I

H. T. HA"'
Hanford Laboratories Operation, General Electric Co., Richland, Wash.

Uranium Oxide-Liquid Metal Slurries

T

IN
POWER REACTOR operation use of a
molten metal as supporting medium for
a fissionable material eliminates expensive fuel cladding, reduces decontamination requirements, permits continual
withdrawal for chemical processing, and
provides freedom from radiolytic decomposition.
The liquid metal must have a relatively low melting point, fairly wide
liquid range, and low neutron cross
section. Because stability of the slurry
varies inversely with differences in density benveen its constituents, the metal
should have a density similar to that of
the suspended material. Bismuth and
lead, \vhich have densities of about 10
grams per cc., are promising.
The molten metal may support the
fissionable material in solution or as a
slurry containing a suspended intermetallic or oxide. Solubilities of the
fissionable metals are limited at practical temperatures, yet high enough to
modify intermetallic particle size by
solution and recrystallization. Oxides
offer a wider temperature range of
particle size stability. However, fissionable oxides may not be wetted by the
liquid metal at reasonable temperatures.
Whether wetting occurs is determined
by the balance of forces at the interface formed by the solid and liquid.
This relationship is described mathematically by
;SA

= ysL

yL cos

(1)

in which the surface tensions are those of

the solid, the solid-liquid interface, and
the liquid, respectively; contact angle
0 is measured through the liquid.
When 0 is 90' C. or less, the surface is
said to be wetted. Wetting is favored
by a high solid surface tension, low liquid
and interfacial tensions, or combinations.
To attain maximum solid surface tension, the surface must be free of adsorbed
atoms.
The contact angle of bismuth on
uranium dioxide decreases from 138.5'
at 380' C. to 92' at 1242' C. in argon
( 4 ) . The wetting temperature is therefore greater than 1250' C. Similar
angles were found for lead on uranium
dioxide.
Initial attempts to wet uranium oxide
with bismuth failed. However, in 1956
experimenters a t the Knolls Atomic
Present address, Phillips Petroleum Co.,

Idaho Falls, Idaho.

P o ~ e r Laboratory dispersed uranium

oxide by adding various metals as oxygen
getters (7), presumably raising the oxide
surface energy. Although no uranium
oxide was visible at the surface of the
product in successful experiments,
patches were noted in the interior.
Magnesium. sodium, titanium, and uranium (as hydride) \\ere satisfactory;
lithium and tin v ere not.
Experimental

Methods and Materials.

Two
methods of slurry preparation were used
successfully, magnesium gettering and
in xitu preparation from uranium and
bismuth sesquioxide. I n the former
sufficient magnesium was added to the
uranium oxide charge to reduce the
oxygen-uranium ratio to 1.6, assuming
that all the magnesium was oxidized.
Preparation temperature was usually
700' C.; higher temperatures were used
for comparison to the in situ method.
The in situ method is based upon the
reaction
U

2/3

Bi203

UOz

"3

Bi;

AF = -150 kcal.

(2)

Above 840' C. the bismuth oxide is

molten. The solubility of uranium in
bismuth at this temperature is about
10 weight 7 0 , so that a completely liquid
reaction path is possible.
Materials were prepared as shown in
Table I.
~~

Table I.
Material

U
UOS
Bi203
Bi, Mg

Preparation of Materials
Preparation
Degreased, cleanedin8S "03,
washed in chilled water and
acetone, dried in air
HZreduction of UOa; sieved to
< 140 microns with mean particle size of 3 microns
Oxidation of Bi in air stream
Reagent grade

Slurries were produced in capsules

machined from 304-L stainless steel?
2 X 1 to 5 X 2I/s inch in inner
diameter. Capsule contents were added
in air, no attempt being made to blanket
the operation with inert gas except when
the lids were welded into place.
Capsules were heated either in a muffle
furnace with rocker agitation or in a
9600-cycle-per-second Tocco induction
unit with manual or rocker agitation.

A specially designed rocker was used

with the Tocco unit ( 3 ) .
T o avoid sectioning of each capsule
and tedious analyses, a gamma absorptiometer was constructed for nondestructive examination of the slurries.
I t consisted of a thulium-170 source, an
automatic traversing platform, a 1/16 X
3/8 inch collimating slit in a lead block,
and a sodium iodide scintillation crystal
with associated electronic equipment.
The instrument was capable of determining the uraniumcontentof a uraniumbismuth capsule within 1% absolute.
Measurements on uranium oxide systems
were invalidai.ed by gas pockets in the
slurry.
The experiments at the Knolls Laboratory were confirmed in these laboratories at higher uranium concentrations. A 50-gram batch of slurry containing 13.2 iveight 7 0 uranium as dioxide, plus sufficient magnesium to reduce the oxygen to uranium ratio to 1.6,
was capsulated and heated 3 hours at
700' C. with rocking. The capsule was
air-cooled and cut open. The oxide had
dispersed within the melt. The casting
was trisected and found to contain 13.3,
12.1, and 14.0 weight 7 0 uranium in
descending order. (Uranium concentrations are henceforth expressed parenthetically as weight per cent in descending order.)
In an identical experiment: save for
replacement of bismuth with the bismuth-45.5% lead eutectic composition,
analysis confirmed (15.2, 12.7, 11.7)
that the oxide could be dispersed by
this technique and that the suspension
was reasonably stable.
Gallium Experiments. To facilitate
observation and measuremen:? a dispersion which is unreactive to\vard glass
is desirable, but free uranium is undesirable in the gettered product.
Gallium will not attack glass, but \vi11
remove oxygen from bismuth and reduce
higher oxides of uranium to the dioxide.
It is low-melting and mobile.
However, a charge of uranium dioxide
in bismuth plus gallium was unwet when
heated 3 hours at 1100' C.: 16 hours
at 900' C.? and 8 hours at 1100" to
1175' C. with periodic oscillation. In
a second attempt a sealed \-).cor tube,
previously evacuated to 3 microns at
500' C., was heated to 850' C. for 6
days with periodic shaking by rods attached to the tube ends. The upper
portion of the casting contained high
concentrations of uranium-up
to 26
weight yO-and a spotty distribution of
VOL. 51, NO. 2

FEBRUARY 1959

197

segregated oxide partially coated with

gallium. I t was concluded that uranium
dioxide itself is not wet by bismuth a t
these temperatures.
In Situ Method. The desire to prepare a dispersion free of additives with
tailored oxygen to uranium ratio suggested a n in situ preparation. A charge
Qf uranium, bismuth oxide, and bismuth
was capsulated and heated (Table 11,
No. 1). When the capsule was opened,
the bismuth oxide had disappeared and
no uranium was found in the original
form. Finely dispersed uranium oxide
was visible as red particles during micro-

scopic examination under polarized light.

I n other experiments in which the
oxygen-uranium ratio was close to 2
(Table 11, No. 2 ) , a considerable amount
of unwetted oxide was found at the top
and throughout the upper portions of
the casting.
The amount of oxygen present in the
free space of the capsule increased the
oxygen-uranium ratio to less than 1.97.
Because lack of wetting was not due to '
oxygen gain, either some uranium was
removed by combination with the capsule, or oxides with oxygen-uranium
ratios close to 2 cannot be wet at these
temperatures. I n experiments in which
the original ratio was 1.95 an essentially
unwet powder was found on the surface
of the product; in no experiment in
which the ratio was 1.67 or less has the
unwet powder occurred. As no magnesium is present in the in situ preparations, and uranium dioxide is not wet
a t these temperatures, it is concluded
that in this case wetting is produced
by reduction of the oxide-uranium ratio.
Magnesium Gettering. As gettering
of excess oxygen is not solely responsible
for the wetting obtained with magnesium, alternatives were sought. From
thermodynamic data ( 2 ) uranium monoxide is more stable than the dioxide in
the range 300' to 800' C. Because of
its oxygen deficiency the surface energy
of the monoxide is probably higher than
the dioxide, a condition which would
promote wetting. However, uranium
monoxide has not been identified in
bulk. Surface films have yielded x-ray
diffraction data, but these lines were not
found in diffraction patterns of the dispersion.
Although no evidence exists of free
reduced forms of uranium, magnesia
may exist as a coating or bridge between
the oxide particles and bismuth-e.g.,
Bi..Mg-O..U-0. This mechanism received support from later experiments ( 3 ) ,
in which wetting did not occur with a
higher oxide (U308), but did occur
with addition of magnesium to reduce
the oxygen-uranium ratio to 2.0.
Stability of Slurries. To estimate the
relative stabilities of slurries prepared by
both methods, capsules containing 8
Table II.

198

Charge, Grams
Biz03
Mg

1
2

30.6
9.40

...
...

20.0
12.0

...

21.6

4
5

18.8
47.0

...

O/U = 1.6
INDUSTRIAL AND ENGINEERING CHEMISTRY

weight % uranium (Table 11, 3 and 4)

were subjected to essentially identical
treatment. T h e .regular heating period
was followed by 30 minutes at 600" C.
with intermittent shaking. After dispersion, the capsule was allowed to stand
10 minutes without movement. The
furnace was then carefully removed and
the capsule quenched by a directed
stream of water.
The magnesium
gettered casting
(Figure 1) showed no oxide powder a t the
surface or in the interior. This particular capsule showed a clean separation o f '
slurry and bismuth
inch from the
base. Chemical analysis (10.5, 11.5,
11.0, 10.1, 5.2, 0.0; weighted average
7.9y0) confirmed that the slurry segregated upward-i.e., the effective density
of the oxide was less than that of the
bismuth. I n Figure 2 the large "balloons" are gas pockets. Microscopic
inspection revealed many small uranium
dioxide particles clustered about the
periphery. The region below the interface was completely free of uranium
dioxide.
The casting from the in situ preparation (Table 11, No. 4) revealed no region

Slurry Preparations W e r e Selected to Illustrate Points

O/U

KO.

Figure 1 . A magnesium-gettered uranium oxide-bismuth slurry showed no

oxide powder

Figure 2. A micrograph of Figure 1

casting shows dispersion in segregated
region (50X)

UOZ

c, %

...
. ..

Bi
182.5
213.5

Wt.
14
4

...

0.78

213.0

...
...

20.5

51.1

...
...

195.7
136.9

8
20

53.3

...

1.916

179.8

20

~~l~
Ratio

1
1.96

Heating Condition.
Hours
c.

8
840
1.5
950-1170
2.0
1170-1220
0.5
600-700
1.60
1
875-930
1
1000-1090
'
1
1090.-1140
As in No. 3
1.67
20
900
1.67
1
1000-1200
1.60
2
850
22
900
72
600

NUCLEAR TECHNOLOGY
of uranium freedom after 10 minutes
a t 600" C. (15.4, 10.3, 5.2, 5.3, 4.9,
5.3; weighted average 7.4%), and no unwetted powder. I t was clear that the
effective density of the oxide was less
than the bismuth.
The stability may be expected to increase with uranium concentration, because of higher viscosity-e.g., a n l 1,570
in situ preparation with an oxygenuranium ratio of 1.67 was heated at
700" to 900' C. for 32 hours with rocker
agitation and 1 hour at 1200' C. with
manual shaking. The capsule was allowed to cool without movement. It
is estimated that the contents were fluid
for 15 minutes. Upon sectioning, no
unwet oxide \vas found and analysis
showed relatively little segregation (12.2,
12.5, 11.3, 11.2, 10.6; weighted average
11.6%).
Figure 3 is a micrograph of upper and
lower segments of this casting. The oxide
particles are 3 to 4 microns. T h e kidney-shaped inclusions have a diamond
hardness of 303 under 50-gram load.
The corresponding value for uranium is
200 and for uranium ferride (UeFe)
is 319. It is probable that the particles
are uranium ferride, a product of corrosion. The lowest segment shows uniformity of dispersion.
To identify the appearance of uranium bismuthide, a casting was prepared
containing 11.77, uranium in bismuth
(Figure 4). The large crystals have a
hardness of 58 and do not resemble any
formations in the oxide systems.
The largest uranium concentration
achieved in a slurry fluid a t 600" C .
has been 20 Lveight 7
' (Table 11, 5 and
6). After a quiescent period of 92
hours at 600' C. the slurries showed surprisingly little segregation-e.g.,
22.2,
25.5, 22.0, 21.1, 19.3, 10.9; weighted
average 19.9%. Though still mobile,
these slurries apparently are slushlike.
Sodium Addition. The upward segregation of oxide apparently results
from incomplete wetting-i.e.,
0" <
0 < 90'. One method of countering
this effect is to add a light metal to the suspending medium, to decrease its density.
Addition of up to 3 weight 70 of sodium
lowers the melting point of bismuth (5).
Sodium has also proved effective as a
wetting agent ( 7 ) .
Addition of 1.76 grams of sodium
to a previously described composition
(Table 11, No. 3) increased stability
markedly after 10 minutes' quiescence
at 600' C. (7.7, 7.3, 7.8, 8.4, 8.3, 7.8 us.
10.5, 11.5, 11.0, 10.1, 5.2; 0.07'). Use
of sodium as a stabilizing agent therefore
looks promising ( 3 ) .
Fluidity. Viscosity determinations
are complicated by the presence of
rising solid particles, high temperature,
reactivity toivard container materials,
and anaerobic requirements. ,4 work-

Figure 3. In situ preparation of uranium oxide in bismuth shows uniformity of

dispersion (250X). Upper segment at left, lower at right

able viscometer has not been developed.

Encouraging results have been obtained with a variable-speed solid cylindrical graphite rotor with tapered base
rotating in a graphite pot. The assembly is set in a vertical tube furnace
enclosed in a helium-swept glove box.
Direct comparison with substances of
known viscosity should reduce systematic errors. Comparison at 400" C.
of the 8% in situ preparation with bismuth indicated a viscosity about 1.5
times greater or approximately 21 mp.
Observation of the 11.5 weight 7 0
uranium slurry in a porcelain crucible
under IO-* mm. of mercury pressure
confirmed the fluidity of the system.
Gentle agitation easily redispersed any
oxide which appeared at the surface of
the melt.
Fluidity is felt during manual shak-

ing of the capsules. Occasionally above

1000" C. movement ceases, but fluidity
returns below red heat.
Conclusions

Dispersions of up to 20 weight (ro

uranium as oxide in bismuth are fluid
at 600" C. and may be prepared by the
magnesium-gettering or in situ procedure. An oxide-bismuth slurry may be
prepared by partial reduction of Us08
with magnesium. Wetting is apparently
not complete, as evidenced by upward
segregation of the solids. The mechanism of wetting varies. I n the gettered
slurry wetting apparently occurs as a
result of magnesia coating or bridging;
in the in situ preparation it results from
reduction of the oxygen-uranium ratio
below 2.
The stability of the slurry increases
with increasing oxide concentration due
to higher viscosity. A promising method
of stabilizing low concentration slurries
involves addition of a light metal, such
as sodium, to reduce liquid metal density.
literature Cited

(1) Davidson, .J. K., others, U. S. Atomic

Energy Comm., Rept. KAPL-1686
(1956).
(2)'-GlaSsner, A , , Ibid., ANL-5107 (1953).
(3) Hahn, H. T., Ibid., HW-57161 (1958).
(4) Livey, D. T., Murray, P., Atomic
Energy Research Estab. (Gt. Britain),
AERE-M-R-1746 (19551.
(5) "Metals Reference Book," Interscience,
New York, 1049.
RECEIVED
for review April 7, 1958

ACCEPTED
November IO, 1358

Figure 4. Uranium bismuthide crystals

d o not resemble oxide formations

Division of Industrial and Engineering

Chemistry, Symposium on Chemistry and
Reprocessing of Circulating h-uclear Reactor Fuels, 133rd Meeting, ACS, San
Francisco, Calif., April 1958.
VOL. 51, NO. 2

FEBRUARY 1959

199