'
.
| OF L.
ORNLP
| 428
:
.
11
-
IPEFEFEE
36
2.0
11:25 || 14 LE
11
MICROCOPY RESOLUTION TEST CHART
NATIONAL QURE AU OF STANDARDS -1963
OTS1
LEGAL NOTICE
This report was prepared as an account of
Government sponsored work. Neither the
United States, nor the Commission, nor any
person acting on behalf of the Commission:
A. Makes any warranty or representa-
tion, expressed or implied, with respect to
the accuracy, completeness, or usefulness
of the information contained in this report,
or that the use of any information, appa-
ratus, method, or process disclosed in this
report may not infringe privately owned
rights; or
B. Assumes any liabilities with respect
to the use of, or for damages resulting from
the use of any information, apparatus,
method, or process disclosed in this report.
As used in the above, “person acting on
behalf of the Commission" includes any em-
ployee or contractor of the Commission, or
employee of such contractor, to the extent
that such employee or contractor of the
Commission, or employee of such contractor
prepares, disseminates, or provides access
to, any information pursuant to his employ-
ment or contract with the Commission, or
his employment with such contractor.
a.
vs.
O ROVz-prj42g
CONF-650104-6
29
Jun 20 1965
MASTER
RHY
.
I
THE CHEMISTRY AND THERMODYNAMICS OF MOLTEN
SALT REACTOR FLUORIDE SOLUTIONS (1)
-
-
C. F. Baes, Jr.
Sr. Research Staff Member
Oak Ridge National Laboratory
Oak Ridge, Tennessee
ABSTRACT
A solvent of lithium and beryllium fluorides (about 2 moles
of LiF per mole of BeF2) is used in the fuel salt, the coolant
salt, and the flush salt of the Molten Salt Reactor Experiment.
As a result of the chemical development work done for this reactor
concept, considerable chemical and thermodynamic information has
been acquired concerning this solvent and its solutions with
actinide, lanthanide, and structural metal fluorides. It is the
purpose of this paper to review this information, much of which
is not yet generally available.
The data were obtained mainly by measurements of heterogeneous
equilibria; i.e., by equilibration of melts with gaseous mixtures
containing hydrogen, hydrogen fluoride, or water and by deter-
minations of solid-liquid phase equilibria.
The results of these measurements gave direct information
about such important chemical problems as: (1) The corrodibility
of structural metals and the reducibility of the structural metal
..
.
-
-
(1) Research sponsored by the U.S. Atomic Energy Commission under.
contract with the Union Carbide Corporation.
-
-
-
-
-
- -
PATENT CLEARANCE OBTAINED. RELEASE TO
THE PUBLIC IS APPROVED. PROCEDURES
ARE ON FILE IN THE RECEIVING SECTION,
-
.
-
-
.
.
ions, Ni2+, Fe2+, Cr2+; (2) reactions with water vapor to form
oxide and hydroxide ions, and the removal of these ions; (3)
the precipitation and solubility of the oxides of beryllium,
uranium, zirconium, thorium and the rare earths; (4) the stability
of uranium(IV) toward reduction to the trivalent state and pos-
sible subsequent disproportionation; and (5) the solubilities
and solid solution formation of rare earth fluorides. Equally
important has been the wider usefulness of this information when
the methods of thermodynamics are brought to bear. Thus the data
obtained could be used to: (1) correlate, revise, and extend
existing thermochemical data on fluorides and oxides; (2) deter-
mine activity coefficients of the components LiF, BeF2, UF4, and
NiF, in these molten salt solutions; (3) calculate electrode
potentials involving a variety of solute ions; and (4) estimate
solubilities and reactivities of compounds not directly investi-
gated.
Thus the chemical development program for the Molten Salt
Reactor concept provides a number of interesting examples of the
interrelationship between thermodynamics and problems in reactor
chemistry and engineering.
--LEGAL NOTICE
The report w opard M o o Gover n or worth Molther the Wall
au, me the correctem, mert porn mother an d the Chania:
A. Men my warranty wropewnontdea, apud or loped with repect to the att
mey,
w
ow, or wefalone the trutin
c u we report, what the
my tubormation, prenata, o d, or procura duchowed wale report my not infringe
mtrainy www ohutus
1. Am o letto ma nepoet wolor hoe meer roots true the
wam tertium, w , me me me mend report
Memed to the whore, porn staan at the Crane " meletwa mno ne
Moyes or contractor to beton, tayo at med motore, the two
wel amployee or montre le of Conducere, plays a we wstructor mepera,
da ntes, or more new me, normation
Memployees a cuatrnce
with the too, wo wo aptent with me citructure.

THE CHEMISTRY AND THERMODYNAMICS OF MOLTEN
SALT REACTOR FLUORIDE SOLUTIONS()
C. F. Baes, Jr.
Sr. Research Staff Membor
Oak Ridge National Laboratory
Oak Ridge, Tennessee
1. Introduction
The Molten Salt Reactor Experiment (MSRE), which went
critical at ORNL in June, 1965, is fueled by a molten LiF-BeFz-
ZrF.-UF, mixture and moderated by graphite. While it will not
demonstrate breeding, this reactor experiment is intended to
show that a molten-salt fueled reactor is a possible, indeed an
attractive, means by which to achieve thermal breeding. A .
recent collection of papers has described this reactor and the
extensive development work which led to its construction and
operation (1).
In considering possible materials which could serve as
constituents in such a thermal breeder reactor, Grimes (2)
reached the conclusion that the choice of major salt components
is largely limited to mixtures of Li'F and BeF2 by the need for
neutron economy, low volatility, and chemical stability. While
in the LiF-BeF2 system (Fig. 1) mixtures with melting points
below 500°C occur in the range 0.33 - 0.73 mole fraction BeF2 (3),
compositions nea: 0.33 BeF2 appear the most suitable. (Here, and
elsewhere, numbers preceding a salt component denote mole fraction).
Research sponsored by the U.S. Atomic Energy Commission under
contract with the Union Carbide Corporation.
The composition of the MSRE fuel salt is 0.65 Lif, 0.291 BeF2,
0.05 ZrFr, 0.009 UF,. That of the coolant and flush salt is
LiF - 0.34 BeF2.
As part of the development of the MSRE, a considerable
amount of chemical and thermodynamic information has been gathered
about LiF-BaFz melts near this composition. Most such information
has been gained by the use of heterogeneous equilibria involving
reactions of gases or solids, or both, with the liquid phase. In
general, these studies gave fairly direct information about impor-
tant chemical questions related to the MSRE.
It is the present purpose to review this information, much
of which is not yet generally available, and to indicate its
application to the chemistry of the MSRE. It will then be sum-
marized by thermodynamic methods as a means of extending its
usefulness. This seems fitting and proper at this stage in the
development of the molten salt reactor concept. Future such
reactors evidently will employ salt mixtures similar to those
of the MSRE considered here; hence, a knowledge of the thermo-
dynamics of these solutions should prove generally useful.
2.
Reactions in Lit - 0.33 BeF2
Table I lists the reactions studied, the form of the equili-
brium constants, and values of a and b in the following expression
which gives the numerical value of the equilibrium constants in
LiF - 0.33 BeFz
log K = a + b(103 /T)
(1)
This expression, implying a constant heat (AH = -2.3Rb) and
entropy (AS - 2.3 Ra) of reaction, adequately approximates the
measured values in the temperature range studied (usually 500 -
700°c). The concentration scale is the mole fraction; e.g.,
(2)
*LF, - DAF,/(MMF, + "lif + "BeF,
Gas pressures are expressed in atmospheres, and at the low pres-
sures and high temperatures involved, gases are assumed ideal.
The standard states for reactants and products generally
can be seen from the form of K. Here and elsewhere for most
solutes the standard state is the hypothetical one mole fraction
-2-
Table I
Reactions in LP-0.33BeF:
[log K = 8 +(10°/T)]
Est. Error
in log K
0.06
Source
Ref. 4
3.37
-3.60
-5,31
0.02
Ref: 4
Reduction Reactions Involving Hydrogen
i Ha(8) + MFzla) = N(8) + 2H(8)
2 H2(g) + FeF2(a) = Fe(s) + 27(8)
3 Hz(8) + GFz(a) = C(s) + 25F(8)
4 172(8) + UF:(a) = Fz(a) + F(3)
5 73(8) + BeF,(a) = Be(s) + 2HF(8)
(Pop)*/(PR.)(*P3!
(Pop)/(Pne)(pl
(PP)/(PRE
(Ppp) (Up,//(Pa)(Xype!
(Ppp)/(Pine!
0.06
5.20
5.12
4.07
7.22
-9.06
-9.33
-21.56
Ref. a
Ref. 5
0.02
0.1
Pefs. 6,7
12/12
0.02
Ref. 9
Re!. 10
Metathesis Reactions Involving Cases
6 H2C!8) + Befz(a) – Beo(s) + 2HF(8)
? 282018) + ZrFe(a) = ZrO2(8) + 4H(8)
8 12018) + 2*(a) = 02-(a) + 27(8)
9 12018) + pº(a) = N*(a) + F(8)
10 828(8) + 28°(a) = 52-(a) + 2(8)
U HI(g) + f(a) = 1*(a) + F(8)
(PF)/(PO)
Pyp)/(P700)(xp)
(Ppp)?(x32-)/(PH30
(Pop) (208-)/(F320)
(PF)+(x32-)/(PAS)
(Pop)(x-)/(Pop)
Ref. 9
4.23
-5.67
21.2 -10.66
6.20 -8.66
-1.03 -2.08
108 K(8730K) <
log k(7639) 5 -3
0.06
0.08
0.04
Rel. 9
Ref. 12
Ref. 13
-
0.05
Marin
Metathesis practions Involving Solid Oddes
12 Zrog(s) + 20eF2(a) = 2rFala) + 2Bc0(8)
13 002(8) + 2BeF2(a) = Ur (a) + 2Bco(s)
14 (4) + (a) + zxF (4) + CO2()
15 Thoz(s) + uraia) Thp (a) + UC2(8)
0.01
MUEL
,
-2.75 -0.69
-2.07 -1.74
-0.67 1.05
log K(1023x) 1.2
6,7; Ref. 15
12,14; Ref. 15
Ref. 15
Ref. 18
0.05
(P./xe.
(x )/(xure)
Table I (cantlaued)
(segust (242.)?
0.09
7,8
(zgo+) (322-32
(orygon lagenda
0.1
198
2.52
da. 12
0.30
0.01
Est. t Error
Solubility Acactions
in log K Source
26 Dedla) Beat(a) +07(a)
-0.04 -2.96 0.08 6,8
17 20zla) - Zop**(a) + 202-(a)
-2.82 -6.62
18 vozlo) Ub(a) +202-(a)
-2.25 -7.66
14,17
19 thozla) sae Doo **(a) + 202-(a)
log k(10239x) = -8.6
15,18
20 0(s) - ***(a) + 02-(a)
(x2+)(562-)
-2.58 1.39
Ref. 32
22 Podlo) - Hea*(a) + 02-(a)
Bqs. 25, 26, Ref. 32
22 mile) - Pala)
Ref. a
23 Pepala) – Pefala)
Ref. h
24 lafz(s) - larga)
Rel. 20
25 Copy(s) - Copyla)
0.02 Ret. 20
26 Surg(s) Stapy(a)
1.97 -3.38 0.02 Ref. 20
27 Pusla) - Pusla)
sur
1.30 -3.15
Ref. 19
"The notations (s), (a), and (8) indicate respectively the solid, dissolved, and gaseous states.
Pia expressed in atmospheres; *, is the male fraction and, for LP-0.33BCP2, 18 equal to poles of 1/68 of salt/30.03.
Members other than reference numbers and numbers of cquations in text refer to reactions in this tatle.
2.45
0.01
1.58
Hoita
Spera
Ears
* Gets
tights
0.02
1.6
solution. Excoptions are Lif, BOP,, Bolt, L1, and pe, for
which the solvent composition LiP - 0.33 Bop, is takon as the
standard State. The notation (s), (d), und (g) donote, respectively,
solid, dissolved, and gaseous states. In the majority of reactions
listed, p is the only anion involved, and for simplicity the
dissolved compononts aro written in the molecular form, though
this is not intonded to imply the actual specios present in
solution. Whon anions other than rappour in the reaction,
ionic specios are writton. In some cases this is done to avoid
the choice of a noutral component in the liquid phase and in
others to indicite that complete dissociation of an anion (0.8.,
02-) and a cation (6.8., 2r+) is assumed.
Estimates of activity coollicionts as a function of solvent
composition and solute concentration are discussed in Section 3.
On the basis of those estimates, the values of K indicated in
Table I include, where necessary, a correction of the original
moasurements to a nolt composition of LiP - 0.33 BOP, .
2.1 Hydrogen Reduction Reactions
The reduction of the bivalent fluorides of nickol, iron and
chronium by hydrogen in BeF:-0.38L1F was investigated by Blood (4)
in a series of careful transpiration experiments. The equilibrium
quotients which were measured for the reaction
H(B) + MF, (d) M(s) + 2HP(8)
(3)
were found to be independent of Xup at the low concentrations
studied. These concentrations were linited by the solubilities
of the fluorides. The reduction of UF, dissolved in Lip-Bop, -UF,
nelts by hydrogen as well as the reduction of UF, solid to UF,
solid has recently been investigated by Long (5). He summarized
his results by the following expressions: For the reduction of
UP. to UF, in solution
log(xur. PHP/XUZ PX, *) - 3.995 - 9.329(109/T) +
3.77Xur. + 2.09(XB61, - .30) (4)
For the reduction of UP, solid to UP, solid
log(Par/Pu,*) - .57 - 8.87(10 /T. (5)
-3
While the reduction of Bel, in Lip-0.33B88, by hydrogen is
inaccessible to direct noasuronent, recently Dirian (6) and
Romberger (7) havo mado nossur onents on the following coll
Pd, H, HP | Lip-0.33B887 | Be
Tho coll roaction was taken to be
H2(g) + Bop, (d) * Be(s) + 2HP(8)
(6)
With Pue and Py. corrected to 1 ata, the potential was found to
bo
E = -2.436 + 0.715(/100) v.
(7)
From this the equilibrium constant expansion in Table I has been
calculated. The two olectrodes - Be2+ | Be and ,H2 , Pd | HP -
woro judged to be reversible from polarization measurements.
These results show that nickel 18 the most noble of the
throc structural notals studied, N12+ being readily reduced by
hydrogen (Fig. 2). Chromium was the least noble, Cr2+ being
reduced by hydrogen with difficulty. The sparging of moltoa
Iluorides with hydrogen to reduce structural metal impurities
is routinely used as a purification stop. The concentrations of
N18+, Felt and Cr2+ and the ratio U4+/US+ in Lip-0.33BP, expected
under two sets of conditions is indicated in Fig. 2. On the left
of the figure, wherein the equilibrium constants are plotted, it
appears that ouly Ni?t is extensivoly reduced by a mixture of H
and HF, both at unit activity (1 atm). Nickel is a relatively
iport container material for Lif-B@F, melts in the presence of
excess hydrogen. On the right of Fig. 2 are much more reducing
conditions which approximate those in the MSRE. The concentration
of Cr3+ is set at 100 ppm in equilibrium with the pure metal.
It is seen that falt and N1?+ should be reduced essentially com-
pletely to the metals and some Ud+ should be reduced to US+. The
reaction of U* in the MSRE fuel with the container metal (INOR-8,
a pickel-base alloy containing about 7% Cr, 5% pe and 16% NO)
Cu(in INOR-8) + 2UF, (d) - 2UP, id) + CrP,(a), log K (873°) - 9.0
(8)
is túe only corrosion reaction expected, and the extent of this
reaction should be small (8). In the absence of UFA (1.e., the
flush salt) no reaciion between the salt and the container metal
is expected.
2.2 Metathesis Reactions Involving HF and H2O
Extensive equilibrium measurements of reactions between
water in hydrogen carrier gas with LiF-BeF, melts were made by
Mathews and Baes (9). Equilibriun quotients for the reaction
: H2O(g) + BeFz id) = Beo(s) + 2HF(8)
(9)
(wherein beryllium oxide was present as a saturating solid phase)
were measured from 500 - 700°c over the composition range
Xper. - 0.3 - 0.8, limited at one extreme by the Lir liquidus
(Fig. 1) and at the other by rapidly increasing viscosity. The
results were summarized by the expression
(10)
log (Php?/PH, o XBef,) = a + b xiif + c *Lif
wherein a, b and c all were linear functions of 1/TºK
a = 3.900 - 4.418(103/T),
b = 7.819 - 5.440(103 /T),
C = -12.66 + 5.262(103 /T)
From this expression the activities of BeF2 and, by a Gibbs-
Duhem integration, the activities of Lif were estimated (Fig. 5).
In the same investigation, measurements were made upon melts not
saturated with Beo. In addition to the reaction
H2O + 2F-(d) = 02-(d) + 2HF(g)
for the formation of oxide ion, it became evident, both from
these measurements and from those upon Beo saturated melts, that
hydroxide ion also was formed
H2O(g) + F"(d) = OH®(d) + HF($)
(12)
While the equilibrium quotients for these two reactions were less
accurately determined than was the previous one for Beo saturated
melts (ca. + 20% and † 1.% respectively compared to + 5% because
of limitations inherent in the transpiration method used, they
were sufficient to show that both oxide and hydroxide increase in
th
(11)
12.
stability with increasing temperature. The stability of hydroxide
with respect to oxide, however, decreases with increasing tem:-
perature and it is low enough that hydroxide can be readily
decomposed in these fluoride melts by sparging with an inert
gas (8.8., hydrogen).
OK™ + F + HF(g) + 02-, log K - 5.23 - 6.56(103/T) (13)
Similar measurements have also been made by Baes and
Hitch (10) in which the LiF-0.33BeF2 contained added ZrF. With
X.rf. > ~ 3 x 10-4, Zroz is the stable saturating oxide solid,
hence the following equilibrium may be written .
2H2O(g) + ZrF. (a) = ZrO2 (s) + 4HF(8) . (14)
From these measurements the activity coefficient of ZrF. could be
estimated (Fig. 5) as well as the solubility of Zroz (Sect. 2.5).
It was also found that the equilibria (11 and 12) for the formation
of oxide and hydroxide ions were shifted to the right with increas-
ing xa. ; i.e., in the direction of greater stability of these
ions.
These results are generally consistent with previous
observations that LiF-BeF2 melts are readily freed of oxide con-
tamination by treatment with gaseous mixtures of Hy and HF,
another routinely used purification step. The measured equili-
brium quotients in LiF-0.33BeF2 were used to calculate the
efficiency of HF utilization in such a treatment as a function of
temperature and HF partial pressure with the assumption that
equilibrium is maintained between the gas stream, the molten salt,
and any BeO solid present. This calculation (Fig. 3) shows that
the efficiency in the removal of oxide to a final value of 16 ppm
(X02- = 3.3 x 10-5) is quite high over a wide range of conditions.
The removal of oxides from LiF-0.33BeF2 melts by H2-HF sparging
during the preparation of flush salts and coolant salts for the
MSRE (11) was performed on 100 kg batches in cylindrical vessels
with a single gas delivery line. It appeared that about 2/3 of
the calculated efficiency was obtained even in this relatively
simple equipment.
-6-
In the presence of ZrFt and Zroz the increased stability
of oxide and hydroxide render more difficult the removal f
oxide from such melts than is the case with the solvent salt.
Because of this, in the purification of LiF-BeF2-ZrFe melts it
was found expedient to decant the molten salt away from the
precipitated Zroz as a simple means of removing most of the
oxide before HF-H2 treatment.
2.3 letat..ws Peactions Involving HF-H2S and HF-HI
An attempt has been macio co study the removal of sulfide
from LiF-0.33BeF2 melts by HF-H, sparging (12)
2HF(g) + 52-(d) = f'(d) + H2S(8), log K > 103 (15)
but, unfortunately, the effluent H2S/H, mixture reacted with
the relatively cooler surfaces of the nickel exit tubing used.
The sulfide was evidently precipitated from the melt, presumably
as Ees, though no direct evidence of this was obtained. In tests
on the removal of iodide by HF-H, sparging (13)
HF(g) + 1 (d) = F(d) + HI(g), log K > 104
there was also interference caused by reaction of the effluent
gases with the metal surfaces in the gas exit:lines. Because of
these difficulties it is possible, at present, only to assign
lower limits to the equilibrium constants of reactions (15) and
(16).
(16)
These preliminary results do serve to confirm, however,
observations that the HF-Hy treatment used in purifying molten
salts seems to remove any sulfide, as well as oxide, which may
be present as an impurity. The use of HF sparging also may prove
a valuable means of removing the iodine precursor of the impor-
tant fission product, Xel 35, in an operating reactor.
2.4 Metathesis Reactions Involving Solid Oxides
By combination of reactions 9 and 14, it is possible to
calculate that both Beo and ZrO2 will co-exist at equilibrium
with LiF-0.33BeF: containing oxide ion
ZrO2 (s) + 2Be2+(a) = Zr4+(d) + 2Beo(s):
(17)
-7-
when ZrF. is present at concentration of approximately 3 x 10
mole fraction. With largor amounts of added ZrF., Zro, becomes
the less soluble (stable) oxide.
It was decided to include ZrF. in the MSRE fuel composition
to prevent the precipitation of UO, which otherwise would result
from inadvertent oxide contamination of the fuel. Measurements
of the metathesis reaction
ZrO2 (s) + UF, (d) – ZrF. (d) + VO2 (s)
(18)
have shown that the mole ratio of ZrF. to UF, at equilibrium with
both Uoz and ZrO2, while varying somewhat with temperature and
melt composition, remains well below that chosen for the fuel
salt (14,15). As a consequence, a considerable amount of Zrtt –
an amount easily detected by chemical analysis of the fuel
salt - would be precipitated by oxide contamination before any
VO, should precipitate.
In connection with these studies, it was ascertained that,
contrary to published UO, -Zr0, phase diagrams (16), (U-Zr) Oz
solid solutions are not formed in the temperature range 500 -
700°c. Because of the obvious importance of this to the MSRE,
experiments have been carried out in which both UOz - Zroz mixtures
and (U-Zr) 02 solid solutions prepared by fusion were equilibrated
with LiF-BeFz melts. At the present writing, no solid solution
formation has been detected below 1100°c (17).
The metathesis equilibrium in Table I involving Voz and
Thoz is an estimate from results of Shaffer et al, in LiF-0.4
NaF (18).
2.5 Oxide Solubilities
The oxide concentration in LiF-0.33BeF2 saturated with Beo
may be estimated by combining the equilibrium results for reactions
(9) and (11). The solubility increased rapidly with temperature,
but no strong dependence on Xper was found. In these measure-
ments (9), the mole fraction of oxide at Beo saturation probably
was less than 0.002. Hence, the activities of BeFz and Lif which
were derived from them are probably not appreciably different
from the corresponding activities in oxide-free mixtures.
-8-
From the similar measurements in ZrFx-containing melts (10)
the solubility product of Zroz could be estimated. With increas-
ing xong, the concentration of oxide at ZrO2 saturation at first
falls as would be expected from the equilibrium
ZrO2 (s) = Zrt+(a) + 202-(d)
(19)
However, it then levels off and subsequently rises with further
increases in Xze. (Fig. 4). This could be caused, at least in
part, by the formation of a complex ion, ZrO2+
Zrt+(a) + 02-(d) = ZrO2+ (d)
(20)
or it could be caused entirely by the influence of the changing
melt composition on the activity coefficients of the species
Zr*+ and 02-. The plot in Fig. 4 indicates approximately the
"oxide tolerance" of MSRE fuel salt-flush salt mixtures; i.e.,
the amount of dissolved oxide these mixtures can contain with-
out oxide precipitation. It is seen that the oxide tolerance
increases rapidly with temperature, especially near the fuel
composition (xzrf. = 0.05), indicating that any excess oxide
present might be removed by collecting Zroz on a relatively cool
surface in the MSRE system.
The solubility products of Nio and Feo, indicated in Table I,
were estimated from the previously cited equilibrium results
(Sect. 4.3).
2.6 Fluoride Solubilities
Table I also summarizes the solubility measurements by
Blood [4] of FeF2 and NiFz, those by Barton (19) of PuFz, and
those by Ward, et al (20) of rare earth trifluorides. These last
investigators found fluoride solid solution formation and pointed
out that this could provide a useable method for removing rare-
earth fission products from a reactor fuel stream by exchange
reactions.
3. Effect of Melt Composition Upon Activity Coefficients
In several of the investigations cited above the effect of
melt composition on equilibrium was studied and from these results
-9.
the corresponding variation of a number of activity coefficients
can be determined. The curves shown in Fig. 5 represent these
variations as a function of the solvent composition (left-hand
figure) and as a function of the mole fraction of solute added
to LiF-0.33BeF2 (right-hand figure). The standard states
remain the same; i.e., for the solvent components it is
LiF-0.33BeF2 and for the solutes it is the hypothetical mole
fraction solution in LiF-0.33BeF2.
The curves (Fig. 5, on the left) for Lif and for Bef, are
derived from Eqn. 10, Yper being obtained directly and Yes
being obtained by a Gibbs-Duhem integration. The curves for
PuFz and CeFz are based on the solubility measurements of Barton
[19] and of Ward, et al (20]. The curves for UF4 are based on Long's
results (Eqn. 4) with the assumption that Yue. is similar to
Ycer and Yout. The curve for ZrF, is from the preliminary
measurements of Baes and Hitch (Egn. 14). The curve for NiFz is
an estimate based on the observation that Yrit. in NaF-ZrF4, (as
indicated by reaction 3 [4]) varies less with ZrF, in this solvent
than does Ycer (as indicated by its solubility [20]).
These activity coefficient curves were used, where necessary,
to correct equilibrium measurements to the reference composition
LiF-0.33 PuFz. It is seen that the variation of activity coef-
ficients with composition is not large and that, in general, the
various curves are ordered according to the ratio of charge to
ion radius (z/r) for the cation. As more data become available,
it will be of considerable interest to test the correlation of
Yue with z/r, since it could provide a simple basis for estimates
of activity coefficients where data on the effect of melt compo-
sition are lacking.
4. Thermodymamic Correlations
4.1 Heats and Free Energies of Formation (act and AGT).
From the equilibrium measurements (Table I) and published
values of art and agt for HF(6) and H2O(g) [21], formation heats
and free energies may be calculated for Beo(s), Berz (5), NiFz(s),
-10-
and FeF2 (5) in the temperature range of interes ü here (Table II).
In addition, Long has calculated she and ag* for UF, from his
measurements (Eqn. 5), adopting the corresponding values given
by Rand and Kubaschewski (22] for UF. (s). In general, the
agreement of these calculations with previous estimates is
satisfactory, except in the case of NiF2 (s). Grimes has previously
indicated (23) that the present values for NiF2 (s), based on
Blood's measurements, differ considerably from - and are probably
more nearly correct than - previous estimates for NiFz (s) based
on measurements of Jellinek and Rudat (24). Long's values for
UF3 (s) agree with a previous estimate by Brewer [25] but differ
considerably from estimates by Glassner (26) and Rand and
Kubaschewski [22].
Included in Table II are other heats and free energies of
formation from the literature (27,28). These were used in the
calculation of AĦ and agt values for dissolved components (Table
III); i.e., the partial molal heats and free energies of formation
for the components in the solution standard states defined pre-
viously.
The Art and Agt values in Table III for Lif(d), BeF2 (d),
NiFz (d), FeF2 (d) and CrF2 (d) come most directly from the equili-
brium measurements, with an and Agt for HF(g), H2O(g), and Lif(s)
only being required from the literature. In the case of the
tetravalent fluorides, literature values of and Act for the
corresponding oxides were utilized, but these in general are more
accurately known than are those for the fluorides. The values of
A and AG for the rare earths are based on the solubility measure-
ments [20] and on Brewer's estimates for the solids [28].
The partial molal free energies of formation (Table III)
indicate the relative stabilities of the dissolved fluorides.
In terms of sgt per g atom of fluoride LiF is the most stable,
followed by the rarü earths. BeF2 and ThF4 next fali together,
then UF4 and ZrF4. Least stable are the structural metal fluorides.
-1l-
Table II
Formation Heats and Free Energies of Fluorides and Oddes
(700-1000°K)
-A3+ (1000°K),
kcal
Est. † Error,
kcal
66.20
0.3
46.04
Source
Ref. 21
Ref. 22
Ref. 21
5,6, (2)
5,6,(1),Eq. .10
123.39
-a", kcal
65.19
59.07
146.50
145.47
237.00
260.4
448.5
258.0
118.84
207.57
216.0
Ref. 21
381.1
Compound
I HF(8)
2 H20(8)
3 L1F(s)
4 Beo(s)
5 BeF2(1)
6 ZrO2(8)
7 UF4(8)
8 002(s)
9 UF3(s)
10 ThOz(s)
21 LaFz(s)
12 CeF3(s)
SmF3(s)
NiF2(s)
15 FeF2(s)
Ref. 22
218.0
Ref. 22
351.9
299.8
(7),(1),[q. 5
293
250
Ref. 27
422
360
416
355
405
344
Ref. 28
Ref. 28
Ref. 28
1,22,(1)
2,23,(2)
156.33
118.69
168.62
135.64
& The notations (s), (1), and (6) indicate respectively the solid,
liquid, and gaseous states.
Numbers not in parentheses refer to AG values for reactions in Table I.
Numbers in parentheses refer to AG values in this table.
Table III
Formation Heats and Free Energies for Solutes in LP-0.33BeF2
(773-1000°K)
-
Est. mm
-
kcal
kcel
*
(1000°K),
kcal
124.79
..
142.70
Solute
I L* + F"
2 La3+ + 3F
Ce 3+ + 3F
405.5
351.8
347.0
400.5
+
389.5
+
242.75
337.5
211.80
424.0
386.21
+
+
451.85
Wholt.61
386.48
á Ã Ã t K E Ő avau AwNr
Est. † Error,
_ kcal Source
(3), Eg. 10
24, (21)
25, (12)
26, (13)
5, (1)
19,(10),(5),13)
17,(6), (5),(13)
18,(8),151, 113]
6, (1),(8)
3,(1)
2,(1)
1,(1)
16,(4)
8,9,(2), (13)
11,111,15),
AG (H)
10,(1), (5),
AG+(H2S)
336.73
8 vht + 4F
9 U3+ + 3F
10 cyat
+ 2F
296.19
171.82
+ 2F
Fe2+
21
154.69
+
150.41
132.92
210.61
105.10
159.35
566 (763°K)
146.87
131.91
270.54
+02-
+ 20H
15 Be2+ + 21"
16 Be2+ + 52-
<79 (873°K)
The standard state of the ions is the hypothetical mole fraction
solution in LiF-0.33BeF2, with the exception of t, Be2+, and F, for which
the standard state is LF-0.33BeF2.
Numbers not in parentheses, numbers in parentheses, and numbers in
brackets refer respectively to items listed in Tables I, II, and in this
table.
0.2 Electrodo Potontials
The rosults in Table III are exprossod perha;s in a more
familiar way in Table IV as electrode potentials for various
halt cell reactions with
HF(g) + O * F(d) + 1/2 H2(8), E° - 0
(21)
The standard states remain the same. Tho manner, in which those
aro calculatod is most easily shown by pointing out that any
combination of half coll reactions which gives a complete reaction
of the form
11(s) + 2F,() JIP,(d), SE° - - nf
(22)
will yield the act value given in Table III (n is the number of
equivalents of charge and F 18 the Faraday). Š
The positions of the various cations in the electrochomical
series in LiF-0.33Bef, is seen to be quite similar to that in
aqueous solutions. This is 11lustrated on the left side in Fig. 6,
wherein the Eº values in L1F-0.33BeF, at 1000°K are plotted vs
the Eº values for the same half-cell reaction in aqueous solutions
at 25°c (29). It is notable from this comparison that the struc-
tural metals are relatively more noble in the molten fluorido than
they are in water. Also U% + is more stable with respect to US +
in the molten fluoride. When the Eº values in the fluoride solvent
are compared to values in Licl-KC1 eutectic given by Liu (30)
(on the right in Fig. 6) the same differences persist. To this
extent it might be said that the mol'cen chloride solver ĉ is . iro
like water than like LiF-0.33 BeFz. These differences - Core
relative nobility of the structural metals Fe, Ni, and Cr and
the relative stability of Us + with respect to U3+ – tend to
favor LiF-0.33BeFz as a reactor fuel solvet. The correlation
in Fig. 6 could prove a useful means of estimating the EC values
(AG values) of reactions for which data are lacking.
A number of other observations, in part evident com the
equilibriu measurements, are more clearly apparent from the Eº
values in Table IV. The stability of the trivalent rare earth
caticas is such that direct reduction to the metals as a means of
-12-
Tablo TV
Condicionated mecz5C ?oentecos.-9.332872"
171..
eactiesܕ ܢܐcc ?ܐܕܕ
zo (2.cC0°E, %.
criture
Coc?!!çient
1/C
-2.561
1* re* u(s)
::34,30 ** Lale)
0637 438 ** Cels)
-2.22
0.222
0.22
0.017
-2.24
Song (E)
0.795
-2.02
-1.721
0.715
-1.73
-2.316
0.755
2.319
0.674
610
0.630
0.207
30272016 3:(s)
mitt be as (8)
2-** , Zís)
w * , be * UCI
ug* 3e ** víz
utt ** 13+
** cz(3)
7e2+ + 2 * 7e(s)
122e = x1(s)
216) + (p ag.2c
{I216) + Iº
$52(6) + 2 * 32-
30216) + 2 * 02-
šo,(&) + 72(3) * 0.
P268) + e as
...Ch,
-0.390
-0.011
0.505
0.526
0.233
-0.473
°
.
I
<-0.45
60.10
0.558
1.734
2.671
0.136
0.472
0.044
a calculated from c values in Table III.
• Standard states for ions are defined in scoote (a) of
Table II.
romoving rare-earth lission products would be accompanied by
reduction of Bod* as well as U$*. Trivalent uranium should
disproportionate.
4UF, (d) - 3UF, (n) + U(s), E°(873°K) - -0.086 v. (23)
However, the difference in E° values for the U+/U3+ and the
Cr*/Cr couples is so large that the amount of US+ produced by
reaction with the container metal should be small (Eqn. 8) and,
in fact, the reduction of U*+ to vº by chromium
2Cr(s) + U*+ 2Cr2+ + U(s), Eº (873°K) - -0.950 v. 1. (24)
should not be significant (8). The strongest reducing agent
which can be used in Lip-0.33BeF, without reducing the solvent
components is beryllium metal, which indeed has been used as a
reducing agent during purification of such melts (11). With
Zrf, present, a weaker reducing agent (e.g., zirconium metal)
must be used.
4.3 Estimated Solubilities of other Oxides
Prom the equilibrium data, the free energy change associated
with the following reaction is known
go + 1/4 02 (3) *1/2 02" + 1/2 F2 (g), SG(1000°K) - 53.34
Kcal (25)
By combining this with ac values for the fluorides in Table III,
& values for the corresponding dissolved oxides are obtained.
These values in turn may be compared to published values for AG
of the solid oxides to yield an estimate of the corresponding
solubility products
log K. - -(act - AG)/2.2 RT
(26)
Such calculations show that, aside from the oxides already con-
sidered (Beo, ZrO2, UO2, Thoz) the oxides of Ni2+ and Fe2+ are
the only others of the cations listed in Table III which are
predicted to have low solubility (Table 1). Tio less soluble
of these two is Nio. The predicted value of xyie, in LiF-0.33BeFz
-13-
saturated with Beo and Ni0 is ~ 100% at 10:0°k. This is higher
than is expected for a typical reactor system, i.e., neither
Nio (nor Feo) is expected as a stable solid phase under normally
reducing conditions.
Less exact but useful estimates of the solubilities of
other oxides can be made from the following reaction
1/2 AF (s) + 1/2 02-(d) - Fº(a) + 1/2 102/2(8) (27)
AG - 1/2 G+ (MO_/2) - 1/2 SG? (MF,) - AG (1/2 02- - 8) (28)
wherein the last ag term corresponds to reaction (25) above. A
survey of available ag data for oxides and fluorides of Group I,
II, III, and IV cations, and of the bivalent first transition
metal cations indicates that aside from the oxides already con-
sidered only A13+ and perhaps Sc3+ should be significantly insolu-
ble. This scarcity of insoluble oxides is caused by the relative
greater stability of fluorides. It offers little help in the
search for refractory materials which mic. prove useful agents
for ion exchange or precipitation reactions which could be made
the basis for fuel reprocessing schemes. For example, an attractive
possibility would be the removal of rare earth fission products
by ion exchange reactions involving rare-earth oxides of suitable
low neutron capture cross section, but these oxides are expected
to be coo soluble in a LiF-0.33BeF2 and should cause the precipi-
tation of Beo (or Voz , if UF, is present).
4.4 Formation of Carbides
Direct reactions between constituents of the HSRE fuel and
graphite to form carbides may be represented by the following
general reaction for which chromium metal was chosen as the
reducing agent.
2/2 MF_(d) + 2y/2 C(s) + Cr(s) = Crf, (d) + 2/2 uCy (29)
AG - AG" (Crf, (a)) - 2/2 sGF (MF_(a)) + 2/2 Act (ucy(s))
(30)
Values for the free energy change associated with this reaction
at 1000°K are listed in Table v. They indicate that Zr, U, and
Th carbides should not be formed.
5.
Conclusions
The mutually re-enforcing relationship between chemistry
and thermodynamics in a reactor development program is, I think,
well illustrated in the case of the molten salt reactor program.
The chemical information most urgently needed in order that
development of the MSRE might proceed was obtained by direct
experiment. But, by conducting these experiments in a sufficiently
controlled manner that the chemical equilibria involved could be
identified and the associated equilibrium constants measured,
each experiment could contribute to a growing knowledge of the
thermodynamics of the pertinent molten salt systems. This
knowledge, in turn, has been useful in several ways. It has
extended the significance of the chemical information upon which
it is based, it has indicated where new chemical problems may
arise and how to deai with them, and it provides a growing structure
which will continue to grow in extent as more data are acquired,
and in usefulness as more problems appear which require solutions.
-15-
.
a.
-
Table V
Carbide Formation
MF_(a) + C() + Cr(s) = cxF2(a) + { mC,,
AG (2000°K), a
ucal
AG (Reaction, 1000°K),
Carbide
• kcal
Zrc(s)
UC2(s)
ThCa(s)
-38
-42
-48
& Ref. 27, page 587.
Figure Captions
Fig. 1.
Fig. 2.
Fig. 3.
The System LiF-BeF2. ORNL-Dwg. 65-2526
Oxidation-Reduction Reactions in LiF-0.33BeF2.
Left; reaction of Cr, Fe, and Ni with HF (1 atm)
and H, (1 atm): right; mole fractions of FeF2,
NiF2 and the Xyre. Nye ratio in equilibrium with
100% mole fraction Crf and chromium metal.
ORNL-Dwg. 65-4608 Rev.
HF-H, Sparging. Calculated efficiency of oxide
removal from LiF-0.33BeF2 by sparging with HF-H,
mixtures. The overall utilization of HF required.
to lower the residual oxide content to 10-3 mole/kg
is plotted vs the initial oxide content. The total
pressure is 1 atm. ORNL-Dwg. 65-4187
Oxide Concentration at Saturation in LiF-0.33BeFz
as a Function of the Concentration of ZrFAdded.
ORNL-Dwg. 65-2542R. .
Variation of Activity Coefficients in LiF-BeF2, 600°c.
Left, as a function of xpon at low solute concentration;
right, Yarf, as a function of xzrf, and Yuf, as a
function of Xyr added to LiF-0.33BeF2. z/r is the cation
charge to radius ratio.
Electrode Potentials in LiF-0.33BeF2 (vs HF/H2 ,F-
electrode) at 1000°K Compared (left) to Aqueous Electrode
Potentials at 25°c and compared (right) to Electrode
Potentials in Licl-KC1 Eutectic (vs Ag* /Ag) at 450°c.
ORNL-Dwg. 65-4607R.
Fig. 4.
Fig. 5.
Fig. 6.
FIGURE 1.
The System LiF-BeF2.
ORNL-Dwg. 65-2526
on;
-on
ode
ORNL-DWG 65-4608 |
TEMPERATURE (°C)
700 600
500
TEMPERATURE (°C)
700 600
| 800
500
800
U4+/3+
ct/3+
-Cr2+/cro
Fe2+/Feo
2 이유 ​4
. +x/+ex so + e^x
Cr2+ (~400 ppm)
EQUILIBRIUM QUOTIENT
6.2+
LNi2+IN
Ni2+
2 HF + MOM+ 2F + H,
HF + u3+ = U4++F+12 He
-
MO+ Cr2+ M+ + CrO |
ust + Cr2+ u4+ +
Cro
40-21
1.46
| 0.9
40
44
42
43
0.9
40
4.
42
43

4000/T (K)
4000/(K)

.
*
"
ORNL-DWG 65-4187
INFLUENT HF PRESSURE (atm)
500°C
= 0.2-04 H
=0.05 +
0.2
-
0.1
0.05
600°C
0.2
HF CONVERTED TO H20 (%)
0.05
700°C/ITT
10-2
2
100

5 10-1 2
5
INITIAL OXIDE CONTENT ( mole /kg)
ORNL-DWG 65-2542
600°C
i BeO SATURATION
ZrO2 SATURATION
OXIDE CONCENTRATION (mole /kg)
+ BeO SATURATIONE
500°C
H
ZrO2 SATURATION -
10-3
0.004
0.01
1
2
0.02 0.05 0.1 0.2 0.5
ZrF4 CONCENTRATION (mole/kg)
is :7


ORNL AIC - OPSICIAL
ORNI - AC - OFFICIAL
1
LIIL
L
.Yulibune
TOTAL
T
ITL.:
0111AHIIHUINI
Milit:18
10
IIIIl llllllll
II:IUT!Nonnu
Iniin mihi
M
WIMMINIMI
Hiir: 11
NIP:10:11MNII
HITIMUN
HINDWA
I!HUHU
AMAMI
THUIIIITA
HUHUHUDI TINHO
URLIOIUM
:!!il!::::
HUDBA
QA
11111111111
UJOUIMUM
MANI
WILLILIITUMU
HunnIHNINH
UIT
Audin WIH
H!! HIPMI
MUU
01.10
OO DHI
ហើយ
MWWW.
ITAHIDID
! !! !
Hil!
IIlIllIIIIIIII
TV
UHU
iluli
Tum
HhMHIMU
illllllllll
unuwun
SUIHI!1
WINWIN
TRAVEL
NDH Mihi
motURO
AHITIMUI
HUTH
INTRO
I
lIllliillil
Julio HII
WikillnIHI
DAMUHITINI
HlllUIIHII
UHON
Hull
WIMUUU
Huuuliiill
QALANMI
Wrail Hill HD
Windo
BWAMI
Millon
WIAHNILMIN
AIO DUNHil
widualnum
MUHUHU
:0 Wien
Jiliil
Hilulilu
HII
TAHIMINIMI
Judulli Huuhul
HU!!
IRINHWAI
TEAU
DUDHI
DUUT
lung!
WIMUI
HAHAHAUPUN
Muundumun
ITIMIIII!!
LUMINIQURMA
MilliNIHUTAN
TITILHUIIIIIQDH
INUIUINNIMUIOIHII
IMUNOHTOTT W
TwinDiRhi MODA
Cilin00H III
THMIDT
HA
iMilita
MAHAL
Willandi:
ieval HA
LILA ! HII
Unin UMUHOT
NHUIVITITIINITI
IBMIIN
Om
..انا
HIINIMnnin
ul:
TAD Kiir
WDR
DAMU
ANIWill
li:
IN
inANIUTIUS11
illiHWIMMI
OHHATT
TRIINI
UNUT
Ö
Ő
a
b
Kuorasfory Burby
I CYCLE IN DIVISION
wew.u. ...
KUFFEL . (ISER CO.
359-1G
3EM-Loaлaiтиміс
ONNI - AEC-O
ORNL - AEC - OKSICIAL
ORNL - AEC - OFFICIAL
OINE - AC - OFFICIAL
SEMI-LOGARITHMIC 959.61LG
Keurich. ISSERCO.
CYCLES N 150 DIVISIONS
ii1
.
X Zrfar
.02
1
:
.
or Xofy
1
VINUOUVEAUDIHOUSUION
.04
..
I
.
boat
.
Liili
.....
E
V
.
.
.
-
Figs, Poojar

.
IN
$
-mas.
D
Vic!!!!!!!:1::!:!1:.!"
:1,I
I I
l I ' iiiiii
OANI - AIC - OKSICIAL
ORNI - AIC - ONNICIAL
ORNL-DWG 65-4607
IN
HF - H2iF"
E° IN 2 LIF - BeFe (volts)
events
* -U3.
Sm 3+
-3
2
-4
-3
2
3
-2 - 0 1
EO IN H20 ( volts )
-2 -1 0
EOIN LIFEKF'EUTECTIC ( volts )
21.


References
10.
Briggs, R. B., ed., Molten-Salt Reactor Program Semi-
annual Progress Report, USAEC Report ORNL-3708, Nov. 1964.
Grimes, W. R., Nuclear News - ANS, May (1964) 3/8.
Thoma, R. E. et al, in Oak Ridge National Laboratory Report,
ORNL-3789, April (1965) 3/7.
Blood, c. N., Oak Ridge National Laboratory Report,
ORNL-CF-61-5-4, Sept. (1961).
Long, G., in ref. 3, 65/72. Papers in preparation.
Dirian, G., in ref. 3, 76/79. Paper in preparation.
Romberger, K. A., private communication.
Grimes, W. R.; in ref. 1, 235/244.
Mathews, A. L., and Baes, C. F., Oak Ridge National
Laboratory Report, ORNL-TM-1129, May (1965). Paper in
preparation.
Baes, C. F., and Hitch, B. F., in ref. 3, 61/65.
Shaffer, J. H., in ref. 1, 288/303; Shaffer, J. H., et al,
in ref. 3, 99/109.
Stone, H. H., and Baes, C. F., in ref. 3, 72/76.
Freasier, B. F., Stone, H. H., and Baes, C. F., Work in
progress.
Shaffer, J. H., in Oak Ridge National Laboratory Report,
ORNL-3122, Feb. (1961) 120/122.
Eorgan, J. E., et al, in Oak Ridge National Laboratory Report,
ORNL-3591, May (1964), 45/46.
Cohen, I., and Schaner, B. E., J. Nuclear Materials, 2,
(1963) 18/52.
Romberger, K. A., et al, in ref. 3, 243/245.
Shaffer, J. A., and Watson, G. M., in Oak Ridge National
· Laboratory Report, ORNL-2931, April (1960) 90.
Barton, c. J., J. Physical Chem., 64 (1960) 306/309.
11.
12.
13.
14.
16.
17.
18.
20.
ORNL-2749, (Oct. 1959). See also, Grimes, W. R., et al,
Chem. Eng. Prog., 55, No. 27 (1959) 65/70.
References - Contd.
21.
24.
25.
26.
JANAF (Joint Army-Navy-Air Force) Interim Thermochemical
Tables, Thermal Research Laboratory, Dow Chemical Co.,
Midland, Mich.
Rand, M. H., and Kubaschewski, O., The Thermal Properties
of Uranium Compounds, Interscience Publishers, New York
(1963) 71,
Blood, C. M., Blankenship, F. F., Grimes, W. R., and
Watson, G. M., Activities of Some Transition Metal Fluorides
in Molten Fluoride Mixtures, Paper 208 (presented by Grimes),
7th International Conf. on Coordination Chemistry,
Stockholm, June, 1962.
Jellinek, K., and Rudat, A., 2. Anorg. u. Allgem. Chem.,
175 (1928) 281.
Brewer, L., et al, MDDC–1543 (Sept. 1945).
Glassner, A., The Thermochemical Properties of the Oxides,
Fluorides and Chlorides to 2500°K, Argonne National
Laboratory Report ANL-5750 (1958).
Smithells, C. J., Metals Reference Book, Interscience
Publisher:3, New York (1955) 591.
Brewer, L., in The Chemistry and Metallurgy of Miscellaneous
Materials; Thermodynamics (Quill, L. L., ed.), McGraw-Hill,
New York (1950) 76/192.
Latimer, W. M., The Oxidation States of the Elements and
Their Potentials In Solution, Prentice-Hall, New York,
2nd Ed. (1952).
Liu, C. H. in Handbook of Analytical Chemistry (Meites, L.,
ed.), McGraw-Hill, New York (1963) 5-218 -- 5-221.
Grimes, W. R., in ref. l, 237.
Elliott, J. F., and Gleiser, M., Thermochemistry for
Steelmaking, Addison-Wesley, Reading, Mass. (1960) 177, 190.
27.
28.
29.
30.
31.
32.
ORNL-DWG 65-2526
650
600
LIQUID
550
548
LiF + LIQUID
TEMPERATURE (°C)
+2LiF. BeF2
+ LIQUID
457.6
BeF2 + LIQUID
LiF + 2LIF.BeF2
360.3
350
2LIF.BeF2 + BeF2
300
LIF

10
20
30
40 50 60
BeF2 (mole %)
70
80
90
BeFz
d."
.
St
--...-
--
.nimi nia
.
qim
9/ 14 / 65
DATE FILMED
END