EIR-Bericht Nr. 238

EIR-Bericht Nr. 238

Eidg. Institut fir Reaktorforschung Wirenlingen
Schweiz

High Temperature Reaction of Uranium Carbides and Transient
Metal Chlorides in Molten Chloride Media

M. Taube, J. Strejcek, F. Dietler

Sily

Wirenlingen, Juni 1973

EIR-Bericht Nr. 238

HIGH TEMPERATURE REACTION OF URANIUM CARBIDES AND TRANSIENT

METAL CHLORIDES IN MOLTEN CHLORIDE MEDIA

M. Taube, J. Strejcek, F. Dietler

June 1973
Contents

page

sSummary 1

1. Introduction, objective, principles 2
2. Experimental techniques 12
3. Kinetics: results and discussion 15
4y, Equilibrium: results and discussion U
5. The Impact of the 'inert' metal chlorides 35

6. General remarks 41
Summary

The reaction of uranium monocarbide with molten metal
chlorides (AlClB, ZnClz, CdCl2, HgCl2,
alkali chlorides (NaCl, KCl, RbCl, CsCl and eutectics

NaCl/KC1, NaCl/CaClZ) in temperature region between

MnClz) in molten

600 to 970 °C in sealed silica ampules was investigated.
The kinetics (half time of dissolution) and the equili-
brium was measured. In the system UC/MnCl-NaCl for the
equilibrium state, the concentrations of uranium in both
solid and liquid phases are roughly equal, which permitted
to allowed the investigation of the influence of the secon-
dary components as small amounts of uranium oxycarbide,
uranium sesquicarbide and the impact of different alkali
metals ions. The activity coefficient of uranium trichlo-

ride in this media was assested.
1. Introduction, objective, principles

The aim of this paper is to discuss some particular problems
concerning the kinetics and equilibrium of the reaction between
uranium monocarbide and transient metal chlorides in molten
chloride media in temperature range up to 1000 °c.

The reason for such a study is: (1). The development of uranium-
plutonium carbide as a solid fuel for fast breeder reactors
(e.g. see Schumacher 1971). (2) The possibility that pyrochemi-
cal processes might be employed with molten chloride media for f,
reprocessing and preparation of uranium carbide fuel (Vogler,
Argonne Nat. Lab. 1965, Beucherie 1966, Taube, Schumacher 1970).
(3) The investigation of the properties of uranium trichloride
in molten chloride media for possible use as fuel or blanket/
coolant material in a molten salt fast breeder reactor (Taube,
Ligou 1972) (see also Harder, Long 1969).

The most intensive investigation of the chlorination of
uranium (or thorium) carbide in molten salt media was carried
out using a strong chlorination agent (Ishihara 1964) or electro-
lysis (Hansen 1963). In the work described in this paper the
use of strong chlorination agents was abandoned since the aim was ‘,
the investigation of more subtle mechanisms which can be con-
veniently studied near the equilibrium state, in which both in
the primary solid phase (uranium monocarbide) and in the secon-
dary liquid phase (uranium trichloride in molten alkali chloride
medium), the uranium concentrations are of the same order of

magnitude.

The most appropriate carriers of chlorine for these investi-
gations are the transient metal chlorides and especially manga-
nese dichloride. The criteria for the selection of the appropriate

metal chloride are discussed later.
Only one earlier paper, so far as is known to the authors,
concerns work which is close to the experiments described here.
It is that of Robinson and Chiotti (1964) in which a system of
uranium monocarbide in molten metallic zinc media is equili-
brated with molten zinc chloride in KC1/LiCl media. The use of
a molten metallic phase in the equilibrium state makes this

investigation less important for the present case.

In this paper the reaction of uranium monocarbide with metal
dichlorides is investigated, which as a first approximation can

be written as follows:

<uc > +% (MeCl) = (UCly) + %(Me) + <CD 1)

where < > is solid phase

( ) is liquid solution in alkali chloride medium.

The next step may be the formation of metal carbide which is

represented by:
{Me> + =z {C) = (MeCZ> 2)

In some cases a further step of chlorination is possible - the

uranium (III) to uranium (IV) oxidation:

{Me > 3)

ol L

1
(UCl3) - (MeClX) = (UClu) +

The uranium monocarbide used is almost always a uranium oxy-
carbide which influences reaction 1) in the following simpli-

fied manner:

y 5 J
L ¢uc 0O.> + = (Me Cl ) = (UC1l,) + == <UC o_ >
e (1-x) % * 7Ty (1-y) 7

r 2 dMe>  + <0 ¥)

Only in extreme cases is the following reaction of uranium dioxide

precipitation also possible
U 0. C > + 2 (Mecl) = (U0l + L <U0,> +

-

S \S)

<Me> + EE (o) 5)

but a farther conversion of uranium dioxide to uranium tetrachlo-

ride is also possible

w| =
o

<U02> + Me C1_ = (UClu) + §—<Me 0, > 6)

The last - but not the least important factor which influences
the reactions is the components of the 'inert' media, in this

case the alkali metals or the alkali earth metal chlorides. In
the present work we investigated:

NaCl, NaCl/KCl eutectic, KCl, RbCl, CsCl and particularly

NaCl/CaCl,. eutectic.

2

Both the molten chlorides - the substrate MeClx and the pro-
duct UCl3 dissolve in this molten chloride media resulting in
more or less stable complexes which of course altered the equi-
librium of the entire system. This effect can be represented by
means of the equation

<ucd> + 'y MCL) + 2 (Me> + <CD T7)

» W

(MeC1 'y MC1) = (UC1,
where MC1l - inert metal chloride with, in most cases M an

alkali metal.

An important role is played by the metal chlorides which in-
fluence both the kinetics and the equilibrium of the system under
investigation. Where we used HgCl,, CdCl,, ZnCl, and MnCl, with

10 .10 .10 = & ) -
5d~7, 4da~", 3d level electrons (in order of decreasing free
enthalpy of formation) and A1C1

(without d electrons).

2

The main object of the investigation was uranium 'monocarbide',
nominally UC and for part of the study also uranium sesqui-

carbide UCl.S (U2C3)°

The temperature range of equilibrium and kinetics study was

550 °c to 970 °cC.

The kinetics and equilibrium measurements were done for 1/4,

1/2, 1, 2, 4, 8 up to 24 hours or longer.
The operations being sensitive against atmospherilia were
carried out in glove boxes under a controlled nitrogen atmo-

sphere with approximately 10 ppM O, and 10 ppM H.O. In some

2 2

cases the O2 concentration was a factor of 10 higher which could

have influenced the results.

The criteria for the selection of the appropriate compounds
for the postulated system represented by equation 1 are as
follows: The desired free enthalpy change A G of this reaction
should be as near to zero as possible in the temperature range
600 °

stant Keq be close to 1 but also the distribution ratio D for

- 900 OC. Not only will the value of the equilibrium con-

uranium, where

molar fraction of UC1l
molar fraction of UC

3

In such a case the determination of uranium in both phases - the
molten (in the form of UClB) and the solid (in the form of UC)
will be most exact, enabling the effect of other relatively
small influences (e.g. 'inactive' molten salt media) to be

readily measured.

From a general point of view the electronegativity of any here
choosen metal chloride should lie between a value of 1.2 (the
electronegativity of uranium and plutonium) and 1.6. For higher

electronegativity values the equilibrium moves towards uranium
o

chloride and makes the value of the equilibrium constant and di-
tribution ratio large and difficult to measure because of very

small amounts of uranium carbide. Selection of a metal of appro-
priate electronegativity thus leads to a choice of the following:

Mg, Al, Mn, Zn, Cd, Hg.

For the same reasons the 'inert' or non-reacting compo-
nent metals were selected from those with an electronegativity
no greater than 1, i.e. alkali metal chlorides NaCl, NaCl/KC1l,
KC1l CsCl and, as an alkali earth chloride CaCl2 (Fig. 1).

For the reaction 1, for the temperature 1100 K we can write:

rRrink 100 & = ae - g, = agl00 K
eq U Me reaction
A6 3 agf(1100K) _ pgF(1100K)
U (U013) {uc D
) £(1100K)
AGMe - AGMeClx

(f = formation)

f
UC1l
surement and excellent critical review of Ferris (1972) and for

T = 1100K

Taking the value of AG from the recent experimental mea-

1100k _ ) -1
MGy = 605 KJ.mol — UC1

3 5

1 2 4 [eV]
H
[} |
Li Be B e N 0 F
2 @ . 2 © 49 . 4 @
| |
NaMg ALsi I[P s ¢l ‘
|
|
K [CaScTi y Cr FeNiGalGe As Se  Br
o090 oo 90e0es o o A A
Mn Co Cu A)
'inert' metal chlorides

with no reaction with UC

B)

Suitable metal chlorides
for the chlorination of UC

C)
unsuitable for these experil
ments metals and non-metals

which chlorides react very
easily with UC and move the
eguilibrium far in the

direction UCl3

Ru Cd
RbSr|Y ZrMo)AgFSn Sb |[Te I
—_ oo *—@ 2
Nb In
Tc
I
Hg .
Cs Bal| Hf W Os|PbBi At
TaRelr Po
|Lanthanide
|
Raj| U
Fr Pu

Fig. 1

formula

Electronegativity calculated using the Allred-Rochov

(according to Cotton, Wilkinson, 1966)
From Potter (1972)

AGééOOK - -107 KJ. mol T uc

then the change of free enthalpy

1100K
reaction

1

-498 KJ. mol t U, say ~500 KJ. mol "

AG

Because of the experimental techniques (fittness of analysis
of uranium in both phases) we postulate as a first approxima-
tion a concentration in both phases equal to the case where the

free enthalpy of reaction is zero

1100K i ) K1100K =
reaction eq

AG
Then the approximate metal chloride must have a free enthalpy of

formation

£(1100K) _ 500 -1
AGMeClx = = KJ .mol o4

167 KJ.mol T C1

n

so that we could say, on the basis of data from figure 2, that
(1) the following metal chlorides will shift the equilibrium so

far that practically no uranium carbide will be measurable:
10

HgCl, (Hg2C12), AgCl, Cd012, ZnCl,. 2) the following will

result in an equilibrium in which the uranium is present in

both phases molten salt and solid (carbide) in clearly measurable
55 AlClB. 3) metal chlorides which practically do
not react with UC in other words 'inert' salts; NaCl, KC1,

RbCl, CsC1l, CaCl2 and BaCl2 (Fig. 2).

amounts; MnCl

On the basis of reaction 1 we write

3/x 3/x
Nyca Nie yUC1 Me
K - 3 . S S e N,y 8)
eq NUC N3/x yUC LT c'cC
MeCl_ YMeClx

molar fraction

where N

activity coefficient

Y

but of course

= 1 Yy T 0L v, =01

and activity coefficient quotient: m =

distribution ratios: D =

Me N

-
AGf [kJI.mol™' CL 3

Fig.

50+

100+

150-

200+

250-

NaCl

KCl

2

11

> A

CaCl2

Pu C\

g AGf for 1
mol metal

UcC

UCls4

UCL3
PuCls

> B

=====23 Ba Cly

Free Enthalpy of formation at 1100 K.

A)
B)

"ITnert" salt compounds

Reactive chloride metals

12

Then K_ - p.-pX* 5 . N 9)

If we use the following composition of substrates

1 mol UC + Q

X\

mol MeCl
X

then in equilibrium we obtain

(L-w) = <UC) + % (Q-w) MeClxz W(UCI3) + -;% *w {Me> +w - <C

10)

2. Experimental techniques

From the point of view of experimental techniques the follo-

wing problems are of importance:

1) The stability of the substrates and products against the con-
tainer materials - in this case silica ampules e.g. (for

sake of simplicity without stoichiometric coefficients)

UCl + Sio0 . uo + SiClu + USi

3 2 2 3

2) The stability of the substrates and products against the atmo-

sphere, in this case oxygen, water vapour:

UCl3 * H20 = U0Cl1l + 2HC1
19

3) The purity of substrates, especially the readily hydro-

lysing and oxidising components: UClB, ucC, MeClX

4) The selectivity of the separation methods: for instance the

selective solution only of UCl, in the presence of uranium

5

carbide.

5) The accuracy of the determination method: here the ampero-
metric titration of uranium chlorides soluted in aqueus

phase.

The problem mentioned above(II,1) of the possible interaction

of UCl3 in molten salt media with silica requires further study.

McIver (1966) writes: "The most obvious choice for a vessel is
silica as this does not react with UClB. However, if free uranium
is present in the melt silica is attacked and mixtures of UO, and

2
USi3 formed."

Brown (1972) states that "uranium trichloride partially with
silica at the temperature 850 © - 900 °C the product being U02."

In these experiments for dry substrates the reaction between
UCl, and SiO, is very slow and only after 6-8 hours is the amount

3 2

of UO2 formed significant.

On point 2. the components used and the method of purifica-
tion (drying) was as follows

- AlClB produced by Fluka AG in sealed ampules - no special

purification - used only for rough tests

- Ca012‘2H20 produced by Merck, purified as follows:

dry HC1l (Molecular sieve -4 R) for five hours from 20 °

to 300 °C;

C up
14

removal of HC1l and cooling in the N
with Oxisorb F, 0 <1l vpM, H

2
0 vapour < 0.5 vpM)

stream. (N2 - purified

2 2

- Cd 012, o H20 - purified as for Ca Cl2 . 2H2O and dried

- Hg Cl, (Merck) no purification - for rough checks experiments

only

2

- MgCl2 6H20 (pro analysis, Merck) purified as for CaCl2- 2H20
but a residue of MgO was left (seen as suspension in molten
salt:) )

- MnCl2 . 4H2O (pro analysis, Merck) purified as CaCl2 . 2H2O

- NaCl (pro analysis Merck) partially dried in vacuum and then

heated after 4 hours from 20 °C up to 200 °c

- ZnCl2 (pro analysis Merck) (SnO ~ 1,2 %) purified as CaCl2

2H2O

- CaClz-NaCI eutectic was electrolysed with tungsten electrode

(8 0.3 mm, 10 cm long) at ~ 1 volt the measured current was:

at 740 °Cc : 15 m A at 550 °C : 2 m A

- UC (produced by Nukem) nominal UC dried in glove box in

0.987

vacuum extractor for 1 hour our analysilis gave UCO.985 00.015.

- Silica ampule - vacuum dried, 2 hours, 150 °C sealed after

filling.

On point 4. For the analytical procedure a very important step
is the separation of both phases: the salt phase and the chloride
phase. Here we choose the method of solution in water at approx.
10 °c for periods of 20 minutes. This problem is partly discussed in
in the literature.
15

Besson (1963) writes: "UC reacts readily with water
only above 60 °C. The dissolution of UC in hydrochloric acid is
not complete when the concentration of acid is too low". Bradley
(1962) has observed during the hydrolysis of UC in water at
25 OC for 1 hour a very small gas evolution. The time required
to hydrolise a 3 - 4 g irregularly shaped specimen at 25 °c was
up to a week. Robinson (1964) separated UC from Zn metal by so-

lution in HNOS. The UC is not dissolved.

In this paper a series of hydrolyses of UC in water and
aqueaous solution of HCl were made. The approximate results are

shown in Fig. 3.

From these it is clear that the effect of a 20 minute attack
by water or by weak solution of hydrochloric acid (to prevent
the hydrolysis of uranium compounds) at 10 °c on the UC was
practically negligible. The flow sheet used in the present work

is shown in fig. 4.

3. Kinetics: results and discussion

The first stage in the investigation of equilibrium is of
course the kinetic measurement. For this purpose the uranium
carbide reaction with a series of the above mentioned metal

chlorides was investigated.

The chlorination by HgCl2 (ngclg) and CdCl2 goes very rapid-
ly. For a sample of solid uranium carbide of approx 0.5 g of
irregular shape in an arbitary choosen ratio of uranium carbide
to the metal chloride;

ucC - 2 Q MeCl
X X

16

100
0.8 M HCl
e 10+
2
c
v 0.6 M HCIL .
-) ¢
©
I
>
@
2 T
time of dissolution
used In this paper
0.4 MHCL
0.2 M HCI
01 , J'wa'ter "
0.1 1 10 70 100 B
Time C hours ]
Fig. 3 The roughly estimated kinetics of solution of uranium carbides

in water and low concentration aqueous HC1l solutions

Specimen 0.5 g UC + 5 ml aqueous solution
T 2 100C + 50C

t < 70 hours

Numbers given represent the molarity of HC1
17
Drying and Drying and Drying of <UC>
purifying of purifying of ~ 500 mg for Q=1
metal chlorides alkalichlorides ~ 250 mg for Q=5

N

Preparation
of melt

X

Reaction - Kinetics
and equilibration

<UCY>+3(MeCly) —
(UCl3)+3 <Me>+<C>

T=600 — ~1000°C
t=01 —= 24 hrs

In sealed silica ampoules

v

Solution in cold HCl
aqueous solution: ~01MHCI
~ 10 cm?
t= 20 min -mixing in air !
T=10°C
v
JZ Centrifuge ~ 5 min l
Aqueous solution Undissolved <UC>
of UCl3 —100 ml and possible other
aliquote sample U - compounds
v v
Amperometric solution of solid residue
titration of Uranium in 20ml 0.8 MHCL
+ HNO3 Boil ~ 3 hrs
v
Fig. 4 The chemical preparation Amperometric
titration of Uranium

- ——— e —y————

18

Where Q = 1 (simply stoichiometry) or Q = 1.5, in the tempera-
ture range 700 “%¢ - 900 OC, the dissolution of approximately

half of the uranium carbide occured in 10-30 minutes.

All the data cited - in fig. 5 - here are roughly interpo-
lated from the series of experiments due to the volatility of
some of the chlorides, and in sealed silica ampules they are

not very reproducible.

Fig. 5 gives the results for a temperature of approximately
800 °c against the free enthalpy change of the reaction 1).
The time of half dissolution t 1/2 (in minutes) is a rather
simple function of the free enthalpy change of the appropriate
reaction AG reaction (see fig. 5).
For NaCl/CaCl, the time of dissolution was too short to be mea-

2
sured.

Fig. 6 shows the kinetics of the chlorination by aluminium
trichloride. The approx. increase of temperature from 1000 K
up to 1170 K that is the inverse temperature from 1 x 10—3
to 0.85 x 1072 k71 results in the half dissolution time t 1/2
for UC falling from 4.5 hours to approx 0.6 hours, that is
7.5 times faster. It must be stressed that the AlCl~3 acts in
a different manner than the other metal chlorides and the
reaction 4). and probably 5). may also occur, influencing the

effective kinetics.

The manganese chloride (fig. 7) shows a considerable lower

reaction rate, and what is of importance achieves equilibrium in

a reasonable time of ~ 18 hours at 730 °C, ~ 2 hours at 900 °c,

equivalent to an 9 fold increase of reaction rate in the same

o
19
2000-
1000+
t/2| logti/2
5004  2.69
2.5
< 2004 2.3
oo |
=
=
S 1004 2
S
Y
S 504  1.69
0
S
w
E
—
204 13
104 1.0
0.9
5 1 1 | 1 1 )
10 200 35 50 100 200 500

Change of Free Enthalpy, AG CkJ /reaction]

Fig. 5 Kinetics of 50 % dissolution of UC in molten chlorides media
at 800 °C as a function of free enthalpy changes in the in-

vestigated reaction. Q = 1
20

50-

0.05-

0.03 T : T -

02 05 1 2 45 10 20
Time [hours]

Fig. 6 Kinetics of dissolution of UC in molten AlClB/NaC1

Q = 1,3
T = 730, 900 °C
21

> X
e
N —

sy

losses

0.05+

0.03 I | | 1 | | L}
01 0.2 05 1 2 < 10 20

Time of reaction [ hours ]

Fig. 7 Kinetics of dissolution of UC in molten MnCl2/NaCl

= 730 © and 900 °c

1 and 5

O =3
"
22

10
5 -
™ UC Q‘ 5
Slo s .
210 24
(n] {
1- S SR UC Q=11
* * ¥ <L
0.54
% % UCis Q=11
0.2 -
0.1 -
* *
0.05 | | 1 I I | | ‘ y
01 0.2 05 1 2 5 10 20
Time Chours ]
Fig. 8 Kinetics of dissolution of UC and UCl.S in molten MnClz/NaC1

1 : 4 T = 900 C
23

100- —k
80
Sl -
Ol
2| 60-
g; n
40+
20+
104 // |
0 / f ) | ) 1 | | } | B | | MV | _—
10 20 40 60 80 100 120 200
[CdCl2]
0
Mol % <UC>S
Fig. 9 The "stoichiometry" of dissolution of UC in molten
CdClz/NaCl

T = 800 "C; t = 1/2 hour
24

temperature range. The clear achievement of an equilibrium state
is shown in fig. 7 where the manganese dichloride was used as
the chlorinating agent. Note that the kinetic slopes differ for

different temperatures.

The kinetics of dissolution of uranium sesquicarbide seem to

give the value of t 1/2 a factor 2.5 lower than for uranium mono-

carbide when the chlorination using MnCl, was measured (fig. 8).

2

From this it is clear that for manganese dichloride the equi-
librium may be achieved in a reasonable time at 1000 K after
~ 20 hours at 1170 K after ~ 2 hours. The larger reaction time
is not appreciable due to loss of uranium trichloride probably
from the reaction with silica.
The correctness of postulated stoichiometry of reaction 1). was

measured in the case of CdCl2 which is shown in fig. 9.

4, Equilibrium: results and discussion

The simplified reaction represented by (1) must be modified
by the assessment of the impact of the metal carbides formation:
reaction 2). The metals used in this work as a carrier for chlo-

rine have a relatively small value of carbide formation.

In the case of manganese chloride the possible reaction as

represented by:

*

uUc > o+

MO J\N

—
(MnClZ) — (U013) +<(Mn3/2.C>

* for a formal stoichiometry, not for chemical individuum.

"
25

The manganese / carbon system includes numerous compounds
(Pascal 1960), the most probable in our case is the formation of
MnBC: the stoichiometry of the reaction given above suggests

such a possibility

cC) * > 1/2 MnBC + 1/2 C

For the temperature 1000 K the free enthalpy of formation
(Zefirov, 1965) equals

=1

f (1000 K) ~ - 14 KJ. mole C

(MnBC >

AG

In our case the change of the full free enthalpy of reaction 2).
may be assessed as approx - 7 KJ/reaction. This value is rather
small and does not strongly influence the calculated equili-

brium state.

This is additionally influenced by the relatively large
constraints in the formation of MnSC because in the system under

study the metallic manganese and the free carbon are in two

separated phases: the carbon (graphite) with a density ~ 2 g.cm-3
is floating on the surface of the molten salt (density ~ 3 g.cm_B).
3

).

and the metallic manganese is at the bottom (density ~ 6 g.cm

We must stress here that the change of free enthalpy for
reaction 1) is rather independet of the temperature in the in-
vestigated region (900 - 1200 K). The data for the calculated

values of AG for reaction 1) are given in fig. 10.
AG[kJ/reaction]

AGLkJI/32mol MnCl2])

AG[kJ/mol UCl3]

26

UCys

I\
600- ut
620 o . A
6604 ® ®
¢ ) ®
700 l . !
900 1000 1100 1200
Temp (K]
10 Calculated data of the influence of temperature on the

dissolution UC and UC

A)
B)

1.5

Ferris 1972

others (see Ferris 1972)

in molten MnCl2/NaCl
L P

27

Table 1 gives the values of the distribution ratio for the
equilibrium state at 900 °c in NaCl - medium. From equation (9)

and (10) one can assess the activity co-efficient quotient Ty

=
<
n
no

~

Now we arbitrarly postulate Y(UCl3)Na01 2= 1 (for a more detailed

discussion of the activity co-efficient of UCl3 in molten chlo-

rides media see section V) in this case we obtain

vy (MnC1 = 0.63

2)Na01

But for the hypostoichiometric ratio: % MeClX/UC = 5

4
-

we obtain: T = 0.74 and Y (el

Y 3)Nac1

The activity co-efficient for manganese chloride equals

v (MnC1 = 1,22

2)NaCl
Table 1 Value of distribution ratio for equilibrium state

28

T = 900 OC in NaCl - eutectic

Carbide Molar ratio: Distribution ratio

ucC MeCl2 (precision is roughly

Y Q= TC estimated).

(nominal) y

UC ~ 1.1 0.95 %+ 0.05

s ~2) SeD + 0.3
UCl.S ~ 1.1 0.25 + 0.2

Experimental results: T = 1100 K;

AG calcul. = 0; K
€q

L2 4
< QJ W Dme Dme Nc 1TY
measured calc. cale. assessed| calc.
1 ~1 (0.95% ~0.5 1 1 0.5 2
+0.005)
5 3.5 + 0.3 ~0.78 0.305 05 0.78 0.74

The validity of this calculation and appropriate assumptions is

shown also for the reaction of manganese chloride with uranium

sesquicarbide UC

1.5°
a3

According to Potter:

1 5) = = 22240 - 1.96 T[Kcal.mol-l] (1100 =~ 1400)K

For 1100 K AG LLU0R (uc

£ ) = - 98 KJ.mol T UC

LaB 1.5

The possible reaction is represented by:

{Mn > + €0 2>

AV
)=
N\

(MnCl2) = (UClB) +

RO\
MO\

The calculated free enthalpy change of this reaction

sgl100 K o L o g

For all other unchanged parameters (activity coefficient etc.)

AG =
1100K

R Tln K

-+

9 KJ

= 2.6

=~
R

The calculated distribution ratio is

~—
1R

0.37

(UC, ) =

The experimentely measured distribution ratio D
0.25 0.2

U 1.5

The discrepancy between the calculated and measured results is
probably due to the oxygen content in the nominal uranium sesqui-

carbide in the form of uranium oxycarbide (see below).
30

An important problem arises due to the occurrence of
bound oxygen in so called uranium carbide which is in fact
uranium oxycarbide.

In this case the reaction is as given in equation (4).

In the MnCl2 system investigated here the reaction is
probably as for equation 4 with X = 0.017 which for the
sake of simplicity is taken for stoichiometry and which
approximately corresponds with our substrate (see list of
substrates) - and y = 0.05 which is here more or less arbi-
trarily choosen on the basis of the data of Anselm (see
Potter 1972) concerning the U-C-0 system at 1400 °c. (See
also the criticism of the U-C-0 system (Potter 1972.)

Our substrate is thus an oxycarbide with x = 0.017
(1-x) = 0.983 and the product according to Anselm may be
choosen as an oxycarbide with y = 0.05 (1-y) = 0.095 (see

reaction 4) which gives

U

NN

2
C.g 983 O~g.017 * 5 MeCl, »UCL, +

For the sake of explanation the following ideal solution is

postulated

5

0 = =
0.017 = 2

18
UCO.983 uc + 3 ucC ) 0]

tJ
s

31

According to Potter (1972) the free enthalpy of formation of

hypothetical uranium monoxide can be calculated from the

reaction
f 1 {4 Y -
AG(UO) = 3 AG<U02> RT1ln X (1-x)E
E = interaction parameter or heat therm for a regular solution;
E = O for an ideal solution
x = mole fraction of UO in monoxycarbide phase.

Here we calculate for simplicity using E = O a value of

(1100 K) =
U9 .05 ©0.95

120 KJ. mol ™t

G

The shift of the equilibrium constant may be estimated from

£ £
RTln K = - AG - AG
U%).05 C0.95 ue
1n k(1100 K) - - - (120 - 107) = =3 - _ 1.y
- 8.3 x 1100 9.2 ° *

=~

n
no
()

which for otherwise unchanged parameters results in decreasing
the distribution ratio for uranium oxycarbide to 0.04 versus

approx 1 for hypothetical uranium monocarbide.
32

Robinson (1964) observed a non soluble residue in the

(molten)/zn012 (molten salt)
a black uranium carbide phase in a zinc matrix without any

system UC in Zn in the form of
uranium - zinc phase present but with 1.5 % oxygen. The

| product of this reaction is uranium oxycarbide (equilibrium
time some hours, temperature ~600 OC). The maximum fraction
of the uranium in uranium carbide which could have been oxi-
dized was 0.91. A high concentration of ZnCl2 in the salt
was necessary to accomplish this 91 % oxidation.

So the present experiments agree with the general results

of Robinson.

Some explanation for the case discussed above can be found
in the short discussion of the U-C-0 system by Anselm (see
Potter 1972) (fig. 11).

The chlorination of UC -xox may be represented by two

1
reactions
ue 0 > 2UCc + Xuc. 0
0.983°0.017 3 3 ~7’0.95°0.05
A) UC + 2 MnCl. - (UCl,) + s a0 E 5 g
3 2 3

2 .

B) UCO.9500.05 t 3 MnCl, =+ 0.95 (UC13) + 0.05 <UO>+ ... 3

AG = + 13 KJ mol

According to our kinetics investigation the reaction A has
a higher rate and of course, also a greater equilibrium constant.

This results in a faster and more complete chlorination of the
e

hypothetical uranium monocarbide than that of the monoxy-
carbide UC0.95 00.05 (Fig. 11).

The reaction between uranium oxycarbide and aluminium
chloride may follow the reaction (5) because of the much
higher free enthalpy formation of aluminium oxide than that

for the chlorides (according to Zefirov 1965)

pgf (1100 K) = 450 x 1.5 = - 675 KJ mol T A1l
Al1.0_ (AlO )
273 1.5
AGf(lloo K) . 185 x 3 = - 555 KJ mol T A1l
Alcl3
1000 K £ £ -1
AGYy 5 AGA12O3 AGAlCl3 - 120 KJ mol — Al

and the same values for uranium (simplified without oxida-

tion).
AGgélOOO K) = 475 x 1.5 = ‘710 KJ mol T U
1.5 (hypot)
AGf(looo K) = 155 x 4.18 = 650 KJ mol © T
UcCl
3
3ql000  _  ,of 1000 K _ , £ 1000 K _ _c0 £7 mo1~L @
U UCl3 o, 5

Following equations 5) and then 6)
34

Substrate for these experiments

/ L{Co.gea 00.017

Direction of Chlorination
C

‘ UC1-x Ox+UC1s

)/ <UC15>+<UCo9500050 CO

+<U02>

<U02>+<UC1-x0x> % CO2

Fig. 11 Phase diagram U-C-0
Anseln -~1400 °C (According to Potter 1972)
35

0.017

U0 . 2-0.017

0.017 O.983>+ (AlC1 EE(UCIB) + <U02>

2-0.017 5

2-2x 0.017
t{Me> + S5 017 <c>

without stoichiometric coefficient for the sake of simplicity
<UO2> + (AlCl.),) = (UCIB) + <Al

2037

5. The Impact of the 'inert' metal chlorides

The reactions governing the chlorination of uranium carbide
discussed above is of course, influenced by the properties
of the molten salt medium. In this investigation the follo-
wing 'inactive' alkali metal and alkali earth metal chlo-
rides were used. NaCl, NaCl/KCl eutectic, KC1l, RbCl, CsCl

and NaCl/CaCl2 eutectic.

The influence of these metal chlorides arises from two mechanisms:

1) interaction with the UCl, which is the product

b
2) interaction with the MnCl2 the substrate.
The UCl3 - MeClX system was rather intensely investigated but

most of the published experiments have been carried out in
LiC1l/KC1l eutectic, which because of the peculiar properties of
the small cations of lithium is rather irrelevant to the

36

studies made for this paper. We need only say that UCl3 in
this eutectic at ~500 °C has a positive deviation from ideal,
the activity coefficient being of the order of 10 for low
concentrations of UCl3 (for literature see Hardy 1966). The
activity of UCl3 in NaCl/KCl eutectic (Flenglas 1962) is rea-

sonably close to that above.

The more exact measurements (Hardy 1966) of the system
UClB/NaCl shows a positive deviation from ideal: the activity
coefficient is essentially greater that 1. The deviation from
the ideal is lower, if the temperature is higher and is extre-
mely sensitive to temperature changes. The deviation is lower
at lower concentrations but according to Kertes (1971) in
all system studied (UClS,
are reasonably constant over the concentration range - usually
below 0.01 mol fraction. Also the results of Krahe (1965) show
3 solid /LiCl - KC1l system.

Harder, Long and Steinaway in their recent publication (1969)

UClu..) the activity coefficients

a positive deviation for the UCl

estimate the activity coefficient for UCl3 in NaCl as excess

partial molar free enthalpy

RT 1n y (UClB) [Keal mol™t]:

from UCl3 liquidus data: - 1.9

from the U/UCl3 potential: - 5.0

3 %3 CeCl3 and the above
data: - 2.7T + 0.6

average for UCl, in relation to PuCl

"4
37

for 0.3 mol fraction UCl_5 in NaCl/UClB.

range 600 - 800 °C the aVerage value for the activity coeffi-

cient of UCl3 is therefore 0.25 * 0.10. Some additional in-

formation may be obtained from the phase diagrams of UCl3/

MeClX, but it must be stressed that such derivations are not

Over the temperature

sufficiently conclusive to prove the existence of complex
compounds and, therefore, the excess enthalpy. A sharp maxi-
mum at the melting point is highly indicative of the for-
mation of a complex compound, a less pronounced maximum
simply means a complex of lower thermodynamic stability
(Kertes, 1971).

From the point of view of complex formation the PuCl

3

systems are similar to those of UCl, as far as the compo-

p!
sition of the double salts and their melting characteris-

tics are concerned.

Usually there is no intermediate compound with NaCl. The

UClB/NaCI system gives an eutectic.

For the UCl3 system it seems that a cautious examination

of the phase diagrams points to the existence of two com-
pounds, K.UCl_ and K UCl6 (see Bagnall, 1969). From this it

2 5 3
can be concluded that the activity coefficient for UCl3 is
essentially lower than 1. In addition PuCl3 has an activity

coefficient in the KC1l medium lower than that for the NaCl

system (Pascal 1960, p. 390) (Fig. 12).

38

100
50+ E
P
20" \‘\
[
\\ -
10- —
*\\\El\
oA
51 N\\\
B
™ \\\\\\\:\\\
g; 2- \\\:::j~12_
x 1-—1 r\\m
054
oC L\\ F
0.2- J\|\ |
L_ G
0.1- — H
‘:J °
500 600 700 800 900 [°C]”
0.01 L S S S— u L
700 800 900 1000 1100
Temperature (K]
Fig. 12 Activity coefficient of UCl3 in molten chlorides
A) Partridge 1961 E) Hardy 1966
UC1l/KC1 in NaCl
B) Ferris F) Harder, Long 1969
LiCl o dilution NaCl
| 1gyUCl3 = 2.18+2190, * 1 43 Fapck 16867
C) Inman 1961 Mg012
UC1/KC1 H) Flenglas 1961
D) Knoch, Krahe KC1/NaCl
LiC1l/KC1 dilute J) Knoch 1967

MgCl2/NaCl/KCl
29

The system under investigation at equilibrium is a tertiary

system (e.g. UClB/MnCl2

which may also be described as quasi-binary system MeXUCI

/ MeCl where Me is an alkali metal)

2+X

Me_MnC1l . The case for Mn UC1l seems much less likely
y 2+y 0 2 3+22z

because the Mn with a half filled d-electron shell and

with a radius of 9 nm has the relation RMnClz/RCl = 0.50

which results in the octahedral co-ordination MnCl6-3. U+3

in a chloride medium has the same tendency for octahedral co-
ordination: UCl6 -3. The cation partner is probably MeCl in
both cases. The systems of MnCl2/Na01, MnCl2/KCI and MnCl2/
CsCl are investigated by Seifert (1966). Among others the

following melting compounds were found KMnClB, NaMnClB.

Of course one must treat these assumptions made from the
phase diagrams concerning complexion in the liquid phase with
some caution. But according to Sandannini (see Lumsden 1966)
the NaCl_,/MnCl, molten system was investigated and the devia-

3 2
tions from ideality in this system found

2

52 7 (2 - Wat)® eal

RT 1ln y = - 14000 (l-NNa

and for NaCl : MnCl2 151

RT 1ln y £ 1000 cal = 4180 J

which for T ¥ 1100 ° results in vy £ 0.2

Lumsden (1966) believes that the interaction here is so strong
that it cannot be assumed that the equation given above for

Yy applies over the whole range of compositions.

40

For the system NaCl/MnCl2 our calculation of the acti-

vity coefficient according to the Bogacz and Trzebiatowski
(1966) methods (see also Ferris, 1972)

1nai B lnNi + lnyi — Agfi (T% - %) 4 E§EE-<%9 - 1 + 1n %3)
where To = melting point of component 1
Jl = melting point of the eutectic here = 428 °c
R = Gas constant
AC_ . = difference of the molar heat of the crystaline
e component i on melting
/_\.Hf,i = heat of fusion of component 1

- gives for the excess free enthalpy

eXCess 1

I = = RTln Y = 5420 KJ mol

AG

which is in good agreement with the above data.

From this it seems that the activity coefficient of MnCl2
in molten chlorides is close to 1 since as stated by Ferris
(1972) the activity coefficient for the uranium trichlorides

which must be taken into account is about

1gy U013 = 0+ 1

that is 10 L 2 y < 10

v
41

The experimental results of the impact of the metal chlorides

investigated here are given below (Fig. 13).

In the series of alkali/metal chlorides: Na, (Na-K), K, Cs,
the decrease of distribution ratio is clearly seen which

could be interpreted that the activity coefficient quotient:

is relatively strongly influenced by the alkali metal com-
ponent. If the ionic radius increase the distribution ratio
decrease and the activity coefficient quotient increases.

This 1s possible when YU013 increases and/or YMn012 decre-

ases.

6. General remarks

The chlorination of nominal uranium monocarbide in mol-

ten chlorides media is influenced by several parameters

- the rate of chlorination increases a bout 100 times
(from 500 min to 5 min at 800 © for half dissolved uc) if
the value of the free enthalpy change for the given metal
chlorides changes from 25 KJ per reaction to more than

400 KJ per reaction - that is 16 times.

- the equilibrium state with K = 1 is for T = 1100 K

achieved for the system UC/MnCl2 in molten NacCl.
42

4 s
3 +
= "
‘- | O
2l
ol X
bl PR Z
OID
Dlv
"
o 4
1 +
05+
0 - : : : t :
1 2 3 A 5 6
Ratio 3/2 MnCl2/UCl3
Fig. 13 Distribution ratio in the system UC + MnCl2

= 2-4 hours

L
T = 900 °C
43

- the thermodynamic data for UCl3 from Ferris (1972) and

ucC, UCl.S uo C,_, from Potter (1972) are in good agree-
ment with our results, if the appropriate activity coeffi-
cient of UCl3 and MnCl2 in molten alkali metal chlorides
are between 0.1 and 10.

- the presence of uranium monoxycarbide strongly influences
the equilibrium obtained for these chlorination reagents
which have relatively low free enthalpy of formation e.g.
MnCl, but not for strong chlorination agents; e.g.

2

Al C1
3

Acknowledgements

The authors gratefully acknowledge the assistance of
Dr. S. Huwyler in carrying out the analyses; Mr R. Stratton
for preparation of the text; and they would also like to
thank Dr. H. Schumacher and Dr. P. Tempus for their support

in the performance of these studies.
Ly

Literature

K.W. Bagnall

The Halogen Chemistry of the Actinides in "Halogen Chemistry"
in V. Gutmann (ed) "Halogen Chemistry", Vol. 3. p. 304
Academic Press, London (1967)

P. Beucherie, M. Payrissat, J.G. Wurm

Preparation des carbures d'uranium et de carbures mixtes par
conversion directe des halogenures en carbures

Comm. Europ de 1l'Energie Atomique, Euratom. EUR 3170 f.
Ispra 1966

J. Besson, P. Blum, J. Sputs

A Study of Hydrolysis of Uranium Carbides in Proceed. Symposium
"Carbides in Nuclear Energy" Harwell, 1963

M.J. Bradley, L.M. Ferris

Hydrolysis of uranium carbide between 25 and 100 .o

I. Uranium Monocarbide, Inorg. Chemistry, 1, 3, 683 (1962)
D. Brown, J. Edwards

Preparation and Crystallographic Properties of the Trichlorides
Tribromides and Tri-Iodides of U, Np, Pu.
J. Chem. Soc. Dalton, Trans 16, 1757 (1972)

P. Chiotti, W.C. Robinson, M. Kanno
Thermodynamic Properties of Uranium Oxycarbides
J. of Less Common Metals, 10, 273 (1966)

L.M. Ferris, J.C. Mailen, F.J. Smith

Estimated free Energies of Formation of some Lanthanide and
Actinide Halides at 600 © - 800 OC using Molten Salt - Liquid
Metal Distribution Coefficient Data.

J Inorg. Nucl. Chem. 34, 491 (1972)

W.N. Hansen

Reprocessing of Uranium Carbide by Molten Salt Electrolysis,
NAA - SR - 7660 (1963)
45

B.R. Harder, G. Long, W.P. Stanaway

Compatibility and Processing Problems in the Use of Molten
Uranium Chloride - Alkali Chloride Mixtures as Reactor Fuels,
Nuclear Metallurgy 15, 405 (1969)

C. Hardy-Grena

Etude thermodynamique sur le trichlorure d'uranium dissous
dans les chlorures alcalins fondus.

Comm. Europ de 1l'Energie Atomique, Euratom, Ispra EUR 3153 f
(1966)

T. Ishihara, K. Hiraua

Chlorination Distillation Processing of Irradiated UO, an UC
Third Internat. Conf. Peaceful Uses of Atomic Energy,
Geneva 1964, Volume 10, 8/798. p. 520

2

C.P. Kempter
Hydrolysis Properties of Uranium Monocarbide and Dicarbide,
J. of Less-Common Metals, 4, 419 (1962)
A.S. Kertes
Complexes of Actinides in Molten Salts
Actinides Reviews, 1, 371 (1971) |
J. Krahe
Aktivitdt von UCl3 und anodische Polarisation von UC in der
Schmelze LiC1l-KC1l
Technische Hochschule, Aachen, Dissertation, 1965
J. Lumsden
Thermodynamics of Molten Salt Mixtures
Academic Press, London N.York, 1966
D.L. Maride, D.N. Hume

A New Method for Preparing OH-free Alkali Chloride Melts,
Electroch. Soc. J. .107, 354 (1960) |

E.J. McIver

A Galvanic Cell for Measuring the Uranium Activity in Carbides
at High Temperature,
Harwell, A.E.R.E. - R. 4983 (1966)

B.A. Partridge

The Thermodynamic Properties of Dilute Solutions of UCl3 in
UCl - KC1l Eutectic,
Harwell, A.E.R.E. - R 3683 (1960)
46

P. Pascal

"Nouveau traite de chimie minerale"

Tome XVI, Masson Cie, Paris, 1960

P.E. Potter

Ther Uranium-Plutonium-Carbon-Oxygen System,

J. Nucl. Met. 42, 1, (1972)

W.C. Robinson, P. Chiotti

Reactions of Uranium Carbide in Fused KCl - LiCl/Zinc System,

J. Less-Common Metals, 10, 190 (1966)

H. Schumacher

Plutonium-Brennstoffprogramm Jahresbericht 1972

EIR-Report 230, Wirenlingen (1973)

R.K. Steuerenberg, R.D. Pierce, L. Burris
Pyrometallurgical and Pyrochemical Fuel Processing
Process Chemistry, 1970 CM 56 : PC U

M. Taube et al.

New Boiling Salt Fast Breeder Reactor Concept

Nucl. Eng. Design 5, 109, (1967)

M. Taube, H. Schumacher

A Carbide Fuelled Fast Breeder Reactor with Molten Chlorides

as Fuel Bonding Material and Chloride Vapour as Coolant.

EIR-Report 167, Wirenlingen (1969)

M. Taube, J. Ligou

Molten Chlorides Fast Breeder

EIR-Report 215, Wirenlingen (1972)

M. Taube, J. Strejcek

Reaction of Uranium Carbides with Transient Metals Chlorides
in Molten Salt Media

(to be published)

R.C. Vogel, M. Levensa, F.R. Masten

Argonne National Laboratory, Chemical Eng. Div. Research
Highlights, April 1965 ANL - 7020

A.P. Zefirov

"Thermodynamic Properties of Some Inorganic Substances"
Atomizdat, Moscow, 1965 (in Russian)