3.03

Carbide Fuel

A. K. Sengupta, R. Agarwal, and H. S. Kamath
Bhabha Atomic Research Centre, Mumbai, India

ß 2012 Elsevier Ltd. All rights reserved.

3.03.1

Introduction

56

3.03.1.1
3.03.1.2
3.03.2
3.03.2.1
3.03.2.2
3.03.3
3.03.3.1
3.03.3.2
3.03.3.3
3.03.3.3.1
3.03.3.3.2
3.03.3.4
3.03.3.4.1
3.03.3.4.2
3.03.4
3.03.4.1

3.03.4.1.1
3.03.4.1.2
3.03.4.1.3
3.03.4.1.4
3.03.4.1.5
3.03.4.1.6
3.03.4.2
3.03.4.3
3.03.5
3.03.6
References

History of Carbide Fuel
Glimpses of Carbide Fuel
Physical Properties
Thermophysical Properties of Carbide Fuel
Thermochemistry of Carbide Fuels
Fabrication of Carbide Fuel
Melting Casting
Hydride/Hydrocarbon Route
Carbothermic Reduction Route
Direct pressing method
Sol–gel (wet) route
Quality Control
Chemical quality control method
Physical quality control
In-Pile Performance
Introduction
Burnup
Swelling

Performance of Na-bonded and He-bonded fuel pins
Irradiation creep
Experience on irradiation performance
Fuel–clad chemical interaction
Effects of Burnup on C/M Ratio and Chemical State of Fission Product
MA-Containing Fuel
Fuel Reprocessing and Waste Management
Summary

56
57
59
59
60
64
64
64
64
65
66
67
68
69
70
70
70
71
72
73
75

78
79
80
82
83
84

Abbreviations
ABR
ADS
BARC

Advanced burner reactor
Accelerated-driven system
Bhabha Atomic Research Centre,
Mumbai
C/M, N/M
Carbon to (U þ Pu) ratio, nitrogen to
(U þ Pu) ratios
CARLO
Carbide low in oxygen
CARRO
Carbide rich in oxygen
CDT
Compounded doubling time
CEA, France Commissariat a` l’e´nergie atomique,
France
EBR I
Experimental breeder reactor I
EBR II

Experimental breeder reactor II

EFR
EPMA
FBR
FBTR
fcc
FCCI
FCMI
FFTF
GEN-IV
GFR
IAEA
ICP/MS

European fast reactor
Electron probe microanalysis
Fast breeder reactor
Fast Breeder Test Reactor,
Kalpakkam, India
Face-centered cubic
Fuel–clad chemical interaction
Fuel–clad mechanical interaction
Fast flux test facility
Generation IV
Gas-cooled fast reactor
International Atomic Agency
Inductively couple plasma mass
spectrometer


55


56

Carbide Fuel

IDMS
IGCAR
ITU
LDP
LHR
LMFBR
LWR
MA
MKI & MKII
NIMPHE
PFR
PGM
PHWR
PIE
PUREX
SEM
SFR
TD
TIG
TIMS
TREAT
TRU
U/M, Pu/M

UREX
VPPM
XGAR
XRD
XRF

Isotopic dilution mass spectrometry
Indira Gandhi Centre for Atomic
Research, Kalpakkam
Institute of Transuranium Elements
Large development plant
Linear heat rating
Liquid metal fast breeder reactor
Light water reactor
Minor actinides
Mark I and Mark II fuels of FBTR,
Kalpakkam, India
NItrure Mixte dans a Phenix
Prototype fast reactor
Platinum group metals
Pressurized heavy water reactor
Postirradiation examination
Plutonium and uranium recovery by
extraction
Scanning electron microscopy
Sodium-cooled fast reactor
Theoretical density
Tungsten inert gas
Thermal ionization mass
spectrometry

Transient-overpower test
Transuranium element
Uranium to (U þ Pu) ratio, plutonium
to (U þ Pu) ratio
Uranium recovery by extraction
Parts per million by volume
X-ray gamma autoradiography
X-ray diffractrometry
X-ray fluorescence spectroscopy

3.03.1 Introduction
3.03.1.1

History of Carbide Fuel

The world energy requirement from nuclear sources
was mostly met by the use of uranium in pressurized
heavy water reactors (PHWR) or light water reactors
(LWR). However, uranium reserve is limited and it
was felt necessary that the fuel cycle based on uranium must be a closed fuel cycle. In the vast majority
of LWRs or HWRs, nuclear power is generated by
the fission of 235U by thermal neutrons. The
235
U content in natural uranium is only 0.7% and
the remaining is 238U which is nonfissionable.
238
U can, however, be converted to fissionable 239Pu,
which can be subsequently used as fuel for fast reactors.
The energy generated from natural uranium using


FBRs is about 60 times greater than that obtained
from LWR, even after allowing for losses in the fuel
cycle.1 Hence, for effective utilization of the limited
resources of U, the fast reactor concept came into existence. Consequently, efforts were in progress to develop
fast breeder reactors (FBRs) in several countries such
as the United States, the United Kingdom, France,
and the erstwhile USSR. In the 1950s, the early fast
reactor fuel developed was metallic (experimental
breeder reactor I, EBR I) and it became the first fast
reactor in the United States to generate electricity.
This led to the development of the second generation
of larger fast reactors, which started operating in
the 1970s. With the inception of fast reactor technology, the power density of the nuclear reactor core
went up manyfold, thus enabling compact reactors
producing larger power. These first few prototype
reactors were BN350 (USSR), PHENIX (France),
and PFR (United Kingdom). Large prototype fast
reactors of capacity 500 MWe were developed in
the USSR (BN600) in 1980 and another one of
1200 MWe capacity was developed in France (Super
Phenix). The first-generation large breeder reactor
used uranium–plutonium-mixed oxide fuels in austenitic stainless steel clad with sodium as coolant.
The technology for mixed oxide fuel fabrication and
its irradiation performance are quite well established
and a burnup of 10 at.% or more has been achieved in
several reactors: for example, PFR (United Kingdom),
Phenix, fast flux test facility (FFTF) (United States),
and JOYO (Japan). The European fast reactor (EFR)
program has also considered mixed oxide fuel for
the first demonstration-type reactor. Although FBRs

based on mixed oxide fuel proved quite successful
in enhancing energy production, this option was not
a plausible futuristic solution particularly for developing countries which have limited uranium reserve
to meet their long-term sustainable energy demand.
The anticipated shortage of fissile material in future
led to the search for a fuel having higher breeding
ratio and lower compounded doubling time (CDT)
(less than 15 years) than oxide (about 25 years).
The CDT is defined1 as the time needed to double
the inventory of fissile material allowing for outof-reactor time, reprocessing, and losses during
out-of-reactor time in a system of breeder reactors
where the excess fissile material is used to start up new
breeders as soon as the fuel from each cycle is reprocessed. The high doubling time of oxide fuel is attributed to its lower heavy-atom density. It also operates
at a lower linear heat rating because of its low thermal
conductivity. These limitations of the oxide fuel led to


Carbide Fuel

the opening up of worldwide research for a suitable
fuel with higher fissile atom density and thermal conductivity. Metallic fuel based on U–Pu–Zr or U–Zr
was presumed to be a possible solution, but it too had
limitations such as low melting point and high swelling
rates. Other possible fuel compositions with higher
fissile atom density were nonoxide ceramics, for example, uranium–plutonium-mixed carbide or nitride.
These fuels have higher thermal conductivity (about
5 times higher than oxide; Figure 1), high fissile
heavy-atom density, and a reasonably high melting
point compared to metallic fuel.2 The higher thermal
conductivity of carbide fuel results in more efficient

heat transfer from the fuel to the coolant compared to
oxide and this, in combination with a high melting
point, makes carbide fuel suitable for operation at
high specific power without causing any fuel central
melting. This also enables the use of large diameter
fuel pins, more fissile material per pin, and more
power generation. Increase in the oxygen/metal ratio
with burnup in case of oxide fuel increases the probability of fuel–clad chemical interaction (FCCI),
whereas for carbide and nitride fuel, the nonmetal to
metal ratio (C/M or N/M) remains constant or may
even decrease depending upon the composition of the
fuel. High-specific-power operation permits fewer
pins, compared to oxide, and a more compact core.
This in totality reduces the plant cost significantly.
These findings were very encouraging and led to
initiation of a large amount of development work on
carbide and nitride fuels from 1960 to 1970, and more
than 5000 advanced fuel pins have been fabricated
and irradiated. However, nitride fuel has an inherent

30

Thermal conductivity λ (Wm-1 K-1)

U–19Pu–10Zr
25

MC

MN


20

15
10
5
0
500

Oxide MO1.98

problem of 14N because of the (n.p.) reaction leading to
formation of 14C. Hence to get the full benefit of a
nitride fuel, 14N should be replaced by 15N, which is
an expensive proposition. The (n.p.) reaction will have
a negative effect on the breeding ratio. However, a fuel
cycle based on carbide fuel could be the most feasible
solution, and extensive research on different aspects of
carbide fuel based on out-of-pile and in-pile experiments has been carried out in countries such as the
United States, Germany, France, and India with the
objective of developing their fast reactor program
based on ‘carbide fuel.’
The development of the carbide-fueled FBR was a
little sluggish during the 1970s. The practical difficulty of fabrication of carbide fuel economically was
probably the cause, since the high-purity inert cover
gas required for fuel fabrication was expensive and
maintenance of C/M ratio was difficult. Apart from
this, the thermo-chemical in-pile behavior of carbide
fuels, e.g., extent of loss of ductility of the cladding
due to carburization was not fully understood. Behavior of the fuel under off-normal conditions, like loss

of sodium bonding during transient-overpower conditions, was also not systematically investigated. In
the absence of this information, utilization of carbide
fuel was limited to test pin or capsule irradiation only.
The Indian FBR program, however, started with the
introduction of plutonium-rich mixed uranium–
plutonium carbide as the driver fuel for 40 MWth
(13 MWe) loop-type reactor (fast breeder test reactor,
FBTR). The design of the Indian FBR was similar to
that of Rapsodie (Rapid Sodium) Fortissimo version
based on oxide fuel. The reactor became critical in
the year 1985 and it is the only reactor operative on a
full core of carbide fuel. Carbide fuel cannot be used
in LWR because of its incompatibility with the coolant. However, it can be safely used with liquid metal
(Na or lead) or gas cooled (CO2 or He) in Generation
IV (Gen-IV) type of high-temperature reactor.
Hence, carbide fuel is considered as an ‘advanced
fuel’ with the basic characteristics of a breeding
ratio of at least 1.30 or more and a doubling time of
15 years or less. The burnup limit could be about
15 at.% or more. Table 1 shows the test irradiation
program of carbide fuels in different reactors around
the world.
3.03.1.2

1000

1500
Temperature (ЊC)

2000


57

Glimpses of Carbide Fuel

2500

Figure 1 Thermal conductivity of uranium–plutoniummixed carbide, nitride, oxide, and metallic fuels.

Many laboratories in the world were engaged in studies
on the development of carbide fuels based on uranium
and plutonium and thorium. The results of these


58

Carbide Fuel

Table 1

History of carbide fuels used in different countries and in different reactors

Fuel type

Reactor

Country/
organization

Bond


Density (% TD)

Burnup (at.%)

Clad

References

MC
MC
MC
MC
MC
MC
MC
UC/MC
MC
MC
MC
MC

RAPSODIE
BOR60
EBRII
RAPSODIE
KNK II
EBR II
EBR II
BOR 60

FFTF
PX
FFTF
FBTR

France/CEA
USSR
United States
TUI
FZK, Germany
United States
United States
USSR
United States
CEA/TUI
DOE/PSI
India

Na
Na
Na
He
He
He
He
He
He
He
He
He


91.5
–
–
86
85
80/87
81/87
–
80
80/82
–
91/86

12
–
12
5
7
12
16–20
10
10
–
10
16

–
OX16H15M3G
PE16

–
1.4970
316.20 cw
316.20 cw
OX16H15M3GD9
15/15 Ti
D9
SS316 cw

3
4
2
5
6
7
2
4
8
9
8
10

Table 2

Properties of mixed uranium–plutonium oxide, carbide, nitride, and metallic fuels for SFR

Properties

(U0.8Pu0.2)O2


(U0.8Pu0.2)C

(U0.8Pu0.2)N

U–19Pu–10Zr

Theoretical density
(g cmÀ3)
Melting point ( K)
Thermal conductivity
(W mÀ1  K)
1000 K
2000 K
Crystal structure
Breeding ratio
Swelling
Handling
Compatibility
Clad
Coolant
Dissolution and
reprocessing
amenability

11.04

13.58

14.32


15.73

3083

2750

3070

1400

2.6
2.4
Fluorite
1.1
Moderate
In air

18.8
21.2
NaCl
1.2–1.3
High
Inert atmosphere

15.8
20.1
NaCl
1.2–1.3
High (?)
Inert atmosphere


40
Multiphase
1.4–1.5
High
Inert atmosphere

Average
Average
Demonstrated on
industrial scale
for aqueous and
pilot scale for
pyro-processes
Large

Carburization
Good
Dissolution not simple.
Process not yet
demonstrated on
industrial scale

Good
Good
Dissolution easy but
risk of C14 in waste
management

Eutectics

Good
Pyro-reprocessing
demonstrated on
pilot plant scale

Limited

Very little

Limited

Fabrication/irradiation
experience

Good

studies were brought together in the International
symposium on ‘Carbide in Nuclear Energy,’ Vols I
and II held at Harwell in 196311 at a time when carbide
was being seriously thought of as a reactor fuel. Subsequently, following the oil crisis in 1974, a national
advanced liquid-metal-cooled fast breeder reactor
(LMFBR) fuels development program was initiated in
the United States on the basis of the data available
in the exploratory years of basic development work
on carbide fuel, and a unified national approach was
pursued to look into all the aspects of carbide fuels.

Both carbide and nitride fuels offer the best potential for LMFBR performance in the long run because
of their higher thermal conductivity, fissile-atom density, and chemical compatibility with liquid sodium.
Mixed oxide fuel has a higher melting point than

carbide or nitride fuels but the higher thermal conductivity of the carbide fuels compensates for it. Some
important parameters of these different types of fuel
are given in Table 2.
There are two concepts available for the carbide
fuel pin depending upon the type of bond between


Carbide Fuel

the fuel pellet and the cladding material: He-bonded
and Na-bonded carbide fuels. The average operating
fuel temperature of the He-bonded pin is high
because of low thermal conductivity of the He bond
compared to the Na bond. This design requires a
larger fuel–clad gap and low fuel density (85%) compared to oxide fuel so that the higher swelling of the
fuel due to high operating temperature can be accommodated. The fission gas release will also be higher
compared to that from a sodium-bonded pin. In case
of the sodium-bonded pin, the fuel–clad gap is larger
and the fuel density higher so that there is enough
radial space for accommodating fuel swelling and the
end of life of fuel is determined by the fuel–clad gap
closure. Sodium-bonded fuel does not undergo
much of fuel restructuring at comparable linear
powers used for oxide-fueled fast reactors because
of the lower central temperature. The purity of the
sodium bond is very important and is limited to a
maximum of 10 ppm. High O in sodium reduces
thermal conductivity, affecting thermal performance
of the fuel.
A hyperstoichiometric (C/M > 1, carbon to

metal ratio) fuel composition is chosen so that it
contains some amount of sesquicarbide M2C3 phase
(M ¼ U þ Pu), which takes care of the decrease in
(C/M) ratio with burnup. This decrease in (C/M)
ratio with burnup may result in the formation of
actinide metal phase, which may form low-melting
eutectic with the cladding, thereby limiting the life of
the fuel pin. This is more so for fuels with the components of high plutonium content because, for Pu
fission by fast neutron, the fission yield curve shifts to
the right, generating more noble fission products.
Depending upon the temperature of operation (type
of bond), they may cause bonding of the fuel with
cladding material resulting in fuel–clad mechanical
interaction (FCMI). O and N impurities also play
important roles, as they act as ‘carbon equivalent’,
which decides the carbon potential of the fuel.
O and N are impurities that are picked up during
fabrication and their contents are very important to
make the fuel of the desired specification. Solubility
of O is more in PuC than in UC-rich carbide fuel.
Hence, plutonium-rich fuel has more O solubility
than UC-rich fuel. The details of all these aspects
are discussed in the following section, which deals
with the fuel properties, characterization, fabrication, and postirradiation examination (PIE). The
reprocessing and waste management of carbide fuel
has also been covered, to give a glimpse of the
overall carbide fuel cycle. Table 3 shows the typical

59


Table 3
Specifications of high- and low-plutoniumcontaining fuels

Plutonium (wt%)
Pu/(U + Pu)
Oxygen (ppm)
Oxygen + nitrogen (ppm)
M2C3 (wt%)
Density (% TD)
Grain size

(Pu0.7U0.3)C

(Pu0.2U0.8)C

66 Æ 1
0.70
6000
7500
5–20
90 Æ 1
$10–12 mm

21.3 Æ 1
0.225
360
$400
12.5 Æ 1.4
80 Æ 1
12 mm


fuel specification for carbide fuels and Figure 2 shows
the design of the Pu-rich carbide fuel pin used in
FBTR.

3.03.2 Physical Properties
3.03.2.1 Thermophysical Properties of
Carbide Fuel
The thermophysical properties that are of importance and affect the fuel performance are solidus/
liquidus temperature, thermal conductivity, coefficient of thermal expansion, elastic/fracture properties, creep, and hardness at ambient and at high
temperatures.
The solidus/liquidus temperatures along with thermal conductivity limit the fuel operating temperature
in terms of linear heat rating (W cmÀ1), and the fuel
center which ‘sees’ the highest temperature does not
exceed the solidus temperature. The liquidus temperature gives some indication of the physical state of the
fuel in case of core meltdown under accidental conditions. Thermal conductivity is an important factor that
determines the rate of heat transfer from the fuel to the
clad. As mentioned above, these properties also put a
limit on the fuel surface temperature. Thermal conductivity, though an intrinsic property, varies with a
number of parameters which are characteristic to the
sample. Some of these parameters are density or porosity (shape, size, and distribution), composition, presence of a second phase, grain size, etc. Coefficient of
thermal expansion is an important design parameter
for the fuel pin (both for fuel and cladding material). It
depends on the composition as well as the extent of
second phase present. The stresses generated by the
fuel over the cladding material are partly due to the
difference in the coefficient of thermal expansion
between the fuel and the cladding material. The elastic
property and the fracture property of the fuel are
primarily responsible for the extent of FCMI. Hardness



60

Carbide Fuel

Angular position
of top plug with
respect to bottom plug

263Њ
View in direction S
Æ5.1

Æ5.1
Top plug

13

13

Leak tight
weld (TIG)
Spring

S
S1

8


Wire weld
on the
top plug

4
106

Spring
support
Insulation
pellet

SS316 Clad

4
3

Æ 0.76 mm wire
wound throughout
the length
pitch = 90 mm

±1.5
320

515.3
531.5

505.5
Fuel pellet

531.5
Insulation
pellet
Disk

3

61.5
1.8

Pellet support
tube
Leak tight
weld (TIG)

3

Bottom plug

13

S2 13

FBTR fuel pin
sectional view

FBTR fuel pin

Figure 2 Schematic diagram of the fuel pin design of a fast breeder test reactor.


of the fuel determines the extent of FCMI: a softer
fuel will exert lower stresses on the cladding, thus
have lower FCMI. Both thermal-induced and irradiation-induced creep of the fuel also determine the
extent of pellet–clad mechanical interaction. Creep
properties also depend on a number of variables such
as composition, presence of a second phase as precipitates, grain size, etc. Thermophysical properties of
actinide carbide have been extensively described and
discussed in Chapter 2.04, Thermodynamic and

Thermophysical Properties of the Actinide Carbides of this Comprehensive. A summary of the data
for high (70%) and low (20%) plutonium-containing
fuel12 is given in Table 4.
3.03.2.2

Thermochemistry of Carbide Fuels

Uranium forms three compounds with carbon, that is,
UC, U2C3, and UC2, out of which only U2C3 is a
stoichiometric compound. UC is stable over a wide


Carbide Fuel

temperature and composition range and melts congruently at slightly substoichiometric composition,
at 2780 K. UC2 is stable in two phases, a-UC2 and
b-UC2. U2C3 decomposes into UC þ a-UC2 on heating from 2096 to 2110 K and into UC þ C below
$1400 K. The Pu–C system has four compounds:
Pu3C2, PuC1–x, Pu2C3, and PuC2. The compound
Pu3C2 decomposes into Pu þ PuC at 848 K. Though
the crystal structures of carbides of U and Pu are

very similar, phase diagrams of U–C and Pu–C
are very different. These differences are mainly due
to (i) the presence of Pu3C2 compound, (ii) the low
stability of PuC compared to UC, and (iii) the
high stability of Pu2C3 compared to U2C3 (Chapter
2.04, Thermodynamic and Thermophysical Properties of the Actinide Carbides).
UC and PuC are highly dense, face-centered
cubic (fcc) packed metal atoms with octahedral
holes occupied by carbon atoms. The brittleness of
carbides is due to alternate close-packed planes

of metal and nonmetal atoms, the latter restricting
the slip and thereby hardening the crystal. In addition, the two p-states of carbon do not favor the
formation of octahedral ligands; hence many of
the fission products are not soluble in monocarbide
lattice. During irradiation, carbide fuels are known
to swell more than the oxide fuel. Swelling is a complicated phenomenon, controlled by many factors.
However, the close-packed structure of monocarbides
is known to contribute to higher swelling in carbide
fuels, especially at low temperatures (T/Tm < 0.3), as
bulky fission product atoms of Xe/Kr cannot be
accommodated in the carbide lattice. The transport
properties of vacancies and interstitials in these structures also add to this problem. The presence of the
more open type structure of the M2C3 phase in hyperstoichiometric mixed carbide fuel reduces the
swelling to some extent. The structures of UC and
PuC are isomorphous with monocarbides of transition metals and some of the actinide metals, and thus
these fission products dissolve in the monocarbide
phase to varying degrees.
An increase in O and N impurities results in an
increase of carbon activity and CO pressure of the

fuel (Figure 3). O impurities are also known to contribute significantly to the actinide redistribution in
the carbide fuels and fuel restructuring during burnup.
Much experimental data as well as assessed and critically reviewed works are available in the literature
investigating the impact of O þ N impurities on the
behavior of carbide fuels.13–17 In order to understand

Table 4
Thermophysical properties of high- and lowplutonium-containing carbide fuels
Properties

(U0.3Pu0.7)C

(U0.8Pu0.2)C

Solidus temperature (K)
Thermal conductivity
(W mÀ1 K) at 1273 K
Coefficient of thermal
expansion (300–1800 K)
Hardness (MPa) at 1250 K

2148
12.0

3023
19.0

13.8 Â 10À6

10.9 Â 10À6


1200

1400

61

6 ´ 10–16

0.011

T = 1000 K
pCO

0.010

aC
aC

Pu/M = 0.55

pCO

2 ´ 10–16

0.009

0.008

pCO

1 ´ 10–16

0.007

8 ´ 10–17

aC of SS-clad at 1000 K

6 ´ 10–17

Carbon activity

4 ´ 10–16

0.006
pCO

4 ´ 10

–17

aC

0.005

Pu/M = 0.7
2 ´ 10–17
600

800


1000

1200

1400

1600

1800

2000

0.004
2200

[O] in ppm
Figure 3 Effect of O content and Pu/(Pu + U) on carbon potential and carbon monoxide partial pressure of carbide fuel.


62

Carbide Fuel

w

0

−45
−50

−55

:8
3

C
2

pm

00
p

60

]:
N
+

−60

985 K

−40

935 K

−35

[O


Carbon potential (kJ mol-1)

−15

−30

)C 1

8

0.

,M

U
2

−10

−25

%
U2 C -U
wt
+x
C
3
:4
)C 1

C3
0.8
U
2
M
%
t
2
0.
,
.
u
w
u0
(P
pm
(P
:8
C3
0p
M2
00
,
6
m
]:
pp
+N
00
[O

]: 2
N
+
[O
x
: 8 wt%
)C 1+
m, M 2C 3
8
.
6000 pp
U0
]:
N
+
2
[O
.
(Pu 0
U )C 1+x
: (Pu 0.55 0.45
MARK II
15 wt%
m, M 2C 3:
N]: 6000 pp
)C 1+x [O +
U
u
0.3
MARK I: (P 0.7

SS
31
6
Pu2C -PuC
3
+x

−5

−20

example, MC, MC2 (with high U), and M2C3 (with
high Pu). Solubility of nitrogen stabilizes UC2 and
PuC2 below their decomposition temperatures. Carbon activity of mixed carbide decreases with increase
in Pu content because of negative Gibbs energy
change for the reaction PuC þ U2C3 ! UC þ Pu2C3.
This also results in Pu enrichment of M2C3 phase.
Opposite effects of Pu and N þ O contents also mean
that Pu-rich fuel can accommodate higher nonmetallic impurities than U-rich fuel (Figure 4). The effect
of Pu and O þ N content on carbon activity can
also be seen in the phase diagram of carbide system
(Figure 5). Carbon precipitation results in shrinking
of the MC þ M2C3 phase field with increase in O þ N
content and its expansion with increase in the Pu
fraction.
Mixed carbide shows actinide segregation, with
a higher plutonium content in the M2C3 phase
than in the MC phase,20,21 as shown in Figure 6.
The segregation of actinides in the two phases
MC and MC1.5 reduces with increase in the nonmetallic impurity contents O and N.12 This can be

explained from the negative Gibbs energy changes
of the reactions Pu2C3 þ UN ! PuN þ U2C3 and
Pu2C3 þ UO ! PuO þ U2C3. Hence, the presence
of O þ N impurities stabilizes more U in M2C3 and
more Pu in MC. Segregation of actinides in two
phases is more pronounced in plutonium-rich fuel
and the effect reduces with increase in temperature
and also with increase in sesquicarbide fraction.

t%

the effect of O and N on the behavior of carbide fuels,
it is important to understand their interactions
with individual carbides. The structures of UC and
PuC are isomorphous with UO/PuO and UN/PuN,
and therefore these compounds show reasonable
solubility in the monocarbide phase. Though UO
and PuO are not stable compounds, O replaces
C and gets stabilized in the monocarbide lattice.
PuC can accommodate more oxygen ( 65 mol% PuO)
than UC ( 35 mol% UO). PuC–PuO and PuC–PuN
follow a near-ideal-solution behavior, whereas the
UC–UO system shows a negative deviation from
ideality and has limited solubility. UC–UN and
PuC–PuN form solutions over the complete composition range. UC–UN solution shows a slight positive
deviation from ideality on the UC side. Because of the
increased ionic character of nitrogen and oxygen
compared to carbon, addition of oxygen or nitrogen
impurity in UC, PuC, or MC results in slight contraction in the lattice with a small positive deviation from
Vegards law.18,19 As discussed in Chapter 2.04, Thermodynamic and Thermophysical Properties of the

Actinide Carbides, nonstoichiometry of carbides
decreases by substitution of C by O or N. MC and
M2C3 phases of Pu-rich mixed carbide fuel become
stoichiometric at high N þ O content (!6000 ppm).
Pu-rich hyperstoichiometric carbide fuel is biphasic
MC þ M2C3 in the temperature range of operation of
reactor. However, U-rich hyperstoichiometric fuel
may have three phases at high temperatures, for

PuC-Pu

−108
−110

UC-U

600

800

1000

1200

1400

1600

1800


2000

2200

Temperature (K)
Figure 4 Comparison of carbon potentials of carbide system for different Pu/(Pu + U) and (O + N) values.


Carbide Fuel

Pu/(Pu+U) = 0.2
Pu/(Pu+U) = 0.55
Pu/(Pu+U) = 0.7

C
0.0

63

1.0

X

M

1000 K
0.8

0.2
(U,

+C
)C 1.5
Pu

0.0

[O
pp
]=
m
] = 200 p
20
p
m
00
[O]
pp
=2
m
0

pp
m

0p
pm

0.4

0.2


0.2

[O

[O

]=

20

[O
]

00

=2

00

0p
pm
00
[O
]

=2

(U,Pu)


XC

(U,Pu) + (U,Pu)CNO

1.0

0.4

C

0.8

O+
)CN

0.6

0.6
Pu
(U,

(U,Pu)C1. 5 + 0.4
(U,Pu)CNO

XMN

0.6

0.8


0.0
(U,Pu)N

1.0

Figure 5 Effect of Pu/(Pu + U) and oxygen impurity on the phase diagram of (U,Pu)–C–N system.

PuC1.5

PuCNO

PuC1.5 PuCNO

PuC1.5
1.0

0.9

0.9

0.8

0.8

0.7

0.7

0.6


0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2
0.1
0.0
UCNO

C/M = 1.005
[N] = 1000 ppm
[O] = 1000 ppm
T = 1000 K

C/M = 1.005
[N] = 1000 ppm
[O] = 5000 ppm
T = 1000 K
UC1.5 UCNO


C/M = 1.005
[N] = 1000 ppm
[O] = 1000 ppm
T = 1500 K
UC1.5 UCNO

xPuC1.5 in MC1.5

xPuCNO in MCNO

PuCNO
1.0

0.2
0.1
0.0
UC1.5

Figure 6 Effect of O and N contents and temperature on segregation of U/Pu in (U,Pu)C and (U,Pu)C1.5 phases.

In case of M þ MC system, when metal phase precipitates, the metal phase is richer in Pu, but this
effect decreases with increase in temperature. In
this phase field, increase in nonmetallic impurities
results in a slight increase in segregation. Segregation
behavior is also important for back-end processing of

the mixed carbide by oxidation and dissolution in
HNO3. Excess oxidation of the carbide leads to the
formation of M3O8 phase, which has limited Pu solubility. This results in excessive segregation of plutonium in the MO2 þ x phase, giving problems during
dissolution in HNO3.



64

Carbide Fuel

3.03.3 Fabrication of Carbide Fuel
Fabrication of carbide fuel on commercial scale is a
difficult task and needs additional care because of its
pyrophorocity apart from high radio toxicity, and the
concern for criticality restricts the batch size. Moreover, carbide powders formed during carbothermic
reduction of oxides are prone to oxidation and hydrolysis. This requires high-purity inert-gas cover
(nitrogen or argon) in the fabrication line consisting
of glove boxes. The O and moisture content should
be less than 25 vppm (each) to minimize O pickup
during the fuel fabrication process and reduce the
possibility of any fire hazards due to pyrophorocity.
A mixed carbide fuel fabrication facility consisting of
a series of interconnected glove boxes, maintained
under once-through inert (nitrogen) cover gas is
shown in Figure 7. In 1960, when research on carbide
fuel was initiated, three different methods were followed, namely melting casting, metal hydriding–
dehydriding, and carbothermic reduction of oxide.
The carbide produced by the latter two techniques
is processed further by powder metallurgy techniques for the manufacture of fuel pellets.
3.03.3.1

Melting Casting

Melting casting process, also known as ‘deep casting,’

was followed during the early days of carbide fuel
development. This process was very reliable because
of some advantages over conventional powder metallurgy route; for example, the products were highly
dense and very pure. In this method, UO2 or U metal
chips with graphite are arc-melted and made into the

form of a button. This button is partially melted
many times for homogenization before finally melting and dropping into a mold. Pellets as large as
1.8 cm in diameter and 15 cm long could be made
by this technique. For larger pellets, the skull-melting
technique was followed, in which molten carbide
forms a shell-type cast in a copper mold and acts as
its own containment. Melting casting route results in
large-grained materials compared to that obtained by
powder metallurgical methods.22 Melting casting
method is, however, uneconomical due to the high
cost of metal fabrication.

3.03.3.2

Hydride/Hydrocarbon Route

This method is followed for small-scale production of
high-purity carbides, where the metal hydride reacts
with graphite. Actinide carbides MC and M2C3 can
be prepared from a mixture of hydride and graphite:
PuH2 þ 0:85C ! PuC0:85 þ H2 ðgÞ

½IŠ


The reaction occurs under vacuum at high temperature (1800–2600  C) and this is followed by sintering
at 1500  C for 4 h.23,24
For making UC, the reaction between uranium
metal and a hydrocarbon gas is carried out at
600–800  C, which yields a fine and easily sinterable
product. However, the reaction needs to be controlled carefully, as continued flow of methane produces UC2. Hence, the methane flow must be judged
fairly accurately to produce only UC. In this process,
a fine powder of UH3 is prepared by reacting bulk
uranium metal with hydrogen at 200–275  C, and
the powder formed is decomposed above 430  C to
produce fine uranium powder. This metal powder
is then reacted with methane (or propane) at
600–800  C25–27 to produce UC.

3.03.3.3

Carbothermic Reduction Route

Uranium monocarbide is produced by carbothermic
reduction of UO2 and carbon following the reaction

Figure 7 Carbide fuel fabrication facility showing chains
of interconnected glove boxes maintained under high-purity
inert gas.

UO2 þ 2C ! UC2 þ CO

½IIŠ

UO2 þ UC2 ! 2UC þ CO


½IIIŠ

In this process, UC2 may be formed as an intermediate product. In this method, a homogeneous mixture
of UO2 and carbon is blended together and the mixture is compacted at 300–600 MPa pressure along


Carbide Fuel

with a suitable organic binder. The pellets are heated
for 2 h at 1700  C in a vacuum induction furnace.
The reacted pellets are ground to a fine powder for
compaction.
Uranium–plutonium-mixed carbide is also prepared by carbothermic reduction of UO2–PuO2.
The physical state and homogeneity of the mixture
influences the reaction rate and the quality of the
final product. With increase in specific surface area,
the reaction rate increases but reaction temperature
decreases, which reduce Pu loss by volatilization.
Agglomeration of powders during blending should
be avoided to ensure good microhomogeneity of the
mixture. The reaction is controlled by monitoring
the evolved CO gas during reaction using an IR
detector in the vacuum pump exhaust. Direct carbothermic reduction of UO2 þ x þ PuO2 mixture
does not give the desired result due to formation
of CO2 at low temperatures.28 The resultant reaction is
ð1 À zÞUO2þx þ zPuO2 þ ð3 þ n þ 0:5 Â ð1 À zÞÞC
! ð1 À 2nÞðU1Àz Puz ÞC þ nðU1Àz Puz Þ2 C3
þ 2CO þ 0:5 x ð1 À zÞCO2


½IVŠ

When the starting material is UO2 þ x, the reaction is
complicated by the uneven distribution of plutonium
in the two carbide phases and the formation of
CO2 and CO. Therefore, for a better control on the
reaction mechanism, UO2 þ x could be first reduced
to UO2 by heating it with carbon in vacuum at 10–2
torr at 850  C for 30 min. A mixture of UO2 and
PuO2 is then dry-blended with carbon. The reaction
is complete after heat treatment for 1–1½ h at a
temperature of 1400–1450  C for smaller pellets
and 1550–1600  C for bigger pellets. Carbothermic
reduction temperature and atmosphere (vacuum or
inert) are important parameters, as they decide the
completion of the reduction reaction and the extent
of residual O. It was observed at the Institute for
Transuranium Elements (ITU), Karlsruhe,29 that
carbothermic reduction under vacuum (1 Pa) yielded
better results compared to that carried out under Ar
atmosphere and the reduction temperature was
found to be about 100  C less. The nitrogen impurity,
the source of which is the Ar cover gas, is also less in
the final product. Carbothermic reduction temperature also depends to a great extent on the Pu content
of the material. While fixing the C content in the
oxide–graphite mixture for carbothermic reduction,
the effect of the binder–lubricant, used during pressing of MC powder, should also be taken in to

65


account. Figure 8 shows the process flow sheet for
the fabrication of high-plutonium (>30% PuC)
mixed-carbide fuel pellet by the conventional powder pellet route. In a recent development at BARC,
India, for the fabrication of high-plutonium mixedcarbide pellet for the FBTR, the powder handling
time during grinding/mixing of feed powder and ball
milling of carbide powder was considerably reduced
by incorporation of a high-energy ball mill and
attritor. This has considerably reduced the total
time for production of a batch and also reduced the
radiation dose for the working personnel. The process control steps for the fabrication are indicated in
blue in Figure 8.
Pu evaporation loss is undesirable, as it may lead
to criticality risk by accumulation in the colder
regions of the furnace. This may also affect fuel
economy. Pu loss can be minimized by increasing
the CO partial pressure. Richter et al.30 found that
increasing the CO pressure from 0.8 to 7 Pa reduced
Pu loss from 7.2% to 0.2%.
A possible process simplification,31 ‘the one-step’
conversion of actinide oxalates precipitated from
nitrate solution (PUREX process) into carbide, was
studied in CEA, France, as a part of the carbide fuel
development program for the GEN-IV reactor. This
work focused on carbothermic reduction of actinide
(III) oxalate. As a developmental effort, neodymium
was chosen as a surrogate of Pu, instead of cerium,
Neodymium in oxidation state þIII, has characteristic violet color, which allows easy visual check on
homogeneity and precipitation.
3.03.3.3.1 Direct pressing method


The conventional powder metallurgy route of carbide fuel fabrication has certain disadvantages, for
example, after the carbothermic reduction when the
carbide clinkers are hammer-milled prior to compaction and sintering, fine carbide powder is generated
which picks up O impurity from the cover gas. Also,
handling of the fine powder is risky because of its
pyrophorocity.
In the ITU, Richter et al.30 developed the ‘direct
pressing’ method, in which carbide clinkers after
carbothermic reduction in a predetermined shape
(40–50% TD) are directly pressed to make pellets
of density 65–75% TD, which are subsequently sintered to obtain the final product.
The advantages of this method are as follows:
 Handling of fine carbide powder is reduced, which
eliminates the chances of O pick-up.


66

Carbide Fuel

Analysis:
Pu, U, C, impurities

UO2
PuO2

Weighing

Tablet


Grinding and mixing

Carbothermic
reduction (1485 ЊC)

Analysis:
U, Pu, O, N, C,
impurities
Phase content

Carbon

Crushing
MC clinkers

Zinc behanate: 1 wt%
Naphthalene :1 wt%

Sintering

Dewaxing

Final compaction

Ball milling

Precompaction

Granulation


U, Pu, O, N, C, impurities,
total gas
Phase contents
Pu distribution
Dimension and density

Pellet inspection
Physical and chemical

Accepted sintered pellets
Reject
recycling

Figure 8 Fabrication flow sheet for production of plutonium-rich mixed carbide fuel for the Indian fast breeder test reactor.

 Self-ignition of fine powder in case of accidental
leakage of air is minimized.
 Production of dust is reduced.
 No impurities are picked up from grinding in the
ball mill during comminution.
 The material flows easily for compaction.
 It takes less time for processing and hence is more
economical.
The direct pressing method has been used for mixed
carbide fuel fabrication method.
3.03.3.3.2 Sol–gel (wet) route

The internal gelation route (wet method)32,33 of particle fuel fabrication is dust-free and provides ceramic
fuel with perfect homogeneity of Pu and U on an
atomic scale compared to the dry powder route. In

addition, the wet route can be successfully used as a
simple, economical, proliferation-resistant, radiologically and critically safe, and remotely controlled fuel
production step by recycling the minor actinides (MA)
from reprocessing step as part of closing the fuel cycle.

The sphere-pac fuel pins prepared at Paul Scherrer
Institute (PSI)34 for the AC-3 irradiation test at FFTF
adopted the wet route for production of sintered fuels
of low density (80–85%) with high open porosity. In
this method, an acid-free, concentrated feed solution of
uranium and plutonium nitrate is mixed with an aqueous solution of hexamethyletetramine (HMTA) urea
and dispersed carbon black. The mixture is cooled and
dropped into a column of hot silicone oil (for making
large particles of $800 mm) or injected into a stream
of hot oil (for making smaller fuel particles of $80 mm).
The temperature rise in the droplets leads to the
decomposition of HMTA and urea to form ammonia,
which, in turn, precipitates ammonium diuranate and
plutonate within the droplet, forming a solid sphere.
The details of the fabrication procedure may be
obtained in Blank5 and Geithoff et al.,6 and is shown
in Figure 9 along with the other methods. About 27
sphere-pac pins, which included spare and archive
pins, were fabricated at PSI for loading in the AC-3
cluster for irradiation in FFTF. The fuel column
density ranged from 78.8% to 80.3%.


Carbide Fuel


Dry route

Wet route
Standard
process

Conversion
to powder

Blending

Conversion
to powder

Blend: Add carbon
Briquette

Direct
pressing
U3O8
PuO2+C

Press
Compaction
Disks

Carbothermic
reduction

Carbothermic reduction


Powder
<63 μm

Press
Compaction
Disks

Vacuum inert
gas or nitrogen
(1400 –1700 ЊC)

Plutonium
nitrate

Uranyl
nitrate

Mix with HMTA solution and
add carbon
Form sphere
(gelation)

Solvent wash

Crush
and
Mill

Powder

preparation

Ammonia wash
Hammer mill
comminution
Ballmill

Dry

Calcine
Pressing
Pelletiz

Carbide synthesis and sphere fabrication

Plutonium
nitrate

Carbide synthesis

Uranyl
nitrate

67

Repressing
recompact
Reaction sinter

Sintering


Press slug

Vacuum inert
gas or nitrogen
(1600–1800 ЊC)

Sieving

Granulate

De-bond
Sinter
Centerless grid
Inspect and size

Pellet fabrication

Dimensions
Press pellets

Pellet control

Vibro-fill pin
Density analysis

Fuel pin preparation

Load, weld final control


Load pin

Figure 9 Comparative flow sheet for conventional carbothermic reduction, direct pressing method, and wet method for
fabrication of carbide fuel pellet.

3.03.3.4

Quality Control

Chemical and physical quality control and characterization of the mixed carbide fuel is essential to meet
the fuel specification and is carried out at different
stages of the fabrication flow sheet. Chemical quality
control involves the estimation of U and Pu, C, O,
and N, as well as phase identification and their
amounts by X-ray diffractometry. The sintered pellets
are also made to undergo various types of physical
quality control, such as visual inspection of the pellet,
estimation of density/linear mass, autoradiography

(for Pu homogeneity), microstructural characterization of fuel, and end plug welds. X-ray gamma autoradiography (XGAR) for checking the correctness of
the loading sequence of internal components of the
pin, and passive gamma scanning for ensuring proper
enrichment of the fuel pin, are some of the techniques
used for production of plutonium-rich mixed carbide
fuel for the FBTR in India.35 The different chemical
quality control steps adopted in the fabrication of
carbide fuel at BARC, Mumbai, India, are shown in
the flow sheet (Figure 8).



68

Carbide Fuel

3.03.3.4.1 Chemical quality control method
3.03.3.4.1.1 Uranium and plutonium analysis

TIMS and ICP/MS are used for isotopic determination of U and Pu. Isotopic dilution mass spectrometry
(IDMS), X-ray fluorescence spectrometry (XRF),
gravimetry, redox methods with dichromate titration,
ceric titration, or high-precision titration, and coulometry are some of the methods used for U and Pu
determination. Gravimetry is a fast method for the
approximate determination of U þ Pu of the carbide
samples by studying the weight change due to complete oxidation of the sample and then reducing the
oxides to O/M ¼ 2.0 using the CO/CO2 equilibrium
method.36 IDMS and the redox method are two
important methods for chemical analysis of samples
for U and Pu determination. IDMS method is free
from chemical interference from other elements
but the redox method is more precise. All redox
methods are based on U(VI)/U(IV), Pu(III)/Pu(IV),
and Pu(VI)/Pu(III), Pu(IV) systems using different
reducing/oxidizing agents. For chemical analysis, the
carbide sample is first converted into its oxide by
heating the samples in air at 500  C, followed by dissolution in HNO3–HF and then removal of fluoride by
evaporation. Direct treatment of carbide with HNO3
does not give a clear solution easily. Moreover, some
organic species such as oxalic acid and mellitic acid
(benzene hexacarboxylic acid) are reported to form
during direct dissolution in HNO3. These species

affect the electrochemical determination of U/Pu.
Different methods have been tried for the dissolution
of UC and (U,Pu)C sintered pellets for destructive
quantitative analysis. Chander et al.37 have reported
that refluxing of carbide with an 18 M H2SO4–15
M HNO3 (1:1) mixture results in complete dissolution
in 1 h. This process also excludes the need to convert
carbide into oxide, thus reducing the time for analysis
and the risks involved in oxidation of pyrophoric carbide at high temperature. Moreover, organic species
such as oxalic and mellitic acids are either destroyed
or do not interfere in the analysis process.
Earlier, uranium determination in solution, in presence of Pu, was done by the modified Davies and
Gray method,38,39 in which U is reduced to U(IV) in
a !10 M H3PO4 medium with Fe(II) and then titrated
with K2Cr2O7 solution. However, during processing of
these Pu-containing analytical wastes, the strong complexing properties of phosphates posed problems. So
another redox method was developed in which TiCl3 is
used as reductant in H2SO4 medium (!6 M), containing millimoles of HNO3.40 All the U is brought to the

U(IV) form by adding excess of TiCl3. The excess
reductant is oxidized by HNO3. The HNO2 produced
in situ oxidizes Pu(III) to Pu(IV), thereby eliminating
plutonium interference in uranium determination.
Sulfamic acid is added to destroy any excess HNO2
and the acidity is brought down to 3 M by dilution with
water. A solution of Fe(III) is added to oxidize U(IV) to
U(VI), and the Fe(II) thus produced is titrated with the
K2Cr2O7 solution. The advantage of this method is that
U can be determined in the presence of Pu and Fe. End
point detection of the titration can be done by

the visual indicator method, amperometric method,
or potentiometric method. Biamperometry is more
sensitive than potentiometry for end point detection;
however, better automation is possible with the potentiometric technique.
For Pu determination in the presence of uranium,
an aliquot is taken in 1 M H2SO4 and an excess of AgO
is added to oxidize all the plutonium in the solution
to Pu(VI). The excess AgO is reduced by adding sulfamic acid, which does not reduce Pu(VI). A known
excess quantity of standardized Fe(II) is added to
reduce Pu(VI) to Pu(IV) and the unreacted Fe(II) is
titrated with standard K2Cr2O7 solution.41
For combined determination of U þ Pu in mixed
carbide or determination of either U or Pu in their
respective pure carbides, a common titrimetric method
used in nitric acid medium is carried out by reducing
U/Pu with titanium(III) solution in the presence of
sulfamic acid, and titrating with cerium(IV) solution42; however, Fe, V, Cu, and Mo, if present, will
show interference. Therefore, IDMS in combination
with the redox method gives more reliable results.
For the determination of U and Pu in powder and
pellet forms, a dry and quick method based on XRF
is followed. First, standard powder samples of mixed
plutonium–uranium carbides are prepared and chemically analyzed. Then a calibration curve is obtained
by plotting the intensity ratios of Pu and U of these
standards against the chemically determined plutonium content. This calibration curve is used for
subsequent determination of Pu/U ratio in unknown
samples. The scatter in the results of determination of
Pu and U by XRF were found to be within Æ0.5 wt%.
3.03.3.4.1.2


Analyses of nonmetallic elements

Commonly, the techniques used for assay of carbon
involve oxidation of carbon to carbon dioxide by
igniting the sample in a stream of oxygen and then
flowing the gases generated, passing through the
Schutze oxidizing reagent (iodine pentoxide on silica


Carbide Fuel

gel) to ensure quantitative oxidation of any CO, collecting and weighing the CO2, and CO2 determination
directly by manometry,43 IR spectrometry,44 conductometry,45 or gas chromatography using a thermal
conductivity detector,46 or indirectly after conversion
of carbon dioxide to methane with subsequent detection by gas chromatography using Flame Ionization
Detector (FID).47,48
For the determination of nitrogen and oxygen
impurities, the sample is fused in a graphite crucible
at very high temperature along with a nickel flux. In
this reducing atmosphere, the gases evolved are carbon monoxide and nitrogen. A gas chromatograph is
used for separating the gases. First, carbon monoxide is
oxidized to carbon dioxide with the Schutze reagent
(measured by an IR detector), which is then removed
by ascarits and anhydrones. Then the pure nitrogen
is measured by thermal conductivity.49 The uncertainty of the methods is Æ5% for oxygen and nitrogen impurities and Æ2% for carbon.
The metallic impurities are normally determined
by ICP emission spectroscopy. A quantitative estimation of MC and M2C3 contents of mixed carbide
samples can be made by the X-ray diffraction line
profile.50 The strongest reflections for MC (200) and
M2C3 (310) are chosen as analytical lines and the

K-values of these lines are calculated from the crystal
structure, Lorentz polarization, multiplicity, temperature, and absorption factors. The K-values, I (measured
integrated intensity), and V (volume fraction) of the
phases are related by the following equation, from
which the volume fraction of the MC and M2C3 can
be calculated (precision Æ5% of the absolute value):
VMC
KM C IMC
¼ 2 3
VM2 C3
KMC IM2 C3

½1Š

69

3.03.3.4.2 Physical quality control
3.03.3.4.2.1

Fuel pellet inspection

Sintered fuel pellets are subjected to visual inspection
for any surface defects. Each pellet is inspected visually and compared with a standard one for classification for acceptance and rejection. The type of pellet
defects are end chipping, surface cracks, etc. Dimensional measurements such as length and diameter are
also carried out to estimate linear mass (g cmÀ1) and
to eliminate oversized and undersized pellets. An
automated pellet inspection system supported by a
software may be employed for this purpose, which
reduces the time of inspection considerably.
Ceramographic analysis: Microstructural analysis of

carbide pellets are carried out to measure grain size,
cracks, pores, and inclusions. Typical microstructures
of mixed carbide pellets containing 70% and 20%
PuC are shown in Figure 10. Quantitative metallography will reveal the extent of sesquicarbide phase
present in the sample and this data could be correlated to X-ray diffractometry results. X-ray diffractometry, however, cannot reveal the second phase
M2C3 if it is less than 5%, which, however, can be
detected by metallography. M2C3 appears as a bright
and discontinuous phase mostly located at the triple
points of the grain structure.
Metallographic sample preparation of sintered
mixed carbide fuel pellets needs utmost control of
atmosphere during etching. Mixed carbide pellets
are prone to oxidation because of their highly corrosive nature. The surface of the polished sample
gets tarnished before it can be etched because of the
presence of O or moisture in the cover gas atmosphere in the glove box. Oxidation of the polished
carbide pellet surface is indicated by the coloration
of the surface.

20 mm

Figure 10 Photomicrograph of mixed carbide fuel with (a) 70% PuC and (b) 20% PuC showing bright/white areas of
sesquicarbide phase. Reproduced from Matthews, R. B.; Herbst, R. J. Nucl. Technol. 1983, 63, 9–22.


70

Carbide Fuel

3.03.3.4.2.2 Fabrication of fuel pins


Fabrication of carbide fuel pins consists of loading the
components sequentially as per the fuel pin design.
The sequence of loading the cladding tube starts with
the lower reflector, followed by bottom insulator
pellets, fissile carbide fuel pellet stack, upper insulator pellets, upper reflector, plenum spring, and,
finally, the plenum spacer tube. The closure of the
cladding tube is carried out at the top end cap using
tungsten inert gas (TIG) welding in a helium-filled
welding chamber. Helium gas acts as an inert cover
gas for the carbide fuel pellets and also as thermal
bond between the fuel pellet and the cladding tube
ensuring heat energy transfer to the coolant. In case
of any faulty welding, end cap should be replaced and
rewelding done. The pins are inspected for dimensional tolerance and bowing. X-radiography, dyepenetrant test, and He leak testing are carried out
to check the integrity of the fuel pin.
The technique of fabrication of sodium-bonded
carbide fuel pin is similar to that of sodium-bonded
metal fuel pin. The environment of the sodium filling
box should be kept free of oxygen or moisture
(<50 ppm O and moisture) to avoid any deleterious
effect on fuel performance because of possible
O pickup by sodium and oxidation of sintered carbide
fuel pellets. The detailed procedure of making sodiumbonded pin is given in Burkes et al.51 In brief, the
procedure of making sodium-bonded pin consists in
loading the carbide pellets in the cladding tube filled
with liquid sodium and allowing the pellets to settle
down by displacing liquid sodium up the cladding tube.
Further heating after loading of fuel pellets is necessary
to remove trapped gases. Care should be taken to avoid
the formation of voids or pockets of trapped gases

between the fuel and the cladding inner surface. The
reason for the formation of such pockets of trapped
gases is the nonuniform flow or flow of sodium in
preferred channels due to nonuniform wetting of the
pellet or the cladding inner surface by liquid sodium.
The upper end plug is subsequently closed. The bond
quality of the sodium-bonded pin is assessed by radiography or eddy current test.

3.03.4 In-Pile Performance
3.03.4.1

Introduction

Studies on in-pile fuel behavior of carbide fuel help
in controlling the fuel operating parameters under
normal or steady-state condition during the useful
life of fuel pin and in setting the new operating

parameters under different conditions. The in-pile
fuel behavior depends to a large extent on the fuel
operating temperature, which in turn depends on the
type of pin or bond concept, namely, He-bonded fuel,
Na-bonded fuel, or He-bonded particle fuels.
A large number of mixed carbide fuels have
been irradiated in countries such as the United
States, France, the United Kingdom, Germany and
Switzerland, and at the ITU, Karlsruhe, Germany.
The summary of all such test pin irradiation results
has been in given Matthews and Herbst,2 Ganguly
et al.,50 Bart et al.,52 European Institute of Transuranium Elements,53 and Kummerer.54 Test pin irradiation was also carried out in Japan at their research

reactors JRR-2 and JMTR at JAERI and subsequently
in the experimental fast reactor (JOYO). In India, fast
reactor program started with the commissioning of the
FBTR, where the driver fuel of high plutonium (PuC–
70% MKI) has seen a burnup of 16 at.%. USSR, too,
has test-irradiated carbide fuel pins in their BR-10
reactor and had a full carbide core (burnup up to
5 at.%), and subsequently experimental pins of both
types, He-bonded and Na-bonded, were irradiated in
BOR60 up to 10 at.% burnup. In all these test irradiation programs, mixed carbide powder was used to
make fuel pellets or vibrocompacted fuel, except for
the Swiss program where sphere-pac fuel pins were
used. The fuel was made at the Paul Sherrer Institute
(Switzerland) and irradiated in the FFTF in the
United States. Thirty-one test pins have been irradiated as part of this joint venture.
The performance analysis of the mixed carbide
fuel can be best understood on the basis of their
burnup period, the structural changes occurring during the burnup and subsequently the swelling of the
pin. The basis of such analysis applies to all types of
pin concept, that is, helium-bonded pins, sodiumbonded pin, or the helium-bonded sphere-pac fuel.
3.03.4.1.1 Burnup

The burnup period could be best understood by
dividing it into three stages based on the fuel surface
temperature which changes with the change in fuel
burnup. The schematic diagram of the changes in
surface temperature and the fuel structural changes
are given in Figure 11.
Stage A: At the beginning of life when the virgin
fuel is introduced into the reactor, it undergoes

cracking. This results in the movement of the fragmented fuel toward the cladding surface due to vibration induced by the coolant flow. The fuel–clad gap
partly reduces but does not close. The fuel surface


Carbide Fuel

FB (II)

ns
De

IS
ta
e
bl

te

ra

el
fu

w

Lo

of

lling

swe

Fuel surface temperature Ts

TS(A)

f
te o
h ra
II Hig

ific

ati

on

B (II)

sw
g
lin
el

TS(C)

A

t


B(I)

FB (I)

C

EOL

Figure 11 Variation of fuel surface temperature with
burnup for a He-bonded mixed carbide fuel; A, B, and C
represent three burnup period. Reproduced from
Blank, H. Material Science and Technology, Nuclear
Materials, Pt I; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.;
VCH Publisher: New York, 1994; Vol. 10A.

temperature reaches its maximum as indicated in
Figure 11. The first stage may last for several days
from the beginning of life.
Stage B: In the second stage, unrestrained free
swelling, crack healing, or resintering starts. Resintering occurs for thermally unstable fuels, which causes
an initial increase in gap size resulting in the rise in
the fuel surface temperature. This occurs in a very
short period in the beginning of stage B. During this
stage, free grain boundary swelling occurs and crack
healing starts, which results in the relocation of fuel
fragments further from the center of the fuel to the
periphery. This causes fuel–clad gap closure, increase
in heat transfer coefficient, and decrease in fuel surface temperature (Ts). The decrease in fuel–clad gap
has a limiting value depending upon the fuel surface
roughness and the cladding inner surface roughness.

The decrease in cracks at the center transforms into
porosity and the formation of wedge type of cracks
begins, and when the fuel–clad gap reduces to the
level of surface roughness of the fuel and clad inside
surface, fuel–clad deforms elastically at first. Stage
B ends after a burnup of approximately 3 at.%, which,
however, depends on the type of fuel; for example,
fuel with a high plutonium and oxygen content is less
stable, as it densifies easily and swells more.
Stage C: This stage starts after fuel–clad gap closure when a steady minimum surface temperature is
reached and it continues till the end of fuel life. For
carbide fuel, this could be up to 15 at.%. At this stage,
more homogeneous fuel–clad interaction starts.

71

Structural zones: The cross section of a dense nonoxide ceramic can be divided into four structural
zones (Figure 12) as given below. These four radial
zones were first conceived by Colin et al.,3 and were
subsequently modified by Blank.55 These zones
become well defined when the fuel sees a burnup of
1.3 at.%.
Zone-I represents the porous fuel at the center
having high temperature and helps in the release of
fission gases. The fuel is also very soft because of the
high temperature. The presence of Zone-II depends
on the linear heat rating and the composition of the
fuel. In carbide fuel, this zone will be visible only in
an LHR of more than 1200 W cmÀ1 (for sodiumbonded fuel). However, for a fuel with high oxygen
content (>1000 ppm), this zone can exist even below

1000 W cmÀ1. Zone-III contributes to high microscopic swelling. Based on linear heat rating, Zone-II
may not exist at all and Zone-I may directly connect
to Zone-III (for stable fuel with linear heat ratings of
1000 W cmÀ1). Zone-IV is the coolest zone of the
fuel cross section, which is next to the fuel cladding.
The as-fabricated (with porosity 10%) structure of
the fuel is more or less retained in this zone.
3.03.4.1.2 Swelling

Swelling of fuel is a manifestation of the effect of
irradiation. As the fuel undergoes irradiation, there is
decrease in density and increase in fuel volume,
which can be assessed by measurement of dimension
of the fuel pellet and clad diameter. The factors that
contribute to swelling are (i) fission gases, for example, Kr and Xe (mostly Xe) having high yield ($25%)
and low solubility; and (ii) volatile fission products
such as Cs, Rb, I, and Br. The contribution to geometric swelling can be assessed by quantitative metallography, SEM, or EPMA. Swelling performance of
dense fuels depends on the fuel operating temperature, smear density, fuel–clad gap, and burnup. The
difference in the operating temperature range for
helium-bonded and sodium-bonded fuel pin is that,
for the former, the working range is at the beginning
of the irradiation as indicated by ‘A’ (Figure 11)
when the fuel clad is still open, and the range changes
to ‘C’ when the gap is closed. The burnup range is
3 < F < 12 at.%. Sodium-bonded fuel is a relatively
cool fuel compared to helium-bonded fuel. It swells
very slowly and comes in contact with the cladding
only after about 11 at.% burnup. However, it has
been found that, beyond a linear heat rating of
1000 W cmÀ1, the cool fuel concept is not valid. For

helium-bonded fuel, it starts with a short period of


72

Carbide Fuel

Cladding material

Fuel

Initial fuel–clad gap

Fragmented fuel
Reduced fuel–clad gap
(b) Beginning of life stage ‘A’

(a) As-fabricated fuel

Porosity

IV
III
I

Zone of transition from
low to high swelling
Wedge-type
crack


Swelled fuel

No fuel–clad gap
(c) End of stage B, Zone I: cracks transformed to porosity, wedge cracks
in Zone III and Zone IV
Figure 12 Cross-section of helium-bonded carbide fuel pin irradiated to 11.2 at.% burnup indicating three structural
zones: (a) as-fabricated fuel, (b) high-density central rich in ‘Pu’ and metallic fission product, and (c) porous finegrained with coarse and fine grain. Reproduced from Blank, H. Material Science and Technology, Nuclear Materials, Pt I;
Cahn, R. W., Haasen, P., Kramaer, E. J., Eds.; VCH Publisher: New York, 1994; Vol. 10A.

T/Tm > 0.5, and after the gap closure the fuel temperature drops and major part of the fuel operates below
0.5. Only the inner part operates at a temperature >0.5.
For helium-bonded pin, the thermal stability limit of
the as-fabricated fuel can be matched to the initial fuel
operating temperature and some advantages of the hot
fuel can be achieved. For the purpose, the porosity in the
central zone can be increased at the cost of fuel–clad gap,
thus enhancing the initial short period ‘A.’ The porosity
for gas release takes place at the later period of ‘C.’ The
outer zone in contact with the clad remains cool during
the period ‘C’, and the swelling is partly accommodated
into the porosity of the as-fabricated fuel.
3.03.4.1.3 Performance of Na-bonded and
He-bonded fuel pins
3.03.4.1.3.1 Sodium-bonded pin

For the sodium-bonded fuel pin, the fuel density is
higher than that of He-bonded fuel and the fuel–clad

gap is more. So, under normal operating conditions
for a linear heat rating <1100 W cmÀ1, the structural

changes for the sodium-bonded pin are more or less
constant and predictable. In the beginning (stage A),
the virgin fuel fractures, reducing thermoelastic
stresses and drastically changing temperature gradient of the fuel. The chips from fractured fuel pin may
relocate causing FCMI. However, the concept of
using a shrouded tube to collect the small fractured
pieces to reduce FCMI is not advisable, as it may
cause unnecessary complications in fuel pin design.
There is no abrupt change in fuel temperature in this
concept. The decrease in thermal conductivity with
burnup is compensated by a slow decrease in linear
rating. Since the fuel density is high for Na bonding,
the gas release is less. FCMI can be avoided till end of
life by proper selection of initial fuel–clad gap.
The sodium-bonded pin has some inherent disadvantages; for example, the long residence time of


Carbide Fuel

fuel may result in loss of Na bond, which results in
increase in temperature, more gas release, and further
reduction of bond quality. Also, Na bond acts like
a medium for transfer of C from the fuel to clad,
causing clad carburization. Fabrication of the Nabonded fuel pin is more expensive than that of the
He-bonded pin. This will also affect the cost of
reprocessing because of the presence of sodium in
the feed stock. However, with the large fuel–clad gap,
which is filled with sodium, FCMI can be avoided
altogether in the case of a sodium-bonded pin. Also,
the low temperature of the sodium-bonded pin keeps

the swelling rate of the fuel low.

73

stresses generated on the clad due to swelling be
circumferential to avoid any failure. Clad carburization is a life-limiting factor, as it makes the clad
surface hard and brittle and cracks form and propagate easily. FCMI can be independent of the fuel–
clad gap size when the rate of free swelling is very
high (3–10% per at.% burnup) for center temperature as high as >2000  C. The high swelling cannot
be fully accommodated by the fuel–clad gap for a
high-burnup fuel. Increasing the cladding thickness
can make the clad stronger to accommodate restraint
swelling.
3.03.4.1.4 Irradiation creep

3.03.4.1.3.2 He-bonded pin

The performance of He-bonded fuel to a great extent
depends on the design parameters: namely, fuel–clad
gap, smear density, type of clad, pin diameter, linear
power, and burnup. He bonding is the most preferred
bonding concept of carbide fuel, partly because Na
bonding, apart from the cost, may deteriorate with
burnup and hence the bond quality. The in-pile performance of the He-bonded pin depends to a large
extent on the porosity of the as-fabricated fuel, which
undergoes structural changes when it passes through
two stages of burnup, ‘A’ and ‘B.’ The structural
changes at the end of stage B decide the safe burning
of the fuel in stage C. For He-bonded fuel, the initial
temperature rise is much higher at the beginning of

life because of the lower thermal conductivity of the
He bond compared to Na bond. The lower thermal
conductivity of the He bond also warrants the reduction of the fuel–clad gap compared to the Na-bonded
fuel. As a consequence, the fuel density for Hebonded fuel is lower and a fabrication porosity of
about 15% is recommended (equivalent to 85% pellet density).
The fuel–clad gap closure in the early burnup
period results in lowering of the fuel temperature,
and free swelling changes into restrained swelling
under the contact pressure developed at the clad–
fuel interface. Hence, the mechanical properties of
the fuel and clad (creep, fracture toughness) to a great
extent predict the fuel behavior during the remaining
period of burnup. It has been observed that the Hebonded pin can be operated safely up to 15 at.%
burnup (peak burnup 20 at.%). The irradiation experience with various design parameters has been summarized by Matzke.29
Cladding breaches due to FCMI or FCCI can be
due to loss of ductility of the clad, carburization of
the clad, or fuel swelling. It is desirable that the hoop

Creep is a time- and temperature-dependent deformation mechanism under stress. The source of stresses in the fuel under irradiation is the pressure
generated by fission gases produced within the
fuel. The growth of fission gas bubbles within the fuel
results in swelling of the fuel by creep deformation.
The fuel–clad gap provided in the fuel pin design is
utilized to accommodate this swelling known as ‘free
swelling.’ However, after the fuel–clad gap closure,
free swelling is restrained and a back stress is generated
by the clad on the fuel. This results in restrained
swelling of the fuel, which is accommodated within
the available pores of the fuel by creep deformation.
Carbide has a close-packed fcc structure of the NaCl

type. Unlike close-packed metals, deformation of carbide requires much higher stresses because of the
strong covalent bonding existing between the metal
atom and the carbon atom. It also needs sufficient
thermal excitation to leave the lattice site. Accommodation of fuel swelling (restrained) within the porosity
of the ceramic carbide fuel results in the extension of
the primary creep region and this causes an increase
in the strain rate before it reduces and attains the
steady-state creep. The tertiary creep region lies outside the life span of the carbide fuel and hence is not
important.
Two types of creep deformation are operative in
the fuel: temperature-dependent thermal creep and
radiation-induced or radiation-assisted creep. According to Seitzer et al.56 and Dienst,57 thermal creep starts
at 1000  C and dominates over radiation-induced
creep. However, it was also inferred that the relatively
flat and lower temperature profile in carbide due to
higher thermal conductivity compared to mixed oxide
and the high neutron fluence provides sufficient means
for the reduction of stresses in the fuel by radiationinduced creep. The details of plasticity of carbide can
be obtained in the reviews by Routbort and Singh58 and


74

Carbide Fuel

Matzke.29 The steady-state creep curve is a function of
stress (s), temperature (T ), composition, grain size,
and impurity content in the fuel. The steady-state
creep rate can be expressed by the relation
e¼ Ad Àm sn expðÀDH =RT Þ


½2Š

where A, m, and n are constants for a particular composition and structure. For hyperstoichiometric fuel, the
constants A, n, and DH were estimated by Hall59 and
found to be 1.57 Â 1011(hÀ1), 2.4, and 506 (kJ molÀ1).
Blank60 recommended that the above equation and the
values of the constants be different for different materials and should not be used as an input data for a
model. These data are specific to the fuel composition
and type. Apart from the porosity correction and presence of a second phase (higher carbides), inhomogeneous deformation across the cross section of the fuel
may also change the mechanisms of the creep deformation. Figure 13 shows the steady-state creep rate of
UC and MC1 þ x at a stress of s ¼ 20 MPa. These data
have been taken from the assessment of Hall59 and
Tokar.61
As indicated by Figure 13, for a small portion of
fuel, that is, fuel center to periphery, there will be creep.
Otherwise, the fuel will mostly behave like a brittle
material. The creep data presented in Figure 13 represent those of a high-density material and need to
be corrected for porosity if they are to be used for
any other material. The horizontal line shows the

temperature-independent irradiation-induced creep
for hyperstoichiometric mixed carbide fuel at a stress
of 20 MPa. The irradiation-induced creep, however,
is a function of fluence. The temperature of operation
of mixed carbide fuel under steady-state condition is
also shown in this figure by a horizontal line indicated
by ‘Ts’ (fuel surface) and ‘Tc’ (fuel center).
For mixed carbide fuel with high plutonium content (>55% Pu), no thermal or irradiation creep data
is available in the literature. However, Tokar et al.62–64

estimated, qualitatively, the creep behavior of high
plutonium carbide fuel from hot hardness data by
drawing an analogy of creep and those data. Sengupta
et al.65 also measured the hot hardness data of carbide
as a function of Pu content and sesquicarbide content.
Hot hardness data are also useful in predicting the
FCMI behavior of the fuel when the fuel–clad gap is
closed. Hot hardness of mixed carbide fuels (PuC: 70
and 55%) were measured in a high-temperature
microhardness tester using Vickers pyramid indenters.65 The result (Figure 14) showed decreases in
hardness with increase in temperature, with MKII
(55% PuC) having higher hardness at all temperatures. MKI (70% PuC) shows a sharp decrease in
hardness at 1123 K ($0.52 Tm; where Tm is the solidus temperature), indicating the onset of creep deformation. For MKII fuel, no such sharp transition was
observed. The data generated for MKI were in close
agreement with those of Tokar et al.62–64 up to 1100 K.
The M2C3 phase is harder than MC and is uniformly

MC1+x

(U0.45Pu0.55)C - MKII10
(U0.31Pu0.69)C0.93 63
(U0.3Pu0.7)C - MKI10
(U0.79Pu0.21)C63

1´10-2

104

1´10-4
UC1–x

1´10-5
Irradiation creep
s

1´10-6
TC
1´10

Log hardness (MPa)

de/dt (h-1)

1´10-3

103

TS

-7

5

6

7

8

9


10

104/T (K-1)

Figure 13 Steady-state creep data of UC1Àx and MC1 + x
at a stress of 20 MPa. Ts and Tc are the fuel surface and
center temperatures, respectively. Reproduced from
Blank, H. Material Science and Technology, Nuclear
Materials, Pt I; Cahn, R. W., Haasen, P., Kramer, E. J., Eds.;
VCH Publisher: New York, 1994; Vol. 10A.

200

400

600

800 1000 1200 1400 1600
Temperature (K)

Figure 14 Hot hardness data of uranium–plutoniummixed carbide fuels.


Carbide Fuel

distributed in the MC phase, hindering dislocation
motion and hence reducing creep deformation. It is
recommended that the amount of M2C3 phase in the
fuel should be optimized to allow creep to occur.
Though Pu-rich carbide fuel is harder than U-rich

fuel, beyond a temperature of 1553 K (average volumetric temperature of fuel), the plutonium-rich fuel
behaves in the same way as the U-rich carbide fuel,
for which in-reactor performance indicated no failure
due to FCMI. Hence, from extrapolation it could be
presumed that the Pu-rich fuel will also behave in a
similar manner.
3.03.4.1.5 Experience on irradiation
performance
3.03.4.1.5.1 US experience

The US experience of carbide fuel irradiation has
been summarized recently by Crawford et al.66
A large number of He-bonded and sodium-bonded
pins were irradiated in EBR II as an initial test for
subsequent loading in FFTF to simulate conditions
for a large development plant (LDP). The irradiation
condition simulates the peak-cladding temperature
condition and the peak power condition of FFTF.
The objective of this study was to see the effects of
pin diameter (7.89–9.40 mm), pellet density (81
and 87% TD), pellet–clad gap (0.13–0.78 mm), type
of cladding alloy, and bond type (He or Na) on
in-reactor lifetime, fuel and cladding swelling, fission
product behavior, fuel–cladding mechanical interaction, and fuel restructuring. About 470 MC fuel rods
with sodium or helium bonding were irradiated in
EBR II with different types of cladding: for example,
SS316, PE-16, stainless steel D9, and D21. Over 200
MC fuel rods were irradiated in FFTF in two assemblies: the ACN-1 experiments with rods fabricated
using 316SS and D9 cladding; and the FC-1 test, a
full-size, 91-rod FFTF assembly with 316SS and D9

cladding and ducts. The AC-3 test consisted of 91
full-size, D9-clad rods of which 25 rods contained the
sphere-pac fuel made in PSI, Switzerland, and 66
rods contained pellet fuel irradiated to 9 at.% burnup
without breach. The higher thermal conductivity of
carbide makes the average fuel operating temperature very low and it behaves like a brittle material
showing cracks. However, the cracks resulted in fuel
relocation and did not cause premature pin failure. In
the EBR II test 21, fuel breaches were observed
before reaching the goal burnup. Out of this, 15
were PE-16-cladded rods, and clad failure was attributed to irradiation embrittlement of the cladding
alloy, rendering it less capable of enduring the stress

75

induced by FCMI and fission gas pressure. The FC-1
FFTF experiment (a full-size, 91-rod FFTF assembly) attained goal burnup with breach. A peak fuel
burnup of 20 at.% in 10 MC fuel rods clad in type 316
stainless steel was achieved in EBR II. Of those rods,
five had experienced a 15% transient-overpower test
in EBR II after attaining 12 at.% burnup. Thirteen
other He-bonded rods and three Na-bonded rods
attained 16 at.% burnup in EBR II without breach.
The FFTF AC-3 experiment results showed that,
for the relatively low-temperature conditions of the
test, the pellet fuel and sphere-pac exhibited only
minor observed differences in behavior, and both
types of fuel performed in a manner consistent with
the rest of the MC fuel database. Carbide fuel failures
typically result from FCMI, due to high fuel swelling,

which leads to early fuel–cladding gap closure. Also,
because it generally operates at relatively low temperature, fuel creep is not effective in relieving cladding stress. For this reason, MC fuel pin design must
incorporate a large fuel–cladding gap and make use
of a low-density fuel to delay the onset of FCMI.
No fuel failures have been attributed to the cladding carburization phenomenon. Ten transientoverpower tests of MC fuels were conducted in
TREAT using fuel irradiated in EBR II (burnups
up to 12 at.%) for the purposes of establishing that
cladding breach would occur above 115 and 125%
overpower. The results suggested FCMI-induced
breaches, but most importantly indicated comfortable margins to failure. The rods indicated only
small cladding strains and small amounts of liquidphase penetration of the cladding. The conclusion of
these tests was that nothing in fuel transient-overpower response would prevent or limit application of
MC fuels to fast reactors. The EBR II tests also
included rods irradiated beyond goal burnup to
breach, and one intentionally defected rod irradiated
for 100 days beyond cladding breach. The defected
rod exhibited a reaction between the fuel and the
coolant (O in the coolant), which resulted in a higher
specific-volume reaction product and caused expansion as well as widening of the defect with no release
of activity into the coolant. Other rods irradiated to
natural breach in EBR II did not exhibit that phenomenon. MC fuels appear to operate benignly after
cladding breach. In another experiment, in EBR II,
the effect of sodium void was tested by irradiating an
MC fuel rod with a purposely induced void, simulating void formation due to Na expulsion during irradiation. The fuel rod exhibited microstructural
changes, reflecting a local high fuel temperature


76

Carbide Fuel


and no loss of cladding integrity. This experiment
indicated that MC fuel would withstand an Na bond
expulsion of some magnitude. The US experience
with MC fuels was not very large; however, it was
sufficient to instill confidence that such fuels have
irradiation performance adequate for use in SFRs. In
particular, He-bonded rods clad in lower swelling
cladding alloys with around 80% fuel smeared density appeared to show the best performance
potential.
Kummerer54 summarized the results of irradiation
experiments under different conditions based on a
reference concept. Altogether, 101 pins were irradiated, out of which 83 pins were irradiated under
steady-state condition, which includes 15 failed pins.
Eighteen pins underwent steady-state and cycling,
which reported one pin failure. The parameters, for
example, pin diameter, type of bond, fuel–clad gap,
smear density, cladding material and wall thickness,
linear power, and burnup, were varied. Based on the
results of these irradiation, a reference pin concept
with cold-worked austenitic steel (1.4970) cladding,
pellet diameter 7.0 mm, pellet density 84% TD,
fuel–cladding gap of 400 mm, helium bond, smear density 75% TD, pin diameter 8.5 mm, and clad wall
thickness of 0.55 mm evolved for further irradiation
for final performance test.
The ‘NItrure Mixte dans a Phenix (NIMPHE)’
fuel irradiation experiments were designed to study
the influence of the fabrication route on the fuel
stability. The direct pressing route, as proposed by
Richter,67 produces pellets with improved thermal

stability; however, the starting materials, the reaction
temperature, and the pressing conditions, all influence
the thermal stability. Also, direct pressing of granules
decreases the densification under irradiation.

by direct pressing and irradiated at a linear power of
730 W cmÀ1 with a burnup of 5.8 at.% is being carried
out and results are awaited.
A short-term irradiation experiment at the beginning of life in the High Flux Reactor Petten was carried
out for He-bonded mixed-carbide low in oxygen
(CARLO) and rich in oxygen (CARRO).53 The U:Pu
ratio was 80:20, the O content was 290 and 1590 ppm,
and the carbon content was 4.85 and 5.05 wt% for
CARLO and CARRO, respectively. The pellet diameter was 6.4–6.2 mm. The clad ovalization was Æ4
and Æ6 mm before irradiation and Æ16 and Æ20 mm
after a burnup of 0.5 at.% for CARLO and CARRO,
respectively. The main objective of this test was to
study the thermal stability of the fuel. Axial g-scanning
of the pins indicated that for CARLO the 137Cs activity
was more at the extremities of the pin having a lower
temperature. It was also observed for CARRO, but the
concentration profile was less pronounced.
For CARLO fuel, wedge type crack was observed
at the periphery, indicating strong difference in the
swelling between the fuel center and periphery.
CARRO did not show any such crack indicating
swelling at the periphery. There was no fuel–clad
gap for CARLO but for CARRO a gap existed, indicating in-pile densification at high temperature.
From the analysis of the pellet geometry after irradiation, it was observed that in-pile densification in
He-bonded carbide fuel is not a general phenomenon

but is restricted to unstable CARRO fuel which is
exposed to high temperature at the beginning of life
and that this condition was expected to exist for fuel
pins with large initial gap and/or higher linear heat
rating. O plays a very vital role in stronger radial
transportation of Pu and less axial Cs transportation
and increase in FCCI by carburization of the clad.
The excess O combines with Cs and forms a complex,
thus reducing the mobility of either O or Cs.

3.03.4.1.5.3 The NIMPHE 2 irradiation experiment

3.03.4.1.5.4

Fernandez et al.28 reported the influence of three
different fabrication routes (conventional carbothermic reduction and direct pressing of pellets and
granules) on the irradiation performance of carbide
fuels having densities between 80% and 85% TD in
the Phe´nix reactor. NIMPHE 2 consisted of a capsule
containing two (U0.8, Pu0.2)C pins along with some
mixed nitride pins irradiated at a higher linear power
of 730 W cmÀ1. The pin geometry was similar to
Superphenix standard, with pellet and pin diameters
of 7.11 and 8.5 mm, respectively. Postirradiation
experiments of NIMPHE 2, a carbide fuel fabricated

MC fuel containing 70% and 55% PuC was used as
the driver fuel for FBTR, India. This unique fuel
composition created many differences in the performance of the fuel. This prompted stringent fuel specification at the beginning of the campaign and was
followed by a very conservative approach of raising

the burnup and LHR with intermittent PIE.68 PIE
started with experimental fuel pin irradiation at the
beginning of life followed by fuel subassembly examination at burnups of 2.5, 5, 10, and 15.5 at.%. Evaluation of the radiograph of the fuel pins after 2.5 and
5 at.% revealed presences of gaps between the pellets

3.03.4.1.5.2 German fast breeder project

Indian experience


Carbide Fuel

(a)

77

(b)
(U0.8 Pu0.2)C1.0

(U0.8 Pu0.2)C1.1

Figure 16 Cross-section of mixed carbide fuel after
1.5 at.% burnup. Reproduced from Suzuki, Y.; Arm, Y.;
Handa, M.; Shiba, K. Research and development of uranium
plutonium mixed carbide and nitride fuels at JAERI; IAEA
TECDOC-466; Vienna, 1988; pp 73–82.

(c)

(d)


Figure 15 Photomicrographs of (U0.3Pu0.7)C1 + x fuel
pin at different burnups: (a) 25 GWd tÀ1, (b) 50 GWd tÀ1,
(c) 100 GWd tÀ1, and (d) 155 GWd tÀ1. Reproduced from
Sengupta, A. K.; Basak, U.; Kumar, A.; Kamath, H. S.;
Banerjee, S. Experience on mixed carbide fuel with high
plutonium content for Indian fast breeder reactor- an
overview; Journal of nuclear materials 385; 2009;
pp 161–164.

and clad. After 10 at.% burnup, the pellet–clad gaps
were not observed at the center of the fuel column.
The maximum increase in stack length was 2.61%.
Radiography of fuel pins after 15.5 at.% burnup
showed pellet–pellet gap at the end of the column,
and the stack length increase varied from 2.7% to
3.7%. Maximum fission gas release was 16% and the
corresponding internal pressure in the fuel pin was
2.1 MPa after 15.5 at.% burnup. The Xe/Kr ratio was
around 13. The cross-sections of the fuel at various
burnups up to 15.5 at.% are shown in Figure 15. It
indicates circumferential cracks due to thermal stresses, and discrete zone free from any porosity is
observed near the outer circumferential area. This
was attributed to creep of the fuel due to FCMI. The
end of the fuel column reveals circumferential cracking and fuel–clad gap closure.
3.03.4.1.5.5 Japanese experience

The capsule irradiation tests of carbide fuels (C/M
ratio of 1.0 and 1.1) started in 1983 in JFR 2 and
JMTR.69 Three capsules of carbide fuels containing

two pins in each were irradiated at about 1.5 and 3 at.%
burnup. The pin diameter was 6.5 and 9.4 mm, and the
linear heat rate was 450 and 650 W cmÀ1. After 1.5 at.%
burnup, characteristic cracking and restructuring of

the pellet (Figure 16) took place at the central part
of the pellets with the formation of large pores. Migration of semivolatile Cs to plenum and other cooler
parts of the pin was detected. Clad carburization at the
grain boundary near the surface was also observed.
The results of the other reported irradiation up to
3 at.% are awaited.
3.03.4.1.5.6

French experience

Between the years 1960 and 1970, about 80 MC fuel
pins (80% sodium-bonded), including two NIMPHE
pins in collaboration with TUI as mentioned above,
were irradiated in MTRs (OSIRIS, SOLOE) and then
in Rapsodie and PHENIX reactors.70 The sodiumbonded fuel had a high density (>90% TD) but low
smear density of about 70–80% TD with a large fuel–
clad gap. The fuel could reach a burnup of 14 at.%
without a limited number of clad failures. The
swelling rate was $1.6%/at.% burnup below
1000  C. They perceived some major design risks,
such as lack of sodium bond, locally resulting in overheating of the fuel, excessive swelling, and hence a
possible clad failure. Clad carburization in case of
sodium-bonded pin was also presumed to be an issue
which might cause clad embrittlement. Also, trapping
of small carbide chips in the large gap between the fuel

and the clad may also cause localized clad straining.
For the He-bonded pins, it was observed that smear
density should be limited to 70–75% TD for high
LHRs (800–1000 W cmÀ1) and high burnup
operation.
In summarizing the international experience on
carbide fuel performance, the following conclusions
can be drawn:
 Sodium-bonded fuel design has some intrinsically
good performance characteristics, such as higher


78

Carbide Fuel

LHR compared to the He-bonded design; however, it has several drawbacks such as a more complex process of sodium-bonding technique, their
more stringent quality assurance, and a greater
extent of clad carburization.
 A He-bonded pin design can achieve better performance in terms of LHR and burnup provided it
fulfills the design criteria concerning carbide fuel
fabrication, pin design, and the balance between
fuel–clad gap width and fuel porosity.
 Swelling of carbide fuel is an issue when compared
with oxide fuel. Both solid fission products as well
as fission gases dissolved in matrix contribute to
swelling. Swelling rate increases by growth and
merging of bubbles and it is a function of temperature. Beyond a critical temperature, swelling
increases drastically.
3.03.4.1.6 Fuel–clad chemical interaction


Carbon activity and partial pressure of CO are
important parameters responsible for clad carburization, which make the clad brittle, resulting in clad
breach. In sodium-bonded fuel, carbon transfer from
the fuel to the clad takes place by dissolution of
carbon in sodium liquid. However, in He-bonded
fuel pins, carbon transfer takes place through CO.
After 2–3 at.% burnup, the fuel may swell and come
in direct contact with the clad. But, transport of
carbon down the temperature gradient within the
fuel also takes place through CO.
Thermodynamic computations and experimental
studies on the chemical compatibility between
mixed carbide, SS316, and sodium coolant have
been restricted to uranium-rich (U/(U þ Pu) > 0.7)
compositions. The results of the out-of-pile and inpile experiments as summarized by Ganguly and
Sengupta71 are given below:
 The chemical compatibility between mixed carbide
fuel and sodium was excellent. Out-of-pile test with
helium-bonded MC1 þ x containing up to 12 wt%
M2C3 at 873 K for 1000 h showed no clad carburization, but at 1000 K carbide precipitation along
the grain boundaries and slip lines in a zone up to
50 mm in depth from the fuel–cladding interface, was
observed. The hardening of the SS316 cladding
up to 100 mm was observed. Similarly, out-of-pile
experiments with sodium-bonded MC1 þ x showed
no carburization at 873 K, but at 1000 K an increase
in carbide precipitation in grain boundaries
and along slip lines was noticed up to 90 mm
of the surface. The depth of carburization for


sodium-bonded fuel was more compared to the
He-bonded pins.
 Out-of-pile burnup simulation experiments showed
that neither individual fission products nor the
global chemical effect of burnup was detrimental
to clad carburization.
 In-pile experiments in EBR II up to 12 at.% burnup
with stoichiometric MC and hyperstoichiometric
MC containing up to 20% M2C3 revealed that the
carburization in sodium-bonded carbide elements
was 2–3 times greater than that in helium-bonded
elements. The carburized layer in sodium-bonded
pins was as high as 150 mm. However, no fuel element failures in both He- and Na-bonded pins had
been attributed to clad carburization.
Saibaba et al.17 have estimated the carbon potential
and Pco of hyperstoichiometric (Pu0.7U0.3)C1 þ x up to
2200 K as a function of O, N, and M2C3 contents, and
Pillai et al.72 have reported the carbon potential data
of SS316 up to 1000 K. These calculations show that
 the carbon potential of hyperstoichiometric
(Pu0.7U0.3)C containing relatively high M2C3
(À20%) is lower than that of (U0.8Pu0.2)C with
very low M2C3($2%);
 the Pco of (Pu0.7U0.3)C1 þ x containing relatively
high oxygen (–1 wt%) is lower than that of its
uranium-rich counterpart with very low oxygen
($0.1 wt%);
 carbon activity of the fuel increases with increase
in temperature but that of SS316 decreases with

temperature.
In the MC þ M2C3 phase mixture, plutonium
concentration is greater in the M2C3 phase. On
extending the same basis of calculations to hyperstoichiometric (Pu0.7U0.3)C1 þ x, the extent of plutonium
segregation in the M2C3 phase has been found to be
more than 90%. The mixed sesquicarbide in this case
could, therefore, be assumed to be pure Pu2C3. On
the basis of thermodynamic data, the calculated carbon potentials of PuC, UC, U2C3, and Pu2C3 as a
function of temperature are shown in Figure 4. The
carbon potentials of PuC and Pu2C3 are lower than
that of SS316 up to 1300 and 1200 K, respectively.
The carbon potential of UC is lower than that of
PuC, Pu2C3, and SS316 at all temperatures but the
values for U2C3 are relatively high. The carbon
potential of hyperstoichiometric plutonium-rich
mixed carbide has been found to be lower than
its uranium-rich counterpart at all temperatures.
Theoretical calculations, therefore, indicate that


Carbide Fuel

79

The cooling time is an out-of-pile phenomenon, and
it has been observed by Blank et al.53 that the amount
of Mo, Pd, and Nd increases while that of Pu and
other fission products decreases. The types of fission
products and their chemical state are given below:





Indentations

Figure 17 Microstructure of SS316 cladding material
after compatibility test. Reproduced from Sengupta, A. K.;
Basak, U.; Kumar, A.; Kamath, H. S.; Banerjee, S. J. Nucl.
Mater. 2009, 385, 161–164.

hyperstoichiometric (Pu0.7U0.3)C containing high
M2C3 and oxygen are not likely to cause any major
carburization of SS316 cladding. This was also
observed in the microstructure of SS316 cladding
(Figure 17) after the out-of-pile experiments carried
out by Ganguly and Sengupta.71
3.03.4.2 Effects of Burnup on C/M Ratio
and Chemical State of Fission Product
The fission of uranium–plutonium MC results in the
formation of different types of fission products, but
mainly lanthanides, rare earths, and noble metals.
The chemical state of the fission products will affect
the in-pile behavior of the fuel to a large extent. The
extent of the fission products formed will depend on
the burnup and the cooling time of the fuel. The
fission products generated may form dicarbides or
monocarbides depending upon the carbon potential
(C/M ratio) of the fuel prevailing in the fuel crosssection and the temperature gradient. This may cause
decrease in the C/M ratio and result in the formation
of metal phase U/Pu, which forms low-melting

eutectic with the components of the cladding material. Fission products and their chemical states will
also depend upon the Pu content of the fuel. Fast
fission of Pu results in the formation of more noble
metals. The chemical state of each element and their
quantity (depending upon their half lives) depend on





fuel composition,
neutron spectrum,
residence time of the fuel in the reactor, and
cooling time.

Xe, Kr: inert gas
Cs, Rb, Te, I: volatile fission products
Sr, Ba: alkaline metals form dicarbide
Zr, Mo, Ru, Pd, Y, Nb, Tc, Rh, Ag: 4d metals;
mono-, di-, and sesquicarbide
 Ce, Nd, La, Pr, Pm, Sm, En, Gd, Tb: monoand sesquicarbide: Lanthanides form mono- and
sesquicarbide.
Hence, the resultant C/M ratio after irradiation will
decrease due to formation of the higher carbides of
alkaline-earth metals, the 4d metals, and the lanthanides. This decrease in C/M ratio may result in
shifting of the hyperstoichiometry fuel to hypo and
the resultant metal phase may lead to formation of
low-melting eutectic. Hence, a hyperstoichiometric
fuel is always preferred in fuel stoichiometry. The
extent of hyperstoichiometry (M2C3 content) in the

initial fuel composition is determined by the target
burnup. In their investigation based on an experimental measurement for (U0.85Pu0.15)C1 þ x, Blank et al.
concluded that at 10 at.% burn up, the mean C/Me,
where, Me stands for sum of the actinide and the
dissolved fission product atom, decreases to less
than 1 in hyperstoichiometric carbide. Beyond 10 at.%
burnup, the decrease in C/M ratio and formation
of the metal phase of actinides will be a serious
issue from the point of view of low-melting eutectic
formation. To have an in-depth understanding of
the variation of C/M ratio with burnup, theoretical calculations were carried out by Agarwal and
Venugopal.73 They observed that the chemical state
of rare earth metals plays a crucial role in controlling
the carbon activity of the fuel with burnup. Rare earth
elements show more solubility in M2C3 than in the
MC phase of the fuel. Lanthanides make stable sesquicarbides and dicarbides. The carbon potential of
Pu2C3/PuC is higher than that of LnC2/Ln2C3, and
therefore LnC2 can coexist with MC þ M2C3 fuel.
But the small difference in Gibbs energies of formation of the compounds is offset by dissolution of
Ln2C3 in M2C3 phase of fuel. However, LnC2 is also
known to dissolve completely in an isomorphous
MC2 phase, but absence of MC2 phase in most of
the carbide fuels destabilizes LnC2. Lanthanides
with reasonable fission yield do not form stable monocarbides, but they get stabilized to a limited extent by