1966).

G. Long, Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1965, ORNL-
3789, p. 65.

C. M. Blood, Solubility and Stability of Structural Metal Difluo-
rides in Molten Fluoride Mixtures, ORNL CF-61-5-4 (Sept. 21, 1961),
and C. M. Blood, F. F. Blankenship, W. R.‘Grimes; and G..M. Watson,
"Activities of Some Transition Metal Fluorides in Molten Fluoride
Mixtures," presented at Tth International Conference on Coordination
Chemistry, T1CCC, Stockholm and Uppsala, Jfine 25-29, 1962.

MSRP Semiann. Progr. Rept. Aug. 31, 1961, ORNL-3215, p. 93.

C. E. Wicks and F.;E}jBlock -"Thérmbdynamic Properties of 65 Ele-

1ments—-The1r Oxldes, Halldes, Carbldes, and Nltrldes," Bureau of

Mlnes Bulletin 605 (1963)
JANAF (Joint Army—NaNy-Alr Force) Interim Thermochemlcal Tables,
Thermal Research Laboratory, Dow Chemical Co., Mldland Mlch.

W. R..Grimes, Radiation Chemlstry of the MSRE System, ORNL—TM—SOO

| (March 3l 1963)

 
 

 

 

 

35.

36.
37.

38.

39.
Lo.
Lh1.

L2,
L3.
Lk,
L5.
L6.
L7,
L8.

L9,

50.

 

 

 

 

 

 

 

136
{;;?
‘W. H. Cook, Metals and Ceramics Div. Ann. Progr. Rept. July 1, 1963,
0RNL—3h20; pp. 165-T.
MSRP Semienn. Progr. Rept. Feb. 28, 1965, ORNL-3812, p. 87.
F. F. Blankenship et al., Reactor Chem. Div. Ann. Progr. Rept.
Jan. 31, 1963, ORNL-3417, p. 17. |
W. R. Grimes, N. V. Smith, and G. M. Watson, J. Phys. Chem. 62, 862
(1958). .
M. Blander et al., J. Phys. Chem. 63, 116k (1959).
G. M. Watson et al., J. Chem. Eng. Data T, 285 (1962).
E. S. Bettis et al., Nucl. Sci. Engr. 2, 841 (1957).
C. F. Baes et al., Reactor Chem. Div. Ann. Progr. Rept. Dec. 31,
1965, 0RNL—3913,.p. 38. |
W. R. Grimes et al., Nucl. Engr. - Part VII, 55(27), 63 (1959). F
W. T. Ward et al., J. Chem. Engr. Data 5, 137 (1960). -

 

C. J. Barton, J. Phys. Chem. 64, 306 (1960).

C. F. Baes, Jr., Reactor Chem. Div. Ann. Progr. Rept. Jan. 31, 1967,
ORNL-4076, pp. 49-50.

E. Greenberg, C. A. Natke, and W. N. Hubbard, J. Phys. Chem. 69,
2089 (1965).

J. L. Settle, H. M. Feder, and W. N. Hubbard, J. Phxs.- Chem. ég,
1337 (1961).

J. L. Settle, H. M. Feder, and W. N. Hubbard, J. Phys. Chem. 67,
1892 (1963). |

K. K. Kelly, U.S. Bur. Mines Bull. 584, 1960.

®
 

C

w

51.

52.

53.

5k,

55-
56.

oT.

58.

59.

60.

61.

. 62,
€3,

6k.

"137

H. A. Porte, E. Greenberg, and W. N. Hubbard, J. Phys. Chem, 69,

2308 (1965).

J. F. Elliott'and_M, Gleiser, Thefmochemistry for Steelmaking,

Addison—Wesley_Publishing Co., Reading Mass., 1960.
Memorandum from C. F. Baes, Jr. to H. F. McDuffie, March 8, 1965,
"Chemical Consequences of Nuclear Fission in Molten Fluorides."

Reactor Chem. Div. Ann. Progr. Rept. Dec. 31, 1966, ORNL-40T6,

"pp. L45-5L,

MSRP Semiann. Progr. Rept. Feb. 28, l967_(in press).

W. L. Carter and M. E. Whatley, "Fuel and Blanket Processing Develop-
ment for Molten Salt Reactors}" ORNL TM-1852 (June 1967).

Reactor Chem. Div. Ann. Progr. Refit. Dec. 31, 1965, ORNL-3913.

J. J. Egan and R. N. Wiswall, Nucleonics 13, 104 (1967).

Mélvin S. Foster, Scott E. Wood, and Carl E. Crouthamel, Inorganic

Chemistry 3, 1428 (196L).

'J. H. Bhaffer et al, Reactor Chem. Div. Ann. Progr. Rept. Dec. 31,

1966, ORNL-40T6, pp. 3k-36.

C. J. Barton et al., Reactor Chem. Div. Ann. Progr. Rept. Dec. 31,

1965, ORNL-3913, p. 45.

J. H. Shaffer et al., Nucl. Sci. Eng. 18(2), 177 (1964).

J. H. Shaffer et al., Rea¢t6r Chem. Div. Ann. Progr. Rept. Dec. 31,
1965, ORNL-3913, p. b2.

C. J. Barton ggng;;,:Réactor Chem. Div. Ann. Progr. Rept. Dec. 31,

- 1266,'0RNL-h076,'pp. 39-41. -

 
 

 

 

65.

66.

67.
68.

69.

 

138

-~ J. P. Young, Angl. Chem. Div., Ann. Progr. Rept. Dec. 31, 1961,

ORNL-3243, p. 30.

J. P. Young, Anal. Chem. Div. Ann. Progr. Rept. Nov. 15, 196L,

'ORNL-3750, p. 6.

D. L. Manning and Gleb Mama,ntov, J. Elecfrdahal. lChem. 1, 102 (1964).
D. L. Manning, "Voltammetry of Nickel in Molten Lithimfi Fluoride-~
Potassium Fluoride-Sodium Fluoride," J. Electroanal. Chem. 7, 302
(1964). -

Gleb Mamentov, D. L. Manning,'apd J. M} Dale, "Reversible Deposition

of Metals on Solid Electrodes by‘ Voltammetry with Linearly Varying

- Potential," J. Electroanal. Chem. 9, 253 (1965).

#
7

c:“

 

s

1-50.
51.

60.
61.
62.
63.
6h.
65.
66.
67.
68.
69.
70.
T1.
T2.
73.
Th.
75.
76.
T7.

19.
80.
81.
82.
83.
8.
85.
86.

88.
89..

- 90.
- 91.

- 92,
- 93.
9li.

' 95.

96.

o97.
98.

139
DISTRIBUTION
MSRP Director's Office 99. H.
R. K. Adams 100-103. W.
G. M. Adamson - 104, A.
R. G. Affel = 105. R.
L. G. Alexander - 106. B.
R. F. Apple o '107. P.
C. F. Baes 108. D.
J. M. Baker , 109. C.
S. J. Ball ' | 110. P.
- W. P. Barthold - 111. F.
H. F. Bauman - ‘ 112. P.
S. E. Beall o 113. J.
M. Bender 114. H.
E. S. Bettis 115. R.
F. F. Blankenship | 116. T.
R. E. Blanco 117. H.
J. 0. Blomeke : 118. W.
R. Blumberg 119. P.
E. G. Bohlmann ' 120. R.
C. J. Borkowski 121. M.
G. E. Boyd 122. M,
J, Braunstein - -123. C.
M. A. Bredig | 124, T.
R. B. Briggs 125. H.
H. R. Bronstein 126. S.
G. D. Brunton 127. A.
D. A. Canonico 128. J.
S. Cantor , 129. C.
W. L. Carter 130. J.
G. I. Cathers ) 131. R.
J. M. Chandler 132. A.
E. L. Compere | 133. M..
W. H. Cook | - 13k, R
L. T. Corbin - ~ -135. H.
J. L. Crowley “““j*-,136;'.R
F. L. Culler - 137. C.
J. M. Dale .. 138. cC.
'D. G. Davis .o . 139. R.
S. J. Ditto ... 1hko. H
‘A. S. Dworkin = . 141, H
J. R. Engel - 142, C.-
E. P. Epler - o 1h3L el
D. E. Ferguson - 1hk, L.
L. M. Ferris 145, A,
A. P. Fraas ... 1k6. R.
‘H. A. Friedman -~ .. .. 17, g,
J. H. Prye, Jr. . . 148, E.
C. H. Gabbard 149, L.
R. B. Gallsher 150. P.

@I maeaH

L.

Goeller
Grimes

. Grindell

. Guymon
. Hannaford

Harley

. Harman
Harrill
Haubenreich

Heddleson

. Herndon

Hightower
Hoffman
Horton
Hudson

Inouye

H'
R.

LI,’

C.

Jordan
Kasten
Kedl
Kelley
Kelly
Kennedy
Kerlin

. Kerr
. Kirslis

Krakoviak

. Krewson

Lamb,
Lane
Lindauer

. Litman
. Lundin

Lyon

. MacPherson
. MacPherson
. Martin =
. Mathews .

McClung

‘MeCoy - .
McDuffie
_McGlothlan

. McHargue
" McNeese

Meyer ™ .
‘Moore

Niqhols' 
Nicholson
Oakes

Patriarca

 
 

 

151.
152.
153.
15k,
155.
156.
157.
158-20T.
208.
209.
210.
211.
212.
213.
21k,
215.
216.
217.
218.
219.
220.
221.
222.
223.
22k,
225.
226.
227.
228.
229.
230,
231.
232.
233.
234,
235.
236.
237.
238.
239.
24o-241.
2h2_243,
2ul_25),
255.

C.

M. Perry
B. Piper

. E. Prince
. L. Redford

Richardson
C. Robertson
C. Roller
W. Rosenthal
C. Savage
E. Schilling

Dunlap Scott

H.
W.
J.
M.
G.
A.
F.
G.
0.
P.
W.
I.
R.
H.
J.
E.
R.
J.
C.
B.
A.
J.
W.
K.
M.
J.
L.
G.
H.

E. Seagren

F. Schaffer

H. Shaffer
J. Skinner
M. Slaughter
N. Smith
J. Smith
P. Smith
L. Smith
G. Smith
F. Spencer
Spiewak

C. Steffy
H. Stone

R. Tallackson

H. Taylor
E. Thoma
S. Watson
F. Weaver
H. Webster
M. Weinberg
R. Weir

J. Werner
W. West

E. Whatley
C. White
V. Wilson
Young

C. Young

140

DISTRIBUTION

256-25T.
258.
259..
260~-2Tk.

275.
276-27T .
278.
279.

280.
281.

282-296.

Central Research Lib.
Document Reference Sect.
Laboratory Records
Laboratory Records - RC

EXTERNAL DISTRIBUTION

D. F. Cope, AEC-ORO

W. J. Larkin, AEC~-ORO

C. L. Matthews, AEC-ORO

T. W. McIntosh, AEC- -
Washington

H. M. Roth, AEC-ORO

M. Shaw, AEC-Washington

W. L. Smalley, AEC-ORO

R. F. Sweek, AEC-Washington

Research and Development Div.
Reactor Division - ORO

DTIE