See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/299999748

Dimensional change, irradiation creep and thermal/mechanical property

changes in nuclear graphite

Article  in  International Materials Reviews · April 2016

DOI: 10.1080/09506608.2015.1136460

CITATIONS READS

62 655

7 authors, including:

Barry J. Marsden Graham Hall

The University of Manchester The University of Manchester
134 PUBLICATIONS   1,739 CITATIONS    56 PUBLICATIONS   567 CITATIONS   

SEE PROFILE SEE PROFILE

Abbie N Jones
The University of Manchester
50 PUBLICATIONS   426 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Nuclear Graphite Independent Research View project

RBMK reactors View project

All content following this page was uploaded by Abbie N Jones on 06 February 2018.

The user has requested enhancement of the downloaded file.

 International Materials Reviews

ISSN: 0950-6608 (Print) 1743-2804 (Online) Journal homepage: http://www.tandfonline.com/loi/yimr20

Dimensional change, irradiation creep and

thermal/mechanical property changes in nuclear
graphite

B. J. Marsden, M. Haverty, W. Bodel, G. N. Hall, A. N. Jones, P. M. Mummery &

M. Treifi

To cite this article: B. J. Marsden, M. Haverty, W. Bodel, G. N. Hall, A. N. Jones, P. M.

Mummery & M. Treifi (2016): Dimensional change, irradiation creep and thermal/
mechanical property changes in nuclear graphite, International Materials Reviews, DOI:
10.1080/09506608.2015.1136460

To link to this article: http://dx.doi.org/10.1080/09506608.2015.1136460

© 2016 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 06 Apr 2016.

Submit your article to this journal

Article views: 53

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=yimr20

Download by: [The University of Manchester Library] Date: 18 April 2016, At: 04:19
 Dimensional change, irradiation creep and
thermal/mechanical property changes in
nuclear graphite
B. J. Marsden∗ , M. Haverty , W. Bodel , G. N. Hall , A. N. Jones ,
P. M. Mummery and M. Treifi
Since the start of the ‘nuclear age’ graphite has been employed as a moderator in around 100
nuclear reactors, and today there are still some 30 graphite-moderated reactors operating and
there are plans for new Generation IV high-temperature reactors. Many of the graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

moderator reactors now producing power are operating beyond their original design life.
Therefore in some cases, to aid the reactor operators and designers, the existing graphite
irradiation databases need to be extended either to a higher temperature or higher neutron
fluence. Furthermore, data are needed for the different grades of graphite that are available at
present. This can either be achieved by expensive, time consuming irradiation programmes or
by improving the understanding of the mechanisms and processes which lead to irradiation-
induced dimensional and property changes in the graphite core components. This review looks
at three of the most important graphite properties which change with exposure to irradiation,
namely dimensional change, irradiation creep and thermal expansion. The behaviour of UK
AGR, Magnox and an experimental grade of German reactor graphite are explored in some
detail. First graphite reactor core design is briefly discussed, giving examples of typical graphite
components and core arrangements. Issues related to aging graphite component and core
behaviour are illustrated through examples of component internal and thermal stress generation,
and issues related to whole core behaviour are also outlined. Second the manufacture and
microstructure of different nuclear graphite grades are discussed, highlighting how the choice of
raw materials and manufacturing technique influences the graphite properties. Third the
coefficient of thermal expansion, dimensional change and irradiation creep are analysed using
microstructural and averaging methods which are used to relate crystal to bulk properties by
accounting for graphite crystal orientation and porosity. These techniques, which were first
applied to nuclear graphite in the 1960s, are extended and discussed with the aim of trying to
lend some understanding to the role the microstructural crystallite and porosity distributions play
in defining the dimensional stability and properties of virgin graphite, irradiated graphite and
stressed graphite.
Keywords: Nuclear graphite, Dimensional change, Irradiation creep, Thermal expansion, AGR, Magnox

Introduction significant amount of power in the UK (15 reactors,

2015) and Russia (11 reactors, 2015). In recent years,
The ‘Nuclear Age’ started with the use of graphite as a the Generation IV initiative has generated interest in the
moderator to sustain a chain reaction in the Chicago use of graphite as a moderator in the USA, Japan,
Piles.1 Since this world changing event, graphite has China and Europe. Two types of graphite-moderated
been used as a moderator in over 100 nuclear reactors, Generation IV reactors are under development. First
many of which are still operating and producing a there is the high-temperature, helium-cooled, graphite-
moderated reactor that is capable of producing gas output
Nuclear Graphite Research group, School of MACE, University of temperatures of ∼1000°C. Second there is the molten salt
Manchester, M13 9LP, UK
reactor which has high thermodynamic efficiency, while
∗
Corresponding author, email [email protected]

© 2016 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution License
(http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Received 14 November 2014; accepted 21 December 2015
DOI 10.1080/09506608.2015.1136460 International Materials Reviews 2016 1
 Marsden et al. Dimensional and property changes in nuclear graphite

operating at low pressure thus giving enhanced safety. Thus

These designs lead to the possibility of directly producing
hydrogen fuel efficiently for a ‘carbon free’ future, and in Equivalent DIDO Nickel Flux at a point of interest
the case of helium cooling, giving the possibility of gener- wd
= wNi = n/cm2 /s
ating electrical power via a gas-turbine.2 1.313 × 10−21
In a thermal nuclear reactor a moderator is necessary to
reduce the speed of (thermalise) fast neutrons so that they or
are capable of sustaining a nuclear chain reaction. When fd = 1.313−21 × 10 × wNi dpa/s (2)
nuclear reactors were first under development the possible
and practical moderator candidates were heavy water EDND is the time integral of EDNF
(D2O), light water (H2O) and artificially produced blocks Other units of graphite damage fluence the reader may
of electrode graphite. The absorption and scattering cross- find in the literature are
sections for these are given in Table 1. . energies above 0.18 MeV;
A suitable moderator requires a high neutron scattering . calder equivalent dose MWd/t.
cross-section and a very low neutron absorption cross-sec- Fortunately, there are conversion factors between these
tion. Thus for the first reactors, which used natural units as given in Table 2; further details are given in
uranium metal as a fuel, the use of hydrogen (in the reference.4
form of light water) with its excellent high scattering From Fig. 2 it can be seen that there are complex cor-
cross-section was not possible because it also has a rela- relations between the dimensional and property changes,
tively high absorption cross-section. Light water only which are still not well understood. This review discusses
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

became a viable moderator later when enriched fuel the present state-of-the-art understanding of this behav-
became available. Deuterium in the form of heavy water iour in relation to three of the most important topics in
is an excellent moderator, but it is expensive to produce nuclear graphite technology: the coefficient of thermal
and was, at first, not available in large quantities. How- expansion (CTE), irradiation-induced dimensional
ever, graphite, in its pure form, has a relatively high scat- change, and irradiation creep.
tering cross-section and a low absorption cross-section, Before discussing these topics, it is necessary to give a
was readily available and relatively cheap. Graphite review outlining the general effects of irradiation on
could also be machined into intricate shapes to provide graphite properties and stress development, followed by
channels for the fuel and the coolant, as illustrated in a description of the manufacture of and microstructure
Fig. 1. of nuclear graphite. The averaging methods used to relate
When graphite components are irradiated in a reactor, crystal to bulk properties first applied to nuclear graphite
significant changes to their dimensions and properties in the 1960s are reviewed and extended; illustrating the
occur, as illustrated in Fig. 2. The graphite dimensional need to account for porosity and orientation.
and properties changes are a function of fast neutron
irradiation, temperature and load and must be accounted
for by the reactor designer and operator. The unit of
Empirical material test reactor data
irradiation exposure EDND used in Fig. 2 is particular Graphite reactor core design has to account for significant
to nuclear graphite technology. Equivalent DIDO Nickel irradiation-induced dimensional changes and material prop-
Dose and Equivalent DIDO Nickel Flux (EDND and erties changes. Most graphite core lifetime assessments are
EDNF, respectively) are defined as the equivalent nickel based on finite element analysis using empirical data
activation at a standard position in the DIDO reactor at obtained from material test reactor (MTR) programmes
Harwell defined as follows: as illustrated later in this review using the case of AGR
and Magnox reactor graphite grades as typical examples.
Linear (as opposed to volumetric) dimensional change
Equivalent DIDO Nickel Flux at point of interest data obtained on AGR Gilsocarbon graphite samples
wNi(s) wd irradiated at various temperatures between 430°C and
= wNi = n/cm2 /s (1) 1430°C are shown in Fig. 3. Within this temperature
wds
range, the higher the temperature the sooner ‘turnaround’
from shrinkage to swelling occurs. This behaviour is typi-
where φds = displacement rate of carbon atoms at the cal for most semi-isotropic, medium and fine-grained
standard position in DIDO = 5.25 × 10−8 displacements graphite grades, although the magnitude of the changes
per atom per second (dpa/s); φNi(S) is a flux measured varies from grade to grade.
by the nickel activation reaction at the standard position Dimensional changes at low temperature (below ∼650°
in DIDO = 4 × 1013 n cm−2 s−1; φd = displacement rate of C) in semi-isotropic graphite sometimes exhibit a low flu-
carbon atoms at the position in of interest. ence delay in shrinkage and even some swelling. This is

Table 1 Moderator elemental nuclear cross-sections3 Table 2 Conversion of various graphite damage units into
EDND4
Scattering cross-section Absorption cross-section
Element (barn) (barn) Unit To convert to EDND multiply by

C 5.551 0.0035 Calder equivalent dose 1.0887 × 1017

H 82.02 0.3326 (MWd/At)
D 7.64 0.00055 En>0.18 MeV (n/cm2) 0.67
O 4.232 0.00019 dpa (atom/atom) 7.6162 × 1020

2 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

1 Typical graphite moderator components. a Calder Hall, Magnox brick and tiles. In the Calder Hall and Chapelcross reactors
graphite components similar to these formed a 24-sided polygon, 11-m diameter, 8-m high with a 0.75-m reflector. There were
1696 channels, 1140 tonnes of graphite, 14 044 graphite bricks and in total 58 140 graphite components. Typical power output
was 270 MW thermal. Later, other physically larger Magnox reactors such as Wylfa produced up to 1875 MW thermal. b The
core of an advanced gas-cooled reactor under construction showing the arrangement of graphite fuel channel bricks typi-
cally 460 mm across flats. The bore of each fuel channel is about 265-mm diameter and there are about 300 fuel channels.
Square interstitial bricks are placed between the fuel channel bricks and form the control rod channels. Note that in this
image, the interstitial bricks have yet to be inserted. A typical AGR core is about 9-m diameter by 8-m high surrounded by
a 0.5-m thick reflector. Typical power output is 1500 MW thermal. c Lower core of the pebble-bed HTR-10 test reactor (Tsin-
ghua, China) while under construction. The internal core diameter is about 1.8 m by about 1.9-m high. Some pebble fuel can
be seen in the centre, the core will be eventually filled completely with around 27 000 fuel pebbles. The holes in the reflector
surrounding the core provide channel for coolant gas, control rods and a boron ball secondary shut-down system. The power
output is 10 MW thermal. d A prismatic block (left) for the high-temperature test reactor HTTR (JAEA, Japan). Each block is
360 mm across flats by 580-mm high. The many holes provide location and coolant channels for the graphite fuel elements
(two are shown in place). The prismatic block on the right provides channels for control rods. These blocks are stacked into
30 columns making an active core 2.3-m diameter by 2.9-m high surrounded by removable and permanent reflectors. The
power output is 30 MW thermal

the case with Gilsocarbon and other nuclear graphite between 150°C and 650°C. This graphite grade is highly
grades in which initially there is a small increase in volume anisotropic due to the method of manufacture and the
between zero fluence and about ∼10 × 1020 n cm−2 raw materials used, as discussed later in this review. The
EDND,5 see Fig. 3. The reason for this initial delay/swel- data do not extend as far as turnaround in this case. Aniso-
ling is unclear, but it has been attributed to machining tropic graphite is not recommended for modern reactor
stresses.6 However there are no microstructural obser- design. However it is worth noting that the structural integ-
vations or other evidence to validate this postulation rity of the graphite components used in the Magnox reac-
and this behaviour does not appear to be prevalent in tors was maintained for over 40 years in some cases.
semi-isotropic graphite irradiated at higher temperatures. If graphite is irradiated under load the dimensional
Linear dimensional changes in Magnox Pile Grade A change data are modified. The difference between loaded
(PGA) graphite are shown in Fig. 4 at various temperatures and unloaded dimensional change is referred to as

International Materials Reviews 2016 3

 Marsden et al. Dimensional and property changes in nuclear graphite

2 Typical dimensional and property changes in an isotropic graphite irradiated at ∼500°C4. a Dimensional change, dimensional
change rate and coefficient of thermal expansion (CTE) as a function of fast neutron fluence. b Factorial change in Young’s
modulus (E/Eo − 1) and thermal conductivity (Ko/K − 1) and irradiation creep (elastic strain units or esu) as a function of fast
neutron fluence
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

irradiation creep. The dimensional changes are increased Irradiated Young’s modulus and CTE property changes
under compression and reduced under tension. Unfortu- for both PGA and Gilsocarbon at various temperatures
nately there are no irradiation creep data for Gilsocar- are given in Fig. 6 and it is clear from these figures that
bon or PGA graphite with fluence greater than ∼60 × the behaviour is complex and differs between anisotropic
1020 n cm−2 EDND.4 However, there are reasonable and semi-isotropic grades. For example in the case of
data for medium-grained graphite, namely ATR-2E. PGA, CTE keeps increasing with increasing fluence
For ATR-2E graphite there are data in tension and com- whereas the CTE of Gilsocarbon first increases then
pression up to ∼200 × 1020n cm−2 EDND, see Fig. 5a reduces.
and b. The creep strain is defined as the difference In the case of Young’s modulus there is a significant
between the strain in the unloaded (control) specimen initial increase in modulus, followed by a secondary
subtracted from the strain in the loaded specimen as increase with increasing fluence. If the graphite is irra-
shown in Fig. 5c. With increasing fluence the creep diated through turnaround and past the initial volume,
strain increases fairly linearly at first, and then the rate significant degradation occurs and the modulus starts to
reduces before finally increasing. If compressive reduce. This behaviour is illustrated for Gilsocarbon irra-
creep is considered positive, the creep behaviour is simi- diated at 600°C in Fig. 6d.
lar in tension and compression, certainly within the data
scatter. However, tensile and compressive behaviour
appear to diverge at high fluence. The reason for Radiolytic oxidation
including the ATR-2E data in this review is that it is The topic of radiolytic oxidation is complex and a
later used to explore the creep behaviour in nuclear detailed discussion of the topic is beyond the scope of
graphite. this particular review; however a brief expansion of the

3 Irradiation-induced dimensional changes in semi-isotropic Gilsocarbon irradiated at various temperatures4

4 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

4 Irradiation-induced dimensional changes in anisotropic PGA irradiated at various temperatures.4 a Perpendicular to extru-
sion. b Parallel to extrusion

phenomenon is given here as weight loss is referred to in As the graphite component dimensional and property
passing later in the text. changes are a function of both irradiation fluence and
When graphite is irradiated in carbon dioxide, as is the temperature, the brick will at first shrink faster at bore
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

case in AGR and Magnox reactors, ionising irradiation than at the outside and there will be an accompanying
dissociates the gas into carbon monoxide and an oxidising variation in properties across the brick. The main proper-
species.8 Within the graphite porous structure carbon ties of concern being Young’s modulus, strength, CTE
atoms combine with the active species resulting in signifi- and irradiation creep. Over time graphite brick internal
cant component weight loss; over 40% in some regions of shrinkage and thermal stresses are generated; first tensile
a higher powered plant. Because the loss of carbon occurs at the bore and compressive at the outside, and then rever-
within the graphite open porosity, visual inspection of a sing in sign towards the end of reactor life. Fortunately
graphite component or sample alone will show no indi- much of this stress is reduced by irradiation creep. In
cation of the loss of mass. addition, the difference in CTE between the bore and
Weight loss significantly reduces Young’s modulus, the outside of the brick, coupled with the influence of
strength and thermal conductivity;4,9 in the case of irradiation creep, causes the stresses to significantly
Young’s modulus by about 50% at 20% weight loss. To change when the reactor is shut-down and the graphite
a lesser extent weight loss influences thermal expansion cools from operating to ambient temperature. This behav-
and dimensional change. Irradiation creep in graphite is iour is illustrated by the finite element stress analysis
modified through the change in modulus with weight results plotted against full power years (fpy) in Fig. 8;
loss.10 both bore and keyway-base stresses are presented at
Radiolytic oxidation is not an issue for Generation IV power and at 2 year shut-downs. The change in stress at
helium-cooled reactors systems such as high-temperature shut-down is mainly due to the temperature difference
reactors (HTRs) or the molten salt graphite-moderated across the brick at power. This temperature difference
reactors. leads to the generation of thermal stresses which creep
out during reactor operation. These stresses return in
the opposite sense at shut-down leading to the spikes in
Implication of dimensional and stress as shown in Fig. 8. Also at shut-down there is a
property change on graphite component of stress associated with the irradiation-
induced change in CTE coupled with a change in CTE
component life due to irradiation creep strain.11
As illustrated in Fig. 1 graphite-moderated nuclear reac- Although the design of the pebble and prismatic HTRs
tors have many fuel and control rod channels. As the are somewhat different from an AGR or Magnox reactor,
graphite structure provides channels for fuel, fuel-cooling stresses will also be generated in their graphite com-
and control rod entry, the structural integrity of the ponents due to similar fluence and temperature gradients.
graphite components is important from safety and life- It is therefore important that the reactor designer has the
time considerations. Throughout reactor life component graphite data available for the particular design to assess
stresses develop which have to be accounted for by the the core safety and lifetime. For further detail on graphite
reactor designer and assessor. As an example, a typical component stress analysis detailed information is avail-
analysis of an AGR graphite brick is presented below. able in reference.11
This analysis has been carried out by the authors and is In addition to considering the generation of internal
purely illustrative and should not be taken as representa- thermal and shrinkage stresses it can be seen from Fig.
tive of any particular reactor. 3 that the dimensional changes are significant and ∼3%
A peak rated AGR graphite fuel brick is considered in shrinkage can be achieved in parts of the core. As graph-
Fig. 7. Locally the high irradiation intensity of the fuel ite-moderated cores are large in size and there is a neutron
within the channel results in the neutron flux (and hence flux gradient across the diameter and along the height of
fluence) exposure to the graphite component being higher the core, components will tend to bow at the core periph-
at the inside than at the outside, see Fig. 7. In addition ery. In the case of channel-type graphite reactors such as
there is a temperature difference between the inside and AGR, Magnox and RBMK reactors, the core rests on
the outside of the brick. steel, or steel and concrete, structures and are restrained

International Materials Reviews 2016 5

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

5 Irradiation creep in loaded and unloaded ATR-2E Graphite, irradiated at 550°C.7 a Compression loading (5MPa). b Tension
loading (5MPa). c Irradiation creep strain in ATR-2E graphite

6 Coefficient of thermal expansion and fractional change in Young’s modulus in PGA and Gilsocarbon graphites as a function
of irradiation.4 WG – With Grain, AG – Against Grain. a CTE PGA. b CTE Gilsocarbon. c Fractional change in Young’s modulus
PGA. d Fractional change in Young’s modulus Gilsocarbon

6 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

7 Finite element mesh of typical AGR graphite brick illustrating fluence and temperature gradients. a Typical fluence distri-
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

bution. b Typical temperature distribution

around the periphery by steel components so there will be is ∼27.5 × 10−6 K−1 in the ‘c’-axis, but approximately
differential movement at start-up and shut-down due to −1.5 × 10−6 in the ‘a’-axis.16 The elastic modulus parallel
the difference in the CTE of steel compared to graphite; and perpendicular to the basal planes are C11 = 106 × 1010
the expansion coefficient of steel could be about five N m−2 and C33 = 3.46 × 1010 N m−2, respectively, while
times that of graphite. It is therefore important that the C12 = 18 × 1010 N m−2 and C13 = 1.5 × 1010 N m−2, with
reactor designer accounts for component interactions a basal plane shear modulus of C44 = 0.45 × 1010 N
during: normal operation, thermal transients and fault m−2.16
conditions (including seismic). This is achieved using These extreme property values parallel and perpendicu-
large computer codes and various scale experimental lar to the basal planes strongly influence the properties of
models; readers interested in this aspect of graphite-mod- artificially produced polycrystalline graphite as discussed
erated design can find more information in references.12–14 below.

Nuclear graphite manufacture and Artificial polycrystalline graphite

microstructure Artificially produced graphite is used in many industries,
In this section the microstructure of PGA and Gilsocar- including the nuclear industry. Petroleum- or pitch-based
bon are discussed in some detail. In the case of ATR-2E filler coke is crushed, blended and then mixed with a bin-
some limited information on manufacture is given by der and formed into blocks. The blocks are first baked and
Haag,7 but unfortunately the authors are not aware of then finally graphitised.16 In a nuclear reactor, the graph-
any microstructural images or detailed microstructural ite properties are changed by fast neutron fluence, which
studies of ATR-2E graphite. This is unfortunate as the modifies the graphite microstructure. In addition, graph-
ATR-2E dataset is the only reasonable source of graphite ite properties can be altered by loading due to the influ-
irradiation creep data irradiated to a high fluence. ence of microscopic strain. To understand the factors
that define graphite properties and irradiation-induced
changes, it is necessary to understand the microstructural
Graphite crystal structure changes that occur from the nano-metre through to the
The most thermodynamically stable form of graphite is a bulk scale.
structure in which the carbon atoms in the basal planes The microstructure of the raw materials and manufac-
are bound by sp2 hybridisation in a planar network to turing process determines: the crystal orientation, the por-
three equidistant nearest neighbours, 120° apart in a osity and the structural features which together determine
given plane to form the hexagonal graphene structure. the material properties of the final product. The main raw
Strong in-plane covalent double bonds of both σ-type material used in graphite manufacture is filler coke, of
and π-type are present.15 However, these sheets of tightly which there are various types. Petroleum coke is a carbon-
bonded hexagonal lattice sheets are only weakly bound aceous solid formed in oil refinery coker units or in other
together with van der Waals type bonds in an ABAB cracking processes, whereas pitch coke is produced from
stacking sequence with an interlayer spacing of 3.35 Å, coal tar during coke production.17,18 In the case of Gilso-
as shown in Fig. 9. The basal plane spacing between carbon graphite a naturally occurring pitch was used.18
atoms is 1.415 Å, which gives unit cell dimensions for c The morphology of the filler coke particles arises during
and a of 6.7 Å and 2.5 Å, respectively. This configuration the refining process; it is determined by the mixing and
has a theoretical density of 2.266 g cm−3. flow as the liquid passes through an intermediate liquid
This lattice structure results in highly anisotropic crys- crystal phase (mesophase), during which spherules and
tal properties, for example the CTE in the range 20–120°C other shapes form at the nano to bulk scales. These

International Materials Reviews 2016 7

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

8 Normalised through-life stresses in a typical AGR graphite moderator brick. The change stress every 2 fpy is caused by ther-
mal stress at shut-down, see text

spherules harden, establishing the principal features of a several micro-metres in length, illustrated in Figs. 10d
lamella-form graphitisable coke morphology containing and 11d, are formed during manufacture on cooling
many fissures, cracks and folds.19,20 It is the selection, from graphitisation (∼2800°C) temperature due to the
by the graphite manufacturer, of the particular structure large difference in the CTE between the crystallite c-
of the filler coke to be used, that is the main contribution axis and a-axis; in the final product these nano-metre
to the properties of the final graphite product. wide cracks provide accommodation for thermal expan-
Two very different examples of coke are the needle- sion and irradiation-induced crystallite dimensional
shaped particles and the spherical particles used for changes.
the production of two nuclear graphite grades, PGA Coke supplies can vary over time as the original source
and Gilsocarbon. Each raw coke particle contains maybe discontinued, for example the cokes used to pro-
many lenticular cracks running parallel to the basal duce Gilsocarbon and PGA are no longer available.
planes as illustrated in Figs. 10 and 11. The Mrozowski Changes in coke supply can significantly change both
cracks21 which are a few 10s of nano-metres in width by the virgin and irradiated graphite properties. Thus, when

9 Graphite crystal lattice

8 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

10 Optical, SEM and TEM images of unirradiated nuclear graphite (Gilsocarbon) microstructure. a Polarised optical image of
Gilsocarbon showing the general orientation of the crystallite. Note the lenticular porosity in the filler and globular porosity
in the binder (porosity is shown as black). The general radial orientation of the crystallite ‘c’-axis within in the filler particle is
radially outwards from the centre. b SEM image of a fissure in a Gilsocarbon filler particle showing that although the general
direction of the crystallites is radial, in the finer structure the geometry is more complex and is made up of many folded
graphite plate structures. c SEM image of Gilsocarbon at a larger scale showing the many thousands of folded graphitic
sheets. d TEM image of Mrozowski cracks in Gilsocarbon, parallel to the basal planes (inside the individual folded graphitic
sheets, nano-metres in width and micro-metres in length, and providing accommodation porosity)

new grades of nuclear graphite are put forward a new used for forming the blocks largely determines the
irradiation programme will be necessary. However, the thermo-mechanical properties of the final product, such
size, scope and cost of such a programme can be signifi- as Young’s modulus, strength, thermal conductivity, and
cantly reduced through understanding the relationships CTE. Producing the blocks by extrusion can result in a
between the graphite microstructure and properties. highly anisotropic (orthotropic) product, whereas with
During manufacture, the coke is crushed and graded the appropriate choice of filler coke along with forming
into the required particle size, hot-mixed with a pitch-bin- techniques such as pressing (vibration moulding) can
der, and formed into billets known as the ‘green-article’. result in a semi-isotropic graphitic product22 as is the
As the pitch-binder is mixed and cooled, it also passes case of vibration-moulded Gilsocarbon graphite. As illus-
through a mesophase forming graphitisable structures, trated in the high-resolution micrographs in Figs. 10 and
binding the filler coke particles together. Gas evolved 11, both the binder and filler phases in the resultant poly-
during a subsequent baking stage (∼800°C) leads to the crystalline product consist of crystallite structures con-
generation of further porosity within the binder phase. taining many lenticular Mrozowski cracks a few nano-
To improve the final properties, there may be subsequent metres wide and micro-metres in length.21
impregnations under vacuum using a lighter liquid pitch The polarised optical image in Fig. 10a clearly shows
followed by re-baking. The final step is graphitisation, the well-defined crystal orientation of a filler particle in
which is conducted at around 2800°C.17 Gilsocarbon as opposed to the more chaotic distribution
The resulting product is a porous polycrystalline arte- in the binder. In addition, the lenticular cracks within
fact.4 The microstructure of the filler coke and the method the filler and the more globular porosity in the binder

International Materials Reviews 2016 9

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

11 Optical, SEM and TEM images of unirradiated nuclear graphite (PGA) microstructure. a Polarised optical image of PGA
showing needle-like filler particles. b SEM image of PGA filler particle. c SEM image looking ‘end-on’ at needle coke, clearly
showing the Mrozowski cracks between the graphitic sheets. d TEM image of Mrozowski cracks in PGA which provide
accommodation porosity

are also clearly shown. The optical polarised and SEM graphitic sheets.23 In addition, there is the possibility of
images shown in Fig. 11a, b and c illustrate the typical shear between graphitic sheet structures as well as ‘kink-
needle coke filler particles found in PGA. The difference ing’ and ‘twinning’ of the structures themselves, as
in nature of the needle coke particle when viewed observed by other authors.24 Compacting or stretching
longitudinally compared to the view ‘end-on’ is clearly areas of the microstructure may also influence the contri-
shown, as is the porosity within the folds given in bution of the a-axis thermal or irradiation shrinkage on
Fig. 11b and c. Longitudinally, the needle filler particle the bulk behaviour. It is also the case that due to the com-
microstructure is relatively stiff due to the covalent bond- plexity of the microstructure, tensile loading will lead to
ing, giving strength to the relatively flat basal planes. both tensile and compressive forces in the graphitic
Transversely, the folded porous structure can be more sheet-like structures, depending on their orientation and
easily elastically deformed on loading. At the nano-scale connectivity with the surrounding microstructure.
the Mrozowski cracks are shown in Figs. 10d and Similarly, both tensile and compressive crystallite forces
Figs 11d; these cracks offer space to accommodate either will also result from compressive loading. These
thermal or irradiation-induced crystallite growth (or forces will have the potential of opening and closing Mro-
both). zowski cracks and reorienting the graphite crystallite
It is the summation of the thermally driven crystallite structures.
expansion (or irradiation-induced dimensional change, During manufacture, the method used to form the
swelling and/or shrinkage) of the folded graphitic struc- graphite billets leads to a bias in the alignment of the filler
ture that governs the bulk thermal expansion and dimen- particles.22 With Grain (WG) is used to define the
sional change observed in a graphite component. direction in the billet in which the direction of most of
However, it is clear there is considerable scope for the the crystals ‘a’-axes are aligned and Against Grain (AG)
c-axis expansion to be accommodated by Mrozowski is the direction in the billet in which most of the crystallite
cracks and larger lenticular cracks between the folded ‘c’-axes are aligned. This bias defines the bulk material

10 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

anisotropy in graphite components. For example in the Nuclear graphite generally falls into one of the latter
case of the UK graphite grades, forming billets of PGA two definitions.
by extrusion lead to a highly anisotropic graphite, as the Also defined by the same two standards25,26 is the iso-
needle-shaped grains tend to be align in the direction of tropic ratio, which the ratio of a given property value in
extrusion, i.e. the extrusion direction is WG. In case of the against-grain direction to its corresponding value in
Gilsocarbon, forming the blocks using pressing (vibration the with-grain direction. Isotropic nuclear graphite is
moulding) tended to align the ‘c’ crystal axes at 90° to the graphite in which the isotropy ratio based on the CTE
pressing direction which is 90° to the brick axis. However measured over the range (25–500°C) is 1.00–1.10. Near-
this bias (and hence anisotropy) is much less pronounced isotropic nuclear graphite is graphite in which the isotropy
in Gilsocarbon than in PGA. based on the CTE measured over the range (25–500°C) is
It is clear from the images in Figs. 10 and 11 that the 1.10–1.15.
lenticular features within the folded graphitic sheet struc- Modern nuclear graphite strives to meet the first defi-
tures are reflected through to the larger scales within filler nition of isotropy ratio. Some of the earlier graphite
structures, while the binder phase contains many globular grades such as PGA are anisotropic in nature with high
gas-evolution pores. The two types of porosity constitute isotropic ratios.
about 20% of the bulk volume in typical polycrystalline Nominal virgin properties of Gilsocarbon, PGA and
graphite; porosity which is accessible to the atmosphere ATR-2E are given in Table 3, along with the grain size
is defined as ‘open’, that which is not, ‘closed’.9 Thus, an isotropy ratio defined by the ratio of against-grain to
typical polycrystalline graphite can be considered to be with-grain CTE. Gilsocarbon can be considered to be iso-
a porous polycrystalline material with crystallites at tropic and ATR-2E near-isotropic while PGA shows con-
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

many orientations biased by the filler coke structure and siderable anisotropic behaviour; all three grades are
the method used to form the graphite billet during manu- medium grained.
facture, bound together by a more randomly orientated Table 4 below gives virgin Poisson’s ratios for
binder phase. There will be a considerable amount of por- Gilsocarbon and PGA measured using digital image
osity, particularly the smaller lenticular (Mrozowski) por- correlation and an ultrasonic technique; the authors
osity lying parallel to the crystallite c-axis, together with do not know of any reference to the Poisson’s ratio of
other porosity having no particular orientation relative ATR-2E.
to the crystallite or fabrication process.
The final product grain size is defined in ASTM D7219
and C709-925,26 as: Averaging methods
. coarse grained — containing grains in the starting mix
that are substantially greater than 4 mm in size. If the effect of porosity is ignored, the linear CTE of
. fine grained — containing grains in the starting mix orthotropic polycrystalline material with a hexagonal
that are generally less than 100 μm in size. crystal structure (such as graphite or zirconium), can be
. medium grained, — containing grains in the starting described by:29
mix that are generally less than 4 mm in size. aVolume 1 2
aLinear = = ac + a a (3)
3 3 3
where αLinear, αVolume, αc and αa are the linear, volumetric,
Table 3 Nominal virgin properties of Gilsocarbon, PGA and crystal c-direction and crystal a-direction CTEs,
ATR-2E4,7,27
respectively.
Graphite Grade ATR-2E PGA Gilsocarbon Both αc and αa vary with temperature as shown in the
experimental data given in Fig. 12.
Coke Unknown Needle Gilsonite Thus if the crystallites were randomly orientated and
Grain size mm 1.0 0.8 0.5 there were no porosity a graphite would be expected to
Forming method Extrusion Extrusion Moulded
have a CTE of ∼8 × 10−6 K−1 at ambient temperature.
Density g/cm3 1.8 1.7 1.8
Young’s modulus GPa— WG 9.6 11.7 10.9 However, most artificial polycrystalline graphite is ortho-
Young’s modulus GPa — AG 8.4 5.4 10.8 tropic with a lower bulk CTE, ∼1 × 10−6 K−1 to ∼4 ×
Flexural strength MPa — WG 23.0 19.0 26.7 10−6 K−1 at room temperature.22 The reason for this is
Flexural strength MPa — AG 18.9 12.0 26.9 that in general, the crystallites are not randomly orien-
Compressive strength 55.9 27.0 70.0 tated and much of the lenticular porosity can absorb the
MPa — WG
large crystal c-axis expansion. A more useful relationship
Compressive strength 57.8 27.0 70.0
MPa — AG which accounts for orientation can be derived (Nye,30
Tensile strength MPa — WG 12.6 17.0 20.3 Tempest29).
Tensile strength MPa — AG 12.4 11.0 19.9
Coefficient of thermal 3.6 0.9 4.7
expansion ×10−6 K−1 Table 4 Poisson’s ratios for Gilsocarbon and PGA28
[20–120oC] – WG
CTE ×10−6 K−1 4.1 3.0 4.9 DIC
[20–120oC] — AG
Isotropic ratio based on CTE 1.14 3.33 1.04 Material and orientation Blocks Cylinders Ultrasonics
ratio
Thermal conductivity 179.0 200.0 137.9 PGA Extrusion (νxy) 0.08 0.09 0.12
W/mK — WG Transverse (νyz) 0.12 n/a 0.21
Thermal conductivity 163.0 109.0 137.9 Transverse (νyx) 0.05 n/a 0.06
W/mK — AG Gilsocarbon 0.19 0.22 0.23

International Materials Reviews 2016 11

 Marsden et al. Dimensional and property changes in nuclear graphite

For an individual hexagonal crystal the magnitude of a (6) is sensitive to reduction in moduli due to crystallite
property M in direction x at angle φ with respect to the imperfections or perhaps deformation of the slender crys-
(0002) basal plane can be defined by: tallite structures interspersed with Mrozowski cracks. Sla-
gle however, presented calculated plane strain and plane
M(w) x = Mc cos2 (w) + Ma (1 − cos2 (w)) (4) stress values of bulk CTE and modulus and compared
them to experimental data.32 He also presented a hybrid
where Ma and Mc are the single crystal properties in the
model that combined plane strain assumptions and
crystal a- and c-directions, respectively.
plane stress assumptions with reasonable success, but
Assuming that the crystallite contribute to a bulk prop-
the exact detail of these calculations were not made
erty of the polycrystalline structure can be defined in pro-
clear enough in his review to repeat them here.
portion to their volume fraction, Vi, then for crystals
It was suggested by Hill36 that the approach proposed
oriented with their [0002] pole c-axis at angle φi to the
by Voigt37 and Reuss31 should be further averaged to
reference direction, their contribution to the bulk prop-
give a better estimate of the mean value of the property.
erty can be written as:
In recent years the effect of texture or crystallographic
M (wi )x = Mc Vi cos2 (wi ) + Ma Vi (1 − cos2 (wi )) (5) orientation may be considered using Orientation Distri-
bution Functions (ODFs).38 ODFs are functions which
where Vi is the volume fraction of M in the ith direction describe the texture of a sample; it is a measure of the
Summing over the total volume: volume fraction of crystallites having a particular orien-


tation.39 Voigt, Reuss and Hill methods are available in
Mx =Mc Vi cos2 (wi ) + Ma Vi (1 − cos2 (wi ))
open source software used to estimate properties from dif-
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

i i
fraction texture measurements40 but it should be noted
or that these techniques do not account for inter-crystalline
Mx = Mc f + Ma (1 − f ) (6) porosity, such as that in nuclear graphite. Recently other
authors41,42 investigated the elastic properties of graphite
where f is a temperature, strain and irradiation-dependent under hydrostatic pressure and presented a self-consistent
function, which defines the crystal orientation distri- method based on averaging methods with some success.
bution of the crystallite and the probability of the individ- Similar methods have been applied to the analysis of the
ual crystal properties in either the a- or c-direction, effect of porosity on the modulus, strength and thermal
contributing to the overall bulk property. conductivity at The University of Manchester34,35,43–45
In the case of graphite it is therefore assumed that all and at present these techniques are being further investi-
parts of the structure contribute equally. That is, the bin- gated by one of the authors of this review (Haverty). Cer-
der phase (which we have called randomly oriented) con- tainly care must be taken when applying these averaging
tributes equally as much as the filler. A corollary to this is techniques to modulus. However, in this review the meth-
that if the binder is truly random, it will not contribute to odologies based on the proposal by Bacon46 are applied to
the texture but just give background intensity. the analysis of CTE and dimensional change behaviour as
The summation above assumes all the properties can be discussed below.
added in series (constant stress) and is often referred to as The relationship given in equation (6) has been shown
the Reuss approximation.31 Other authors32 have investi- to work very well for zirconium, a hexagonally closed
gated summing the properties in parallel (constant strain) packed metal which normally exhibits strong texture.
and in a combination of both parallel and series32 often Kearns47 showed that for Zircaloy-2 and Zircaloy-4
referred to as the Voigt approximation.33 there is a linear relationship between CTE and f. However
This leads to the following ‘plane strain’ expression (i.e. due to the porosity and complex microstructure of graph-
the contribution of the crystallites summed in parallel) for ite, application of this relationship is not straightforward,
bulk CTE: as discussed below. In addition, the validity of equation
p (6) relies on Mc and Ma being able to act independently.
2 Considering the folded plate structures interspersed with
I (w)E(w)a(w)sin(w)d w
0 Mrozowski cracks shown in Figs. 10a, b, and Fig. 11a
ax = p (7) and b, it can be seen that connectivity is mainly along
2 the a-axis while there is little connectivity along the crystal
I (w)E(w)sin(w)d w
0 c-axis. Thus this gives significant scope for independence
of Mc and Ma particularly at low crystal strains at
where I(φ) is a distribution function and E(φ) and α(φ) are which crystallite deformation can be accommodated.
orientation functions of crystallite modulus and CTE, Several authors have defined the CTE in polycrystal-
respectively. For further details the reader should refer line graphite using various averaging techniques based
to Slagle.32 on the above relationships through a variety of ‘poros-
Thus, in applying ‘plane strain’ it is necessary to relate ity and structure factors’ mostly stemming from the
crystallite modulus to bulk, macroscopic properties. This work of Bacon:46 Sutton and Howard48 took account
poses a difficulty as Young’s modulus is not only a func- of porosity in defining relationships for CTE in
tion of the magnitude and orientation of the crystal direc- PGA, Slagle32 developed constant stress and constant
tion and shear moduli but also the associated crystal strain models for Young’s modulus and CTE, Mor-
accommodation porosity.34 Bulk graphite modulus is gan49 proposed a two-phase model, Simmons50 devel-
also strongly related to the larger and more globular por- oped models derived from first principals for CTE
osity.35 In addition, if the theoretical values for the crystal and dimensional changes, and Bokros et al. 51,52 devel-
moduli are used, unrealistically high values for bulk mod- oped models for CTE and dimensional change for
ulus are obtained.35 This probably indicates that equation HTR application. Averaging techniques have also

12 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

12 Temperature dependence of graphite crystal coefficient of thermal expansion. Modified from data given by Kelly16. a ‘a’-
direction. b ‘c’-direction

been used to devise methodologies which are used to data for this purpose has become easier via the use of
predict the change in CTE as a function of pole figure techniques.61
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

temperature.53,54 To help the reader understand the basis of the Bacon46

It has also been observed that under load the CTE of techniques used by Sutton and Howard48 the experimen-
graphite changes significantly, from ∼4 × 10−6 K−1 to tal technique is briefly outlined below. However the reader
∼8 × 10−6 K−1 in Gilsocarbon graphite;55–57 increasing is encouraged to refer to a recent paper by Campbell62
with compressive loading and decreasing with tensile which describes the experimental technique and method
loading. In addition, changes to CTE of the same order of analysis in detail.
also take place when graphite is irradiated under load.58 A thin specimen is mounted with its extrusion (WG)
Fast neutron damage significantly changes the CTE, axis positioned as shown in Fig. 13a, which sets the pre-
and in semi-isotropic graphite such as Gilsocarbon, with ferred (0001) plane as a sample reference with direction
increased fluence the CTE first rapidly increases,59,60 as indicated by [0002] in the diagram. The incident x-
remains constant for a short period and then decreases. ray beam enters the rectangular specimen from the
The behaviour is more complex in highly orthotropic right. The Bragg diffraction ring pattern giving the
graphite, such as PGA4 where with increasing irradiation [0002] pole intensity is shown on the left of the figure sus-
the CTE increases, plateaus and then increases again. pended by the Bragg angle uB, in this case it is a maximum
This review revisits one of the main averaging methods at J and K and a minimum at F and G. The Debye–Scher-
often referred to in the literature, highlights some rer ring is projected around the azimuthal angle g, but it is
deficiencies, proposes a new variation to the relationship [0002] pole intensity as a function of the dispersion angle
and then explores the influence this has on CTE versus φ (i.e. the angle between the c-axes of the graphite crystals
temperature methodologies and the change in CTE with and the extrusion or pressing direction) which is required
load. The methodology is then successfully applied to (as shown in Fig. 13b). The derivation of the relationship
irradiation-induced dimensional change and irradiation between φ and g is not included here and any reader
creep in graphite. requiring further detail should consult reference.62
The graphite grades investigated are two UK nuclear Using pole intensity data such as that shown in Fig. 13b,
grades, PGA and Gilsocarbon, which are used as modera- Sutton and Howard48 defined the relationship between bulk
tors in the Magnox reactors and the Advanced Gas- CTE and crystal CTE using the following relationship:
cooled Reactors, respectively. However, due to the lack aWG = K1 Pc ac + K2 Pa aa
of high fluence UK data for irradiation creep, German (8)
aAG = K3 Pc ac + K4 Pa aa
ATR-2E graphite is examined.
where αWG and αAG are the WG and AG instantaneous
CTE of PGA graphite, αc and αa are the CTE of the indi-
Averaging porosity and orientation vidual crystallites16 in the c- and a-directions, respectively,
and K1, K2, K3 and K4 are orientation functions46 defined
relationships by the equations given below:
One of the most important texts on this subject is con- 
I (w) cos2 (w)sin(w) dw
sidered to be that of Sutton and Howard48 who used K1 = 
graphite orientation factors as defined by Bacon46 to pre- I (w)sin(w) dw
dict CTE in extruded anisotropic PGA graphite (both 
I (w) sin3 (w) dw
WG and AG). K2 = 
The methodology proposed by Bacon46 determines the I (w)sin(w) dw
intensity of the crystal [0002] poles in a graphite sample 
1 I (w) sin3 (w) dw
using XRD. This was done using transmission XRD, K3 = 
2 I (w)sin(w) dw
although modern equipment is mostly based on reflective
  
XRD; the main difference is that the former technique 1 I (w) sin3 (w) dw
samples a volume of material whereas the latter is a sur- K4 = 1 −  (9)
2 I (w)sin(w) dw
face sampling technique. Also, the analysis of XRD

International Materials Reviews 2016 13

 Marsden et al. Dimensional and property changes in nuclear graphite

13 Experimental setup and typical output used to determine the Bacon Anisotropy Factor (BAF) in graphite. a Transmission
XRD experimental set up due to Campbell.62 b Typical [0002] pole intensity data for PGA graphite, 0–90o shown as a function
of dispersion angle ∅. This is repeated as a sinusoid pattern around the 360o
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

where φ is the dispersion angle between the c-axes of the Hence the BAF at 400°C is given by:
graphite crystals and the extrusion direction, and I(φ) is
2+x
the solid angular density of the basal [0002] poles obtained BAF400 =
using XRD in transmission as described above. To maintain 2 − 2x
(13)
symmetry the orientation factors have the properties K1 + 2(BAF400 − 1)
x=
K2 = 1, K3 + K4 = 1 and K2 = 2K3. 2BAF400 + 1
The BAF is defined as
However, unlike the BAF, the orientation factor, ‘x’ is
invariant with temperature, thus is a more useful measure
aAG K3 ac + K4 aa
BAF = = (10) of anisotropy.
aWG K1 ac + K2 aa Therefore solution of equation (8) becomes

Examination of equation (10) shows that due to the aAG K2 − aWG K4

Pc =
crystallite CTE temperature dependence, BAF must also ac (K2 K3 − K1 K4 )
be temperature dependent varying from ∼2.2 at 100°C (14)
aAG K1 − aWG K3
to ∼1.9 at 600°C. However BAF is usually quoted at Pa =
aa ( − K2 K3 + K1 K4 )
∼400°C where αa = 016 thus:
Sutton and Howard48 also noted that as αa is zero at
aAG(400) K3 1 K2 approximately 400°C, both x and Pc could be calculated
BAF400 = = or (11) at this temperature as
aWG(400) K1 2 K1

aAG
aWG − 1
Sutton and Howard48 introduced the factors Pc and Pa x=
aAG
aWG + 2
to account for accommodation porosity aligned with the 1

crystal c- and a-directions. A value of unity implies zero (15)

porosity, and a positive value allows crystal growth due and
to an increase in temperature to be accommodated by
aWG + 2aAG
Mrozowski cracks21 in the c-axis direction. Presumably Pc =
a negative value allows for crystallites to shrink as at ac
low temperatures (below 400°C), the CTE in the a-axis Sutton and Howard48 measured αWG and αAG as a func-
is negative. tion of temperature on two samples of PGA and used
The Bacon46 relationships were also simplified by values for graphite crystal CTE αc and αa due to Nelson
Sutton and Howard48 using a single orientation factor and Riley63 to calculate Pc and Pa by solving equation
‘x’ as: (8), as given in Table 3.
The present authors used the orientation function given
1 in Fig. 2 of reference48 along with equation (12) to recalcu-
K1 = (1 − x) late K1, K2, K3 and K4 as 0.206, 0.794, 0.397 and 0.603,
3
respectively, which are identical to those in previous refer-
1
K2 = (2 + x) ence;48 these values lead to a BAF400 = 1.92. Solving
3 equation (8) at different temperatures results in the accom-
  (12)
1 1 modation factors Pa and Pc, see Table 5 and Fig. 14.
K3 = 1+ x
3 2 The reason for the discrepancy in results is due to a
  singularity in equation (8) at ∼400°C as αa approaches
1 1 zero, see Fig. 12. Sutton and Howard48 make no reference
K4 = 2− x
3 2 to the singularity in their paper, other than not giving
values of Pa near to 400°C. It is also unclear why their

14 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

Table 5 Re-calculation of porosity factors from reference48

Recalculated
Crystal expansion Bulk expansion Accommodation accommodation
× 10−6 °C48 × 10−6 °C factors48 factors
Mean temperature
(+/−50°C) αc αa αWG αAG Pc Pa Pc Pa

125 27.5 −0.93 1.04 2.92 0.320 1.93 0.322 1.07

175 27.8 −0.73 1.20 3.14 0.332 2.04 0.334 1.23
225 27.9 −0.55 1.30 3.34 0.350 2.15 0.351 1.65
275 28.0 −0.36 1.48 3.66 0.376 1.89 0.377 2.44
∗
325 28.0 −0.17 1.66 3.82 0.379 0.380 3.98
∗
375 28.0 −0.01 1.86 4.02 0.386 0.388 47.58
∗
425 28.1 0.13 2.20 4.20 0.373 0.375 0.29
475 28.2 0.26 2.04 4.32 0.407 1.40 0.409 −1.64
525 28.3 0.45 2.22 4.44 0.404 1.22 0.405 −0.40
575 28.5 0.47 2.34 4.48 0.393 1.20 0.395 0.05
625 28.7 0.56 2.28 4.32 0.374 0.43 0.375 0.13
675 28.9 0.65 2.38 4.50 0.386 0.77 0.388 0.13
∗
Denotes calculated values that approach infinity.
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

other values of Pa do not tie in with those calculated by the Solving equation (16) gives two possibilities for P
author. Clearly a discussion on this difficulty in solving dependent on grain direction:
their equations is an important omission from their paper.
From the low value of Pc it is clear that the porosity aWG − K2 aa
PWG =
between the microstructural folded graphitic plate struc- K1 ac − K2 aa
tures significantly reduces the influence of the relatively (17)
aAG − K4 aa
large crystal c-axis expansion, whereas the solution for PAG =
the porosity in the a-direction is clearly unsatisfactory. K3 ac − K4 aa
An alternative approach proposed by Reynolds64 and
and thus, the problem of the singularity has been
applied by Price and Brokos52 is to only apply a porosity
removed.
factor in the c-direction. The justification for this given by
To investigate the relationship given in equation (16)
Price and Brokos52 is that the contribution to the bulk
the instantaneous CTE of both PGA and Gilsocarbon
CTE from the crystal a-direction is small compared to
were measured using a Netzsch Dilatometer DIL 402C,
that from the crystal c-direction. However, the present
following the procedure suggested by British Standard
authors have taken an alternative approach by associating
BS EN-821-1.65 Mean curves were obtained from twelve
the porosity factor ‘P’ with the crystal c-axis and ‘1-P’
measurements taken from four Gilsocarbon samples
with the crystal a-axis. It can be argued that the assump-
and twenty PGA samples, as shown in Fig. 15. The refer-
tion of linking the contributions cannot be strictly justi-
ence standard used was well-characterised Al2O3, sup-
fied, but it is later shown that the application of this
plied by the dilatometer manufacturer. The criterion for
method results in linear small variations of P with temp-
acceptance of the results was that the standard deviation
erature and loading which are similar in trend for Gilso-
of the run was less than the accuracy of the technique:
carbon and PGA, graphite grades with very different
0.1 × 10−6 K−1. The instantaneous CTE was used to
microstructures. Another advantage of this approach is
solve the equations for PWG and PAG, see Fig. 14. Values
that a well-behaved set of equations are derived that can
of ‘x’ for Gilsocarbon and PGA were calculated from pole
be solved for both P and also for x if it is unknown.
figure data obtained using reflective XRD in a Philips
Thus equation (8) becomes
X’Pert-1 Texture XRD in u − 2u configuration, as 0.05
and 0.38 respectively, for which the intensity of the
[0002] pole peak is dominant, see Fig. 16. The data
were collected in 5° increments of tilt angle from 0 to
aWG = K1 Pac + K2 (1 − P)aa 85°. Defocusing is observed at high tilt angles (>75°).
(16)
aAG = K3 Pac + K4 (1 − P)aa The pole figures were corrected for background and defo-
cusing errors. An empirical defocusing correction was
applied by measuring the defocusing on POCO graphite,
which represented a texture-less graphite. The flake-par-
It should be noted that P does not represent the total ticle graphite microstructure would make it difficult to
graphite porosity as measured by helium pycnometry or produce a completely random powder. From these data
some similar method, but is the combined accommo- orientation functions K1, K2, K3 and K4 were calculated
dation and microstructural requirements to account for using equation (12) to be 0.316, 0.684, 0.342 and 0.658,
crystallite thermal growth and shrinkage in the bulk respectively, giving a BAF of 1.08 for Gilsocarbon,
graphite, and hence the probability that the c-axis CTE which are close to the one-third, two-third values for an
contributes to the bulk CTE. It is also important to isotropic material. The corresponding values used for
note that some porosity, particularly the globular porosity PGA are given above.
in the binder, will not play a part in the relationship given The calculated values for PWG and PAG are plotted in
in equation (16). Fig. 14. As Gilsocarbon is almost isotropic the WG and

International Materials Reviews 2016 15

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

14 Accommodation factors; original Sutton and Howard and revisited calculation. Note that the original and new Pc values
overlap each other. The ‘new’ data refers to the revised approach as discussed later in this paper

AG P values are almost the same in magnitude slightly where αT1 is the known instantaneous CTE at tempera-
higher than the Sutton and Howard48 values for Pc; ture T1, αcT1, αaT1, αcT2 and αaT2 are the crystal CTE in
whereas for PGA the values of P in the two directions the a- and c-directions at temperatures T1 and T2. αT2
are similar in magnitude and almost the same in value is the instantaneous CTE at the required temperature.
as Pc. The so-called structure factor A, which is invariant with
temperature, does not distinguish between accommo-
dation porosity and orientation as do equations (8) and
Methodology for calculating CTE over (16), or with anisotropy of the graphite, such as PGA
where a different value of the structure factor A is
various temperature ranges required for WG and AG.
The methodology used for calculating CTE over various Rearranging equation (18) gives
temperatures ranges54,57 is similar to that given in
equation (6), but is based on a thermodynamically aT 2 = AaT1 + Bi
based derivation due to Simmons50 and Hall et al.66 as or as mean CTE for engineering purposes:
given below for isotropic graphite.
T
2 (19)
aT1 = AacT1 + (1 − A)aaT1 1
(18) a(20−T2) = (Ai aT1 + Bi )dT
aT2 = AacT2 + (1 − A)aaT2 (T2 − 20)
20

where

(acT2 − aaT2 )
Ai = and Bi = aaT2 − aaT1 Ai (20)
(acT1 − aaT1 )

In nuclear graphite assessments T1 is usually taken as

70°C and mean CTE data are tabulated over the range
20–120°C, although measurements of CTE made over
this low and small temperature range are not rec-
ommended as the accuracy obtained can be poor.67
Using the crystal data given in,48 Ai and Bi have been
plotted in Fig. 17. As αa << αc, Ai approaches 1.0 × 10−6
K−1. However, Bi is a strong function of temperature indi-
cating that graphite temperature dependence is a strong
function of αa only. The reason for this may be attributed
15 Instantaneous CTE curves for Gilsocarbon and PGA to the significant relative change in crystal αa compared
Graphite with the smaller relative change in αc between T1 and T2.

16 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

16 Typical PGA and Gilsocarbon XRD pole figures (The images are of regenerated pole figures of the [0002] peak). The colour
scale used for each is the same (0.5–2.5 multiples of random distribution or m.r.d). a PGA (AG). b Gilsocarbon
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

Applying these equations, the instantaneous CTE for a loss. It also indicated that crystallite CTE is not changed
polycrystalline graphite at any temperature can be with increasing fast neutron fluence confirming measure-
obtained from a measurement made at only one tempera- ments made by Kelly et al.69 at irradiation temperatures
ture using the crystal αc and αa data available from the lit- above 300°C.
erature.16 This methodology works well for all graphite
grades22 and can even be applied to irradiated, radiolyti-
cally oxidised graphite as shown in Fig. 18 below (the
curves are normalised to 4 × 10−6 K−1 at 120°C). The
The effect of strain on CTE
data given in Fig. 18 was obtained from measurements In virgin graphite a significant change in CTE can be pro-
made on virgin, medium fluence and very high fluence duced due to loading.57,70 Whereas uniform change in
irradiated Gilsocarbon graphite samples, including some temperature may not be expected to significantly change
highly radiolytically oxidised samples irradiated at NRG crystal orientation in unrestrained graphite, the appli-
Petten in a recent Materials Test Reactor programme. cation of stress or strain may change CTE as reported
The MTR CTE data are compared to the theoretical by Oku et al.71 on stressed graphite and Zukas et al.72
CTE calculated using equation (19) and the graphite crys- on thermally crept graphite, as discussed later.
tal data given in Fig. 12. Considering the scatter in the Figure 19a shows the change in CTE in Gilsocarbon as
crystal CTE data and the extreme irradiation conditions a function of load taken from Preston and Marsden.57
the mean temperature-dependent CTE relationship is Recently Marrow et al. 73 measured the change of the lat-
adequate for reactor graphite structural integrity assess- tice ‘d’ with load. Taking the maximum change in the ‘d’
ments. However, in the case of highly anisotropic PGA given in Fig. 10 of their paper implies a crystal strain of
larger discrepancies have been reported68 which merit ∼284με. The crystal modulus, c33 in this direction is
further investigation. ∼36 GPa16 which implies a pressure on the basal plane
The observation that the temperature dependence of of ∼10MPa. From this the change in c-axis CTE can be
irradiated and well-oxidised graphite is not changed calculated using the methodology described by Kelly74
suggests that the crystallite connectivity remains largely which is about 0.2 × 10−6 K−1 or 1% which would not
intact despite the loss of carbon, in this case ∼ 45% weight explain the large changes in CTE given in Fig. 19a.

18 Mean CTE temperature dependence in irradiated Gilso-

17 Ai and Bi values based on crystal data taken from48 carbon graphite, normalised to 4 × 10−6 × K−1 at 120°C.

International Materials Reviews 2016 17

 Marsden et al. Dimensional and property changes in nuclear graphite

19 Change in CTE, orientation factor and porosity factor in virgin Gilsocarbon graphite as a function of stress. CTE data taken
from reference,57 measurements taken at room temperature. The average unloaded CTEs were 5.34 and 5.17 × 10−6 K−1 in
the loaded and unloaded directions, respectively. a Change in CTE due to loading. b Orientation factor ‘x’ and porosity factor
‘P’ as a function of load
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

Figure 19b shows the orientation factor x and the fac- reported by Oku et al. 71 They used XRD to measure
tor P for the data as a function of load calculated using the BAF in pre-stressed graphite, in samples both com-
equation (16). pressed at room temperature (25°C) and also compressed
Both the change in orientation and porosity factor are at 2010°C, a temperature at which thermal creep is signifi-
predicted to increase with increasing compressive load cant in graphite. On removal of the loading they obtained
and decrease with tensile load, although there is signifi- BAF values for isotropic IG-430 graphite of 1.06, 1.45 and
cant scatter in the tensile data, see Fig. 19b. 2.52 for as received, compressed at room temperature, and
This would indicate that under at least compressive compressed at 2010°C respectively.75 These results are
loading, both changes in porosity and orientation influ- remarkable in that the orientation factor more than
ence CTE. Preston et al.57 did not measure the change doubled on compression at 2010°C. Unfortunately, they
in either porosity or orientation as a function of load. did not measure the resultant CTE on these specimens.
However the implication of the significance of the The capacity for graphite to re-orient is illustrated by
changes given in Figs. 19a and b can be investigated by the SEM image of raw Gilsocarbon coke in Fig. 21a
comparison with the change in porosity factor P due to showing the tortuous arrangement of folded graphitic
temperature and by investigating the change in crystal sheets with significant porosity available for reorientation
orientation required to account for the large change in under loading. This structure is very obvious in the raw
CTE due to compressive load, as described below. coke but is still maintained in the final graphitised pro-
First, the change in porosity factor due to compressive duct, as shown in Fig. 21b.
loading is of a similar order to the change found for While these images show that polycrystalline graphite
Gilsocarbon with increasing temperature, as given in can accommodate strain the exact mechanism by which
Fig. 14. However, the effect of stress on CTE over the this is achieved require further study using techniques
loading range −50MPa to 10MPa (−5% to 0.1% strain) such as tomography.
is significantly greater than that for temperature change This ability for polycrystalline graphite microstructure to
in the range 100–700°C (∼0.3% unrestrained thermal deform and recover under loading has recently been illus-
strain), as shown in Table 6. This may indicate that the trated at the University of Manchester76 by subjecting
change in CTE is unlikely to be due to change in porosity graphite samples to very high hydrostatic pressure. The
alone. samples were compressed to around 80% of their original
Second, in order to explore the change in crystal orien-
tation required to achieve large changes in CTE of
−50MPa and 12MPa, theoretical pole density distri-
butions as a function of the dispersion angle φ have
been calculated using equation (9) and are shown in
Fig. 20. Such changes in orientation distributions are cer-
tainly achievable as can be seen from the measurements

Table 6 Comparison of the changes in CTE with thermal or

stress loading

Temperature Stress Strain CTE × 10−6

Loading o
C MPa % K−1

Stress 20 −50 4 8
Stress 20 10 −0.5 ∼4.5
Thermal 100–700 0 ∼0.3 ∼4.5 to ∼6.0
20 Predicted change in [0002] pole intensity due to loading

18 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

21 SEM images of Gilsonite coke and Gilsocarbon graphite. a Gilsonite coke. b Gilsocarbon graphite

volume thus utilising most of the internal porosity. Incred- undertaken at within the Nuclear Graphite Research
ibly the samples recovered almost completely after the Group at the University of Manchester.
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

hydrostatic pressure was removed, as shown in Fig. 22.

However, it is interesting to note that if deviatoric pressure
was applied the samples exhibited significant permanent set.
The reason graphite has this ability to deform and Observations on the effect of neutron
recover is probably related to the ability of the graphite
sheets to twin and kink24 allowing many millions of gra-
irradiation on graphite crystal and
phitic sheets to rearrange under external loading resulting polycrystalline CTE
in closure of porosity and crystal reorientation. The work Measurements of crystal CTE of irradiated Highly Orien-
of Oku et al. discussed previously gives some credibility to tated Pyrolytic Graphite (HOPG) imply that αc and αa
this possibility.71 In addition Zukas et al. 72 thermally between 300 and 650°C can be considered to be invariant
crept semi-isotropic POCO graphite to 54.3% of the orig- to fluence, up to a fluence of 300 × 1020 n cm−2 EDND
inal length resulting in a change in BAF from 1.0 to 2.88. (Fig. 23).
Unfortunately, again they did not measure the change in The irradiation-induced changes in CTE of polycrystal-
CTE. As well as recoverable deformation, kinking, and line PGA and Gilsocarbon graphite are given in Fig. 6a
twinning, there is the possibility that the compression and b in the temperature range 300–650°C. The CTE of
applied by Oku et al. 71 or the creep deformation of PGA increases slightly with irradiation; the increase is
Zukas et al. 72 might have damaged graphite crystals in more significant AG at the lower temperatures, but at
some orientations (e.g. by plastic deformation), which the higher temperatures 450°C and above, the CTE
would affect their scattering intensity. The investigation both WG and AG is considered to be invariant to fast
of this possibility is important as it would also affect the neutron fluence for practical purposes. The reasons for
pole figure intensities. this may be the relatively large inter-crystalline accommo-
It is clear that further investigation of crystal orien- dation in the PGA filler particles.
tation distribution due to loading is required and at pre- The CTE behaviour of Gilsocarbon, and other semi-
sent, orientation factor change due to thermal and isotropic graphite grades, is more complex as there is an
irradiation creep is one of the main themes being initial increase followed by a decrease to around half the
unirradiated value. A recent detailed statistical analysis
of this and other CTE data59,60 has shown the initial
rise to be almost instantaneous.
It may be expected that an increase in CTE would be
correlated to a decrease in Young’s modulus.78 However
in both PGA and Gilsocarbon, fast neutron irradiation
causes an initial increase in modulus and CTE, see Fig.
6. In the case of Gilsocarbon, after this initial increase
in CTE and modulus, the modulus starts to rapidly rise
again and there is a fall in CTE which does appear to
be correlated with modulus as expected. At very high flu-
ence as the structure of the graphite degenerates, the mod-
ulus falls and the CTE stabilises at a value of about half
the initial CTE. In the case of PGA, over the data
range available the modulus continues to increase but
there is significant scatter in the data. It may be inferred
that as there are no driving forces to significantly change
22 Volumetric strain in PGA graphite subject to very high the crystal orientation and there is an observed invariance
hydrostatic pressure.76 The original volume was almost of crystal CTE to fast neutron fluence, see Fig. 23, the
completely recovered on unloading (load path shown changes in CTE in polycrystalline graphite are mainly
by arrows)

International Materials Reviews 2016 19

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

23 Crystal CTE as a function of fluence measured on HOPG graphite samples irradiated between 300 and 650°C.69,77 a αc. b αa

related to porosity changes due to crystal dimensional

changes as discussed below.

Analysis of irradiation-induced
dimensional change
Using the same considerations as given by equation (6)
the linear dimensional change dl/l in an orthotropic
material should be given by:
 
dl 1 DXc DXa
= +2 (21)
l 3 Xc Xa
24 Experimental dimensional change data measured on where ΔXc/Xc and ΔXa/Xa are the irradiation-induced
HOPG samples irradiated at 430°C and 600°C79 crystal dimensional changes perpendicular and parallel

20 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

25 Application of averaging models to MTR Gilsocarbon graphite data, irradiated at 400°C and 600°C.80 a Prediction using
equation (21). b Prediction allowing for crack closure at 30 × 1020 n/cm2 EDND. (Experimental data at 600°C shown in
blue80). c Prediction using equation (22). Fit defined using the exact function and a simplified version of ‘F’, along with
the experimental irradiation data at 600°C.80 d Prediction using equation (22). Fit defined using the exact and a simplified
version of ‘F’ along with the experimental irradiation data at 430°C80

to the basal planes as a function of fluence and tempera- Fortunately, a good set of irradiation data on crystal
ture. This approach to predicting dimensional change in dimensional change has been obtained by irradiating
polycrystalline graphite was used by Price and Bokros.52 HOPG79 at 430°C and 600°C as given in Fig. 24.
However their analysis did not extend as far as dimen- Applying equation (21) directly predicts constant
sional change ‘turnaround’, i.e. the change from shrink- expansion as the significant c-axis growth dominates,
age to growth as illustrated in Fig. 3. Here, that analysis see the red curve in Fig. 25a.
is taken well beyond this point for Gilsocarbon and Assuming instant crack closure at the dimensional
PGA graphite. change inflection point (∼30 × 1020 n cm−2 EDND)
gives the prediction in Fig. 25b. This underestimates the
fluence at turnaround and overestimates the dimensional
change after turnaround, as well as giving a step change in
dimensional change rate.
From microstructural observation81 and the use of
small angle neutron scattering (SANs)82 it is known that
there is a distribution of lenticular cracks from the
nano-scale through to the micro-scale capable of accom-
modating c-axis expansion. This is believed to be confir-
mation of the presence of Mrozowski21 cracks. In
addition, mercury porosimetry appears to indicate that
it is changes to the small pores, less than ∼1μm, that influ-
ence volumetric change in irradiated graphite.83 The
cumulative closure of this accommodation plus other
structure effects will enhance the influence of the crystal
c-axis dimensional changes. Thus equation (21) has
been modified by the inclusion of an influence function
‘F’ (a function of both fluence and temperature) to give:
26 Influence functions ‘F’ for Gilsocarbon specimens irra-  
diated at 430°C and 600°C80 and ATR-2E7 creep control
dl 1 DXc DXa
= F +2 (22)
specimens irradiated to 500°C l 3 Xc Xa

International Materials Reviews 2016 21

 Marsden et al. Dimensional and property changes in nuclear graphite

27 Prediction of dimensional change in PGA graphite irradiated at 600°C and 430°C, using equation (22), compared to the PGA
experimental irradiation data.80 a 600°C. b 430°C

The function F has been obtained by fitting curves to The orientation factor ‘x’ was taken as 0.4 and was
the Gilsocarbon irradiation data80 and the appropriate obtained using XRD pole data.
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

HOPG experimental data79 and then iterating between An important conclusion may be drawn from this
equation (22) and a statistical fit made to the experimental analysis of PGA, i.e. the analysis indicates that similar
dimensional change irradiation data. At low values of flu- ‘F’ functions are appropriate for both Gilsocarbon and
ence the iteration produces unrealistically high values, PGA graphite implying that in both cases the amount
and therefore back extrapolation was used to determine of porosity associated with crystal c-axes is similar in
the F function below a fluence of ∼30 × 1020 n cm−2 volume.
EDND.
The resulting prediction of the irradiation-induced
dimensional change in Gilsocarbon graphite irradiated Implications for irradiation creep in
at 600°C compared to actual irradiation data80 is given
in Fig. 25c. A similar prediction for Gilsocarbon irra-
graphite
diated at 430°C is given in Fig. 25d. Irradiation creep in graphite is defined as the difference in
The influence functions F for Gilsocarbon are plotted dimensional change in a loaded specimen compared to
in Fig. 26 along with fits to data for two creep control the dimensional change in an unloaded specimen irra-
ATR-2E specimens irradiated at ∼500°C in HFR Petten.7 diated under the same conditions. Irradiation creep is
The F function for Gilsocarbon at 600°C tends to zero at very important in graphite-moderated reactor design
zero fluence, whereas the Gilsocarbon 430°C specimen at because it relieves significant thermal and shrinkage stres-
zero fluence has a finite value, indicating that c-axis crys- ses that would lead to component failure after a few years
tal swelling needs to be accounted for at the lower operation without the presence of irradiation creep.11
temperature. Irradiation creep has been attributed to basal plane slide
The shape of the F function for the ATR-2E com- due to pining and un-pinning of dislocations in the crystal
pression control specimen is similar to the Gilsocarbon basal planes,84 but as yet there is no convincing micro-
600°C at low fluence but diverges slightly at high fluence. structural evidence of this mechanism.
However, the ATR-2E tension control specimen at low The analysis of CTE and dimensional change discussed
fluence has a finite value, which implies that the ATR- above has reflected the strong geometric anisotropy of the
2E tension control was irradiated at a lower temperature graphite crystal structure with weak bonding perpendicu-
than the compression control. This creep experiment was lar and strong bonding parallel to the basal planes that
not properly reported due to the closure of the German leads to significant anisotropy in properties and crystal
HTR reactor programme so it is not possible to check if behaviour, both in the virgin and irradiated condition.
this was the case. The function F tends to unity at very In addition, the microstructure of the polycrystalline
high fluence implying the influence of the c-axis expansion graphite provides significant porosity associated with the
dominates. crystal c-axis which is available to accommodate both
From equation (21) and examination of Fig. 25 it can thermally- and irradiation-induced c-axis expansion,
be observed that the initial dimensional change rate in iso- thus allowing properties associated with the crystallite a-
tropic Gilsocarbon graphite is approximately one-third of and c-axes to be considered to be largely independent of
the crystal a-axis dimensional change rate (as defined by each other in averaging crystal properties to derive bulk
irradiating HOPG) which explains the reasonable fits properties. The approach used for dimensional change is
obtained before ‘turnaround’.52 The c-axis crystal expan- taken one step further below to examine irradiation creep.
sion progressively increases with increased fluence, but has Irradiation creep per unit elastic strain in graphite was
not reached unity in the extent of the polycrystalline data. defined by Kelly and Brocklehurst10 by the following
The same approach was applied to anisotropic PGA relationship:
graphite. The 430°C and 600°C functions for ‘F’ used s s
are those obtained for Gilsocarbon, a very different 1cr = [1 − e−4g ] + 0.23 g (23)
Ecr Ecr
microstructure to PGA. However, the model fits corre-
spond remarkably well to the MTR data,80 see Fig. 27. where εcr is creep strain defined in units of elastic

22 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

28 Difference between loaded and unloaded CTE in for irradiation creep data and data on virgin graphite under stress. a Data
plotted against irradiation creep strain and for virgin graphite against elastic strain. b Data plotted against elastic strain only

deflection i.e. normalised by dividing the measured strain and Bradford86 who added an extra term to account
by the creep modulus Ecr and multiplying by the applied for data that inferred the presence of high-fluence, reco-
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

stress, σ. The fast neutron fluence is γ. verable creep.

The first term in equation (23) accounts for short term It had also been observed that in irradiated graphite
non-linear transient recoverable creep which is one elastic CTE was modified by creep strain58 (Fig. 28a). The mag-
deflection and the second term is linear non-recoverable nitude of this change in CTE was of the same order as that
creep. observed by Preston et al. 57 in loaded virgin graphite, but
The creep modulus is defined as the magnitude of creep strain leading to the change in
  CTE in irradiated graphite was much larger than the elas-
E tic strain in virgin graphite, see Fig. 28a. However when
Ecr = (24)
E0 s both sets of data are plotted against elastic strain, both
the irradiation and virgin data are reconciled, see Fig.
where E is the instantaneous modulus and Eo is the initial
28b. Note that although Davies and Bradford86 include
modulus.
The subscript
  s refers to the so-called structure recovery strain as well as elastic strain, only elastic strain
term E
E0 s = E
E0 / E
E0 p which accounts for the fall in is considered here.
creep rate at medium fluence and the rise in creep rate It has been previously demonstrated in this review that
at high fluence.4 This term is derived from the Young’s the change in CTE in virgin graphite measured under load
 may be attributed to crystal re-orientation. This re-orien-
modulus data given in Fig. 6. EE0 is the initial rise in tation in loaded virgin graphite, along with the change in
p
modulus attributed to dislocation pinning.10 CTE, is removed on unloading apart from a small
Up to a fast neutron fluence of ∼60 × 1020 n cm−2 residual.57 It is postulated here that the same mechanism
EDND, which is the extent of the UK Gilsocarbon occurs in irradiation crept specimens, leading to a recov-
and PGA creep data, this equation was found to be ery of the frozen-in primary creep and hence CTE.58
independent of loading direction, i.e. tension or As previously stated, unfortunately there are no
compression, and applicable to many grades of nuclear irradiation creep data for Gilsocarbon or PGA graphite
graphite.10 However, at higher fluence the equation with fluence greater than ∼60 × 1020 n cm−2 EDND.
did not fit irradiation data for other non-UK However, there are reasonable data for another med-
graphite grades. Other authors85 modified equation ium-grained graphite, namely ATR-2E; there are data in
(23) using dimensional change instead of modulus. tension and compression up to ∼200 × 1020 n cm−2
More recently equation (23) was modified by Davies EDND, see Fig. 5.

29 F and G functions for tension and compression loading. a F functions. b G functions for ATR-2E Graphite

International Materials Reviews 2016 23

 Marsden et al. Dimensional and property changes in nuclear graphite

30 Predicted and experimental irradiation dimensional change and creep curves in ATR-2E graphite. a Model predictions (solid
lines) compared to the unloaded and loaded experimental dimensional change data (symbols). b Model predictions (solid
lines) compared to the irradiation creep data (symbols)

By examining Fig. 5 it is clear that the effect of stress is shown in Fig. 30b, along with the UKAEA creep law
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

to accelerate the initial dimensional change rate, which as defined by equation (23). The ATR-2E creep curves
discussed above in unloaded graphite is set by one-third of given in Fig. 30 are identical to those given elsewhere.7
the crystal a-axis rate. This rate has been significantly Considering the simplicity of the averaging orientation
increased or decreased due to compressive or tensile load- equations, the similarities between the experimental and
ing, respectively. predicted irradiation creep behaviour are remarkable.
Thus taking account of the independence of the a- and If after a certain fluence, to simulate unloading, the
c-axes, the irradiation creep equation (22) can be rede- function G is relaxed back to unity the dimensional change
fined as returns back towards the unloaded dimensional change
curve by the remaining offset, see Fig. 31a. This behaviour
dl DXc DXa
=F K1 + 2G K2 (25) can be compared with experimental data given in Fig. 31b
l Xc Xa and may possibly be explained by the remarkable ability of
K1 and K2 are the usual anisotropy factors defined by graphite to recover deformation as illustrated in the high
the CTE elastic strain relationship given in Fig. 28b, and compression experiments, see Fig. 22. However, in the
in this case the orientation factor x defining K1 and K2 is compression experiments the recovery was non-linear
0.046 and −0.056 for compression and tension, respect- ‘elastic’ behaviour whereas in the case of irradiation, con-
ively. It is this behaviour that may explain the difference tinued fast neutron fluence is required to ‘unlock’ the
in the compressive and tensile irradiation creep change internal stresses that have been accumulated under load.
behaviour observed in ATR-2E graphite.7
The F function, as previously described, allows for the
delay in the influence of the c-axis crystal expansion. It
Implication of radiolytic oxidation
has been recalculated by iteration between a fit to the Both dimensional change and CTE are modified by
experimental ATR-2E unloaded control sample data, weight loss4 but not as much as may be expected when
compressive and tensile, and equation (22) as described considering the large change in Young’s modulus which
previously for Gilsocarbon graphite above. The resultant can be equated to loss in section.4 This may imply that
F curves are shown in Fig. 29 below. At low fluence these it is changes to the finer porosity, i.e. Mrozowski cracks
curves diverged because of the initial swelling and delay in that are mainly associated with changes to dimensional
shrinkage as discussed previously. Above fluences of change and CTE, while the larger globular-shaped poros-
about 60 × 1020 n cm−2 EDND the change defined by ity is associated with the large changes in Young’s mod-
the F function is similar for both ATR-2E and Gilsocar- ulus due to radiolytic oxidation. However further
bon in the temperature range 430–600°C. studies are required to confirm this observation. The
The function G is a function of load which increases the irradiation creep rate is increased by radiolytic oxidation
influence of the a-axis dimensional change in the loaded but data are sparse.10
crystallite. The G function was obtained by iterating
between equation (25) and a fit to the unloaded dimen-
sional change curve, as with the F function the values of
Discussion
G, below a fluence of 30 × 1020 n cm−2 EDND were calcu- The highly anisotropic nature of the graphite crystal
lated using back iteration to avoid unrealistic high values. coupled with the complex porous microstructure of
The influence of the function G on loaded dimensional nuclear polycrystalline graphite gives rise to complex
change behaviour is the greatest at low to medium fluence. thermal and irradiation-induced property changes. The
Its value depends on whether the loading is tension or approach taken in this review is based on averaging
compression, changing from a value of ∼0.3 or ∼1.4 to methods. The applicability of these simple methods lies
a value of ∼0.9 or ∼1.1 for compression and tension, in the strong crystal anisotropy and the considerable por-
respectively, see Fig. 29b. The resulting dimensional osity available to accommodate c-axis thermal and
change behaviour for ATR-2E is shown in Fig. 30 and irradiation expansion. This porosity has to be accounted
the irradiation creep curves are comparable to those for when assessing thermal and irradiation-induced

24 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

31 Dimensional change predictions of loading and unloading compared to irradiation data.87 a Unloading prediction (solid
lines) compared with experimental data (symbols). b UKAEA irradiation data loading and unloading

changes to CTE, dimensional change and creep. The ther- tool to obtain further understanding. In addition further
mal and irradiation-induced property changes discussed development of the use of finite element techniques used
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

here are shown to be a primary function of crystal prop- to model irradiation-induced microstructural changes
erties, crystal orientation and porosity associated with pioneered by Hall et al.66,88 and more recently by Delan-
the c-axis. nay et al.89 using a modified crystal plasticity model may
In the case of CTE the averaging technique has been further enhance mechanistic understanding of these com-
shown to be capable of accounting for changes due to plex processes in respect to the properties investigated
temperature and loading. However, in the case of changes here as well as others such as modulus and thermal
due to fast neutron irradiation the initial increase in both conductivity.
CTE and modulus is difficult to explain and requires
further investigation. The analyses clearly show the influ-
ence of porosity and crystal orientation on properties; this
Conclusions
sensitivity is probably related to the very high value of The influence of the microstructure of polycrystalline
crystal CTE in the c-axis direction compared to the crys- nuclear graphite on dimensional stability and material
tal a-axis direction. properties has been discussed in detail. It is shown that
The application of the averaging method to dimen- different choices of raw materials and manufacturing
sional change clearly shows the change from the influ- technique can lead to significant differences in graphite
ence of crystal a-axis shrinkage to that of the c-axis properties and irradiation behaviour. This is important
expansion. One of the remarkable observations is the because when raw materials are discontinued both graph-
similarity in the accommodation function for semi-iso- ite virgin and irradiated properties will change, leading to
tropic Gilsocarbon, anisotropic PGA and German a need for the necessity of time consuming and expensive
ATR-2E; three graphite grades with very different graphite irradiation programmes. However, the size, scope
microstructures. The initial dimensional change swelling and cost of such programmes can be significantly reduced
observed at certain temperatures clearly merits further through a better understanding of the relationships
investigation. between the graphite microstructure and properties.
The application to irradiation creep may appear, at first It is demonstrated that dimensional and property
reading, to stretch the application of the methodology. changes to graphite core components can lead to the gen-
However, if as it appears from observation of the poly- eration of significant internal and thermal stresses and
crystalline graphite microstructure the crystal behaviour that dimensional stability can influence whole core behav-
in the a- and c-axes can be considered independent, iour, thus challenging the structural integrity of the graph-
then the analysis is valid. The creep mechanism appears ite core.
to change the initial dimensional change rate. This may Examination of graphite using microscopy reveals a
be due to microstructural deformation made possible by complex microstructure containing a significant amount
the significant available porosity and the ability of the gra- of porosity. Microscopy shows sheets of graphite crystal-
phitic structure to twin, kink and recover. It is possible lites that form twisted structures that can deform, and
there may be some basal plane shear, but it is difficult to recover, to a certain extent, under load. Irradiation causes
see how this mechanism would recover. It also appears the individual crystallites to swell in one direction and
that deformation, and accompanying stresses, in the pres- shrink in the other. However significant swelling and
ence of fast neutron irradiation may be locked-in, and will load deformation can be accommodated by porosity
only recover on the removal of loading while still subject which extends in size from the nano-scale to the micro-
to irradiation. The energy input of fast neutron scale.
irradiation may provide the driving force to activate The thermal expansion, dimensional change and
recovery mechanisms. irradiation creep are analysed using modified averaging
It is clear that further microstructural investigation of methods accounting for graphite crystal orientation and
the evolution of the microstructural and property changes porosity. The methods are applied to Gilsocarbon, PGA
in irradiated nuclear graphite is desirable. Hopefully the and ATR-2E nuclear graphite. The methodology
techniques described in this review will provide a useful accounts for crystal properties and orientation change

International Materials Reviews 2016 25

 Marsden et al. Dimensional and property changes in nuclear graphite

and porosity accommodation in the crystallite c-axis. It is 2. R. Hino, M. A. Fütterer and Y. Hassan: ‘Special edition on HTR
shown that the effect of temperature on CTE can be 2012 conference’, Nucl. Eng. Design, 2014, 271, 1.
3. National Institution of Standards and Technology US Department
related to changes in the c-axis accommodation whereas of Commerce, Gaithersburg MD: ‘Neutron scattering lengths and
changes in CTE due to loading are mainly related to cross sections’, 7 January 2013; Available from: http://www.ncnr.
changes in crystal orientation. Dimensional changes in nist.gov/resources/n-lengths/.
the irradiation temperature range 400–600°C are shown 4. B. J. Marsden and G. N. Hall: ‘4.11 — Graphite in gas-cooled reac-
tors’, in ‘Comprehensive nuclear materials’, (ed. R. J. M. Konings),
to be initially driven by crystal a-axis shrinkage and 325–390; 2012, Oxford, Elsevier.
later by c-axis growth, as porosity is taken up by crystallite 5. E. D. Eason, G. Hall and B. J. Marsden: ‘Development of a model of
irradiation swelling. One of the remarkable observations dimensional change in AGR graphites irradiated in inert environ-
is the similarity in the accommodation function for ments’, Spec.l Publ.-Roy. Soc. Chem., 2007, 309, 43.
semi-isotropic Gilsocarbon, anisotropic PGA and Ger- 6. T. D. Burchell and L. L. Snead: ‘The effect of neutron irradiation
damage on the properties of grade NBG-10 graphite’, J. Nucl.
man ATR-2E; three graphite grades with very different Mater., 2007, 371, (1–3), 18–27.
microstructures. 7. G. Haag: ‘Properties of ATR-2E Graphite and property changes due
Irradiating graphite under load is shown to change the to fast Neutron irradiation’, Vol. Jul-4183; 2005, Jülich, FZJ, Institut
influence of the a-axis shrinkage either due to microstruc- für Sicherheitsforschung und Reaktortechnik.
8. J. V. Best, W. J. Stephen and A. J. Wickham: ‘Radiolytic graphite oxi-
tural deformation, or basal plain shear. However, it is dation’, Prog. Nucl. Energy, 1985, 16, (2), 127–178.
unclear how the latter mechanism could be recoverable. 9. B. T. Kelly, P. A. V Johnson, P. Schofield, J. E. Brocklehurst and M.
The main conclusion of the paper is that closure of por- Birch: ‘U.K.A.E.A. northern division studies of the radiolytic oxi-
osity and possible reorientation of crystals can have a sig- dation of graphite in carbon dioxide’, Carbon, 1983, 21, (4), 441–
nificant effect on the response of the material to 449.
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

10. B. T. Kelly and J. E. Brocklehurst: ‘UKAEA reactor group studies of

irradiation, oxidation and stress. irradiation-induced creep in graphite’, J. Nucl. Mater., 1977, 65, 79–
It is clear that in order to verify and extend the obser- 85.
vations in this review that there is a need for more detailed 11. D. K. L. Tsang and B. J. Marsden: ‘Constitutive material model for
examination of irradiated graphite specimens both the prediction of stresses in irradiated anisotropic graphite com-
ponents’, J. Nucl. Mater., 2008, 381, (1–2), 129–136.
unloaded and crept. Characterisation of irradiated speci- 12. G. B. Neighbour: ‘Management of the Ageing process in graphite
mens is required using techniques such as XRD, tomogra- reactor cores’; 2007, Cambridge, The Royal Society of Chemistry.
phy and the various types of microscopy available, along 13. G. B. Neighbour: ‘Securing the safe performance of graphite reactor
with property measurements made in three-orthogonal cores’; 2008, Cambridge, The Royal Society of Chemistry.
directions. In particular a well-characterised set of irra- 14. G. B. Neighbour: ‘Modelling and measuring reactor core graphite
properties and performance’; 2011, Cambridge, The Royal Society
diated crept specimens along with their control specimens of Chemistry.
would lend much to understanding. Using such methods 15. T. D. Burchell: ‘4.10 — Radiation effects in graphite’, in
it should be possible to improve the understanding of ‘Comprehensive nuclear materials’, (ed. R. J. M. Konings), 299–
the mechanisms and process which lead to irradiation- 324; 2012, Oxford, Elsevier.
16. B. T. Kelly: ‘Physics of graphite’; 1981, London, Applied Science.
induced dimensional and property changes in graphite 17. W. P. Eatherly and E. L. Piper: ‘Chapter two — manufacture’, in
core components in order to help to extend the life of ‘Nuclear graphite’, (ed. R. E. Nightingale), 21–51; 2013, New
existing graphite-moderated reactors and to design high- York, Academic Press.
temperature Generation IV reactors. 18. H. F. Kretchman: ‘The story of Gilsonite’; 1957, Salt Lake City,
Utah, American Gilsonite Company.
19. H. Marsh, J. W. Akitt, J. M. Hurley, J. Melvin and A. P. Warburton:
Acknowledgements ‘Formation of graphitisable carbons from gilsonite pitch and polyvi-
nyl chloride — a mass spectrometric and N.M.R. study’, J. Appl.
This work was made possible through funding provided Chem. Biotechnol., 1971, 21, (9), 251–260.
by the EPSRC grant EP/I003312/1, ‘Fundamentals of 20. J. D. Brookes and G. H. Tayor: ‘The formation of some graphitizing
current and future uses of nuclear graphite’ and the Euro- carbons’, in ‘Chemistry and physics of Carbon’, (ed. P. L. Walker
pean Commission FP7-Euratom-Fission project Jr), 243–286; 1968, London, Marcel Dekker INC.
‘Advanced High-Temperature Reactors for Cogeneration 21. S. Mrozowski: ‘Mechanical strength, thermal expansion and
structure of cokes and carbons’, in ‘First biennial conference on car-
of Heat and Electricity R&D’ (ARCHER). Project refer- bon’; 1953, Buffalo, American Carbon Society, University of
ence: 269892. The authors also acknowledge funding by Buffalo.
the UK Office for Nuclear Regulation (ONR). Any 22. D. J. Page: ‘The industrial graphite handbook’; 1991, Danbury, CT,
views expressed in this review are those of the authors UCAR Carbon Company Inc.
23. K. Wen, J. Marrow and B. Marsden: ‘Microcracks in nuclear graph-
and not necessarily those of ONR. ite and highly oriented pyrolytic graphite (HOPG)’, J. Nucl. Mater.,
2008, 381, (1–2), 199–203.
24. P. J. Heard, M. R. Wootton, R. Moskovic and P. E. J. Flewitt:
ORCID ‘Deformation and fracture of irradiated polygranular pile grade A
B. J. Marsden http://orcid.org/0000-0002-5139-0251 reactor core graphite’, J. Nucl. Mater., 2011, 418, (1–3), 223–232.
25. ASTM: ‘Standard terminology relating to manufactured carbon
M. Haverty http://orcid.org/0000-0003-2553-2743 and graphite’; 2009, West Conshohocken, PA, ASTM International.
W. Bodel http://orcid.org/0000-0002-8339-7968 26. ASTM: ‘Standard specification for Isotropic and near-isotropic
G. N. Hall http://orcid.org/0000-0002-8692-4110 nuclear graphites’; 2014, West Conshohocken, PA, ASTM
A. N. Jones http://orcid.org/0000-0002-1350-4617 International.
27. M. Haverty: ‘The microstructure, texture and thermal expansion of
P. M. Mummery http://orcid.org/0000-0003-0128-4319 Nuclear graphite’, PhD thesis, University of Manchester,
M. Treifi http://orcid.org/0000-0002-5820-9152 Manchester, 2015.
28. J. L. Lord, Maria, K. Grants, R. Morrel and J. Jiang: ‘Experimental
study of the poisson’s ratio of graphite’, in ‘EDF Graphite Research
References Meeting, University of Manchester’; 2008, NPL.
1. R. E. Nightingale: ‘Graphite in the nuclear industry’, in ‘Nuclear 29. P. A. Tempest: ‘Preferred orientation and its effect on bulk physical
graphite’, (ed. R. E. Nightingale), Chap. 1, 1–20; 2013, New York, properties of hexagonal polycrystalline materials’, J. Nucl. Mater.,
Academic Press. 1980, 92, (2–3), 191–200.

26 International Materials Reviews 2016

 Marsden et al. Dimensional and property changes in nuclear graphite

30. J. F. Nye: ‘Physical properties of crystals’; 1964, London, Oxford 54. W. H. Martin and M. F. Entwisle: ‘Thermal expansion of graphite
University Press. over different temperature ranges’, J. Nucl. Mater., 1963, 10, (1),
31. A. Reuss: ‘Berechnung der Fließgrenze von Mischkristallen auf 1–7.
Grund der Plastizitätsbedingung für Einkristalle’, ZAMM — 55. P. E. Hart: ‘Experimental relation between Dubinin-Polanyi
J. Appl. Math. Mech./Zeitschrift für Angewandte Mathematik und equation and langmuir equation’, Carbon, 1972, 10, (2), 231–233.
Mechanik, 1929, 9, (1), 49–58. 56. I. W. Gazda: ‘Variations in CTE as a function of prestressing’,
32. O. D. Slagle: ‘Averaging techniques in polycrystalline graphites’, Carbon, 1970, 8, (4), 511–515.
Carbon, 1968, 6, (1), 111–121. 57. S. D. Preston and B. J. Marsden: ‘Changes in the coefficient of ther-
33. W. Voigt: ‘Lehrbuch der Kristallphysik’; 1928, Leipzig Berlin, B.G. mal expansion in stressed Gilsocarbon graphite’, Carbon, 2006, 44,
Teubner, [Ann Arbor, Mich.], [J.W. Edwards], 1928. [1946], Series:B. (7), 1250–1257.
G. Teubners.; Sammlung von Lehrbü chern auf dem Gebiete der 58. B. S. Gray, J. E. Brocklehurst and A. A. McFarlane: ‘The irradiation
mathematischen Wissenschaften, mit Einschluss ihrer induced plasticity in graphite under constant stress’, Carbon, 1967,
Anwendungen. 5, (2), 173–180.
34. C. Berre, S. L. Fok, B. J. Marsden, P. M. Mummery, T. J. Marrow and 59. M. R. Bradford and A. G. Steer: ‘A structurally-based model of irra-
G. B. Neighbour: ‘Microstructural modelling of nuclear graphite diated graphite properties’, J. Nucl. Mater., 2008, 381, (1–2), 137–
using multi-phase models’, J. Nucl. Mater., 2008, 380, (1–3), 46–58. 144.
35. C. Berre, S. L. Fok, B. J. Marsden, L. Babout, A. Hodgkins, T. J. 60. E. D. Eason, G. N. Hall, B. J. Marsden and G. B. Heys: ‘Models of
Marrow and P. M. Mummery: ‘Numerical modelling of the effects coefficient of thermal expansion (CTE) for Gilsocarbon graphites
of porosity changes on the mechanical properties of nuclear graph- irradiated in inert and oxidising environments’, J. Nucl. Mater.,
ite’, J. Nucl. Mater., 2006, 352, (1–3), 1–5. 2013, 436, (1–3), 191–200.
36. R. Hill: ‘The elastic behaviour of a crystalline aggregate’, Proc. Phys. 61. B. D. Cullity and S. R. Stock: ‘Elements of x-ray diffraction’; 3rd
Soc. Sect. A, 1952, 65, (5), 349. edn, 2001, Upper Saddle River, NJ, Prentice Hall.
37. R. S. Igor and L. I. Alexander: ‘Elastic properties of mono- and 62. A. A. Campbell, K. B. Campbell and G. S. Was: ‘Anisotropy analy-
polycrystalline hexagonal AlB 2-like diborides of s, p and d metals sis of ultra-fine grain graphite and pyrolytic carbon’, Carbon, 2013,
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

from first-principles calculations’, J. Phys.: Condens. Matter, 2008, 60, 410–420.

20, (41), 415218. 63. J. B. Nelson and D. P. Riley: ‘The thermal expansion of graphite
38. C. N. Tomé: ‘Preferred orientations in polycrystals and their effect from 15°c. to 800°c.: part I. Experimental’, Proc. Phys. Soc., 1945,
on materials properties’, in ‘Texture and anisotropy’, (eds. C. T. U. 57, (6), 477.
F. Kocks and H. R. Wenk), 282–324; 1988, Cambridge, 64. W. N. Reynolds: ‘Anisotropy factors of polycrystalline graphites’,
Cambridge University Press. Carbon, 1968, 6, (3), 277–282.
39. J. S. Kallend: ‘Determination of the orientation distribution from 65. British Standards Institution, London. ‘Advanced technical cer-
pole figure data’, in ‘Texture and anisotropy: preferred orientation amics. Monolithic ceramics. Thermo-physical properties.
in polycrystals and their effect on materials properties’, (eds. Determination of thermal expansion’, BS EN 821–1:1995;
C. T. F. Kocks and H. R. Wenk), 102–125; 1998, Cambridge, September 1995.
Universal Press. 66. G. Hall, B. J. Marsden, S. L. Fok and J. Smart: ‘The relationship
40. D. Mainprice, R. Hielscher and H. Schaeben: ‘Calculating anisotro- between irradiation induced dimensional change and the coefficient
pic physical properties from texture data using the MTEX open- of thermal expansion: a modified Simmons relationship’, Nucl. Eng.
source package’, Geol. Soc., London, Spec. Publ., 2011, 360, (1), Des., 2003, 222, (2–3), 319–330.
175–192. 67. ASTM: ‘Standard guide for measurements on small graphite speci-
41. T. Lokajicek, P. Lukas, A. N. Nikitin, I. V. Papushkin, V. V. Sumin mens’, in ‘D7775 — 11e1’; 2011, West Conshohocken, PA, ASTM
and R. N. Vasin: ‘The determination of the elastic properties of an International.
anisotropic polycrystalline graphite using neutron diffraction and 68. M. P. Metcalfe, N. Tzelepi and J. F. B. Payne: ‘Evaluation of instan-
ultrasonic measurements’, Carbon, 2011, 49, (4), 1374–1384. taneous coefficient of thermal expansion of PGA graphite’, in
42. S. Matthies: ‘GEO-MIX-SELF calculations of the elastic properties ‘Modelling and measuring reactor core graphite properties and per-
of a textured graphite sample at different hydrostatic pressures’, formance’ (ed. G. B. Neighbour), 17–22; 2013, Cambridge, The
J. Appl. Crystall., 2012, 45, (1), 1–16. Royal Society of Chemistry.
43. C. Berre, S. L. Fok, P. M. Mummery, J. Ali, B. J. Marsden, T. J. 69. B. T. Kelly and J. E. Brocklehurst: ‘High dose fast neutron
Marrow and G. B. Neighbour: ‘Failure analysis of the effects of por- irradiation of highly oriented pyrolytic graphite’, Carbon, 1971, 9,
osity in thermally oxidised nuclear graphite using finite element (6), 783–789.
modelling’, J. Nucl. Mater., 2008, 381, (1–2), 1–8. 70. H. Wang, X. Zhou, L. Sun, J. Dong and S. Yu: ‘The effect of stress
44. C. Berre, C. Berre, P. M. Mummery, B. J. Marsden, T. Mori and P. J. levels on the coefficient of thermal expansion of a fine-grained iso-
Withers: ‘Application of a micromechanics model to the overall tropic nuclear graphite’, Nucl. Eng. Des., 2009, 239, (3), 484–489.
properties of heterogeneous graphite’, J. Nucl. Mater., 2008, 381, 71. T. Oku, A. Kurumada, Y. Imamura, K. Kawamata and M.
(1–2), 124–128. Shiraishi: ‘Effects of prestresses on mechanical properties of isotro-
45. L. Babout, B. J. Marsden, P. M. Mummery and T. J. Marrow: pic graphite materials’, J. Nucl. Mater., 1998, 258–263, (Part 1).
‘Three-dimensional characterization and thermal property model- 72. E. G. Zukas, J. A. O’Rourke, P. E. Armstrong and W. V. Green: ‘The
ling of thermally oxidized nuclear graphite’, Acta Mater., 2008, influence of creep strain on the elastic modulus, crystallographic
56, (16), 4242–4254. orientation and density of a poco graphite’, Carbon, 1973, 11, (4),
46. G. E. Bacon: ‘A method for determining the degree of orientation of 317–321.
graphite’, J. Appl. Chem., 1956, 6, (11), 477–481. 73. T. J. Marrow, D. Liu, S. M. Barhli, L. Saucedo Mora, Ye.
47. J. J. Kearns: ‘Thermal expansion and preferred orientation in Vertyagina, D. M. Collins, C. Reinhard, S. Kabra, P. E. J. Flewitt
Zircaloy (LWBR Development Program)’; 1965, Report of and D. J. Smith: ‘In situ measurement of the strains within a
Westinghouse Electric Corporation, Bettis Atomic Power mechanically loaded polygranular graphite’, Carbon, 2016, 96,
Laboratory, Pittsburgh. 285–302.
48. A. L. Sutton and V. C. Howard: ‘The role of porosity in the accom- 74. B. T. Kelly: ‘The effect of pressure and fast neutron irradiation on
modation of thermal expansion in graphite’, J. Nucl. Mater., 1962, the thermal expansion coefficients of graphite crystals’, Carbon,
7, (1), 58–71. 1981, 19, (2), 71–73.
49. W. C. Morgan: ‘Notes on the two-phase model for radiation-induced 75. C. Se-Hwan and K. Gen-Chan: ‘Mrozowski cracks and oxidation
dimensional changes in graphite’, Carbon, 1966, 4, (2), 215–222. behavior of IG-110 and IG-430 nuclear graphites’, in
50. J. H. W. Simmons: ‘Radiation damage in graphite’; 1965, Oxford, ‘Transactions of the Korean nuclear society Spring Meeting’;
Pergamon Press. 2006, Chuncheon, Korea’.
51. J. C. Bokros: ‘Absorption factors for a modified bacon preferred- 76. A. T. Bakenne: ‘Deformation and modulus changes of nuclear
orientation technique’, Carbon, 1965, 3, (2), 167–174. graphite due to hydrostatic pressure loading’, in ‘School of mechan-
52. R. J. Price and J. C. Bokros: ‘Relationship between preferred orien- ical, aerospace and civil engineering’, 2012, PhD thesis, Manchester,
tation, thermal expansion, and radiation-induced length changes in The University of Manchester.
graphite’, J. Appl. Phys., 1965, 36, (6), 1897–1906. 77. B. T. Kelly, W. H. Martin and P. T. Netley: ‘Dimensional changes in
53. D. K. L. Tsang, B. J. Marsden, S. L. Fok and G. Hall: ‘Graphite pyrolytic graphite under fast-neutron irradiation’, Philos. Trans.
thermal expansion relationship for different temperature ranges’, Roy. Soc. London. Series A, Math. Phys. Sci., 1966, 260, (1109),
Carbon, 2005, 43, (14), 2902–2906. 37–49.

International Materials Reviews 2016 27

 Marsden et al. Dimensional and property changes in nuclear graphite

78. S. Yoda and K. Fujisaki: ‘An approximate relation between Young’s 84. B. T. Kelly and A. J. E. Foreman: ‘The theory of irradiation creep in
modulus and thermal expansion coefficient for nuclear-grade graph- reactor graphite — The dislocation pinning-unpinning model’,
ite’, J. Nucl. Mater., 1983, 113, (2–3), 263–267. Carbon, 1974, 12, (2), 151–158.
79. J. E. Brocklehurst and B. T. Kelly: ‘The dimensional changes of 85. C. R. Kennedy, M. Cundy and G. Kleist: ‘The irradiation creep
highly-oriented pyrolytic graphite irradiated with fast neutrons at characteristics of graphite to high fluences’, in ‘International confer-
430°C and 600°C’, Carbon, 1993, 31, (1), 179–183. ence on carbon’, 443–445; 1988, UK, University of Newcastle upon
80. J. E. Brocklehurst and B. T. Kelly: ‘Analysis of the dimensional Tyne.
changes and structural changes in polycrystalline graphite under 86. M. A. Davies and M. Bradford: ‘A revised description of graphite
fast neutron irradiation’, Carbon, 1993, 31, (1), 155–178. irradiation induced creep’, J. Nucl. Mater., 2008, 381, (1–2), 39–45.
81. A. N. Jones, G. N. Hall, M. Joyce, A. Hodgkins, K. Wen, T. J. 87. J. E. Brocklehurst and B. T. Kelly: ‘Review of irradiation creep in
Marrow and B. J. Marsden: ‘Microstructural characterisation of graphite under CAGR conditions’, UKAEA Report ND-R-1406
nuclear grade graphite’, J. Nucl. Mater., 2008, 381, (1–2), 152–157. (S), Springfields, 1989.
82. Z. Mileeva, D. K. Ross and S. M. King: ‘A study of the porosity 88. G. Hall, B. J. Marsden and S. L. Fok: ‘The microstructural model-
of nuclear graphite using small-angle neutron scattering’, Carbon, ling of nuclear grade graphite’, J. Nucl. Mater., 2006, 353, (1–2),
2012, 50, (14), 5062–5075. 12–18.
83. B. W. Ashton, J. E. Brocklehurst and B. T. Kelly: ‘The effect of fast 89. L. Delannay, P. Yan, J. F. B. Payne and N. Tzelepi: ‘Predictions of
neutron irradiation on the open pore volume of MK. 2 moderator inter-granular cracking and dimensional changes of irradiated poly-
graphite’, 1972, UKAEA, TRG M 5910(S), AB7/21948, crystalline graphite under plane strain’, Comput. Mater. Sci., 2014,
Springfields. 87, 129–137.
Downloaded by [The University of Manchester Library] at 04:19 18 April 2016

28 International Materials Reviews 2016

View publication stats