You are on page 1of 10

Journal of Nuclear Materials 442 (2013) 133–142

Contents lists available at ScienceDirect

Journal of Nuclear Materials


journal homepage: www.elsevier.com/locate/jnucmat

Mechanical characteristics of SiC coating layer in TRISO fuel particles


P. Hosemann a,⇑, J.N. Martos a, D. Frazer a, G. Vasudevamurthy b, T.S. Byun c, J.D. Hunn c, B.C. Jolly c,
K. Terrani c, M. Okuniewski d
a
University of California Berkeley, Nuclear Engineering, CA, USA
b
Virginia Commonwealth University, Mechanical and Nuclear Engineering, VA, USA
c
Oak Ridge National Laboratory, TN, USA
d
Idaho National Laboratory, ID, USA

a r t i c l e i n f o a b s t r a c t

Article history: Tristructural isotropic (TRISO) particles are considered as advanced fuel forms for a variety of fission plat-
Received 29 May 2013 forms. While these fuel structures have been tested and deployed in reactors, the mechanical properties
Accepted 23 August 2013 of these structures as a function of production parameters need to be investigated in order to ensure their
Available online 30 August 2013
reliability during service. Nanoindentation techniques, indentation crack testing, and half sphere crush
testing were utilized in order to evaluate the integrity of the SiC coating layer that is meant to prevent
fission product release in the coated particle fuel form. The results are complimented by scanning elec-
tron microscopy (SEM) of the grain structure that is subject to change as a function of processing param-
eters and can alter the mechanical properties such as hardness, elastic modulus, fracture toughness and
fracture strength. Through utilization of these advanced techniques, subtle differences in mechanical
properties that can be important for in-pile fuel performance can be distinguished and optimized in iter-
ation with processing science of coated fuel particle production.
Ó 2013 Elsevier B.V. All rights reserved.

1. Introduction pressure vessel to withstand internal gas pressure buildup. Finally,


the outermost layer consists of a high-density pyrolytic carbon
TRISO fuel particles have long served as the fuel form of choice (OPyC) layer that serves to protect the SiC layer during handling
in high temperature gas reactors (HTGRs) [1–7] and their use is and compaction. Both IPyC and OPyC shrink during early stages
now considered to be extended into fluoride salt high temperature of irradiation to induce compressive stress on the ceramic pressure
reactors (FHRs) [3–6] and light water reactors (LWRs) [2]. For vessel [9].
HTGR applications, a large number of these TRISO fuel particles The silicon carbide layer of the TRISO particle is considered to
are combined in a graphite matrix of either spheres for pebble be the most important structure since it prevents fission product
bed fuel or cylindrical compacts for prismatic fuels [8]. These fuel release and therefore is an essential safety feature. Maintaining
forms have been successfully deployed in Germany in the 1960’s the integrity of the SiC layer throughout the operating lifetime is
[9] and are now being produced and evaluated in the US national essential to retaining fission products within the fuel particle and
laboratory system [4] as well as in China and other countries. prevents contamination of the reactor and potentially the outside
The typical TRISO particle has five layers. In the center is the atmosphere. The exact conditions of TRISO fuel manufacturing var-
fuel kernel, which is often uranium oxide (UO2) but can be an ies and depends on the actual use. The TRISO fuels examined in this
oxide, carbide, or oxide/carbide mix with uranium, plutonium, tho- project had a ZrO2 kernel directly coated with a PyC layer and an
rium or transuranic elements as the fuel [10]. The fuel is sur- outer SiC layer. The non-radioactive surrogate kernel was chosen
rounded by a low-density pyrolytic carbon buffer layer that for this study in order to make the material easier to handle in a
serves as a porous medium in order to absorb fission products non-radiological facility. For the purpose of characterizing the SiC
and accommodate kernel swelling. Next, there is a high-density in- this should not have a significant effect on the properties.
ner pyrolytic carbon (IPyC) that is impermeable to gaseous fission A typical TRISO particle is a multilayered, <1-mm-diameter
products, and serves as the substrate for SiC deposition. Surround- structure making localized investigation a challenge. The spherical
ing the IPyC layer, is a high density and high strength silicon car- shape of TRISO particles and the small diameter of the SiC layer,
bide (SiC) layer providing a diffusion barrier to prevent the usually 30–50 lm, makes determining mechanical properties
release of gaseous and metallic fission products, and acts as using traditional methods difficult.
Silicon carbide is a material of general interest due to its supe-
⇑ Corresponding author. Address: 4169 Etcheverry hall, 94720 Berkeley, CA, USA. rior irradiation tolerance [3]. It is often used for high temperature
E-mail address: peterh@berkeley.edu (P. Hosemann). environmental applications because of its high thermal

0022-3115/$ - see front matter Ó 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jnucmat.2013.08.041
134 P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142

conductivity, low thermal expansion, good thermal shock resis- 2. Experimental


tance, and good chemical inertness [3]. Important to nuclear appli-
cations, SiC has a high resistance to radiation damage due to its In this study two ZrO2 kernel sizes, with diameters of 520 lm
small-neutron capture cross section [3]. For these reasons, SiC is of- and 670 lm, were coated using typical AGR IPyC coating condi-
ten considered as a candidate for advanced nuclear reactor materi- tions in the 150-mm-diameter coater at Babcock and Wilcox Nu-
als even as a composite [11–13]. As described above, the TRISO clear Operations Group (B&W-NOG) [18]. These particles were
particle uses a SiC coating as the main structural layer of the fuel sent to Oak Ridge National Laboratory (ORNL) and SiC was chemi-
particle and its properties have therefore been subject to research cal vapor deposited in a 50-mm-diameter coater using MTS.
over the past decade. Hydrogen was used as the fluidization medium and mixed with
The mechanical properties of the SiC layer of TRISO particles the MTS. The processing parameters, such as SiC deposition dura-
have been explored with different techniques including the ring tion, coating temperature and MTS flow rate were varied as shown
test method [14,15], the crush test [16–19], the micro-cantilever in Table 1. A roller micrometer was used to separate the two differ-
test [20,21], and indentation methods [22,23]. Fracture toughness ent kernel sizes after each batch was coated. Within each batch,
properties have also been explored previously [24]. In this work, particles with small kernels were separated into subsample A
effects of the manufacturing variable on Young’s modulus, hard- and particles with large kernels were separated into subsample
ness, and fracture toughness is presented utilizing a suite of small C. Each subsample was then separated again by size distribution
scale testing techniques including nanoindentation, microcantile- into smaller groups and the group at the peak of the Gaussian dis-
ver testing, shear crush testing, and indentation fracture. Indention tribution in the particle size (A2 and C2) was used for material
techniques allow mapping the properties as a function of location characterization [18]. The mean and standard deviation of each
across the different layers and have been found useful on nuclear layer’s thickness was determined using optical imaging [35]. The
materials in the past [25–27]. larger diameter kernels (C2 batches) typically had a slightly thin-
The TRISO fuel particles are manufactured using fluidized-bed ner IPyC layer, because coating rate is inversely related with the
chemical vapor deposition (FB-CVD), an efficient method to coat particle size due to the geometrical effect on surface flux [18].
particles using reactant gas. Particles are typically suspended in a
deposition zone by flowing a gas upwards through a particle bed.
2.1. Sample preparation for nanoindentation
Pyrolytic carbon is deposited from acetylene, or a mixture of acet-
ylene/propylene, at a deposition temperature ranging from 1250 to
For each of the ten different samples, 24 individual TRISO parti-
1450 °C, and at a concentration of acetylene/propylene of 20–40%
cles were mounted in a two-part cold mount epoxy. Double sided
[28]. The ZrO2 (surrogate) or heavy metal fuel kernel is manufac-
tape was centered and fixated to the bottom of a mounting cup and
tured using a gel precipitation process where the final step is sin-
individual TRISO particles were placed in a 5  5 matrix. One posi-
tering in a H2 atmosphere [28]. Silicon carbide is deposited by the
tion on the outside of the matrix was left open to allow for index-
decomposition of methyltrichlorosilane (MTS) in a hydrogen envi-
ing. The array of particles were closely spaced and centered in the
ronment at temperature between 1300 and 1600 °C [29]. MTS is
1.25’’ diameter mount as shown in Fig. 1a. Arranging the particles
used for depositing silicon carbide on isotropic graphite and its
in a tightly packed matrix helped keep the cross sections flat with
equivalent ratio of Si to C makes it easy to deposit stoichiometric
sharp edges at the interfaces and also helped keep an even planar
films [29]. Hydrogen is usually used as a carrier gas to transports
surface for all particles.
the MTS from the bubbler to the reactor, and also functions as a
The mounted particles were then polished to near mid-plane.
dilution gas [30]. The ideal deposition temperature for
By observing the mounted particles in an optical microscope, one
stoichiometric cubic SiC (b-SiC) has been found to be in the range
can see that a shadow around the edge of the SiC layer decreases
of 1450–1550 °C, which results in the highest strength and Young’s
in thickness as the mid-plane is approached. Additionally, since
modulus of SiC [31]. Grain size is primarily a function of the
the average TRISO particle diameter is known, measuring the ini-
deposition temperature. Higher deposition temperature results in
tial thickness and then periodically measuring the amount re-
a larger grain size. It is important to note that using FB-CVD
moved using a high precision micrometer, can give a good
technique, the grain size is not constant but instead grows
estimate as to when a near mid-plane polish is approached. Polish-
considerably with thickness [32–34]. The focus of this paper is
ing very close to or past mid-plane of the TRISO particles would
on demonstrating the application of these new mechanical testing
cause the ZrO2 kernel to separate and fall out from the rest of
methods to TRISO fuel particles; however, some evaluation of the
the particle, so this was avoided.
different processing parameters and their effects on the
The embedded TRISO particles were manually polished in suc-
mechanical properties is included.
cession using Buehler SiC abrasive paper starting at coarse P600

Table 1
Samples provided by ORNL for this study. All relevant processing parameters and specimen geometries are listed.

Sample Kernel diameter (lm) PyC thickness (lm) SiC thickness (lm) SiC deposition time SiC deposition temperature MTS flow
Mean St. Dev. Mean St. Dev. Mean St. Dev. (min) (°C) (cc/min)
AGRB-2T-A2 529 9 46 3 37 1 122.0 1500.0 179.0
AGRB-2T-C2 717 9 43 3 34 1 122.0 1500.0 179.0
AGRB-1W-A2 540 10 46 3 35 1 131.0 1450.0 157.0
AGRB-1W-C2 718 9 43 3 31 1 131.0 1450.0 157.0
AGRB-1V-A2 528 9 46 3 27 1 131.0 1450.0 138.0
AGRB-1V-C2 719 8 43 2 24 1 131.0 1450.0 138.0
AGRB-3R-A2 540 9 46 3 35 1 127.0 1550.0 148.0
AGRB-3R-C2 712 10 44 3 29 1 127.0 1550.0 148.0
AGRB-3S-A2 529 11 46 3 41 1 129.0 1550.0 174.0
AGRB-3S-C2 711 9 44 3 32 1 129.0 1550.0 174.0
P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142 135

Fig. 1. SiC Triso spheres mounted in epoxy and polished (a). Optical micrograph of the spear not polished to the sphere centerline (b). Example hardness map obtained using
cross section nanoindentation on a sample (c).

grit down to P4000 grit. Buehler diamond paste of 3 lm, 1 lm, and imaging and indentation. Care had to be taken to avoid kernel rota-
4 lm in diameter on hard polishing clothes were used. Extreme
1=
tion during grinding, which could result in broken layers. Grinding
care had to be taken when polishing the TRISO particles due to was stopped close to, but before passing through, the particle mid-
the very different material properties between the layers, as well plane. Attempting to grind past the midplane was more likely to
as the mounting epoxy. Low force and fine grinding steps were result in broken coatings due to kernels becoming loose and even
necessary in order to have intact TRISO particles. If high speeds falling out of the mounts. Acetone was used to dissolve the Crystal-
or high forces were used, fracture of one of the coatings could oc- bond-509 and recover the ground hemispheres. Samples were then
cur, which would then destroy the rest of the particles in the sam- placed in a muffle furnace and heated for 3 h at 750 °C. This was
ple mount as coating pieces fracture off during grinding. As a last sufficient to remove the IPyC layer, leaving the ZrO2 kernel hemi-
step, the sample mounts were polished with 0.05 lm colloidal sil- spheres free to fall out of the SiC hemispherical shells. Fig. 3 shows
ica on a Buehler VibroMet polisher. In between each polishing step, a scanning electron image of a typical SiC hemispherical shell. A
the sample was thoroughly cleaned with an ultrasonic cleaner with similar approach as described above was used to manufacture C-
ethanol as a cleaning medium. It was found that as the mid-plane Rings as evaluated in the discussion section.
of the TRISO particles was approached; using the ultrasonic cleaner
would quickly separate the ZrO2 kernel from the particle. Therefore
not polishing all the way to the mid-plane was more desirable to 2.3. SEM investigation
avoid the removal of the kernel. Having the kernel separate from
the TRISO structure can lead to easy fracture of the remaining shell SiC layer thickness and grain size were investigated utilizing
due to handling and can make the indention experiment difficult SEM. The resulting images can be found in the results section. A
since a deep gap will remain next to the edge. Quanta 3D field emission gun (FEG) scanning electron microscope
with focused ion beam (SEM/FIB) at UC Berkeley was used to ob-
tain the images. FIB surface milling provided a clean surface for
2.2. Sample preparation for crush testing better imaging of the grains in the SiC.

A modified crush testing and evaluation method for hemispher- 2.4. Nanoindentation and indentation fracture toughness
ical shell specimens was applied to obtain the fracture stress data measurements
for selected samples. For each sample, 24–47 hemispherical shell
specimens were selected and tested. Approximately 200 particles Nanoindentation measurements were performed at the Center
were bonded to an 1.2500 diameter aluminum cylinder using for Advanced Energy Studies (CAES) in Idaho Falls, ID and the
Crystalbond-509 mounting adhesive. This adhesive melts at Materials Laboratory at the University of California, Berkeley. The
121 °C to allow particles to be mounted in a single layer and sets nanoindentation measurements at CAES were conducted on the
upon cooling to a hard form suitable for grinding. Enough epoxy Hysitron Triboindenter, where the nanoindenter tip was mounted
was applied to cover the particles, which provided suitable edge on a transducer allowing lateral scanning of the sample, as well as
support during grinding. The grinding procedure was similar to precise motion using the piezoelectric and capacitor actuators. The
that described above for preparation of cross sections for optical nanoindentation measurements at UC Berkeley were conducted on

(a) (b) (c)

Fig. 2. Optical microscopy image of a indent set into the SiC shell. The cracks used for fracture analysis are clearly visible (a). SPM scan over a cracked shell for fracture
toughness measurements (Z-scale = 8 lm a) (b). SPM image of another indent the crack originating from the indent corners is clearly visible going through the grains and not
going in the radial direction visible (z-scale 48 lm) (c).
136 P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142

Loading
Plunger

Insert material
(Soft metal)

SiC
specimen

Testing Bed
(a) (b)

Fig. 3. SEM image of a SiC hemispherical shell (AGRBW-1W) prepared for crush testing (a). Schematic sketch of the crush test as developed by ORNL (b).

the Micro Materials NanoTest™. The NanoTest is a horizontal tip- This implies that the values obtained here might not be directly
sample arrangement in contrast to the vertical arrangement of comparable to are not bulk values but are in reasonable good agree-
the Hysitron device [36]. This system allows higher loads of up ment with what is found elsewhere and discussed in the later sec-
to 500 mN without changing the transducer. In addition the Micro tion of this paper.
Materials system is capable of heating the sample and tip up to The radial cracks, initiated by indenting, were measured with
750 °C and performing high temperature tests. However, the effect the in situ scanning probe microscopy (SPM) imaging capability
of temperature was not part of this study. The reason for utilizing of the Hysitron TriboIndenter as well as an optical microscope.
two different indenters was machine availability. Both systems The SPM allowed for immediate post-test observation of the indent
were always calibrated against fused silica before each indentation produced on the sample and eliminated the need to locate the in-
run in order to allow cross comparison between samples and ind- dent with its associated cracks with a different instrument such as
enters. The indents were typically 200 nm deep and 4 lm apart an SEM.
from each other. A cross comparison on selected TRISO particles
was conducted ensuring that the same data was measured using 3. Crush test measurements
both systems. The indentation size effect was not part of this study
therefore the values measured should be compared to indents of In a study performed in 2008 by ORNL [19], a modified crush
similar depth. testing technique for hemispherical shell SiC specimens was devel-
A nanoindenter was also used to measure the fracture tough- oped. As described by the ORNL study [19], the modified crush test
ness of a given material. For these measurements, a Berkovich dia- used a soft metal (either copper or brass) at the specimen–plunger
mond indenter tip was used to perform fracture toughness contact to generate a uniformly stressed area at the inner surface of
measurements. Fracture toughness was calculated based on the the shell specimens. A schematic testing diagram shown in Fig. 3
crack length starting from the corners of the imprint using the shows the hemispherical shell specimen diametrically loaded be-
widely known the following equation: tween a bottom base and a loading plunger with a blanket metal
 12 inserted in between. The hemispherical shell specimens were
E P tested in a 40 kN capacity Tinius-Olsen static testing machine
Kc ¼ a ð1Þ
H c3=2 using the plunger-jig assembly. While the detailed theoretical
approach of how the measured load is related to fracture stress
where Kc is the fracture toughness of the material, P is the maxi-
is given in [19], a summary is also given here.
mum load, c is the crack length measured as shown in Fig. 2, E is
The maximum membrane stress and bending stress in a partial
the elastic modulus, H is the hardness, and A is an empirical con-
spherical thin shell loaded by a load, F is calculated using
stant corresponding to the indenter shape, a = 0.016 for a Berkovich
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
tip [37]. In the literature extensive discussions on indention fracture F 1  t2
toughness methods are published [51,52] including a comprehen- rmembrane ¼ C 1 ð2Þ
t2
sive review of the technique and the calculations utilized today.
In our work we utilize Eq. (1) since it is the equation given in the Fð1 þ tÞ
textbook on nanoindentation [53] and is rather simple to use. In rbending ¼ C 2 ð3Þ
t2
addition this equation does not utilize the distance from the inden-
tation center to the edge of the imprint but rather uses the distance where m is the Poisson’s ratio, t is the thickness of the shell, and
from the crack tip to the center of the indent. While being an older coefficients, R is the outer radius of the shell, C1 and C2, are obtained
approach it does have the advantage that a sometimes difficult to using below,
determine edge of the imprint is not needed and the question of
C 1 ¼ 0:2205  0:04l  0:0115l2 ð4Þ
where exactly the imprint starts becomes irrelevant. It has to be
pointed out to the reader, not familiar with indentation fracture
C 2 ¼ 1:2044 expð1:2703lÞ ð5Þ
toughness, that several different approaches exists as outlined in
[52] where most equations are based on empirical data generation. With
P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142 137

1=4
l ¼ r0 ½12ð1  t2 Þ=ðR2 t2 Þ ð6Þ 4. Results

The maximum tensile stress is given in Eq. (26), which occurs at the Images of the SEM analysis are displayed in Figs. 4–8. A low
inner surface of the shell [19]. magnification and a higher magnification image are presented for
each specimen investigated. The images obtained were used to
rmax ¼ rmembrane þ rbending ð7Þ characterize the grain structure on each sample. Some samples
show a more uniform grain structure while others show a radially
Since the ratio between mean fracture strengths of two specimens oriented grain texture. Using the intercept method in ImageJ soft-
having different sizes can be correlated with the ratio of the effec- ware on the SEM images, the grain sizes were quantified for the
tive surface area [38,39], the mean fracture stress for a full-size TRISO variants at three different locations across the SiC layer in
spherical particle can be determined using Eq. (7) [31]. two directions and the average of each location was calculated
!1=m !1=m with the appropriate error as listed in Table 2 – at the IPyC/SiC
SLE pr20 interface, the middle of the SiC layer, and at the outer edge of
rFf ¼ rLf ¼ rLf ð8Þ
SFE 4pðR  tÞ2 the SiC layer. A summary of these results is shown in Table 2 be-
low. All the TRISO variants have very similar small grain structure
where rLf and rFf are the fracture stresses for partially loaded and at the IPyC/SiC interface. Progressing towards the outside of the SiC
full spherical shell specimens, SLE and SFE are the effective surfaces, layer, the grain size was found to increase in all the samples, and
m the Weibull modulus, and r0 is the radius of effective area which the grains were radial oriented starting at the IPyC/SiC interface.
is given by the measured radius of indentation impression. The At a deposition temperature of 1450 °C there was a finer grain
modified crush test technique was performed multiple times on structure throughout the layer. With the increase of the deposition
each variant to determine statistical results. temperature to 1500 °C and 1550 °C, there was a more pronounced

Fig. 4. SEM image of SiC grain structure for AGRBW-1W TRISO batches with SiC deposition duration = 131 min, MTS flow = 157 cc/min, and SiC deposition
temperature = 1450 °C. (a and b) small kernel’s (c and d) large kernels.

Fig. 5. SEM image of SiC grain structure for AGRBW-1V TRISO batches with SiC deposition duration = 131 min, MTS flow = 138 cc/min, and SiC deposition
temperature = 1450 °C. (a and b) small kernel’s (c and d) large kernels.
138 P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142

Fig. 6. SEM images of SiC grain structure for AGRBW-2T TRISO batches with SiC deposition duration = 122 min, MTS flow = 179 cc/min, and SiC deposition
temperature = 1500 °C. (a and b) small kernel’s (c and d) large kernels.

Fig. 7. SEM images of SiC grain structure for AGRBW-3R TRISO batches with SiC deposition duration = 127 min, MTS flow = 148 cc/min, and SiC deposition
temperature = 1550 °C. (a and b) small kernel’s (c and d) large kernels.

Fig. 8. SEM pictures of SiC grain structure for AGRBW-3S TRISO batches with SiC deposition duration = 129.2 min, MTS flow = 174 cc/min, and SiC deposition
temperature = 1550 °C. (a and b) small kernel’s (c and d) large kernels.
P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142 139

Table 2
Summary table of the results obtained from nanoindentation and grains structure analysis including the basic production parameters.

Sample SiC SiC MTS Average H. ± Average H. ± Average H. ± SiC outer SiC H SiC grain size Average red ±
dep. dep. flow outer edge outer edge SiC/IPyC edge grain gradient diameter (lm) modulus
time temp. SiC SiC interface structure across SiC
(min) (°C) (cc/ (GPa) (GPa) (GPa) (GPa) (GPa) (GPa) IPyC/ Middle Outer (GPa) (GPa)
min) SiC edge
AGRB-2T-A2 122.0 1500.0 179.0 30 2 34 2 28 3 Medium Slight 2.0 2.1 3.0 220.0 11.0
AGRB-2T-C2 122.0 1500.0 179.0 31 2 33 4 30 3 Medium Slgiht 1.5 1.8 2.3 195.0 8.0
AGRB-1W-A2 131.0 1450.0 157.0 39 1 38 1 40 1 Fine None <1 <1 <1 335.0 5.0
AGRB-1W-C2 131.0 1450.0 157.0 40 2 41 2 38 3 Fine None <1 <1 <1 328.0 13.0
AGRB-1V-A2 131.0 1450.0 138.0 41 3 42 3 41 3 Fine None <1 <1 <1 318.0 8.0
AGRB-1V-C2 131.0 1450.0 138.0 38 4 39 4 36 3 Fine Slight 1.1 1.3 1.8 322.0 16.0
AGRB-3R-A2 127.0 1550.0 148.0 36 2 39 1 32 3 Large Large 2.2 2.7 3.3 304.0 10.0
AGRB-3R-C2 127.0 1550.0 148.0 36 3 38 2 34 3 Large Large 1.2 2.4 2.7 298.0 14.0
AGRB-3S-A2 129.0 1550.0 174.0 42 3 41 4 42 3 Large None 1.7 3.2 4.6 340.0 13.0
AGRB-3S-C2 129.0 1550.0 174.0 39 2 41 1 36 2 Large Slight 1.4 2.4 2.7 314.0 10.0

was found from sample to sample. The summarized hardness data


in Table 2 includes two columns, where the average hardness is
listed as a function of location in the SiC layer (near the IPyC/SiC
interface and near the outer edge). The nanoindentation results
for all variants show that some of the fuel particles had a hardness
gradient in the SiC layer, where a higher hardness value at the out-
er edge of the SiC was found. This gradient is qualitatively summa-
rized in Table 2.
The indentation fracture toughness data are displayed in Table 3
for selected samples. Most samples experienced an anisotropic
fracture behavior. The cracks propagated significantly further in
the azimuthal direction than in the radial direction, as can be seen
in the AFM image in Fig. 2. This is particular interesting since the
Fig. 9. Compiled hardness values across the three layers for all TRISO variants. columnar grains tend to be oriented in the radial direction. It
was also observed that the cracks are of a transgranular rather than
intergranular nature. The difference in crack length resulted in dif-
Table 3 ferent fracture toughness values in the two directions (Table 3),
Results obtained from indentation fracture toughness measurements. with a significantly higher fracture toughness in the radial direc-
Sample Tangential average fracture Radial average fracture tion. It is also interesting that the azimuthal cracks were not
toughness (MPa m1/2) stress (MPa m1/2) straight, but rather followed the circumferential path of the
AGRBW 1W-A2 1.5 ± 0.6 1.7 ± 0.3
particle.
AGRBW 1W-C2 1.1 ± 0.8 2.1 ± 0.8 The crush test results are reported in Table 4. For samples 2T-A2
AGRBW 2T-A2 1.1 ± 0.6 1.6 ± 0.3 and 2T-C2, a significant difference in the fracture stress was found,
AGRBW 2T-C2 1.2 ± 1.5 3.6 ± 3.6 while the production parameters were the same but not the kernel
AGRBW 3S-C2 1.0 ± 0.5 1.8 ± 0.8
diameter. This observation is discussed in more detail in the dis-
cussion section of this paper.
increase in the grain structure size progressing towards the outer
part of the SiC layer. This is reflected by the three columns in
Table 2 showing the average grain size measured near the IPyC/ 5. Discussion
SiC interface, the middle of the SiC layer and the outer edge.
The nanoindentation results for all the samples are plotted in This study on the mechanical properties of TRISO fuel particles
Fig. 9. There, the complete hardness profiles for the SiC layer, IPyC delivers a range of different parameters as a function of processing
layer, and kernel are displayed. As expected the SiC is the hardest condition which are discussed in this section. The data suggests a
structure while the IPyC is the softest; however, sample to sample strong influence of the kernel diameter and SiC growth condition
variations were found. It can be seen that the IPyC layer has a on the grain structure, and, therefore, possibly some influence on
rather homogeneous hardness profile, indicating that the measure- the mechanical properties.
ments are reasonably reliable since the hardness numbers show Most specimens investigated showed the often reported grain
minimum sample to sample variation. The same is true for the structure of fine grains at the IPyC/SiC interface and larger grains
surrogate fuel kernel, where no significant hardness variation at the outer areas [40]. As expected, due to their low energy of for-

Table 4
Crush test fracture stress results.

Sample SiC dep. MTS flow SiC dep. Local fracture Mean fracture Avg. hardenss ± (GPa) Avg. hardenss ± (GPa) Avg. red ± (GPa)
temp. (°C) (cc/min) duration (min) stress (MPa) stress (MPa) across SiC (GPa) SiC/IPyC modulus (GPa)
interface (GPa)
AGRB-1 W-A2 1450 157 131 1182 356 39 1 40 1 335 5
AGRB-2T-A2 1500 179 122 1074 196 30 2 28 3 200 11
AGRB-2T-C2 1500 179 122 1481 109 32 2 30 3 196 8
AGRB-3R-A2 1550 148 127 1047 203 36 2 32 3 304 10
AGRB-3S-A2 1550 174 129 1560 461 42 3 42 3 341 13
140 P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142

mation stacking faults were observed similar to what was reported A small but measurable gradient can be found from the IPyC/SiC
in [41]. The data also suggests that higher deposition temperature interface to the outer surface. The SiC near the IPyC/SiC interface
results in larger grains and causes a larger grain size gradient shows higher hardness values than the indents placed at the out-
throughout the sample, while lower deposition temperature pro- side area especially on specimens which also experience a larger
duces a finer and more homogeneous distribution of grain size. It microstructural gradient throughout the SiC layer. On the very fine
appears that samples with longer deposition times (indicating low- grained material this was not observed. It appears that the grain
er deposition rate) can be seen as having a finer grain structure structure is directly linked to the hardness within the same mate-
than the samples with a shorter deposition time. No clear trend rial and that the finer grain structure leads to higher hardness
can be seen from the variation in the MTS flow. Samples with lar- within the same SiC layer. Investigations of pore densities might
ger grains on the outside of the TRISO fuel particle still maintain a be of value as a function of layer thickness, but were not conducted
fine grain structure near the IPyC interface, leading to a gradient in here. However, considering the rather different production param-
average grain size and a radial arrangement of the grains. Similar eters it was found that the hardness is rather similar across all
results have been reported in [42,27] where even a 1300 °C depo- samples.
sition temperature was reported. A detailed study outlining the A significant observation from the fracture toughness measure-
grain structure and type of grain boundaries has recently been ments was that the cracks induced with the Berkovich tip had an
published by [41]. The results found here agree well with the grain anisotropic nature. The cracks found in the radial direction were al-
sizes found in that study. ways shorter than the cracks in the azimuthal direction, and the
The nanoindentation results obtained draw a similar picture. A cracks in the this direction followed the curvature of the SiC shell.
hardness of 36–42 GPa and an E-modulus between 304 and This means that the fracture toughness was higher in the radial
341 GPa was found on most samples, except the specimen direction than in the azimuthal direction. Utilizing the calculations
AGRBW-2T-A2 and AGRBW-2T-C2, which showed a lower hard- mentioned in the experimental section of this paper it was found
ness and a modulus around 200 GPa. The most significant differ- that the fracture toughness was nearly twice as high in the radial
ence on these samples was that they had the shortest deposition direction (Table 3). In order to avoid any effect of the 3 sided pyr-
time of 122 min and the highest MTS flow rate (179 cc/min). amid, up to 10 indents were placed on a SiC shell, changing the an-
The indentation results presented here are similar to what was gle of the three sided pyramid to the outer edge as the three sided
found elsewhere [43]. In [23] it was found that different moduli pyramid (indenter) was moved around the circumferential direc-
can be found as a function of grain structure in respect to the tion. This did not impact the anisotropic cracking. It appears that
indentation direction, so that indents in grain axis showed a mod- the cracks go through the grains and do not follow any particular
uli of 440 GPa, while grains tilted 15° towards the indenter showed grain structure, which is in contrast to [23] but can be seen in
a moduli of 400 GPa. This was not reproduced here, since the sam- Fig. 2. However, since no TEM was conducted, it is not known if
ple was always cut near a 90° axis (cross section) in this case. there are finer nano-cracks underneath the indents. The question
The hardness of the IPyC was consistently between 4 and 6 GPa, remains, why did the cracks in the azimuthal direction propagated
and the reduced modulus between 20 and 30 GPa, which is in good further than in the radial direction. The idea of having remaining
agreement with data reported in [44]. stress in the sample from processing might be viable, however,

Fig. 10. SiC Average hardness and average reduced modulus vs. SiC deposition temperature (a). Average hardness and average reduced modulus vs. SiC Deposition duration
(b). Average hardness and average reduced modulus vs. MTS gas flow (c). Average hardness vs. mean fracture stress obtained from crush testing (d).
P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142 141

Table 5
results from microcantilever experiments.

Sample Bend bar a (m) b (m) w (m) L (m) P (N) Fracture toughness (MPa m1/2)
IV-C2 2 1.22E-06 7.13E-06 6.07E-06 1.20E-05 1.85E-02 7.99
1V-C2 4 5.00E-07 5.53E-06 7.89E-06 9.20E-06 2.04E-02 5.24
1V-C2 7 1.31E-06 8.05E-06 5.44E-06 1.30E-05 8.41E-03 3.79
1W-A2 1 1.50E-06 6.45E-06 4.13E-06 1.30E-05 4.76E-03 4.47
1W-A2 2 1.50E-06 6.82E-06 3.46E-06 1.30E-05 4.70E-03 5.03

[41,45] recently showed that it appears there is no significant played in Fig. 10d it is not entirely clear what causes the softer
strain in the lattice on a TRISO particle cross section. High resolu- and weaker properties (see Figure 11d)
tion microbeam XRD or Raman might be able to investigate this In the past, it was reported that fracture toughness can be mea-
question in more detail in future studies to come. sured via a microcantilever bend bar testing technique [46,21,48],
Considering the fact that the TRISO shell has higher fracture [50] where one manufactures a microbend bar utilizing a FIB and
toughness in radial direction than in circumferential direction performs the associated test utilizing a nanoindenter. In this work,
can be discussed in the light of in reactor service. In actual opera- initial efforts were made to perform similar tests. The FIB quanta
tion this anisotropic property can be desirable. A radial fracture pe- 3D FEG was used to manufacture bend bars, and initial tests were
netrating from the IPyC to the outside of the particle can lead to made. Since this research is in its early stage we report these re-
immediate fission gas release while a circumferential crack will sults under discussion in order to give a comprehensive view of
still prevents this. In fact a circumferential crack might help releas- the techniques and measurements but not to fully document the
ing stress in the SiC particle and therefore makes radial cracks less results at this time. Fig. 11 shows images from the microcantilever
likely which improves the properties. Also radial cracks potentially bend bar test and an initial dataset; the results of first tests are re-
generated during operation can hit a preexisting circumferential ported in Table 5 utilizing the equation proposed in [21]. It was
crack and get deflected which also helps to prevent fission gas found that these initial datasets show a 2–3 times higher fracture
release. toughness than measured using the indentation technique and as
The crush tests conducted on these coating batches lead to the reported in the literature. It appears that this is due to the fact that
data reported in Table 4. Five specimens were selected. The mean it is not a simple task to manufacture a sharp notch on the speci-
fracture stress obtained by crush tests correlated well with the men using the FIB and the notches manufactured in this first at-
average hardness data, and formed a linear correlation (Fig. 10). tempt were too dull. A pre-notching and fatigue cracking
The relationship between hardness and mean fracture stress can technique like that proposed in [49] might be a more desirable ap-
be calculated to be 0.0316 + 27.376. Comparing the local and proach. However, with the development of high temperatures ind-
the mean fracture stress, it is found that sample 2T-A2 and 2T-C2 enters this technique when improved so sharper notches are made
experience a lower mean fracture stress which correlates well with can be also conducted at higher temperatures.
the lower hardness numbers found. While the data between mean Initial attempts were also made to simplify the crush testing. In-
fracture stress and hardness is consistent to each other as dis- stead of a half sphere, a C-Ring was manufactured and tested. The

Fig. 11. Microcantilevers manufactured in the 1V-C2 specimen (a). Fractured cantilever in cross section (fracture surface) (b). Load displacement curve obtained from
microcantilever experiments (c). Experimental setup of in situ ring push testing (d).
142 P. Hosemann et al. / Journal of Nuclear Materials 442 (2013) 133–142

tests were performed in situ in collaboration with LANL on a cus- [9] D.A. Petti, J.T. Maki, J. Buongiorno, R.R. Hobbins, G.K. Miller, Key Differences in
the Fabrication, Irradiation and Safety Testing of U.S. and German TRISO-
tom-built mesoscale mechanical testing stage. While the sample
coated Particle Fuel and Their Implications on Fuel Performance, Idaho
handling is significantly more delicate, the benefit of such in situ National Engineering and Environmental Laboratory Bechtel BWXT Idaho,
testing is clearly visible, since one can observe the failure directly. LLC; INEEL/EXT-02-00300.
However, great care has to be taken to not damage or pre-crack the [10] J. Powers, B. Wirth, J. Nucl. Mater. 405 (2010) 74–82.
[11] D. Frazer, C.A. Back, C. Deck, H. Khalifa, P. Hosemann, Microsc. Microanal. 18
100 lm wide C-bend specimens. In addition, the outer layer of the (2012) 1420–1421.
SiC is in tension and the inner layer in compression, which is not [12] T. Hinoki, E. Lara-Crzio, L. Snead, Fus. Sci. Technol. (2003) 211–218.
the same loading condition as in service. Nevertheless, as a quan- [13] L.L. Snead, R.H. Jones, A. Kohyama, P. Fenici, J. Nucl. Mater. 26 (1996) 233–237.
[14] K. Bongartz, E. Gyarmati, H. Schuster, K. Tauber, J. Nucl. Mater. 62 (1976) 123–
titative measurement and quality control method, these types of 137.
tests might still be useful. [15] K. Bongartz, E. Gyarmati, H. Nickel, H. Schuster, W. Winter, J. Nucl. Mater. 45
(1972) 261–264.
[16] M.W. Kim, J.H. Kim, H.K. Lee, J.Y. Park, W.J. Kim, D.K. Kim, J. Ceram. Process.
6. Conclusions Res. 10 (2009) 373–377.
[17] T.S. Byun, J.D. Hunn, J.H. Miller, L.L. Snead, J.W. Kim, Int. J. Appl. Ceram.
Technol. 7 (2010) 327–337.
Mechanical and microstructural characteristics of SiC and PyC
[18] J.D. Hunn, K.A. Terrani, T.S. Byun, G. Vasudevamurthy, J.H. Miller, B.C. Jolly,
shells in TRISO particles were characterized in this study utilizing ‘‘Fabrication and Characterization of Sixteen SiC Variants Deposited on the
a range of different small scale mechanical testing techniques. It Same IPyC Substrate for Fracture Strength Testing,’’ Oak Ridge National
was found that the grain structure gradient correlates with the Laboratory Research Report. ORNL/TM-2009/324, December 2009.
[19] T.S. Byun, J.W. Kim, I. Dunbar, J.D. Hunn, ‘‘Fracture Stress Data for SiC Layers in
nanohardness across the SiC layer, where the IPyC/SiC interface TRISO-Coated Fuel Particles’’, Oak Ridge National Laboratory Research, Report,
was harder (smaller grains) and the outside rim was softer (larger ORNL/TM-2008/167, September, 2008.
grains). Columnar grains were found on the samples manufactured [20] X. Zhao, R.M. Langford, J. Tan, P. Xiao, Scripta Mater. 59 (2008) 39–42.
[21] D. Di Maio, S.G. Roberts, J. Mater. Res. 20 (2005) 299–302.
with higher deposition temperature while the lower temperature [22] E. Lopez-Honorato, P.J. Meadows, J. Tan, P. Xiao, J. Mater. Res. 23 (2008) 1785–
samples showed smaller grains and more homogeneous 1796.
properties. [23] J. Tan, P.J. Meadows, D. Zhang, X. Chen, E. Lopez-Honorato, X. Zhao, F. Yang, T.
Abram, P. Xiao, J. Nucl. Mater. 393 (2009) 22–29.
Indentation hardness and crush test strength correlated well. It [24] H. Zhang, E. López-Honorato, A. Javed, I. Shapiro, P. Xiao, J. Am. Cer. Soc. 95
was found that a linear correlation between these two techniques (2012) 1086–1092.
existed. Fracture toughness measurements were also conducted [25] P. Hosemann, D. Kiener, Y. Wang, S.A. Maloy, J. Nucl. Mater. 425 (2012) 136–
139.
and it was found that the samples had an anisotropic fracture
[26] P. Hosemann, C. Vieh, R.R. Greco, S. Kabra, J.A. Valdez, M.J. Cappiello, S.A.
behavior leading to higher fracture toughness in the radial direc- Maloy, J. Nucl. Mater. 389 (2009) 239–247.
tion, compared to the azimuthal direction. Also the cracks in azi- [27] P. Hosemann, J.G. Swadener, D. Kiener, G.S. Was, S.A. Maloy, N. Li, J. Nucl.
Mater. 375 (2008) 135–143.
muthal direction follow the SiC curvature. It is important to note
[28] E. Lopez-Honorato, J. Tam, P.J. Meadows, G. Marsh, P. Xiao, J. Nucl. Mater. 392
that the cracks propagated perpendicular to the grain structure (2009) 219–224.
rather than with the grain structure, and the cracking was trans- [29] C. Tang, Y. Tang, J. Zhu, Y. Zou, J. Li, X. Ni, Nucl. Eng. Des. 218 (2002) 91–102.
crystalline. Recent approaches and ideas for in situ C-bend testing [30] J. Kim, H. Lee, J. Park, W. Kim, D. Kim, J. Kor. Cer. Soc. 45 (2008) 518–523.
[31] S.J. Xu, J.G. Zhou, B. Yang, B.Z. Zhang, J. Nucl. Mater. 224 (1995) 12.
and microcantilever testing were also introduced. [32] K. Minato, K. Fukuda, J. Mater. Sci. 23 (1988) 699.
[33] D. Helary, O. Dunge, X. Bourrat, P.H. Jouneau, F. Cellier, J. Nucl. Mater. 350
Acknowledgements (2006) 332.
[34] L. Tan, T.R. Allen, J.D. Hunn, J.H. Miller, J. Nucl. Mater. 372 (2008) 400.
[35] J.R. Price, D.B. Aykac, J.D. Hunn, A.K. Kercher, in: Proceedings of Machine Vision
The authors want to thank the NRC for providing funding for Applications in Industrial Inspection XV, SPIE, vol. 6503, 2007, p. 650302.
this work, through NRC faculty development grant number NRC- [36] Z.C. Duan, A.M. Hodge, JOM (2009) 32–36.
[37] G.M. Pharr, D.S. Harding, W.C. Oliver, Mechanical Properties and Deformation
38-09-948. In addition, we want to thank the DOE/AGR program Behavior of. Materials Having Ultra-Fine Microstructures, in: M. Nastasi et al.
at ORNL and the NRC safeguards program for student support. (Eds.), Kluwer Academic Press, 1993, pp. 449–461.
Instrument access at CAES was supported by the ATR-NSUF user [38] G. Hong, T.S. Byun, R.A. Lowden, L.L. Snead, Y. Katoh, J. Amer. Cer. Soc. 90
(2007) 184–191.
program at Idaho National Laboratory, which is supported by the
[39] T.S. Byun, E. Lara-Curzio, L.L. Snead, Y. Katoh, J. Nucl. Mater. 367–370 (2007)
US Department of Energy, Office of Nuclear Energy under DOE Ida- 653–658.
ho Operations Office Contract DE-AC07-051D14517. [40] D. Helary, O. Dugne, X. Bourrat, P.H. Jouneau, F. Cellier, J. Nucl. Mater. 350
(2006) 332–335.
We also want to thank N. Mara and W.M. Mook for performing
[41] R. Kirchhofer, J.D. Hunn, P.A. Demkowicz, J.I. Cole, B.P. Gorman, J. Nucl. Mater.
an initial ring crush feasibility test utilizing the in situ loading 432 (2013) 127–134.
stage in the SEM at Los Alamos National at the Centre for Inte- [42] Jong Ho Kim, Hyeon Keun Lee, Ji-Yeon Park, Weon-Ju Kim, Do Kyung Kim, J.
grated Nanotechnologies, a US Department of Energy. Kor. Cer. Soc. 45 (2008) 518–523.
[43] C. Bellan, J. Dhers, Thin Solid Films 469-470 (2004) 214–220.
[44] E. Lopez-Honorato, P.J. Meadows, P. Xiao, G. Marsh, T.J. Abram, Nucl. Eng. Des.
References 238 (2008) 3121–3128.
[45] R. Kirchhofer, B. Hansford, I.E. Reimanis, B.P. Gorman, Microsc. Microanal. 16
[1] S. Saito et al., Design of High Temperature Test Reactor (HTTR), JAERI-1332, (2010) 1616–1617.
Japan Atomic Energy Research Institute, 1994. September. [46] X. Zhao, R.M. Langford, J. Tan, P. Xiao, Scr. Mater. 59 (2008) 39–42.
[2] K.A. Terrani, L.L. Snead, J.C. Gehin, J. Nucl. Mater. 427 (2012) 209–224. [48] D.E.J. Armstrong, M.E. Rogers, S.G. Roberts, Scr. Mater. 61 (2009) 741–743.
[3] L.L. Snead, T. Nozawa, Y. Katoh, T.S. Byun, S. Kondo, D.A. Petti, J. Nucl. Mater. [49] K. Matoy, H. Schönherr, T. Detzel, T. Schöberl, R. Pippan, C. Motz, G. Dehm, Thin
371 (2007) 329. Solid Films 518 (2009) 247–256.
[4] P. Mills, R. Soto, G. Gibbs, et al., Next Generation Nuclear Plant Pre-conceptual [50] N.A. Mara, J. Crapps, T.A. Wynn, K.D. Clarke, A. Antoniou, P.O. Dickerson, D.E.
Design Report, INL/EXT-07-12967, Idaho National Laboratory, November 2007. Dombrowski, B. Mihaila, Phil Mag (2013). <http://dx.doi.org/10.1080/
[5] C.W. Forsberg, P.F. Peterson, P. Pickard, J. Nucl. Tech. 144 (2003) 289–302. 14786435.2013.786192>.
[6] C.W. Forsberg, P.F. Peterson, H. Zhao, J. Energy Eng. 132 (2006) 109–115. [51] G.R. Anstis, P. Chantikul, B.R. Lawn, D.B. Marshall, J. Am. Ceram. Soc. 64 (1981)
[7] International Atomic Energy Agency, Fuel Performance and Fission Product 533–538.
Behavior in Gas Cooled Reactors, IAEA-TECDOC-978, IAEA, Vienna, 1997. [52] C.B. Ponton, R.D. Rawlings, Mater. Sci. Tech. Ser. 5 (1989) 865–872.
[8] F.A. Silady, L.L. Parme, ‘‘The Safety Approach of the Modular High Temperature [53] A. C. Fischer-Cripps; Springer; ISBN 978-1-4419-9871-2, 2011 (Chapter 9).
Gas Cooled Reactor (MHTGR),’’ 11th International Conference on the HTGR,
Dimitrovgrad, June 13–16, 1989.

You might also like