Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Elaboration and behavior under extreme irradiation conditions of nano- and micro-structured TiC S. Gavarini a,c,⇑, N. Millard-Pinard a, V. Garnier b, M. Gherrab a,b, J. Baillet a, L. Dernoncourt a, C. Peaucelle a, X. Jaurand c, T. Douillard b a b c
Université de Lyon, Université Claude Bernard Lyon 1, CNRS/IN2P3, UMR5822, IPNL, F-69622 Lyon, France INSA de Lyon, MATEIS CNRS UMR5510, 7 Avenue Jean Capelle, F-69621 Villeurbanne, France Université de Lyon, Université Lyon 1, Centre Technologique des Microstructures (CTl), 5 rue Raphael Dubois, 69622 Villeurbanne Cedex, France
a r t i c l e
i n f o
Article history: Received 17 February 2014 Received in revised form 2 March 2015 Accepted 24 April 2015
Keywords: Ceramic Carbide Irradiation Noble gas Grain size
a b s t r a c t Titanium carbide samples were prepared by spark plasma sintering. Three different microstructures were prepared with average grain sizes of about 0.3, 1.3 and 25.0 lm. Each microstructure was irradiated with either 500 keV 40Ar+ ions or 800 keV 129Xe++ ions. The irradiation fluence varied from 6 1016 to 3.2 1017 at.cm 2. Irradiation was carried out at room temperature (RT) or at 1000 °C. Post-irradiation annealing was performed on some samples to follow the surface modification. In fact, clusters and nanocracks were observed at depth in the nanometric grains (1 lm). Microcracks can induce localized surface blistering after irradiation at RT and for the highest fluencies. The size, shape and density of the blisters were proposed to depend on the crystallographic orientation of each grain. The microstructure with sub-micrometric grains exhibited increased surface roughness after irradiation, with grain removal and grain boundary abrasion but no blistering. Surface blistering is not observed after irradiation at 1000 °C but the complete delamination of extended areas containing large grains occurs. In this article, we highlight the role played by gastight grain boundaries and porosity to explain the distinct behavior of microstructures. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Due to their thermophysical and nuclear properties, refractory ceramics are attractive materials for applications in future nuclear power reactors (fusion [1–3] or fission concepts [3,4]) [5]. Since the Fukushima accident, ceramic cladding has been the subject of substantial research as it could offer improved safety and performance as well as addressing the current problems with zirconium alloy cladding in light water reactors (LWRs) [6–11]. For example, under severe accident scenarios in LWRs, the increase in temperature causes oxidation of the Zr alloy cladding (inside by UO2 but primarily outside by steam [11]) and results in rapid production of hydrogen. However, highly refractory ceramic clad fuel could survive much higher temperatures. As an illustration of the possible high performance of ceramic cladding, silicon carbide (SiC) fiber-reinforced SiC-matrix composites (SiC/SiC composites) were obtained by Japanese researchers using a Nano-powder Infiltration and Transient Eutectoid (NITE) ⇑ Corresponding author at: Université de Lyon, Université Claude Bernard Lyon 1, CNRS/IN2P3, UMR5822, IPNL, F-69622 Lyon, France. http://dx.doi.org/10.1016/j.nimb.2015.04.064 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
process that enabled production of an almost fully-dense SiC matrix [12]. Using this technology, SiC/SiC composites exhibit acceptable strength at room temperature (RT), pseudo-ductile fracture mode and extremely low gas permeability. Another technology called SiC Triplex nuclear fuel was developed recently in the USA to obtain cladding with three distinct layers [13]. This SiC Triplex design allows independent optimization of 3 properties: the inner monolith for fission gas retention, the fiber-reinforced matrix for overall mechanical performance, and the outer monolith for corrosion resistance. In France, ceramics are being studied for years as cladding in some fission reactors of Generation IV such as high temperature reactor (HTR) or gas cooled fast reactor (GFR) [14–16]. Recently, SiCf/SiC has also been proposed for hexagonal wrapper in sodium cooled fast reactor (SFR) [17]. A multi-layer design, called Sandwich structure, was proposed with an inner metallic liner and SiC/SiC layers (French atomic agency ‘‘CEA’’ Patent [14]) allowing a gastight material even when the SiC matrix is cracked. One of the possible issues of using SiC in cladding could be the relative decrease of its thermal conductivity at high temperature, under irradiation and also because of SiC matrix cracking in
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
SiC/SiC composites [18–20]. Other highly refractory carbide [21] and nitride [22,23] ceramics have been studied over the past few years as possible alternatives to SiC, either in the form of thin coatings (diffusion barriers [24]) or as possible matrices in fiber-reinforced composites [18]. Titanium carbide (TiC) is among the candidate materials because of its typical ceramic properties (very high melting point and hardness [5]), but also because of
Table 1 Densification ratio and average grain size for each of the three microstructures. Microstructure
SPS cycle
Densification ratios
Average grain size ± SD
M1
Cold uni-axial compacting step at 80 MPa + Temperature ramp of 5 °C/min up to 1300 °C under 17 MPa + Dwell for 1 h at 1300 °C under 80 MPa Cold uni-axial compacting step at 80 MPa + Temperature ramp of 5 °C/min up to 1975 °C at 17 MPa + Stage level for 1 h at 1975 °C and 80 MPa Cold uni-axial compacting step at 80 MPa + Temperature ramp of 5 °C/min up to 1975 °C at 17 MPa + Stage level for 10 h at 1975 °C and 80 MPa
90.0% ± 2.0 Geometrical method
340 ± 120 nm
97.3% ± 0.5 Archimedes’ method
1.34 ± 0.66 lm
95.7% ± 0.5 Archimedes’ method
24.80 ± 8.30 lm
M2
M3
115
its good thermal conductivity. Experiments carried out in France on the GANIL (Large National Heavy Ion Accelerator) facility at 500 °C showed that the thermal conductivity of TiC was higher than that of SiC [18,19]. Promising results were also obtained concerning TiC resistance to irradiation with krypton ions at RT [25,26]. In fact, a slight variation of thermal conductivity was observed but the surface topography remained unchanged after irradiation. TiC was not susceptible to amorphization during He ion irradiation over a wide temperature range 12–1523 K [27]. Upon annealing, several defect recovery stages were identified, and the specific diffusion processes responsible for the recovery were speculated as the diffusion of interstitial clusters. Other high temperature recovery and annealing experiments on TiC have indicated the possibility of activation of C vacancy migration at 600 °C and Ti vacancy migration above 1000 °C [21,28,29]. We showed in previous work that the grain size obtained after sintering of TiN and TiC influences their behavior under irradiation and oxidation at high temperature [30–33]. Some other studies observed a potential gain in terms of resistance to irradiation of non-oxide ceramics due to submicron or nanometric grains [23,34]. This specific behavior of nanostructured ceramics is linked to the high density of grain boundaries (GBs) which favor the evacuation of some structural defects during irradiation [35]. These GBs are also expected to play a significant role in gas mobility and release in nuclear applications [36]. In fact, ceramic cladding must be a retention barrier to the fission products (FPs) during the fuel cycle as indicated previously. Among the FPs are inert gases which are mainly composed of Xe and Kr [35] (and He gas atoms from a-decay). These noble gases are known for their ability to form small clusters and gas bubbles at high concentrations. Bubble nucleation and growth can lead to substantial modification of the properties: decrease of thermal conductivity, crack propagation, creation of migration channels for other FPs. Although numerous noble-gas implantation studies have been performed on metals,
Fig. 1. SEM micrographs (secondary electrons) of the three initial microstructures: (a) M1, (b) M2 and (c) M3.
116
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Table 2 Summary of the irradiation conditions. Incident ion 40
+
Ar Ar+ Xe++ 129 Xe++ 40
129
*
Fluence (at.cm
2
)
Ar 17 1 = 1.6 ± 0.2 10 Ar 17 2 = 3.2 ± 0.2 10 Xe 16 1 = 6.0 ± 0.2 10 Xe 17 2 = 1.2 ± 0.2 10
U U U U
Beam intensity – energy
Projected range ‘‘Rp’’ – straggle (nm)*
Max concentration near Rp (at.%)*
Max dpa production*
1 lA – 500 keV 1 lA – 500 keV 20 lA – 800 keV 20 lA – 800 keV
263.5 – 64.7
9.9 18.0 5.0 9.4
137 274 211 423
162.3 – 45.5
SRIM 2008Ò code.
17 Fig. 2. SEM micrographs (secondary electrons) of the three microstructures after argon irradiation at a fluence of UAr 1 = 1.6 ± 0.4 10 : (a) M1, (b) M2 and (c) M3.
Fig. 3. TEM images of M3 cross section after argon irradiation at RT and at the UAr 1 fluence.
117
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Fig. 4. (a) STEM image of the cracks and (b) argon X-ray cartography for the M3 cross section (white zone is argon-enriched).
5,0
M1 M2 M3 SRIM Profile
4,5
Xenon concentration (at. %)
4,0 3,5 3,0 2,5 2,0 1,5
Depth resolution
1,0 0,5 0,0
0
50
100
150
200
250
300
350
400
Depth (nm) Fig. 6. As-implanted xenon depth profiles under UXe at RT (experimental and 1 SRIM2008Ò profiles).
irradiated with Ar or Xe ions and submitted to thermal treatment. The mobility of the implanted rare gas was studied as well as its impact on the evolution of surface morphology and composition. Specific attention was paid to the influence of the following parameters: irradiation temperature, nature of the rare gas and fluence. Influence of the grain size of the samples on the implanted rare gas behavior is also discussed.
2. Experimental 2.1. Sintering cycles and resulting microstructures Fig. 5. STEM-HAADF images of (a) M1 cross-section after argon irradiation at RT and at the UAr 1 fluence; microcracks are only present in the largest grains and (b) zoom on nanometric grains that exhibit elongated (black arrows) and rounded shapes (black circles).
relatively little is known about microstructural effects of simultaneous high displacement damage and gas concentration in non-oxide ceramics [37–39]. Further information on the effects of high fluxes of noble gases is useful for determining material parameters which provide the best radiation resistance. In the present study, three different microstructures of TiC were obtained using spark plasma sintering. Sintered samples were then
The starting material is commercial TiC nanopowder (NanoStructured & Amorphous Materials Inc., USA) with a particle size of 40 nm and purity of 99% [40]. The main impurities of TiC are known to be oxygen and nitrogen compounds such as TiN, TiC and TiO which are isomorphous compounds [5]. The composition (major components) of the commercial powder was determined using elemental chemical analysis (see details in ref. [41], powder Nano-2), and the following composition was found TiC0.9O0.23N0.1. Spark Plasma Sintering (SPS) was used to obtain sintered bodies, and different cycles were implemented to obtain three distinct microstructures that will be called M1, M2 and M3 thereafter.
118
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
After sintering, carbon content was found to be comparable to the value found for the initial powder whatever the considered microstructure. The oxygen content has dropped to a value of about 4 at.% for both M1 and M2, and less than 2 at.% for M3 [42]. The resulting densification ratios are given in Table 1. The sintered densities of M2 and M3 were measured by Archimedes’ principle whereas that of M1 was estimated by geometric measurements as it has open porosity. Several samples were cut from each sintered body and these were then mechanically polished down to micron scale with diamond paste. The finishing step of this protocol consisted in vibrational polishing on colloidal silica. The samples were then heated at 1000 °C for 10 h under secondary vacuum ( M2 > M1 (i.e. 1.00 > 0.94 > 0.82). TEM images of M3 cross-sections are shown in Fig. 3 and show the formation of nanocracks at a depth between 250 and 325 nm (i.e. slightly beyond initial Rp, see Table 2). These nanocracks are interconnected in certain zones, which results in extended
2.4. Ion beam analysis RBS experiments were done using a 4 MV Van de Graaff Accelerator to determine xenon depth profiles. As indicated above, the RBS technique is not fully suitable in our case to obtain argon depth profiles because of the low mass of argon (argon signal located on titanium ‘‘step’’ signal). Measurements were therefore obtained with a 4He+ beam of a few MeV with an intensity of about 20 nA. Beam spot dimensions on the sample were 1 mm2 and the detection angle was 172°. Micro-RBS was also done using 3.05 MeV 4 He+ ions delivered by the HVEE in-line 3.5 MeV Singletron accelerator at the CENBG (Bordeaux Gradignan Nuclear Study Center) on AIFIRA (Interdisciplinary Applications of Ion Beams in the Aquitaine Region) platform. Data treatment, chemical map reconstruction and definition of regions of interest were made using Supavisio software (provided by the CENBG). For these specific experiments, the beam size was 0.6 0.6 lm2, micro-beam intensity was about 500 pA and an annular detector covering a large solid angle (about 170°) was used. With these specific conditions and relatively long acquisition times, it was possible to obtain xenon and even argon profiles on some samples irradiated at the highest fluence. Then, SIMNRA 6.04Ò software [46] was used to deduce Ar and Xe depth profiles from experimental scans and spectra. The metric depth scale was calculated using the density of pure TiC as an approximation to the unknown density in gas doped region. Note that the depth resolution of the RBS technique using a few MeV a particles and TiC samples is about 15–20 nm from the surface (from RESOLNRA 1.0 program [47]). 3. Results Xe 3.1. RT irradiation at lowest fluence (UAr 1 and U1 )
The surfaces of the three microstructures after RT irradiation under UAr 1 are presented in Fig. 2.
Fig. 8. (a) SEM-SE and (b) TEM image of M3 cross-section obtained by FIB after argon RT-irradiation at a fluence UAr 2 . The upper left inset figure shows the electron diffraction pattern for the irradiated band.
120
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Fig. 9. SEM micrographs of the (a) M2 (SEM-SE), (b) M2 (SEM-BSE) and (c) M3 (SEM-SE) microstructures after xenon irradiation at UXe 2 .
Fig. 10. Xenon depth profiles (implantation under UXe at RT) after thermal 1 treatment at 1000 °C for 10 h.
grain size is well below 100 nm (un-coarsened grains previously mentioned), see the central zone of Fig 5a. However, zooming in on this area (Fig 5b) enables us to observe elongated shapes (black arrows) and rounded shapes that could be nanometric clusters (black circles). Some transversal cracks were also observed (Fig 5a) but these could have been created during the FIB process because of the relative fragility of this particular microstructure. For xenon irradiation, no evidence of blistering was found in any of the microstructures, even on large grains. Xenon depth profiles deduced from RBS analysis are shown in Fig. 6. When we compare the Xe distribution predicted by SRIM code with the experimental data, we see that the xenon penetration depth is slightly overestimated by the code and the Full Width at Half Maximum (FWHM) is slightly larger for the experimental data. M2 and M3 microstructures exhibit similar profiles, closest to the theoretical prediction, whereas M1’s profile is lower. Xenon relative loss from the M1 microstructure compared to the other two microstructures is estimated at about 10%.
Xe 3.2. RT irradiation under the highest fluence (UAr 2 and U2 )
microcracks (zoom in Fig. 3a). Nanometric shapes that could correspond to nano-bubbles are also visible on each side of these cracks (zoom in Fig. 3b). Argon X-ray cartography was also performed on this cross-section (STEM mode) and it revealed the presence of argon on each side of the microcracks as shown in Fig. 4. Similar microcracks were also observed by STEM-HAADF for the other microstructures M1 and M2. However, these were only visible on the largest grains, i.e. above 100–150 nm typically as visible on Fig. 5a. Such features were not observed in the zones where the
M2 and M3 microstructures were also irradiated at higher flu17 17 ences (i.e. UAr at.cm 2 and UXe 2 = 3.2 ± 0.4 10 2 = 1.2 ± 0.4 10 at.cm 2). In the case of Ar-irradiation, localized blisters formed on the M3 surface (Fig. 7a, white circles), whereas no blisters were observed on the M2 surface (grain size > . This trend was confirmed on other analyzed zones.
123
4. Discussion 4.1. RT-Irradiation The presence of blisters on the surface of large grains after irradiation at RT and high fluence (therefore mostly for M3) is coherent with the presence of localized cracks beneath the surface and parallel to it as observed by TEM (Fig 8). Surface blistering was also observed in other materials after irradiation with hydrogen, helium or xenon ions at high fluence [37,48–57]. This phenomenon has been particularly studied for plasma-facing regions of fusion reactors, as plasma-facing materials will be exposed to intense fluxes of helium and hydrogen ions [37]. According to Reboh et al. [49,54,55], nanometric platelets (diameters around 10 nm and 1 nm) containing gas have been shown to be probable precursors for local blistering after H/He implantation in silicon. These platelets would act as nucleation sites for nano-cracks and then micro-crack formation extending in a lateral direction. At RT, cavity formation via long-range migration of vacancies or gas vacancy complexes would not be expected to occur in TiC since vacancies are immobile. The short-range migration and clustering of gas atoms might be enhanced by localized lattice strain in the vicinity of the implanted Ar or Xe. In addition collisional displacement
Fig. 15. (a) Xenon linear cartography (lRBS) with the corresponding SEM-SE micrograph and (b) xenon depth profile for the four analyzed grains.
124
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Fig. 16. SEM-SE micrographs of the three microstructures after argon irradiation at 1000 °C and a fluence UAr 2 : (a) M1, (b) M2 and (c)M3 microstructure.
damage during prolonged irradiation could induce short-range diffusion (also called ‘‘sputtering coalescence’’ [58,59]) and thereby facilitate the formation of small cavities. According to the few comparative studies found in the literature, similar blistering mechanisms can be reasonably assumed for all noble gases, including high-Z ones. As an illustration, Weber et al. [56] compared surface modifications induced by He, Kr and Xe implantation in metals and concluded that the same mechanisms were responsible for the saturation of the material resulting in bubble and blister formation at high energies and fluences. Similar behavior was also highlighted by Michel et al. [60,61] for Kr and Xe ions implanted in UO2 at relatively low fluence (equivalent to 1 at.%) concerning bubble nucleation. Argon platelets were observed by TEM during 50 keV Ar+ ion implantation in aluminum by Nair et al. [62]. A stress model and a gas pressure model have been proposed in order to explain blister formation [37,39,63,64]. According to some authors [37,64], both models are important to explain blistering and exfoliation. In our case, elongated nano-shapes that may correspond to nano-cracks have been observed by TEM in the case of nanometric grains (Fig 5b). For other microstructures (grain size P1 lm), the presence of more extended cracks (called microcracks) beneath and parallel to the surface could correspond to a more advanced stage. These cracks are located at a depth slightly beyond the theoretical ion range calculated by SRIM code in the case of argon for instance. However, this difference could be due to the fact that the thicknesses are measured directly in microns by scanning electron microscopy, whereas theoretical ion range is calculated assuming a normal density of the material (and no surface swelling), which is not the case in the area where there is a high concentration of gas. Note that the lateral stress model predicts a plane of fracture located towards the boundary between the material containing gas and the underlying undamaged material – i.e. at the end of the ion range rather than at the peak [64]. The absence of blisters on the surface of Xe-irradiated samples 16 at the lowest fluence (UXe at.cm 2, i.e. 5 at.% Xe peak con1 = 6.10 centration) could be an indication of the threshold concentration allowing visible Xe-induced blistering in our experimental
conditions. In fact, doubling the fluence (i.e. 9.4 at.% Xe peak concentration) results in the formation of blisters covering almost the entire surface. Such blisters had already been observed by Weber et al. [57] on the surface of TiN thin films irradiated with 360 keV Xe+ ions at fluences higher than 6 1016 ions.cm 2 (i.e. concentrations exceeding 7 at.% at Rp). TiN and TiC being isomorphous compounds, the threshold concentration for the formation of Xe-blisters near the surface of these two materials is suggested to be close to 7 at.%. A similar threshold is awaited for argon as both gaseous species have limited matrix solubility and diffusivity in the ceramic at RT. As a comparison, Weber et al. [56] found a critical gas concentration for blistering in aluminum implanted with Xe and Kr ions of about 14 at.% (bubble formation above 2 at.%). Threshold helium concentration needed to produce surface blistering and exfoliation was shown to be higher than 25 at.% in the case of a-SiC at RT (about 1.7 at.% for visible cavity formation) [37]. It is interesting to observe that, for two comparable gas peak Xe concentrations (i.e. UAr 1 and U2 ), argon-induced blisters are statistically larger than Xe ones (average diameter 3 lm for Ar blisters against 1 lm for Xe ones). Previous studies on metals implanted with helium ions have reported that blister size increases with increasing implantation depth, with a roughly linear relationship. An approximately correlation of r/R 5 was observed where r is the blister radius and R is the ion range [65,66]. The lateral stress model predicts r R1.5 [37]. One or both of these relationships may be appropriate in our case but it is difficult to prove their correctness because of the large variation in blister size on a given specimen. For both species, no blisters were observed on the surface on nano-grained and porous specimen. The explanation for this behavior is that the blisters are probably not stable when their contours intercept a porosity which may then play the role of an outlet. This is concordant with the higher gas release measured in the case of M1 just after RT-irradiation. Similar conclusions were obtained by comparing nano-porous, polycrystalline large-grain and single crystal of tungsten implanted with He ions at 850 °C [67]. Indeed, nano-porous W with 5 1021 He/m2 did not show surface blistering or exfoliation whereas polycrystalline large-grain tungsten exhibits blistering at 2 1021 He/m2 and exfoliation at 1 1022 He/m2.
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
125
Nano-porous tungsten appears to have significantly less retention than both polycrystalline and single crystal tungsten, supporting reduced ‘‘path length’’ argument for mitigating helium retention. In our case, a better controlled elaboration process in order to reduce (or to close) the porosity and improve GBs quality for the nano-grained microstructure could result in a more gastight microstructure and possible surface blistering. Indeed, for M3 microstructure, some blisters are formed very close to the GBs (see Fig. 2c). In fact, for this microstructure, sintering conditions led to significant grain coarsening and a residual porosity mostly located at triple points. This grain growth also results in the formation of stronger and very sharp GBs which contribute to a gastight microstructure. Unlike other polycrystalline ceramics implanted with helium [36,37], no continuous cavity layer is formed along grain boundaries that may have suggest enhanced gas mobility in the grain boundary compared to the crystalline matrix.
In the present study, irradiations at room temperature resulted in the formation of cavities approximately parallel to the irradiated surface mostly independent of the specific crystallographic orientation of the polycrystalline grains. As an illustration, elongated cracks are visible on both sides of the grain boundary in Fig. 8a. This strong geometric relationship between cavity habit plane orientation and the specimen surface is an indication of an important role of lateral stress [37]. Without consideration or implantation stresses, cavity formation would be expected to occur preferentially on the low index plane (low surface energy plane, i.e. (1 0 0) plane for TiC), that is most nearly parallel to the irradiated surface [37]. However, some specific grain orientations could actually have a better resistance to blistering as some grains do not show surface blistering for the M3 microstructure (see Fig. 9c). Sousbie et al. [68] have shown in the case of silicon monocrystals implanted with H or He ions that blistering depends on the face considered. In fact, many authors suggested that this phenomenon is strongly related to the mechanical properties of the material which are in turn correlated to the lattice planes in fcc structures [49,68–73]. The conditions required for blistering of silicon monocrystals implanted with hydrogen and deuterium (D) were also studied by Mountanabir et al. [52]. They concluded that the
Fig. 17. STEM (Bright field) images on the cross-section of M3 microstructure after Ar irradiation at high temperature.
Fig. 18. (a) Argon linear cartography (lRBS) with the corresponding SEM-SE micrograph and (b) argon depth profile for the different analyzed zones.
126
S. Gavarini et al. / Nuclear Instruments and Methods in Physics Research B 356-357 (2015) 114–128
Fig. 19. Inverse pole figure of undelaminated grains after argon irradiation at 1000 °C and a fluence UAr 2 . The top left corner shows bright undelaminated grains which were protected from the Argon irradiation. Those grains were not considered for the pole figure analysis.
density, size and degree of rupture of the D blisters depended on the crystal orientation. The (0 0 1) face was shown not to exfoliate because of a lack of gas pressure in the blisters. 4.2. High temperature irradiation and post-irradiation annealing The post-irradiation annealing at 1000 °C is supposed to thermodynamically increase the gas pressure entrapped in the localized cracks observed beneath the surface by TEM (see Fig. 8). Blister exfoliation occurs under the effect of this pressure as cracks become connected to the surface. Surface cracking is responsible for xenon depletion as observed in Fig. 14 (double peak profile at high temperature). Note that this typical phenomenon had already been observed in a previous work leading with titanium nitride irradiated at high fluence with Xe ions [74], and also by Weber et al. [56] on polycrystalline aluminum implanted with xenon ions. Indeed, the highest probability for blister formation is at the depth of highest gas concentration. Therefore, blister exfoliation leads to loss of xenon mainly out of this depth. A noticeable difference between post-irradiation annealing and irradiation at 1000 °C is that no blisters were observed in the latter. However, extended delaminated areas corresponding to the entire surface of initial grains was seen (Fig 16). One can conclude from these observations that the irradiation temperature is a key parameter influencing the lateral distribution of the implanted species by crack propagation. A corollary effect is that the rare gas is released in huge quantities from the grains as a result of surface delamination. Bulk diffusion is generally activated at approximately half of the melting temperature in units of K (Tm 3067 °C for TiC [5]). Above this temperature, thermal vacancies are mostly available. In our case, Ti-vacancies are supposed to be weakly mobile at 1000 °C in TiC and gas bubbles must have difficulty in obtaining vacancies; they are therefore likely to be over-pressurized as at RT. According to Evans et al. [63] in the case of metals implanted with helium, the only difference between surface blistering at RT and exfoliation (also called flaking) at higher temperature (
