3.14

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

Yeon Soo Kim
Argonne National Laboratory, Argonne, IL, USA

ß 2012 Elsevier Ltd. All rights reserved.

3.14.1

Introduction

392

3.14.1.1
3.14.1.2
3.14.1.3
3.14.2
3.14.2.1
3.14.2.2
3.14.2.3
3.14.2.3.1
3.14.2.3.2
3.14.2.4
3.14.2.4.1
3.14.2.4.2
3.14.2.4.3
3.14.2.5
3.14.3
3.14.3.1

3.14.3.2
3.14.3.3
3.14.3.4
3.14.3.4.1
3.14.3.4.2
3.14.3.4.3
3.14.3.5
3.14.4
3.14.4.1
3.14.4.2
3.14.4.3
3.14.4.3.1
3.14.4.3.2
3.14.4.4
3.14.4.4.1
3.14.4.4.2
3.14.4.4.3
3.14.4.5
3.14.5
References

Background
Historical Evolution of U Intermetallic Fuels
Performance Topics of U Intermetallic Fuels
U–Al
U–Al Fuel Properties
Thermal Conductivity of U–Al Alloy and UAlx–Al Dispersions
U–Al Fabrication
U–Al alloy
UAlx

U–Al Irradiation Performance
Fuel swelling by fission products
Interaction between U–Al and Al
U–Al blister threshold temperature
Summary for U–Al
U–Si
U–Si Fuel Properties
Thermal Conductivity of (U–Si Intermetallic)–Al Dispersions
U–Si Fabrication
U–Si Irradiation Performance
Fuel swelling by fission products
Interaction between U–Si and Al
U–Si blister threshold temperature
Summary for U–Si
U–Mo
U–Mo Fuel Properties
Thermal Conductivity of (U–Mo Alloy)–Al Dispersions
U–Mo Fabrication
U–Mo alloy powder fabrication
U–Mo dispersion plate fabrication
U–Mo Irradiation Performance
Fuel swelling by fission products
Interaction between fuel particles and Al matrix
U–Mo alloy blister threshold temperature
Summary for U–Mo
Summary and Outlook

392
392
394

395
395
396
397
397
397
397
397
400
400
401
401
401
401
402
403
403
406
410
411
411
411
413
413
413
415
415
415
416
419

419
420
420

Abbreviations
ANL
ATR
BU
EFPD

Argonne National Laboratory
(Argonne, IL)
Advanced test reactor (at INL)
Burnup
Effective full power days

EOL
EPMA
ETR
FD
HEU

End of life
Electron probe microanalysis
Engineering test reactor (at INL)
Fission density in fuel phase
High-enrichment uranium (usually $ 93
wt%235U)

391



392

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

IL
INL
LEU
MEU

Interaction layer (reaction layer)
Idaho National Laboratory (Idaho Falls, ID)
Low-enrichment uranium (<20 wt%235U)
Medium-enrichment uranium (35–45
wt%235U)
MTR
Material testing reactor (at INL)
OM
Optical microscopy
ORNL Oak Ridge National Laboratory (Oak
Ridge, TN)
ORR
Oak Ridge research reactor (at ORNL)
PIE
Postirradiation examination
RERTR Reduced enrichment for research and
test reactors (program)
SEM
Scanning electron microscopy

SEM
SEM backscattered electron image
BEI
SEM
SEM secondary electron image
SEI
TEM
Transmission electron microscopy

3.14.1 Introduction
3.14.1.1

Background

Uranium intermetallic fuels such as U–Al, U–Si, and
U–Mo are chiefly meant for research and test reactors in which neutron production, instead of power
generation, is the main purpose. The operation temperatures of these fuels are lower than those of uranium ceramic fuels used for power generation such as
UO2. In general, the U intermetallic fuels can achieve
much higher fission densities than can the oxide fuel.
Currently available research reactor fuels are predominantly in a dispersion form that is composed of
fuel particles dispersed in an inert matrix. Figure 1
illustrates the cross section of a dispersion fuel plate.
The fueled zone in a dispersion fuel plate, that is,
the fuel particles–matrix mixture zone, is frequently
called the ‘fuel meat’ or ‘fuel core’ and is metallurgically bonded to the cladding. Throughout this chapter, ‘fueled zone’ is used instead of fuel meat.
Aluminum is the most popular choice for the
matrix because of its low neutron absorption cross
section, low cost, and good fabricability. It also has

Fueled zone (or fuel meat)


adequate mechanical, physical, thermal, and chemical
properties for cladding material. Aluminum alloy as
cladding material also has good corrosion resistance to
any slightly acidic coolant. Another useful attribute of
aluminum is its relative compatibility to reprocessing.1
3.14.1.2 Historical Evolution of
U Intermetallic Fuels
The U intermetallic fuels for use in research and test
reactors are, in the order of earliest to latest, U–Al,
U–Si, and U–Mo. The basic driving force for the
development of a new fuel is to obtain a higher
uranium density in the fuel phase. Uranium metal
has the highest uranium density, but it is not usable
because of poor irradiation stability. The uranium
intermetallics were introduced to achieve stable irradiation performance of uranium metal. The uranium
densities for the candidate fuels are given in Table 1.
Because of its structural similarity to matrix aluminum, the first uranium intermetallic fuel chosen
for research and test reactor purposes was U–Al alloy.
U–Al alloy has a well-established performance history as the fuel for the materials testing reactor
(MTR) and engineering test reactor (ETR).
Fabrication of U–Al alloys with high uranium contents poses difficulties during the rolling process, and
uranium inhomogeneity increases proportionally with
uranium content. The typical picture-frame method of
fuel assembly and the related rolling fabrication method
are illustrated in Figure 2. When alloys of greater than
25 wt% uranium contents are needed, dispersions of
UO2 instead are used.2 (Full use of UO2–Al dispersion
has been limited in the United States, chiefly because of
swelling problems due to a reaction between UO2 and

Al encountered early in its development, although this
is not the case for Russian-built reactors. Instead, a little
less dense U3O8–Al dispersion has been used.) Table 1
gives the basic properties of fuels currently used (or
candidate fuels) for research and test reactors.
The application of monolithic U–Al alloy in higher
power reactors such as the advanced test reactor
(ATR) at INL and high-flux isotope reactor (HFIR)
at ORNL was deemed limited because of the fabrication limitations inherent for a high U-density fuel,

Cladding

Figure 1 Schematic of the cross section of a dispersion fuel plate frequently adopted in research and test reactors.
The fueled zone (or fuel meat) is composed of fuel particles (darker phase) dispersed in an Al matrix (brighter phase).


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

Table 1

Basic properties of uranium intermetallic fuels

Fuel

Melting point
( C)

Physical
density
(g cmÀ3)


Uranium
loading
(g cmÀ3)

U
U–7Mo
U–10Mo
U6Mn
U6Fe
U3Sia
U3Si2a
USi
UAl2a
UAl3a
UAl4
U0.9Al4a
UAlxc
UC
UN
UO2a
U3O8a
Ald

1133
1145
1150
726
815
930b

1665
1580
1590
1350b
731b
641b
NA
2500
2630
2875

19.1
18.4
18.2
17.8
17.7
15.6
12.2
10.96
8.1
6.8
6.1
5.7
6.4
13.6
14.3
10.96
8.4
2.7


19.1
17.1
16.4
17.1
17.0
15.0
11.3
9.8
6.6
5.0
4.2
3.7
4.5
13.0
13.5
9.7
7.1
0

b

660

a

Currently used Al dispersion fuels.
Decomposes.
This case is for the mixture of 60 wt% UAl3 + 40 wt% U0.9Al4.
d
Al is included for reference.

b
c

high fuel swelling, and the need for adding B-10
burnable absorber. The use of U3O8–Al dispersion,
that is, U3O8 particles dispersed in an Al matrix in
high-power reactors, was first considered and abandoned due to concern about the exothermic reaction
between U3O8 and Al3 and the interdiffusional reaction growth between U3O8 and Al, although this fuel
form is still used in some other reactors.
The fuel form of U–Al alloy with a U density high
enough to satisfy the need for high-power rectors
is a mixture of UAl2, UAl3, and UAl4, known as
UAlx. It develops when the U weight fraction is
pushed beyond $62 wt%. The exact fractions of the
compounds included in UAlx depend on the fabrication process. Whitacre4 reported that typical powder
lots used in the ATR contained phase fractions of
7.6 wt% UAl2, 78.6 wt% UAl3, and 13.8 wt% UAl4.
UAlx has several positive features that enable its
superior performance in high-power reactors. Fuel
swelling can be reduced by accommodating fission
product swelling in the powder dispersions, which
include pores left during fabrication. It also has exceptional resistance to fission gas bubble formation.
In addition, fabrication with a uniform distribution
of burnable absorbers is possible.5
To achieve a higher U-density fuel, UAlx composed of dominantly UAl2 instead of UAl3 was

393

tested.6,7 When handled in air, however, UAl2 is
more pyrophoric than UAl3, and this leads to complications in fabrication and the potential for oxygen

impurities in the fuel.
The US Department of Energy (US DOE) initiated
the RERTR (Reduced Enrichment for Research and
Test Reactor) program in 1978 to convert the world’s
research and test reactors using high-enrichment uranium (HEU) to those using low-enrichment uranium
(LEU). An enrichment in 235U of 20 at.% is the threshold between HEU and LEU. To use a fuel with reduced
enrichment, keeping the fuel phase volume the same in
the fueled zone (see Figure 1), requires using a fuel
having a higher uranium density to compensate for the
reduced fissile fraction in LEU.
In the RERTR program, the fuel form developed
to accomplish this is U3Si2, which allows the highest
possible uranium loading among the qualified fuel
types. This fuel showed excellent stability during
irradiation. Fission gas bubble swelling is of no concern for fuel dimension expansion at typical research
and test reactor applications. This fuel enabled LEU
core conversion of $60% of the research and test
reactors worldwide. Another U–Si intermetallic fuel,
U3Si, can achieve even a higher U loading than U3Si2,
but in plate-type designs, it shows unstable growth of
fission gas bubbles at high temperatures and burnups.8 In addition, the interaction layer (IL) between
U3Si and Al grows faster than that of U3Si2–Al. In
a rod-type design, U3Si–Al dispersion, however, is
known to have acceptable performance due to the
ability of the pin to constrain fission gas bubble
swelling in a more stable manner.9
Failure to convert high-power research reactors
using HEU to LEU U3Si2 called for fuels of even
higher uranium density. Given the unstable irradiation
behavior of the high-uranium-content compounds

(e.g., U3Si and U6Fe), the fuel development effort has
shifted to uranium–molybdenum alloys with Mo content ranging 6–10 wt%, in both monolithic and dispersion fuel forms. Since 1997, the U–Mo alloys have
been irradiation-tested under the auspices of the U.S.
RERTR program and other programs in Argentina,
Canada, France, South Korea, and Russia. These tests
have shown that U–Mo alloy has stable irradiation
behavior.
A major complication in U–Mo alloy dispersion
in Al is the reaction between U–Mo and Al. Under
certain irradiation conditions, fission gas bubble
growth in this reaction product is sufficient to cause
fuel plate failure. A small amount of silicon added
to the matrix aluminum has been found to be a


394

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

Cover
plate
Fuel bearing
alloy or meat
compact
Frame

Cover
plate
(a)


Al powder
Fuel compact

Al-alloy frame and
cover for cladding

Fuel powder
Assemble
and weld

* The hot roll procedure is typically
composed of several passes of rolling.
Before the first roll pass, annealing for
about 1 h is performed.
Between passes, additional annealing
is performed for ~15 min.
The annealing temperature is
determined by the hardness of
cladding material.
For Al6061 cladding, it is ~485 ЊC.
The softer the cladding material,
the lower the annealing temperature
is used.

Hot roll*
# The rolled plate is

Preirradiation
blister test#


annealed at ~485 ЊC for
1 h for blister test.
If blisters form, the
plate is disqualified.

X-ray fuel
geometry
Cold roll

Shear excess
material to desired
plate dimensions
(b)
Figure 2 Illustration of plate fabrication. (a) Exploded view of the dispersion fuel compact assembly by the
picture-frame method. (b) Flow diagram of the hot-rolling fabrication method.

promising remedy to this problem. The U–Mo
monolithic fuel, in which a U–Mo thin foil is sandwiched between cladding and directly bonded to
cladding, is currently under development and has
the advantage of providing higher U density than
the dispersion form while essentially eliminating the
problem related to reaction products between the
fuel and matrix. However, the problem of gap formation between fuel and cladding must be solved before
this fuel form is usable.

3.14.1.3 Performance Topics of
U Intermetallic Fuels
The performance of U intermetallic fuels is closely
related to whether they are crystalline or amorphous
during irradiation. The U intermetallic fuels tend to

be amorphized by damage in the crystal structure
caused by highly energetic fission fragments. The
viscosity of an amorphized material is lower than
when it is crystalline: in other words, the fluidity


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

of the material increases when it becomes amorphous. Amorphization of a crystalline material to
metallic glass is usually accompanied by an increase
in volume – a quantity called ‘free volume’ – which
facilitates atomic mobility, enhancing diffusion.10
Fission gas mobility is also high in amorphous material and the fuel material is more readily deformed
by the growing gas bubbles. Hence, overall, fission
gas bubble growth in an amorphous material is faster.
The three U–Al intermetallics undergo amorphization depending on the fission rate and temperature. The lower the irradiation temperature and the
higher the fission rate, the more readily the fuel
becomes amorphous. Among the three, UAl4 amorphizes most readily and UAl2 the least. The reaction
products between the fuel and matrix are also uranium aluminides and undergo amorphization. The
U–Si intermetallics (U3Si and U3Si2) also become
amorphous during irradiation, and the reaction product between the fuel and matrix, U(Si, Al)3, also
undergoes amorphization. The U–Mo alloy is not
amorphized during irradiation, but the reaction product between the fuel and the matrix, (U, Mo)Alx,
becomes amorphous.
In the following sections on U–Al, U–Si, and U–Mo
fuels, the areas of (1) physical properties, (2) fabrication
methods, and (3) irradiation performance are discussed. Each of the three review areas are described
in detail. The physical properties section discusses the
phase diagram, lattice structure of important compositions, and density. The section on fabrication methods
discusses relevant fuel particle fabrication processes.

The section on irradiation performance includes fuel
swelling, IL growth between fuel particles and matrix
aluminum, and blister threshold temperature.
Fuel particle swelling and IL growth are two major
fuel performance topics in research reactor fuel plates.
Both work to increase the fueled zone volume. In plate
geometry, fueled zone volume expansion is transferred
directly to plate thickness increase because the
restraint in this direction is the weakest. Monitoring
plate thickness is an effective method of tracking the
fueled zone swelling. An excessive plate thickness
increase is the indicator for potential fuel plate failure.
A unique measure for sound fuel performance
considered in research and test reactors is ‘blister
threshold temperature’ testing with irradiated plates.
In typical research and test reactor fuel designs,
because there is neither a gap between fuel and
cladding nor a plenum, no fission gas release and
collection is possible outside of the fueled zone.
This is another advantage of using dispersion fuel.

395

Fission gas and any gas included during fabrication
remain in the fueled zone; in particular, fission gases
are contained in pores or fission gas bubbles. Gas
pressure in large pores and fission gas bubbles, which
may be insufficient to cause detrimental creep or yielding of fuel, could instead result in blistering of a fuel
plate when the plate is heated to a certain temperature.
Two types of mechanisms can be considered for blistering. One is pore (or void) connection, and the other

is pressure rupture of fission gas bubbles. Figure 3
shows the images of a typical blister-tested plate. In
the typical blister test, the sample plate is held at a
specified temperature for 30–60 min during each
annealing step.
The temperature at which blisters form is termed
the ‘blister temperature.’ The higher the blister temperature, the more resistant the fuel is to blistering.
As a design requirement, the minimum blister temperatures, also called ‘blister threshold temperatures,’
are typically tested with irradiated plates for the
anticipated power excursions.

3.14.2 U–Al
3.14.2.1

U–Al Fuel Properties

The U–Al phase diagram is shown in Figure 4. There
are three intermetallic compounds in the U–Al system: UAl2, UAl3, and UAl4. UAl2 forms directly from
the liquid, but UAl3 and UAl4 form by peritectoid
reactions with aluminum as follows:
UAl2 þ Al ! UAl3

½IŠ

UAl3 þ Al ! UAl4

½IIŠ

(a)


(b)
Figure 3 Images of a U3Si2–Al dispersion fuel plate after a
postirradiation blister test at 450  C. (a) Blistered plate
surface morphology. (b) Cross section of a blistered plate in
the plate thickness direction.


396

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

5

Weight percent Al
10
20

30 40

60 80

1800
1620

L

1600

L + UAl2
1350


1400

1200
1105

γ
UAl2 + g

800

UAl2
+
UAl3

UAl3 + L

758
UAl2 + b

776

UAI2 + a

UAl3

668

600


UAl4

1000

400

731
UAl4 + L
U0.9Al4

1135

UAl2

Temperature (ЊC)

UAl2 + L

641

660

UAl4

+

Al
U

20


40

60

80

AI

Atomic percent Al
Figure 4 U–Al phase diagram. Redrawn from Okamoto, H. In Binary Alloy Phase Diagrams; Massalski, T. B., Ed.; ASM
International: Materials Park, OH, 1990.

UAl2 has a face-centered cubic structure with a ¼ b ¼
c ¼ 0.776 nm (the MgCu2-type crystal structure). UAl3
has a simple cubic structure with a ¼ b ¼ c ¼ 0.426 nm
(the Cu3Au-type structure).11 UAl4 has a bodycentered orthorhombic structure with a ¼ 0.441,
b ¼ 0.627, c ¼ 1.371 nm.12 UAl4 is found as U-lattice
deficient, and therefore U0.9Al4 is frequently used to
designate this compound. This compound is, however,
expressed stoichiometrically as UAl4.9.
The densities of the compounds are 8.14 g cmÀ3
for UAl2, 6.80 g cmÀ3 for UAl3, and 6.06 g cmÀ3 for
UAl4. However, it is reasonable to assume the density
of UAl4 as 5.7 g cmÀ3 considering its U-deficiency
structure.
3.14.2.2 Thermal Conductivity of U–Al
Alloy and UAlx–Al Dispersions
The thermal conductivity of U–Al alloy depends on
the uranium composition in the alloy and on the

temperature. At 65  C, the thermal conductivities of
the as-cast U–Al alloys are a linearly decreasing
function of the uranium concentration13:
kT ðU À Al alloyÞ ¼ 225 À 2:9CU

½1Š

where kT is the U–Al alloy thermal conductivity in
W mÀ1 KÀ1 and CU is the uranium content in the
alloy in wt%.
Thermal conductivity of UAl2 is not available,
though it can be reasonably estimated as a value
lower than that of UAl3. In general, the higher the Al
content in a U–Al intermetallic, the higher the thermal conductivity because Al is more thermally conducting than is U. Among the U–Al compounds, UAl4
has the lowest thermal conductivity due to its defective structure. Because of peritectoid reactions given in
eqns [I] and [II], during heating a UAlx–Al dispersion
the UAl4 fraction increases while that of UAl3 remains
nearly unchanged and that of UAl2 decreases. Hence,
the overall thermal conductivity of a UAlx–Al dispersion decreases during heating, but the thermal conductivity of UAlx–Al is determined chiefly by the Al
matrix. For a UAlx (60 wt% UAl3 and 40 wt% UAl4)
volume fraction of 35% in the fueled zone, the thermal conductivity of the fueled zone is $62 W mÀ1 KÀ1,
considering a porosity of $6% in the fueled zone,
which is typical for this fuel. This is similar to that of
U3O8–Al but slightly lower than that of U3Si2–Al with
the same fuel volume fraction and porosity.


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

The recommended data for the thermal conductivities of UAlx–Al dispersions obtained from

Hofman and Snelgrove14 are used to fit an equation
as follows:
2
kT ðUAlx À AlÞ ¼ 225 À 1:61vUAlx À 0:121vUAl
x
3
þ 1:45Â10À3 vUAl
x

½2Š

where kT is in W mÀ1 KÀ1 and vUAlx is the UAlx vol%
(for the 60 wt% UAl3 and 40 wt% UAl4 mixture)
including typical fabrication porosity in the fueled
zone.
3.14.2.3

U–Al Fabrication

This section deals with the fabrication methods used
for the fuel phase made of U–Al intermetallics. The
U–Al alloy is in the form of a single slab, whereas
UAlx is in the form of a particle powder. The fueled
zone of the fuel plate containing U–Al alloy is therefore monolithic in that no distinct aluminum matrix
is involved, whereas that of UAlx is composed of fuel
particles in a matrix of aluminum. In the overall fuel
plate fabrication, both fuels have the same procedure
shown in Figure 2, except for the difference that lies
in whether the fueled zone is a monolithic alloy or a
dispersion compact.

3.14.2.3.1 U–Al alloy

U–Al alloy is produced directly by melting and casting proper amounts of U and Al metal together,
which determines the U density in the alloy. For a
plate-type fuel, the alloy melt is poured into graphite
molds to produce cast alloy slabs. Each monolithic
slab of U–Al alloy is then sandwiched between Al
alloy cladding slabs and hot-rolled into dimension to
form a plate-type fuel element.
This method can be used for fabrication of U–Al
alloys with up to $25 wt% uranium. Up to 40 wt%
uranium, U–Al alloy can be fabricated in the same
way with a small amount of UAl4 precipitates in the
alloy. Alloys with such content ratios are liable to
contain small fractions of metastable UAl3.
3.14.2.3.2 UAlx

UAlx designates a mixture of UAl2, UAl3, and UAl4:
The exact composition varies depending on the powder fabricator. Fabrication of UAlx is determined by
its composition.
The first step of the powder fabrication of U–Al
compounds with the high U weight fraction is
arc-melting of the mixture of U and Al metals.

397

Since U aluminides are brittle, the typical method
for powder fabrication is mechanical pulverization by
the use of jaw crushers and hammer mills. U–Al
intermetallics are pyrophoric. In particular, UAl2 is

highly pyrophoric; hence fabrication of this compound is more difficult, although it is the highest in
U density among the three U–Al intermetallics. The
desired particle size is controlled by the use of metallic sieves. The undersized or oversized particles are
recycled. The comminuted particles are irregular.
They have sharp corners, cracks, and high surface-tovolume ratio. The fuel particles are also brittle, which
causes higher porosity in the fueled zone after plate
fabrication. Since the comminution is performed in air,
oxygen is absorbed into the fuel particles. The oxygen
inclusion has an effect on fuel performance (to be
discussed later).
It is difficult to fabricate pure U aluminide compounds. Typically the product powder is a mixture of
UAl2, UAl3, and UAl4. If necessary, pure compounds
can be prepared by the use of uranium hydride,15 but
this method is not a commercially viable option
because of high cost, the hazardous process involved,
and difficulty of reprocessing the scrap materials.
Porosity is common in as-fabricated UAlx–Al and
U3Si2–Al dispersions, with the amount depending
on the fabrication method. For all fabrication methods,
however, it typically increases with the fuel phase volume fraction in the fueled zone. For example, in UAlx
composed of 60 wt% UAl3 and 40 wt% U0.9Al4, the
density is rUAlx ¼ 0:6ð6:83Þ þ 0:4ð5:7Þ ¼ 6:4 gcmÀ3
and the average stoichiometry x ¼ 0.6(3) þ 0.4(4.9) ¼
3.8. For this case, as-fabricated porosity as a function
of UAlx volume fraction can be estimated using information given by Beeston et al.,5 as shown in Figure 5
along with the data for U3Si2–Al.
3.14.2.4

U–Al Irradiation Performance


3.14.2.4.1 Fuel swelling by fission products

Fuel swelling by fission products is conveniently
divided into two distinct parts. One is fuel swelling
by solid and liquid fission products due to the difference between the volume of a uranium atom and solid
fission products, and the other is by gaseous fission
products that form bubbles. The former takes place in
the matrix of fuel particles and is due to the atomic
volume difference between the uranium atom and
solid fission product atoms. (Fuel matrix here stands
for the solid part in the fuel particles, excluding the
fission gas bubbles.) This also includes liquid fission
product atoms. Most fission gas atoms remain in the


398

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

gap material in fast reactor fuel pins. However, in the
U–Mo particle cases, this loss is largely negligible
because no sink for gas release is available, although
some occur by fission recoils to matrix aluminum.
Counting in this portion changes the value in eqn [3]
by $0.5% per 1027 fissions mÀ3. Consequently, the
fuel swelling by the solid (plus liquid) fission products
can be expressed by
 
DV
¼ 4:0fd

½4Š
V0 s

14

UAlx
U3Si2

12
Porosity (%)

10
8
6
4
2
0

0

10

20
30
UAlx volume (%)

40

50


Figure 5 Fabrication porosity in a UAlx–Al dispersion
plate as a function of the volume of UAlx. Information
on U3Si2–Al dispersion was obtained from Matos and
Snelgrove,16 and the fabrication data for UAlx–Al were
from Beeston et al.5

matrix of fuel particles particularly at low burnup.
This also contributes to fuel matrix swelling, with
the rate proportional only to burnup or fission density, independent of the fabrication method, fuel type,
and fuel temperatures. Hence, it is commonly applicable to other U intermetallic fuels also. The latter
is due to the fission gas bubble growth. As the gas
bubble swelling increases at higher burnup, however,
more gas atoms move into and collect in gas bubbles.
The resulting solid swelling rate is therefore proportionately lower. Modeling this effect in detail has not
been tried due to the lack of accuracy in relevant
measurements.
3.14.2.4.1.1 Fuel swelling by solid fission
products

Fuel swelling by solid (including liquid fission products also) fission products is a result of difference in
atomic volumes between uranium atoms destroyed by
fission and solid fission products and how the fission
products stay in the alloy. Hofman and Walters17 estimated burnup-dependent fuel swelling by solid fission products for U–Zr alloy. Their estimation is
1.2% per at.% burnup in U–10Zr, which can be transformed in terms of fission density of U–10Mo as
follows:
 
DV
¼ 3:5fd
½3Š
V0 s

where fd is the fission density in 1027 fissions mÀ3 .
However, this estimation is applicable only for the
situation in which alkali and alkaline-earth fission
products release and dissolve in the liquid sodium

where fd is the fission density in 1027 fissions mÀ3 in
fuel particles.18,19
3.14.2.4.1.2
products

Fuel swelling by gaseous fission

Fuel swelling by fission gas bubbles is more difficult
to quantify. The fission gas bubbles that form in UAlx
are so small that they were hardly observable by
authors in the literature. Francis20 observed fission gas
bubbles in UAl3 with burnup of 60%, or 6.5Â1027
fissions mÀ3 (in fuel particles). He also reported that
no gas bubbles were observed in reaction layers,
which were UAl4. He suggested that this was due to
the lower uranium density in UAl4 and the lower fission
density of $4.6Â1027 fissions mÀ3 than in UAl3 and
also suggested that the defect structure in UAl4 might
accommodate extra fission gas, delaying bubble formation. Miller and Beeston21 found no visible fission gas
bubbles in UAl2 up to 4.6 Â 1027 fissions mÀ3. Hofman,22
too, found no fission gas bubbles in UAlx up to 7 Â 1027
fissions mÀ3. Miller and Beeston21 and Hofman22 commonly observed fission gas bubbles only in uranium
oxide clusters, but none in UAl2 or UAlx. The source of
oxygen in the clusters is the air included in the fueled
zone from both the fuel powder and the later plate

fabrication performed in an air environment. Figure 6
shows a scanning electron microscopy (SEM) image
of the fracture surface of UAlx irradiated at $130  C
to 56% burnup of 93% enriched fuel (or 7 Â 1027 fissions mÀ3). It is unclear whether the oxide clusters acted
as reservoirs absorbing fission gas, or whether UAl4
in UAlx helps retard bubble formation. More recent
high-resolution transmission electron microscopy
(HR-TEM) investigation of irradiated uranium intermetallic fuels has revealed the presence of small fission
gas bubbles in the IL.23,24 Thus, it may be possible that
very small micro bubbles are present in irradiated UAlx,
but were undetected due to limits on the resolution of
microscopy techniques at the time.
Recent TEM work by Gan et al.23 shows that small
bubbles, visible only in TEM, are in fact formed in


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

399

U-oxide phase including
fission gas bubbles
Al
U–7 wt%Mo

Interaction
zone

1 μm


5 μm

UAlx

Figure 6 Scanning electron microscopy image of a 93%
235
U-enriched UAlx–Al dispersion irradiated to 60%
235
U burnup. Fission gas bubbles are visible in U oxide
phase and no gas bubbles are visible in UAlx.
Reproduced from Hofman, G. L. Nucl. Technol. 1987,
77, 110–115.

Holes
Interaction product

Gas bubbles

Aluminum
0.2 mm
Figure 7 Transmission electron microscopy (TEM) image
showing bubbles formed in interaction layer between
U–7Mo and Al. Fission gas bubbles are visible as gray
spheres in the interaction product with a size range of
0.01–0.2 mm. Some conspicuous bubbles are marked with
arrows. The bright upper left corner and a bright hole in the
interaction product were formed during preparation of the
TEM sample. Reproduced from Gan, J.; et al. J. Nucl. Mater.
2009, 396, 234.


Figure 8 Scanning electron microscopy image showing
fission gas bubbles formed in interaction zones between
U–7Mo and Al irradiated to 4.1 Â 1027 fissions mÀ3.
Reproduced from Jue, J. F. Private communication;
Idaho National Laboratory; 2009.

interaction products of U–7 wt% Mo dispersion in Al
(see Figure 7). Since the interaction products are
probably similar to those in UAlx–Al (even though
Mo is involved), this TEM work suggests that fission
gas bubbles also form in UAlx but would be too small
to be seen in an SEM, particularly in older, lower
resolution devices. For example, the image in Figure 8
is a recent SEM work by Jue,24 which indeed shows
fission gas bubbles with a maximum size of $0.2 mm in
the IL between U–7Mo and Al after irradiation to
4.1 Â 1027 fissions mÀ3 in fuel particles. To summarize,
fission gas bubbles form in UAlx but have not been
consistently observed.
A direct quantification of the gas bubble swelling
rate is currently impossible. Instead, the gas bubble
swelling rate is estimated by subtraction of solid
fission product swelling from the total swelling.
The total swelling rate, that is, contribution by solid
fission products and fission gas bubbles, obtained for
ATR fuel tests is
 
DV
ð%Þ ¼ 6:4fd
½5Š

V0 total
where the swelling is for the fuel particle, and fd is
fission density in 1027 fissions mÀ3 in fuel particles.5
The difference found from eqns [4] and [5], 2.4%
per 1 Â 1027 fissions mÀ3, approximates the contribution by fission gas bubbles. The contribution of fission
gas bubbles to total fuel swelling is considerable, even
when nearly invisible. In addition, fission gas bubbles
in the oxide clusters also contribute.
In general, the overall plate thickness increase
in UAlx–Al dispersions because fueled zone swelling


400

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

is lower than in any other fuel dispersions. This
advantage can be attributed to several factors, including higher as-fabricated porosity, the defect structure
of UAl4, and the high resistance of UAlx to large
bubble formation.
3.14.2.4.2 Interaction between U–Al and Al

UAlx and Al react during irradiation even at low temperatures due to irradiation-enhanced interdiffusion.
The U–Al phase diagram shown in Figure 4 shows
that UAl2 and UAl3 react with matrix Al according to
the peritectoid reactions given in eqns [I] and [II].
Because there is no higher compound in terms of Al
content than UAl4, only UAl4 stays stable with Al.
Reaction products on U aluminides are discernable
in optical micrographs due to color difference, as

shown in Figure 9. The volume fraction analyses of
UAlx–Al before and after irradiation showed that the
volume fraction of UAl4 increases while that of UAl3 is
nearly unchanged and that of UAl2 decreases.
The U–Al diffusion-couple tests in the temperature range of 100–600  C showed that UAl3 is the
dominant phase created. UAl4 is less prevalent, and
UAl2 is present the least.12 At lower temperatures,
there was a greater tendency to form UAl2.
In U–Al diffusion-couple tests at 400–600  C, however, Castleman25 did not observe UAl4 and rarely
observed UAl2. This is because UAl2 is unstable around

Al as thermodynamic data indicate (see Table 2).
The diffusion layers for both works by Kiessling and
Castleman showed a considerable amount of pores.
Out-of-pile measurements by Nazare et al.28
yielded activation energies of 220 kJ molÀ1 for the
reaction between UAl3 and Al, and 180 kJ molÀ1
for the reaction between UAl2 and Al. The slightly
higher value for UAl3–Al dispersion will result in a
slower reaction rate than UAl2–Al dispersion.
Measured reaction data of UAlx–Al from in-pile
tests are scarce because reactions between the fuel
and matrix are not an irradiation performance issue
with this fuel. Relevant changes in volume fractions
due to fuel–matrix reactions are also generally small.
All three uranium aluminides undergo amorphization, depending on the fission rate and temperature.
Therefore, the reaction products of these fuels with Al
are also subject to amorphization. The lower the irradiation temperature and the higher the fission rate,
the more the tendency of amorphization. Among the
three, UAl4 is the easiest and UAl2 is the hardest for

amorphization. Crystalline material is more stable
during irradiation than amorphous material.
3.14.2.4.3 U–Al blister threshold temperature

No blister threshold temperature data were found
for U–Al alloy fuel, which is probably because this
fuel is more prone to failures by breakaway swelling.
UAlx–Al dispersion fuels have high blister threshold temperatures due to their high resistance against
formation of fission gas bubbles. Beeston et al.5 proposed that the blister threshold temperature could be
measured as a function of fueled zone fission density,
and Nazare29 fitted Beeston’s data, adding more data
to give the following correlation:
For a fuel plate without B4C,

C

TB ¼ 921 À 59:2fd

B

½6Š

For a fuel plate containing B4C,
TB ¼ 907 À 54fd À 224B

A
Table 2

D


10 mm

Figure 9 Scanning electron microscopy image of a 93%
235
U-enriched UAlx–Al dispersion after a 60% 235U burnup.
A is UAl2, B is UAl3, C is the interaction product (UAl4), and
D is U oxide. Adapted from Hofman, G. L. Nucl. Technol.
1987, 77, 110–115; Ryu, H. J.; Kim, Y. S.; Hofman, G. L.
J. Nucl. Mater. 2009, 385, 623.

Thermodynamic properties of U–Al compounds

Compound DG0 (J molÀ1)a
UAl2
UAl3
UAl4
a

À95 420À9.738T lnT + 76.91T
À110 070À5.464T lnT + 44.38T
À134 150À31.48T lnT + 247.3T

Chiotti and Kateley.26
Kubaschewski and Alcock.27

b

½7Š

DH0298

(kJ molÀ1)b
98.8
114.0
130.0


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

where TB is the blister temperature in K, fd is
the fueled zone fission density in 1027 fissions mÀ3,
and B is the B4C content in wt%.
B4C is added for reactivity control by absorbing
neutrons in some plates. Neutron absorption in
boron atoms results in generation of helium gas,
which additionally lowers the blister temperature
as in eqn [7].
3.14.2.5

Summary for U–Al

U–Al alloys and uranium aluminides clad with aluminum alloy have advantages for water-cooled thermal research and test reactors operating at low
temperatures primarily due to the favorable characteristics of aluminum. U–Al alloys are monolithic
fuels, while uranium aluminides are in dispersion
form in an aluminum matrix.
The three uranium aluminides undergo amorphization depending on the fission rate and temperature.
The lower the irradiation temperature and the higher
the fission rate, the faster the amorphization. Among
the three, UAl4 amorphizes most readily and UAl2
least readily. Crystalline material is more stable during irradiation than amorphous material.
UAlx–Al dispersions, where UAlx denotes a mixture of UAl2, UAl3, and UAl4, have lower fueled zone

swelling than any other type fuel dispersions due to
low fission gas bubble swelling. Large fission gas bubbles greater than the resolution limit of an SEM do not
form in UAlx because of the defective structure in UAl4
and high resistance to large bubble formation.
Measured interaction data of UAlx–Al from in-pile
tests are scarce because reactions between the fuel and
matrix have not been an irradiation performance issue
for this fuel. Relevant changes in volume fractions
due to fuel–matrix reactions are also generally small
because the interaction products between the uranium
aluminides and the aluminum matrix are also uranium
aluminides. The volume fractions of UAl4 increases,
while that of UAl2 decreases, and the change in the
volume fraction of UAl3 depends on the kinetics of
other compounds. Because UAl3 has the highest thermal conductivity of the three, and UAl2 and UAl4 have
similar thermal conductivities, the evolution of volume fractions in UAlx partly counteracts the decrease
in overall thermal conductivity.
UAlx–Al dispersion fuels have high blister temperatures due to their high resistance against formation of
fission gas bubbles. The blister threshold temperature
is a function of fission density.

401

3.14.3 U–Si
3.14.3.1

U–Si Fuel Properties

In the U–Si system, U3Si, U3Si2, and USi are the
compounds of interest for candidate fuels chiefly

because of their high uranium density (see Table 1).
The physical densities of U3Si, U3Si2, and USi are
15.3, 12.2, and 10.96 g cmÀ3, respectively. Clearly,
U3Si is the most favorable fuel among U–Si intermetallic fuels for the same reason. The U–Si phase diagram is shown in Figure 10. U3Si2 and USi form
directly from liquid, but U3Si forms only by the following peritectoid reaction at 925  C:
U3 Si2 þ U ! U3 Si

½IIIŠ

At room temperature, U3Si has a body-centered
tetragonal structure. It undergoes transformation to a
face-centered cubic crystal structure at 765  C, and
this structure is maintained up to 925  C. Hence, at
usual reactor operation temperatures, U3Si is in a
tetragonal structure. The U3Si structure is classified
as a distorted variant of Cu3Au-type. No close Si–Si
bonding occurs; only U–U and U–Si bonds are present. This is the reason for the unusual ductility of
the intermetallic compound.11 The lattice parameters
of the tetragonal structure are a ¼ 6.029 A˚ and
c ¼ 8.696 A˚.11,30
U3Si2 has a congruent melting point at 1665  C
and has no transformations in the solid state. It has a
primitive tetragonal structure with a0 ¼ 7.3299 Æ 4 A˚,
c0 ¼ 3.9004 Æ 5 A˚, c/a ¼ 0.532, and ten atoms per unit
cell. The crystal structure is a deformed Cu3Au-type
with pairs of Si atoms replacing a single Si atom.
These close Si–Si bonds impart brittleness to the
compound.11
The crystal structure of USi is controversial in the
literature. The most recent data suggest that the

structure is tetragonal with the lattice parameters of
a ¼ 10.5873 A˚ and c ¼ 24.3105 A˚.31

3.14.3.2 Thermal Conductivity of
(U–Si Intermetallic)–Al Dispersions
Thermal conductivities of U3Si–Al and U3Si2–Al
dispersions are similar to each other. Considering
the uncertainties related to the fueled zone porosity
and nonstoichiometric nature of these fuels, that is,
miscibility of U3Si and U3Si2, it is reasonable that the
same thermal conductivity is used. The difficulty
of fabricating a stoichiometric U–Si compound is


402

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

Weight percent silicon
5

10

1800

20
1770

1665


1600

30 40

60 80

1710

L

L

1510

1580
1540

1414

1315

1200
1135

γ + U3Si

USi3

985


1000

930

USi

β + U3Si

400

U

20

450

USi2

U3Si

α + U3Si

U3Si2

668

600

USi1.88


776

800

U3Si5

Temperature (ЊC)

1400

40
60
Atomic percent silicon

80

Si

Figure 10 U–Si phase diagram. Redrawn from Okamoto, H. In Binary Alloy Phase Diagrams; Massalski, T. B., Ed.;
ASM International: Materials Park, OH, 1990.

discussed in Section 3.14.3.3. The thermal conductivities of U3Si–Al and U3Si2–Al dispersions are
fitted on the basis of the data given in Matos and
Snelgrove16:
kT ðU3 Si2 À AlÞ ¼ 225 À 3:22vU3 Si2
À 4:48 Â 10À2 vU2 3 Si2
þ 6:76 Â 10À4 vU3 3 Si2

½8Š


where vU3 Si2 is the U3Si2 or U3Si volume fraction in
% including typical fueled zone porosity.
3.14.3.3

U–Si Fabrication

U3Si2 is used for plate-type fuels, whereas U3Si is
currently applied only for rod-type fuels. In practice,
however, it is almost impossible to fabricate the exact
stoichiometric form of one of these compounds. This
is why fuel manufacturers instead start with slightly
Si-rich alloys, which lead to final products more
abundant in higher Si content compounds. For
example, the stoichiometric U3Si2 requires an Si
composition of 7.3 wt%. To suppress the formation

of U solid solution and U3Si, typically 7.5 wt% Si
is added to the U–Si alloy. A prolonged heat treatment at 925  C has shown to remove the presence of
U solid solution.32 The secondary phases typically
reside inhomogeneously in a fuel particle, which
causes inhomogeneous size distributions of fission
gas bubbles inside the fuel particles. In some cases,
U3Si occupies whole fuel particles in a nominal U3Si2
fuel. As a result, postirradiation images show anomaly
in fission gas bubble size between fuel particles.
Alloy ingots of U–Si are made by mixing and
melting of uranium and silicon with a desired Si/U
ratio. The ingots are sometimes annealed in an inert
atmosphere to complete compound formation. These
ingots are then broken into smaller particles by a

powder fabrication process.
Two types of powder fabrication methods for
U-silicide fuels are most commonly used. The relatively brittle compounds of U3Si2 and USi are
obtained by comminution of the alloy in a glove box
in a nitrogen atmosphere with a hardened steel mortar and pestle. U3Si powder fabrication, because U3Si
has greater toughness than U3Si2 and USi, is typically


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

carried out by grinding in a ‘shatter box.’ This
method calls for laborious processes because of the
toughness of U3Si. Typical particle sizes range
between 40 and 150 mm.33
An atomization technology widely used in powder
metallurgy is applied to fabricate spherical powders
of U3Si2 and U3Si.9 This method typically uses a
rotating disk in a vacuum chamber. When a liquid
fuel melt is poured on the disk, the centrifugal force
of the disk produces liquid fuel droplets that are
cooled in the process of flying away. The size of fuel
particles is determined by the disk rotation speed.
Figure 11 shows a comparison between atomization and comminution powders. Atomized powder has
several advantages over comminuted powder. First, the
surface-to-volume ratio of the atomized particles is
smaller, so reaction product volume between fuel particles and matrix aluminum is smaller. Second, atomized particles have higher homogeneity in silicon
content and fewer impurities because they are rapidly
solidified from the liquid and are uncontaminated by
mechanical pulverization. Finally, atomized particles
have lower residual stresses and defects, which is an

advantage from the view point of fuel swelling.
The typical plate fabrication method shown in
Figure 2 is also used for the fabrication of U–Si
powder dispersion plates.
3.14.3.4

U–Si Irradiation Performance

3.14.3.4.1 Fuel swelling by fission products

Fuel swelling by solid fission products discussed in
Section 3.14.2.4.1 is also applicable for U–Si in that
this kind of fuel swelling is dependent only on
burnup, regardless of the fuel kind. The swelling
rate given by eqn [4] can be used. However, fuel
swelling kinetics by fission gases, that is, fission gas
bubble growth, is different and discussed more in
depth below.
The fuel swelling by solid fission products is
given by
 
DV
ð%Þ ¼ 4:0fd
½4Š
V0 s
where fd is fission density in 1027 fissions mÀ3 in fuel
particles.18,19
Fuel swelling kinetics of U–Si fuel particles is well
documented in the literature.14 In Figure 12, the fuel
swelling kinetics of U3Si and U3Si2 are plotted

together with the fuel swelling by solid fission products calculated with eqn [4]. For each fuel type, the
fuel swelling by fission gas bubble growth can be

403

(a)

(b)
Figure 11 Examples of comminuted and atomized
powders of U3Si2. (a) Scanning electron microscopy (SEM)
image of U3Si2 powder fabricated by the comminution
method. (b) SEM image of U3Si2 powder fabricated by the
atomization method. Reproduced from Kim, C. K.; Kim,
K. H.; Jang, S. J.; Jo, H. D.; Kuk, I. H. In Proceedings of the
1992 International Meeting Reduced Enrichment for
Research and Test Reactors (RERTR), ANL/RERTR/TM-19,
CONF-9209266, Sept 27–Oct 1, 1992; Argonne National
Laboratory: Argonne, IL, 1992.

estimated by subtracting the fuel swelling by solid
fission products from the total fuel swelling. The
data included in the graph were obtained from tests
at temperatures below 110  C, where fission gas bubble growth, and therefore fuel swelling, have been
considered athermal and dependent only upon the
burnup.
U3Si and U3Si2 are known to become amorphous
under irradiation34,35 due to fission damages. The
primary damage to the crystal structure is caused by
highly energetic fission fragments. In the amorphous



404

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

140
120

U3Si

(ΔV/V0)f (%)

100
80
60
40

U3Si2

20
0

100% LEU BU

(ΔV/V0)s

0

2
4

6
Fission density (1027 fissions m–3)

8

Figure 12 Fuel particle swelling of U–Si intermetallic
fuels.

fuel, fuel swelling depends on the viscosity of fuel. The
viscosity of an amorphized material is lower than when
it is crystalline: in other words, the fluidity of the
material increases when it becomes amorphous. Fission
gas mobility is also high in amorphous material and the
fuel material is more readily deformed by the growing
gas bubbles. Hence, overall fission gas bubble growth in
an amorphous material is faster. Figure 13 shows fission gas bubble morphology of amorphous fuels. Bubbles are large and interconnected, which is observable
in breakaway swelling.
Amorphization is clearly a low-temperature
phenomenon, as amorphized materials devitrify
(recrystallize) at the so-called glass transition temperature. Above this temperature, amorphization is
not possible and the fuel in question exhibits the
familiar crystalline irradiation behavior. U–Si fuels
are normally amorphous during irradiation because
the glass transition temperature for U-silicides is
much higher than typical fuel operation temperatures ($120  C).
These fuels show that they preserve their preirradiation hardness and brittleness. The observed fluidlike behavior thus only exists during irradiation.
Figure 14 shows fuel microstructures and the
fission gas bubble morphology of irradiated U3Si
and U3Si2. Although both are amorphous during irradiation, there are inherent differences: notably, fission
gas bubble growth in U3Si is high and unstable,

whereas that of U3Si2 is generally low and stable.
For this difference in fission gas bubble growth,
Hofman and Kim8 offered an explanation by evoking
the correlation between free volume and viscosity,
which was first developed by Doolittle36:

50 μm
(a)

50 μm
(b)
Figure 13 Optical microscopy images showing unstable
fission gas bubble growth shown in amorphous fuels at
fission density of 4.5Â1027 fissions mÀ3. (a) U3Si. (b) U6Fe.


 ¼ 0 exp

C
DVR


½9Š

where C is a constant and DVR is the part of the
quenched-in free volume associated with structural
relaxation that is recovered during annealing of the
glass prior to recrystallization. Hofman and Kim8
noted that U3Si has larger free volume than U3Si2.
It has been shown in the literature that the free

volume of a glassy metal is strongly affected by
composition, since the short-range bonding character
of an alloy is maintained in the glass state.37
Apparently, the additional Si bonds in U3Si2 have
a large effect on the amount of free volume in the
glassy state, and therefore also on the fluidity of


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

the fuel – the fission gas diffusivity – and the resulting
swelling behavior. Although amorphization is a prerequisite for low-temperature high-swelling behavior,
it needs to be accompanied by an increase in free
volume.

405

The bubble morphology from higher temperature
tests is available in the literature.38,39 Figure 15
shows the fission gas bubbles at different temperatures. Compared to Figure 14(a), Figure 15(c) suggests that bubble growth in U3Si2 can be enhanced
to the level of U3Si if the temperature is increased
by $60  C. U3Si2 appears to experience a bubble
growth phenomenon at high temperatures similar to
that of U3Si at low temperatures; the low bubble
growth advantage of U3Si2 provided by the high
Si/U ratio is negated if the temperature is increased.
A mechanistic rate-theory model demonstrates that
the bubble coarsening process in irradiated amorphous
materials such as U3Si and U3Si2 depends on their
(a)


(a)

50 μm
(b)

50 μm
(c)

50 μm
(b)
Figure 14 Fission gas bubble morphology of U-silicide
fuels (19.5% 235U enriched) irradiated at temperatures
($100  C) in the Oak Ridge research reactor. (a) U3Si
irradiated to 15 at.% BU, (b) U3Si2 irradiated to 19 at.% BU.
Notice the difference in magnification. Notice that the scale
in (b) is a factor of 10 greater than that in (a).

Figure 15 Optical microscopy images of U3Si2
(75% 235U enriched). (a) T ¼ 105  C and FD ¼ 3.2 Â
1027 fissions mÀ3 (13 at.% U total BU), (b) T ¼ 136  C and
FD ¼ 5.4 Â 1027 fissions mÀ3 (22 at.% U total BU), (c)
T ¼ 160  C and FD ¼ 6.1 Â 1027 fissions mÀ3 (25 at.% U total
BU). Adapted from Kim, Y. S.; Hofman, G. L.; Rest, J.;
Robinson, A. B. J. Nucl. Mater. 2009, 389, 443; Kim, Y. S.;
Hofman, G. L.; Yacout, A. M. J. Nucl. Mater. 2009, 392, 164.


406


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

viscosity.40,41 Estimated irradiation-induced viscosity
values were obtained by comparing the calculated
bubble size distribution with the observed average bubble size and density as a function of fission rate and
burnup. This model predicts the viscosity values
determined from the bubble size distribution analysis.
The calculated temperature dependence of the
viscosity depends on the assumption that the rate of
change of the calculated formation enthalpy with
respect to temperature is symmetric with respect to
the uranium concentration. In addition, the temperature independence of certain material properties
(such as thermal expansion coefficient) has also been
assumed. Thus, only the trend of the calculations
should be meaningful. It is important to note that, as
U3Si2 is irradiated, the Si/U ratio shifts to the right. In
any event, the calculations show that a $30 K increase
in temperature results in a viscosity for U3Si2 that is
similar to that of U3Si irradiated at the lower temperature (see Figure 16). In addition, the calculated viscosity of U3Si2 is much more sensitive to temperature
than that of U3Si.
3.14.3.4.2 Interaction between U–Si and Al

U3Si, U3Si2, and USi react with Al to form a single
intermetallic compound, U(AlSi)3. The solubilities of
Al in U3Si, U3Si2, and USi are very small ((1 at.%).11
U(AlSi)3 has a composition intermediate between UAl3
and USi3, both of which are mutually soluble. The
1012
1011
1010

Viscosity (Poise)

109
108
107
106
105
104
103
102
101
100

0.3

0.4
U3Si

0.5
Si/U ratio

0.6

0.7
U3Si2

Estimated viscosity based on swelling data
Calculated viscosity at 370 K
Calculated viscosity at 400 K
Calculated viscosity at 430 K


Figure 16 Viscosity of U–Si intermetallic fuels during
irradiation versus Si/U ratio for three temperatures. Adapted
from Kim, Y. S.; Hofman, G. L.; Rest, J.; Robinson, A. B.
J. Nucl. Mater. 2009, 389, 443; Kim, Y. S.; Hofman, G. L.;
Yacout, A. M. J. Nucl. Mater. 2009, 392, 164.

Al/Si ratio in the IL is the highest for U3Si–Al, lower
for U3Si2–Al, and the lowest for USi–Al. The compositions of the compounds lie on the tielines between the
uranium silicides and Al, as shown in the isothermal
section of the ternary phase diagram (see Figure 17).
For all cases, the reaction products U(AlSi)3 have a
density of $7.1 g cmÀ3, and approximately equal
volumes of uranium silicide and Al are used. Only a
small volume change, $4%, is involved in the reaction.
The compositions of the IL from in-pile tests
of U3Si2–Al are also shown in Figure 17. The compositions deviate from the exact stoichiometry, that
is, (Al þ Si)/U ¼ 3. This indicates that the reaction
products become amorphous during irradiation, as
has been observed in in-pile tests reported previously.45,46 Since the IL is amorphous, U, Al, and Si
atoms exist in a mixture without crystalline restriction of stoichiometry.
When matrix Al exists, the IL of U3Si2–Al is rich
in Al with the Al/Si ratio $3.5 and the (Al þ Si)/U
ratio is $5.3.47 Leenaers et al.45,44 reported a larger
($5.0) Al/Si ratio and a smaller ($4.6) (Al þ Si)/U
ratio than those by Kim. Kim’s (Al þ Si)/U ratio is
larger than Leenaers’ due to higher burnup. Using
the IL physical density of 7.1 g cmÀ3 and assumption
of a linear time-dependent burnup profile in the IL,
the (Al þ Si)/U ratio increases to $4.5 at a fission

density of 1.43 Â 1028 fissions mÀ3. This suggests that
the (Al þ Si)/U ratios from both the present test
and Leenaers’ test are higher than the theoretical
assessment.
The (Al þ Si)/U ratio was seen to decrease to $3.3
with Al/Si $0.29 with depletion of the matrix Al
around the U3Si2 particles during irradiation.47 A twophase mixture of USi2 and U(AlSi)2 was also observed.
This is consistent with the findings of Nazare,48 who
observed the formation of U(Al, Si)2 in annealing tests
of U3Si/Al and U3Si2/Al dispersion fuels at 600  C,
when the Al matrix was completely consumed. This
probably occurs when Al atoms continue to diffuse into
the fuel from the IL while no further Al flux exists from
the Al matrix. Subsequently, the Al/Si ratio continuously decreases, and eventually Si becomes more prevalent than Al. The compositions move toward the Si
corner and U–Si side of the ternary diagram.
The formation of gas bubbles in the ILs is important because of its potential effects on the IL growth
rate. The gas bubbles in the IL, on one hand, reduce
the effective diffusion area and thereby reduce the IL
growth rate. On the other hand, they increase the
IL volume itself, which results in a higher measured
IL growth rate.


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

407

Si
From out-of-pile tests
[p,D,R]

From in-pile tests [p]
Reaction product composition
from in-pile tests [L]
From in-pile tests after Al depletion
[p]
From in-pile tests after Al depletion
(two-phase mixture) [p]

USi3 80

Si

USi2

At

USi
40

p

ga

U3Si2

at.%
lity
ibi
sc
Mi


om

%

60

U3Si
20
U(AlSi) 3

U

20

60

40

UAI2

Al

80
UAI3

UAI4

Atom % Al
Figure 17 U–Si–Al ternary phase diagram showing reaction products between U, Si, and Al. p – Hofman et al.,42

D – Dwight,11 R – Rhee et al.,43 L – Leenaers et al.44

Micrographs of irradiated LEU U3Si–Al and
U3Si2–Al are shown in Figure 18. The ILs of both
fuels are generally uniform in thickness and free of
visible fission gas bubbles. Therefore, the gas bubbles
in the unreacted fuel serve as a boundary between
the unreacted fuel and the ILs. The gas bubbles are
remarkably different in size between U3Si and U3Si2.
Because both samples in this figure have similar
burnup and temperature, this comparison shows that
gas bubble swelling in U3Si is larger than in U3Si2.
Gas bubbles are found in ILs of high-burnup HEU
U3Si–Al and U3Si2–Al fuels, as shown in Figure 19.
The fission density of the fuel particles of this sample is
2.4 Â 1028 fissions mÀ3, which is about 4.5 times larger
than the LEU samples presented in Figure 19. This
comparison shows that the formation of visible fission
gas bubbles in the ILs depend on the fission density.
U3Si2–Al has similar gas bubble behavior in the ILs, but
these appear at higher burnup than those in U3Si–Al.
IL growth is a typical interdiffusion process
including U, Si, and Al. Out-of-pile tests are typically
performed at high temperatures $600  C and have
shown that interdiffusion is the rate-controlling process in IL growth of silicide–Al dispersion. Therefore,
by applying Fick’s law, we have J ¼ ÀD@C=@x
where J is the Al flux, D is the Al diffusion coefficient, and C is the concentration of Al in the IL.

The Al flux is expressed by J ¼ ðr=MÞ@x=@t,
where r is the physical density of the IL and M is

the molecular mass of the IL compound. Equating
these two equations and integrating gives
Y 2 ¼ kt

½10Š

where Y is the IL thickness, k ¼ ð2M=rÞDCD. Here k
is the IL growth coefficient and DC is the absolute
value of the Al concentration difference across
the IL. Because the reaction product is invariant,
ð2M=rÞDC % constant. As D is expressed in an
Arrhenius-type equation, so is k:


Q
½11Š
k ¼ A exp À
RT
where A is a preexponential constant, Q is the activation
energy for interdiffusion in U(AlSi)3, t is the time, T is
the absolute temperature, and R is the gas constant.
IL thickness data for out-of-pile diffusion tests of
U3Si–Al are reported by Rhee et al.43 Their measurements of IL thickness data are best represented by
Q ¼ 220 kJ molÀ1 and A ¼ 1:5 Â 1013 sÀ1 , giving Y in
micrometers. For U3Si2–Al, IL thickness data from
out-of-pile tests were obtained at the Argonne
National Laboratory (ANL).42 A data fit of these
data gives Q ¼ 354 kJ molÀ1 and A ¼ 2:6 Â 1021 sÀ1
with Y in micrometers.



408

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

20 μm

40 mm

(a)

(a)

U3Si2
IL

40 mm

(b)

Figure 18 Optical microscopy images of U3Si–Al and
U3Si2–Al dispersion fuels (low-enrichment uranium)
irradiated at 100  C in Oak Ridge research reactor (ORR).
(a) U3Si–Al irradiated for 319 days to a fission density of
5.3 Â 1027 fissions mÀ3 in ORR (A105). The dark phase
shows the U3Si fuel particles, the bright phase is the Al
matrix, and the gray phase in between is the interaction
phase. The fission gas bubbles are shown as black
circles in fuel particles. (b) U3Si2–Al irradiated for 300 days
to a fission density of 4.7 Â 1027 fissions mÀ3 in ORR (A34).

The dark phase shows the U3Si2 fuel particles, the bright
phase is the Al matrix, and the gray phase in between is
the interaction phase.

Fuel temperatures of typical in-pile tests are
much lower (<200  C) than the out-of-pile tests
($600  C). Simple extrapolations to the lowtemperature regime of the IL growth correlations
for out-of-pile tests yield orders of magnitude smaller IL thickness values. This implies that thermally
activated diffusion must be augmented by fission
enhanced diffusion during irradiation. It is known
that at temperatures below approximately one-half
of the absolute melting point of a solid (in the case

5 μm

(b)
Figure 19 Scanning electron microscopy (SEM) images
of high-burnup high-enrichment uranium fuel showing
gas bubble formation in fuel and interaction layer (IL).
(a) SEM backscattered electron image of U3Si–Al
dispersion fuel irradiated for 23 days to a fission density
of 2.4 Â 1028 fissions mÀ3 (fuel particle) at 110  C in the
high-flux isotope reactor (H3-2) showing large fission
gas bubbles in the IL. (b) SEM secondary electron image
of U3Si2–Al dispersion fuel irradiated for 130 days to a
fission density of 9.4 Â 1027 fissions mÀ3 (fuel particle) at
100  C in Oak Ridge research reactor showing small
bubbles in the IL. Notice different scales in the micrographs.

of U(AlSi)3 this is 835 K), diffusion is enhanced by

irradiation, primarily due to fission fragments.
A fission fragment with a high kinetic energy at
birth slows down while imparting energy to atoms in
the ‘displacement spike,’ producing vacancies and
interstitial atoms. In the displacement spikes, the
initial slowing down of fission fragments involves
a process of rapid heating and quenching. In a more
thermally conductive medium such as a metallic fuel,
however, dissipation of the thermal displacement
spike to the surrounding medium and the restoration


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

of local thermal homogenization will take place in a
shorter time.49 Consequently, the contribution to diffusion enhancement by thermal activation of a fission
event is less important than that from the generation
and annihilation of defects that directly influence
interdiffusion.
The defect generation rate is proportional to the
fission rate. The reaction rate coefficient given by eqn
[11] for out-of-pile data can be modified by including
the fission rate:


Qirr _ p
½12Š
f
k ¼ Airr exp À
RT

where f_ is the fission rate, p is the exponent of the
fission rate dependence, Airr is a constant, and Qirr the
effective activation energy for irradiation tests, which
is related to the mobility of vacancies.
To determine p, the measured IL thicknesses were
normalized with the fission rate (p ¼ 1) and its square
root (p ¼ 0.5), respectively. The result is compared in
Figure 20. When normalized with the fission rate with
a power p ¼ 1 or 0.5, Y2 must be linear with respect to
time. However, the data given in Figure 20(a) with
p ¼ 1.0 deviate from a straight line. For U3Si2–Al, the
data increase somewhat in a parabolic manner,
whereas the data given in Figure 20(b) with p ¼ 0.5
follow straight lines for all fuels. This indicates that
p ¼ 0.5 is a better fit. This result suggests, in analogy
with the crystalline irradiation-enhanced diffusion
theory of Sizmann,50 that recombination of fissionfragment-induced defects may be the underlying
mechanism. Hofman et al.42 developed a correlation
with p ¼ 1 with acceptable accuracies because of its
composite modeling. They applied a thermal activation term and an irradiation term separately, which
fortuitously reduced the significance of the linear
dependence of the fission rate.
The data-fitting for U3Si–Al results in
Airr ¼ 4:3 Â 10À6 and Qirr ¼ 56 kJ molÀ1, where the
fission rate is in fission cmÀ3 sÀ1 and Y in mm. For
U3Si2–Al, following the same procedure,
Airr ¼ 2:2 Â 10À8 and Qirr ¼ 41 kJ molÀ1 are found
for U3Si2–Al.
The activation energy for U3Si–Al obtained from
irradiation test data is larger than that for U3Si2–Al.

This is the reverse of the out-of-pile data, so it may
be concluded that the mechanism for irradiationenhanced diffusion is different from the thermally
activated diffusion. For both U3Si–Al and U3Si2–Al,
the activation energies for the in-pile data are about
an order of magnitude smaller than those for the outof-pile data because out-of-pile diffusion is more

409

temperature dependent than in-pile diffusion.
In-pile diffusion relies on vacancy migration that
occurs at lower energies than the atom activation in
out-of-pile diffusion.
Since the higher U/Si ratio is the cause for a
faster IL growth rate for U3Si–Al than U3Si2–Al
(see Figure 20), the effect of U/Si ratio on IL growth
with burnup, particularly at high burnup, is worth
further discussion. The U/Si ratio in fuel is the
same in the IL as in the fuel particle. Therefore,
the higher the U/Si atom ratio in the IL, the higher
the growth rate. In this context, the growth rate ought
to decrease as the U/Si ratio of the IL decreases with
burnup. The IL U/Si atom ratios are calculated for
U3Si–Al, U3Si2–Al, and USi–Al and plotted versus
the fission density in Figure 21.
The U/Si ratio of IL in U3Si–Al decreases to
that of U3Si2–Al at a fission density of $1.8 Â 1028
fissions mÀ3 and subsequently to that of USi at
2.5 Â 1028 fissions mÀ3. For LEU fuels, however,
these fission densities are unreachable. In Figure 22,
only the H3-3 sample having HEU has a fission

density of 2.4 Â 1028 fissions mÀ3. However, the IL
growth rate does not appear to slow in this sample,
as shown in Figure 22. Samples A115 and A116 have
relatively high fission densities 1.6 Â 1028 and
1.2 Â 1028 fissions mÀ3, respectively, but the fitted
curves in Figure 22 rule out a decrease in the IL
growth rates at high burnups. Instead, a closer examination reveals that U3Si–Al tends to have faster than
the average IL growth at high burnups.
The U/Si ratio of U3Si2 drops to that of USi at a
fission density of 9.5 Â 1027 fissions mÀ3. HEU samples
A121 and A122 were irradiated to fission densities
of 9.4 Â 1027 fissions mÀ3 and 1.4 Â 1028 fissions mÀ3,
respectively. Neither shows any sign of a decrease in
IL growth with burnup. Instead, they experienced
above-average growth, consistent with the observation
for U3Si–Al, which is the opposite of observations
regarding the effect of U/Si ratio on IL growth.
Possible reasons for this contradiction are discussed
below.
Fissions in the IL and in the fuel particles irreversibly decrease the U/Si ratio by U depletion, but
they increase fission product concentrations in the
IL and the unreacted fuel. The transition-metal
elements including rare-earth elements are produced at a rate of $1.3 atoms per fission. These
are known to take U atom sites in the U–Al IL and
react with Al, and they have low chemical affinities
to U. Consequently, these fission products contribute to IL growth, compensating for the reduction in


410


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

16
14

.
Y 2/f (10-22 cm5 s)

12

U3Si–Al (HEU)
U3Si–Al (MEU)
U3Si–Al (LEU)

10

U3Si2–Al (HEU)
U3Si2–Al (MEU)
U3Si2–Al (LEU)

8
6

USi–Al (MEU)
USi–Al (LEU)

4
2
0
0


100

200

(a)

300
Time (day)

400

500

25

Y2/fr0.5 (10-15 cm3.5 s0.5)

20

15

U3Si–Al (HEU)
U3Si–Al (MEU)
U3Si–Al (LEU)

10

U3Si2–Al (HEU)
U3Si2–Al (MEU)

U3Si2–Al (LEU)
USi–Al (MEU)
USi–Al (LEU)

5

0
0
(b)

100

200

300

400

500

Time (day)

_ (b) Y2 normalized with
Figure 20 Y2 versus time plots to show the effect of the fission rate. (a) Y2 normalized with f,

The interaction layer thickness data included in the plots are at 100 C, but H3-3 (high-enrichment uranium) at 225  C
is normalized to 100  C.

the U atom concentration by burnup. The 1.3 yield
ratio explains the higher than average IL growth

rates for the high fission density samples (A115
and A116 for U3Si–Al and A121 and A122 for
U3Si2–Al), which is apparently contradictory to the
observation that the lower the U/Si ratio, the lower
is the IL growth.
3.14.3.4.3 U–Si blister threshold temperature

For typical fuel particle loadings, miniature scale
plates of U3Si2 and U3Si were blistered in the
range of 515–530  C. When the fuel loading was
very highly increased, they blistered at 450–475  C.

pffiffi
f_.

The full-scale plates were in the range 550–575  C.
These data show that the blister threshold temperatures for U–Si compound plates are similar to those
for UAlx plates.51 From these data, the blister threshold temperatures are typically set in the range
525–550  C.14
When boron is added, the blister threshold temperature decreases by about 100  C, which is
observed in a similar pattern as with UAlx–Al
dispersion fuels (see Section 3.14.2.4.3). The blister
threshold temperature for U–Si intermetallic dispersion fuels is less sensitive both to burnup and to
fuel volume loading than UAlx–Al dispersion fuels.52


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

3.5
3.0


U/Si atom ratio

2.5
U3Si

2.0
1.5
1.0

U3Si2

0.5
USi
~18.5

~9.5

0.0
0

5
10
15
20
Fission density (1027 fissions m–3)

25

30


Figure 21 U/Si mole ratio changes in U3Si, U3Si2, and USi
versus fission density.

Y 2 (μm2)

60

U3Si–Al (HEU)

U3Si2–Al (HEU)

U3Si–Al (MEU)

U3Si2–Al (MEU)

U3Si–Al (LEU)
USi–Al (MEU)

U3Si2–Al (LEU)
USi–Al (LEU)

A115

40
A116
A122

20
H3-3


0

0

2

A121

4
6
f t, 1014 (fissions cm–3 S)0.5 S

8

pffiffi
Figure 22 Y2 versus f_ t for U3Si–Al, U3Si2–Al, and
USi–Al. All samples were tested at 100  C, H3-3
(high-enrichment uranium) at 225  C was normalized to
100  C. The dashed lines are the fitted average lines
for the corresponding fuel.

3.14.3.5

Summary for U–Si

U3Si2 is presently considered the best qualified fuel
in terms of uranium loading and performance for
research and test reactors. Although U3Si is unsuitable for a plate-type geometry because of unstable
swelling, it is still applicable for fuel rods. In U3Si2/

Al dispersion fuel, ILs grow more slowly than in
U–Mo/Al dispersions. The ILs in U3Si2/Al are free
of porosity formation, in contrast to U–Mo/Al.
Fission gas bubbles in the unreacted fuel particles
are generally small and stable except under extremely
high-burnup and/or high-temperature conditions.
Two methods are most commonly used for
U-silicide fuel fabrication: ground powder fabrication
and atomization powder fabrication. The ground

411

powder method uses a mechanical comminution process of the U–Si alloy. The typical atomization
method adopts an atomization technology widely
used in powder metallurgy. The atomized powders
are spherical, whereas the comminuted powders are
irregular in shape.
U3Si and U3Si2 are amorphized under irradiation
due to damage induced by fission. In the amorphous
fuel, fuel swelling depends on the viscosity of fuel.
The viscosity of an amorphized material is lower than
when it is crystalline: in other words, the fluidity of
the material increases when it becomes amorphous.
Hence, fission gas bubbles grow faster when fuel is
amorphized. Although both are amorphous during
irradiation, U3Si and U3Si2 have inherent differences
in fission gas bubble growth in that U3Si has unstable
swelling, whereas U3Si2 has stable swelling. The
cause for the difference is attributed to the lower
free volume in U3Si2. Apparently, the additional Si

bonds in U3Si2 have a large effect on the amount of
free volume in the glassy state.
In out-of-pile tests, U3Si, U3Si2, and USi react
with matrix Al to form a single intermetallic compound, U(AlSi)3. This has a composition intermediate between UAl3 and USi3, which are mutually
soluble. The Al/Si ratio in the IL is the highest
for U3Si–Al, lower for U3Si2–Al, and the lowest
for USi–Al. For all cases, the reaction products
U(AlSi)3 have a density of $7.1 g cmÀ3, and approximately equal volumes of uranium silicide and Al are
used. Only a small volume change, $4%, is involved
in the reaction.
The compositions of the IL from in-pile tests of
U3Si2–Al show composition deviations from the exact
stoichiometry, that is, (Al þ Si)/U ¼ 3, indicating that
the ILs become amorphous during irradiation. Since
the IL is amorphous, U, Al, and Si atoms exist in a
mixture without crystalline restriction of stoichiometry. The IL growth is expressed as a function of the
fission rate, temperature, and irradiation time.
The blister threshold temperatures are in the
range 525–550  C. The blister threshold temperature
for U–Si dispersion fuels exhibits a similar trend to
that observed with UAlx–Al dispersion fuels.

3.14.4 U–Mo
3.14.4.1

U–Mo Fuel Properties

The high-temperature body-centered cubic g-phase
of uranium has shown superior performance to the
low-temperature orthorhombic a-phase in terms of



412

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

fuel swelling. It was recognized early on in the development of fast reactor fuels that molybdenum is one
of the strongest g-stabilizers of the transition-metal
elements, even stronger than Zr, and that it enables
alloys with U to have relatively high U density (see
Table 1).8 A disadvantage of Mo as an alloying element is that it has higher neutron absorption cross
sections than Si and Al. Even though it has not been
dealt with here, physics designs showed that the
penalty caused by the parasitic neutron absorption
was insignificant.
As can be seen in the U–Mo phase diagram shown
in Figure 23, the solubility of Mo extends to 22 wt%
(or 41 at.%) in the high-temperature body-centered
cubic g-phase, but it is limited to a few percents in
the a- and b-phases. The g-phase undergoes a eutectoidal decomposition at 565  C, transforming to the
dual-phase mixture of the orthorhombic a-phase and
the ordered tetragonal g0 -phase which has the nominal stoichiometry of U2Mo. This transformation is
slow when the molybdenum content is more than
about 6 wt%, so a g-phase metastable U–Mo alloy
typically with 6–12 wt% Mo can be obtained by
quenching the alloy melt into the g-phase. This

Weight percent Mo
10
20


0

30

1400
B A

L

1300
1200

Temperature (ЊC)

1100
1000
γ

900
β+γ

β

800

γ+δ

α+β


700
α+γ

600

γ + γЈ

alloy remains in the g-phase indefinitely at room
temperature.
Another advantage of the g-phase uranium over
the a-phase lies in its slower reaction with matrix
aluminum than a-phase uranium (see e.g., Park et al.53
and references therein).
The radiation stability of U–Mo alloy depends to
a considerable extent upon its ability to retain the
g-phase during irradiation. During irradiations at temperatures below the g- and a þ g0 -phase boundary,
which is 565  C for U–10 wt% Mo, the alloy tends
to transform to the a- and g0 -phases because these are
the thermodynamically stable phases at these lower
temperatures. The effect of fission-induced displacements and thermal spikes is to oppose this tendency,
the spikes tending to disorder the g0 -phase and to
produce a homogeneous composition of uranium
and molybdenum in the g-phase. The critical fission
rate is the rate at which the minimum number of
displacements and thermal spikes that maintain the
g-phase are in balance, with the thermodynamic tendency to transform to the a þ g0 -phases. Willard and
Schmitt54 reported a critical fission rate–temperature
correlation above which the g-phase U–Mo is stable
and so fuel swelling is low. Their results are given
in Table 3.

The typical Mo content is in a range of 6–12 wt%,
and 10 wt% Mo in the U–Mo alloy is optimal for
low fuel swelling. Fission gas bubble swelling appears
to be minimum at this content. Coincidentally, the
10 wt% Mo in the U–Mo alloy is close to the eutectic
composition.
The variation of the lattice parameter of U–Mo
alloys with the Mo concentration has been provided
as follows55:
a0 ¼ 3:4808 À 0:314xMo
½13Š
where a0 is in A˚, and xMo is the Mo content in U–Mo
alloy in mole fraction.
The U–Mo alloy density can be estimated by the
rule of mixture
rUÀMo ¼ ð1 À xMo ÞrU þ xMo rMo

580

½14Š

565

500
α

α + γЈ

400


γЈ

γЈ + δ

Table 3
Calculated critical fission rate for stabilizing
g-phase U–Mo

300
0

10

20
30
40
Atomic percent Mo

50

Figure 23 Partial phase diagram of the U-rich side
U–Mo system. The arrows marked by A and B represent
U–10 wt% Mo and U–7 wt% Mo alloys, respectively.

Temperature (K)

Critical fission rate (mÀ3 sÀ1 )

644
658

672
686

8.8 Â 1017
2.2 Â 1018
4.8 Â 1018
9.2 Â 1018


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

where xMo is the Mo content in mole fraction. If the
unit cell information is used, the density of U–Mo
can also be obtained by
rUÀMo ¼

n½ð1 À xMo ÞMU þ xMo MMo Š
NA ð3:4808 À 0:3314xMo Þ3

½15Š

where n is the number of atoms in the unit cell (which
is 2 for the body-centered cubic g-phase), xMo is the
Mo concentration in mole fraction, MU and MMo are
the molecular weights of U and Mo, respectively, and
NA is the Avogardro number.
At room temperature, the U–Mo alloy density is
given by an empirical equation56:
rUÀMo ¼ 19:1 À 9:11xMo


½16Š

where xMo is the Mo content in mole fraction.
Equation [15] predicts a density larger by 1% than
by eqn [16], and also larger by 0.7% than by eqn [14].
The lower predictions by eqn [16] are probably due
to the measurement data that include small pores
in the fabricated samples. Equation [16] is recommended for practical use, however, because it is more
likely to fit the real alloys.
3.14.4.2 Thermal Conductivity of
(U–Mo Alloy)–Al Dispersions
The thermal conductivity of U–Mo alloy is given
approximately as follows56:
kTUÀMo ¼ 2:2 þ 0:032T

½17Š

where k is the thermal conductivity of U–Mo alloy
in W mÀ1 KÀ1 and T is the temperature in K in the
range of 298 < T < 773 K. The thermal conductivity
of (U–10Mo)–Al dispersions, where 10 is for 10 wt%
Mo, is given by fitting data for 40% fuel volume
fraction in the dispersion57:
ðUÀ10MoÞÀAl

kT

¼ À 18:75 þ 0:70T À 1:24 Â 10À3 T 2
þ 7:53 Â 10À7 T 3


½18Š

where k is in W mÀ1 KÀ1 and T is in K in the range
298 < T < 773 K.
3.14.4.3

U–Mo Fabrication

3.14.4.3.1 U–Mo alloy powder fabrication

Like U–Si intermetallic fuels, the powder fabrication
methods for U–Mo alloys widely adopted currently
are the comminution and atomization. The comminution powder method uses a mechanical and/or

413

chemical grinding process of the alloy ingot, and the
atomization method typically includes pouring the
U–Mo alloy melt onto a rotating disk in an inert
atmosphere or using a rotating consumable electrode.
The U–Mo alloy is ductile, which poses difficulty
for comminution similar to U3Si. To overcome the
toughness of the fuel alloy, slight oxidization of the
fuel alloy is allowed during the comminution process.
The comminuted powders have more equiaxially
shaped grains and a more homogeneous distribution
of grains than the atomized powder fuel because
there is no thermal process involved during fabrication. In addition, they are heavily cold-worked and
contain a high concentration of dislocations. During
hot-rolling and subsequent irradiation of the fuel

plates, this dislocation structure will undergo polygonization. The final subgrain structure is similar to
that of the atomized powder, providing nucleation
sites for gas bubbles. Comminuted powders are made
from well-homogenized cast alloy rods, and therefore
do not contain the ‘cored’ cellular structure typical of
the rapidly solidified atomized powders.
The microstructure of the atomized powder consists of a ‘cellular’ solidification structure which is
commonly found in rapidly cooled alloys that have a
pronounced solidus–liquidus gap. When the U–10Mo
melt (or U0.78Mo0.22) in the U–Mo phase diagram
shown in Figure 23 is cooled, the melt follows the
right arrow (line A). When it meets the liquidus line,
U0.64Mo0.36 solidifies as solid islands. As the cooling
progresses, the solid phase volume increases, while,
simultaneously, the Mo content in the solid phase
decreases. In the atomization process, however, the
cooling does not follow this equilibrium process.
Instead, once it meets the solidus line, the remaining
liquid phase, which consists of a lower Mo content
than the solid islands, solidifies abruptly, leaving an
interconnected network with a lower Mo content.
This structure, when etched, may appear as a thick
boundary because of its low Mo content and obscures
the thinner grain boundaries. The boundaries between
the cells most likely form low-angle grain boundaries.
Because the swelling rate of U–Mo alloy increases
as the Mo content decreases starting from 10 wt%,
one can argue that the lower Mo content region near
the grain boundaries has a more favorable environment for bubble nucleation. A similar process for
a lower Mo content alloy, such as U–7Mo, is shown

by the arrow B in Figure 23. The gap between the
solidus and liquidus lines on arrow B is slightly
smaller than that on arrow A. Thus, U–7Mo will
have slightly smaller solid islands, that is, smaller


414

Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

CL

CL

GB

Grain

CMo (at.%)

22

5
C

17

0
0


(a)

CL = center of grain boundary
GB = grain boundary
b = half of GB thickness
L = distance between two neighbor grain centers

b
Distance from GB center

L/2

b

0

L

Distance from GB center

(b)

Figure 24 Model of the initial condition of the grain boundary and Mo content. Schematic of (a) a grain boundary, and
corresponding Mo content and (b) Mo content versus distance from grain boundary center.

@Cðx; t Þ
@ 2 Cðx; t Þ
¼D
@t
@x 2


½19Š

with the boundary and initial conditions,
@C
¼ 0 at x ¼ 0; L=2
@x
Cðx; 0Þ ¼ 0 0 x b; Cðx; 0Þ ¼ 5 b

x

L=2 ½20Š

The solution is similar to that obtained by Bleiberg,59
which is as follows:

23
CMo (1.5,t)
22
Mo concentration

grains, than U–10Mo. In addition, the grain boundary
Mo content of U–7Mo is lower than that of U–10Mo.
Bleiberg showed phase reversal of a þ g0 to g
during irradiation to a burnup of $0.1 at.% total uranium burnup (LEU equivalent 0.5% 235U burnup).58,59
(LEU equivalent BU is defined as 235U burnup based
on 20% enrichment. For example, 50% LEU equivalent BU ¼ 10% total U burnup.) He concluded that
phase homogenization was sensitive to the fission rate
and the distance between lamella centers. By using this
result, the ability of cell boundaries to homogenize

during irradiation can be analyzed.
A model to analyze grain homogenization time is
schematically shown in Figure 24. The as-fabricated
Mo concentration at the grain interior is estimated to
be 22 at.% and that at the grain boundary 17 at.%.
For simplicity, the diffusion coefficient of Mo is
assumed to be constant with respect to Mo concentration and is dependent only on the fission rate,
given by D ¼ 1=12fr V 5=3, where fr is the fission rate,
which is assumed to be constant throughout irradiation, and V is the volume affected by a fission spike
(2.01 Â 10À18 cm3). Then, the diffusion process is
governed by Fick’s equation:

21
CMo (0,t)

20
19
18
48 days

17
16

0

100

200
300
Time (day)


400

500

Figure 25 Comparison of the Mo concentration at the
grain center (CMo(1.5, t)) and at the grain boundary center
(C(0, t)) as a function of time. Input data used for the
calculation are fission rate = 6 Â 1014 fissions cmÀ3 sÀ1,
D ¼ 1.6 Â 10À16 cm2 sÀ1, grain size (L) ¼ 3 mm, and
grain-boundary width (2b) ¼ 0.5 mm.

2
Cðx; t Þ ¼
L

L=2
ð

Cðx; 0Þdx
0

L=2
1 ð

4X
x
þ
Cðx; 0Þcos 2np dx
L n¼1

L
0



x
t
cos 2np exp À4Dn2 p2 2
L
L

½21Š

Truncating the series in eqn [21] to n ¼ 10, we
obtain the change in Mo concentration as a function
of time at the grain boundary and grain interior, as
shown in Figure 25. As seen in Figure 25, the Mo
concentrations at the grain center and at the grain


Uranium Intermetallic Fuels (U–Al, U–Si, U–Mo)

boundary converge with time. Complete homogenization, however, takes $2000 days. This is far beyond
the irradiation time of a typical research reactor. The
concentration difference between the grain center
and the grain boundary is still 2.5 at.% at end of life
(EOL) (or 48 effective full power days, EFPD). For
this analysis, we used a grain size of 3 mm and grainboundary ‘width’ of 0.5 mm. This analysis leads to the
conclusion that, for the given irradiation condition of
typical plates, the grain boundaries remain at a lower

Mo concentration to the end of the test.
3.14.4.3.2 U–Mo dispersion plate fabrication

Time to start γ --> α + γЈ transformation (h)

The U–Mo fuel uses the same plate fabrication method
as other U intermetallic fuels, which is hot-rolling
(see Figure 2 for details of the fabrication method).
However, unlike other fuels, the hot-rolling procedure
(including the follow-on blister test) has significant
effects on the performance of this fuel because of the
heat processes involved. The heat processes change
particle characteristics and enhance the reactions
between the fuel particles and matrix aluminum.
When the as-fabricated U–Mo particles in the
metastable g-phase are heated during fabrication at
$485  C, the fuel is forced to undergo a a þ g0 transformation. The time to start this transformation is
governed by the temperature and is typically given
by a TTT (time–temperature–transformation) diagram. The time taken for the typical hot-rolling
fabrication and follow-on blister test is enough to produce partial transformation reactions. Times to transformation can be assessed for typical heat processes
as shown in Figure 26.
As Bleiberg58,59 has shown, irradiation quickly
reverses the partly transformed phases a þ g0 to g.
100

425 ЊC
10

500 ЊC


1

Symbols: measured data
Lines: data fitting

0.1

4

5

6

7
8
9
Mo content in U (wt%)

10

11

Figure 26 Time to start g ! a + g0 -phase transformation
at two temperatures.

415

However, during the heat processes, the reaction
between the fuel particles and matrix aluminum
increases at the fuel surface, which is in the a-phase.

The reaction rate of the a-phase with the matrix is
considerably higher than that of the g-phase.
3.14.4.4

U–Mo Irradiation Performance

3.14.4.4.1 Fuel swelling by fission products

Fuel swelling by solid fission products discussed
in Section 3.14.2.4.1 is also applicable for U–Mo.
The swelling rate given by eqn [4] can be used.
However, fuel swelling kinetics by fission gases, that
is, fission gas bubble growth, is different and discussed more in depth below.
As discussed in Section 3.14.2.4, the solid fission
product swelling is a linear function of burnup and
given by
 
DV
¼ 4:0fd
½4Š
V0 s
where the swelling rate is in % and fd is the fission
density in 1027 fissions mÀ3.
U–Mo swelling, specifically, swelling by gas bubble growth, is known to have two distinct rates: slow
at low burnup and much faster at high burnup. The
phenomenon underlying the transition is grain refinement or ‘recrystallization’ of the g-phase U–Mo. After
this transition, gas bubble agglomeration accelerates,
resulting in faster swelling.60
In Figure 27, evolution of fuel microstructure by
fission gas bubble formation and growth is shown

with three different burnups. Fission gas bubbles
first appear with an SEM resolution of 0.1 mm on
grain boundaries, heterogeneously over the fuel
cross section, as shown in Figure 26(a). There are
virtually no bubbles in the interior of the grains.
Van den Berghe et al.61 reported the observation of
about 2-nm size bubbles in a spacing of 6–7 nm, as
shown in Figure 28. As the bubbles are small, even
though their number density is large, these bubbles
are too small to produce much fuel volume increase,
so we include the effect in the solid fission product
swelling. As burnup increases ($2.5–3.5 Â 1027 fissions
mÀ3), the bubble population increases in the grain
boundaries and additional bubbles progressively
appear at newly formed grain boundaries as grain
refinement continues (shown in Figure 27(b)). At
this stage, the average bubble size also increases with
fission density as the number density increases, both
of which increase the fuel swelling rate. Eventually, at