- Email: [email protected]
Thermal migration of deuterium implanted in graphite: Influence of free surface proximity and structure
Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Thermal migration of deuterium implanted in graphite: Influence of free surface proximity and structure M. Le Guillou a, N. Moncoffre a,⇑, N. Toulhoat a,b, Y. Pipon a,c, M.R. Ammar d, J.N. Rouzaud e, D. Deldicque e a
Institut de Physique Nucléaire de Lyon, CNRS/IN2P3 UMR 5822, Université Claude Bernard Lyon 1, Université de Lyon, F-69622 Villeurbanne cedex, France CEA/DEN – Centre de Saclay, F-91191 Gif-sur-Yvette cedex, France c Institut Universitaire Technologique, Université Claude Bernard Lyon 1, Université de Lyon, F-69622 Villeurbanne cedex, France d CNRS, CEMHTI UPR3079, Université Orléans, CS90055, F-45071 Orléans cedex 2, France e Laboratoire de Géologie de l’Ecole Normale Supérieure, Paris, UMR CNRS ENS 8538, F-75231 Paris cedex 5, France b
a r t i c l e
i n f o
Article history: Received 30 June 2015 Received in revised form 12 November 2015 Accepted 17 November 2015 Available online 22 December 2015 Keywords: Nuclear Reaction Analysis Graphite Deuterium Temperature
a b s t r a c t This paper is a contribution to the study of the behavior of activation products produced in irradiated nuclear graphite, graphite being the moderator of the first French generation of CO2 cooled nuclear fission reactors. This paper is focused on the thermal release of Tritium, a major contributor to the initial activity, taking into account the role of the free surfaces (open pores and graphite surface). Two kinds of graphite were compared. On one hand, Highly Oriented Pyrolitic Graphite (HOPG), a model well graphitized graphite, and on the other hand, SLA2, a porous less graphitized nuclear graphite. Deuterium ion implantation at three different energies 70, 200 and 390 keV allows simulating the presence of Tritium at three different depths, corresponding respectively to projected ranges Rp of 0.75, 1.7 and 3.2 lm. The D isotopic tracing is performed thanks to the D(3He,p)4He nuclear reaction. The graphite structure is studied by Raman microspectrometry. Thermal annealing is performed in the temperature range 200–1200 °C up to 300 h annealing time. As observed in a previous study, the results show that the D release occurs according to three kinetic regimes: a rapid permeation through open pores, a transient regime corresponding to detrapping and diffusion of D located at low energy sites correlated to the edges of crystallites and finally a saturation regime attributed to detrapping of interstitial D located at high energy sites inside the crystallites. Below 600 °C, D release is negligible whatever the implantation depth and the graphite type. The present paper clearly puts forward that above 600 °C, the D release decreases at deeper implantation depths and strongly depends on the graphite structure. In HOPG where high energy sites are more abundant, the D release is less dependent on the surface proximity compared to SLA2. In SLA2, in which the low energy sites prevail, the D release curves are clearly shifted towards lower temperatures when D is located close to free surfaces. Extrapolating our data to Tritium mobility in irradiated graphite, we show that thermal selective extraction of T would be all the more so efficient as the graphite structure is more disordered, which means in the most irradiated and damaged graphite zones in the reactor. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Graphite has been used as neutron moderator in many types of nuclear reactors due to its ability to slow down fast neutrons without capturing them. Whatever the reactor design, the irradiated graphite waste management has to be faced sooner or later regarding the production of long lived radioactive species or radionuclides that might be dose-determining for disposal such as 14C, 36 Cl or 3H. Thus, all over the world, around 250,000 tons of ⇑ Corresponding author. E-mail address: [email protected] (N. Moncoffre). http://dx.doi.org/10.1016/j.nimb.2015.11.034 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
irradiated graphite have been produced through commercial and military nuclear power operation. Many of these reactors are now being decommissioned. Whatever the management options, a particular attention should be paid to Tritium. This radionuclide that was mainly produced through thermal neutron activation 6 Li(nth,a)3H of 6Li impurities [1], has a short life time (around 12 years) but it is a major contributor to the initial radioactive dose [2]. In particular, T release might impact safety of the dismantling operation stage as well as the radioactive waste package management at the operational stages of disposal. Indeed, according to its location and speciation in the irradiated graphite waste, it might be potentially released through isotopic exchange with water or
308
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
water vapor. Thus, in order to anticipate eventual T release, its radioactive inventory, location and speciation in the irradiated graphite should be previously assessed and its behavior during dismantling and waste management foreseen. The protocol used in this work is based on the use of D implanted graphite samples in order to simulate the presence of T, displaced from its original structural site through recoil during reactor operation. NRA (Nuclear Reaction Analysis) is particularly well adapted to the analysis of D. The implanted D profiles are analyzed using the D(3He,p)4He nuclear reaction. First results on the effects of temperature on the behavior of D in nuclear graphite have already been reported in previous papers [3,4]. The aim was to elucidate the behavior of T during reactor operation and to explore options of thermal decontamination in inert gas or oxidizing conditions. The results show that the D release is strongly correlated to proximity of the free surfaces and also to the nature and relative abundance of the different D trapping sites. Thus, higher temperatures are necessary to detrap D located at high energy sites inside the crystallites compared to D located at the coke grain edges or in the porous binder. It is important to remind that nuclear graphite has a polycrystalline structure and contains micrometer sized coke grains that are blended with coal tar pitch acting as a binder. The grains are formed by several more or less oriented crystallites with a size of a few hundreds nm and each crystallite is formed by a triperiodic stacking of graphene planes. The structural organization and the graphitization level of nuclear graphite are thus strongly heterogeneous at different scales [5,6].Therefore, in order to get more insight into the D release mechanisms, we decided to investigate also the role of graphite structure using Raman microspectrometry. For this purpose, two kinds of graphite have been considered: (i) Virgin nuclear graphite provided by EDF (Electricité De France) with an average porosity of 32%; (ii) Highly Ordered Pyrolytic Graphite (HOPG) Grade SPI-1graphite whose micrometer sized crystallites are all oriented parallel to the HOPG platelet, giving it a perfect lamellar structure; this HOPG is highly crystalline and not porous. Then, in order to study the role of surface proximity on the D release, we implanted both kinds of graphite samples respectively at three different implantation depths. 2. Experimental 2.1. Sample preparation and deuterium ion implantation The virgin nuclear graphite is issued from the UNGG (Uranium Naturel-Graphite-Gaz) CO2 cooled reactor Saint-Laurent A2 (EDF, SLA2, St-Laurent-des-Eaux, France), currently under dismantling. The samples are cut to a size of about 9 5 4 mm3 with a diamond saw and one face is polished down to the micrometer with diamond paste. HOPG is obtained from SPI Supplies (West Chester, US) through Neyco SA (Paris, France). We received the HOPG samples as 1 mm thick 10 10 mm2 plates. The plates are cut with a diamond saw into approximately 10 5 1 mm3 plates. Both nuclear graphite and HOPG samples were annealed at 1000 °C– 1200 °C for 8 h either in high vacuum (P ffi 107 mbar) in order to desorb most of the gaseous impurities and also in order to anneal at least partially the defects induced by polishing (for the nuclear graphite samples). The graphite samples are implanted with D+ ions at room temperature (RT) under vacuum either using the 400 kV ion implanter IMIO400 of the Institut de Physique Nucléaire of Lyon (IPNL, France) or the 200 kV ion implanter EATON 200MC of ICube laboratory of the University of Strasbourg, France or the 500 kV ion implanter KIIA of the University of Helsinki, Finland. Deuterium implantation was carried out at three different energies, 70, 200 and 390 keV. A density of 2.2 g cm3 was taken for HOPG samples (density of
perfect graphite) whereas for SLA2, we have assumed a density of 1.92 g cm3 which will be discussed in the next section. Hence, the respective projected ranges (Rp) deduced from SRIM 2011 [7] are respectively of 0.75, 1.7 and 3.2 lm for SLA2 and 0.65, 1.45 and 2.8 for HOPG. Since the D implantation profiles were subsequently analyzed by NRA with a detection limit around 1 at.% in our experimental conditions, D+ ions were implanted at a fluence of 5 1016 D+ cm2. On the basis of SRIM calculation, the D concentrations at Rp are respectively of 3.4, 2.9 and 2.5 at.% and the number of displacements per atom (dpa) at the projected damage range (Rd), using the Kinchin–Pease calculation mode, are 0.3, 0.28 and 0.25 for both types of graphite. It can be noticed that the implanted D concentrations are higher than those of T produced in graphite moderator during UNGG reactor operation which range from some ppb to several tenths of ppm. In spite of the relatively high D+ fluence, the maximum amount of displaced atoms generated by the implantation process in the graphite matrix remains quite low, around 0.3 dpa. 2.2. Analysis of the implanted deuterium The D profiles were analyzed using the D(3He,p)4He nuclear reaction with an incident 3He beam. This reaction has already been used in particular in the context of studies devoted to fusion reactors [8–10].The analyses were carried out with a millimetric beam at the 4 MV Van de Graaff facility of IPNL. The reaction produces 13 MeV protons and 2 MeV 4He+ particles. The incident beam was normal to the sample surface and only protons were detected at 155° using a 2000 lm thick Si surface barrier detector with a solid angle of 12 msr. A 23 lm thick MylarÒ screen was placed in front of the detector to stop the emitted 4He+ particles as well as the backscattered 3He+ particles. Experimental details can be found in Refs. [3,4]. Depending on the increasing implantation depths, the D profiles were analyzed using increasing 3He+energies of 900, 1150 and 1600 keV. These energy values allow adjusting the maximum of the D(3He,p)4He reaction cross section rNRA located around 620 keV at a detection angle of 135° [11], as close as possible to the calculated D projected ranges. As mentioned previously, a density of 2.2 g cm3 was assumed for HOPG due to its highly ordered state and its perfect lamellar structure whereas for the SLA2 samples, a lower density of 1.92 g cm3 was chosen due to its porous structure. Fig. 1 represents the D distribution profiles for both graphite grades as a function of depth for the implantation energy of 390 keV. This figure shows the implantation profiles
Fig. 1. D distribution profiles in graphite as a function of depth for an implantation energy of 390 keV: experimental (black triangles) and fitted SIMNRA profile (black dotted line) and SRIM calculated profiles for respective densities of 1.92 g cm3 (green dashed dotted line) and 2.2 g cm3 (red dashed dotted line). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
obtained by SRIM for both sample densities as well as the experimental data and the SIMNRA fits for the SLA2 sample (1.92 density). Using the RESOLNRA code [12] of SIMNRA [13], we calculated the D(3He,p)4He nuclear reaction depth resolution values for the three incident particle energies (900, 1150 and 1600 keV) for both sample densities at the sample surface and at the respective projected ranges Rp. Calculations show that, within 10%, these values are similar for both densities. The following mean values have been obtained: (i) at 900 keV:270 nm (surface) and 190 nm (Rp), (ii) at 1150 keV:370 nm (surface) and 185 nm (Rp) (iii) at 1600 keV:600 nm (surface) and 180 nm (Rp). Considering our experimental conditions (in particular a detector resolution of 18 keV), these depth resolutions are in rather good agreement with those obtained by Mayer et al. in carbon samples [14]. 2.3. Heat treatments Heat treatments of graphite samples in inert atmosphere were performed either in high vacuum (P ffi 107 mbar) or in inert gas flow (Ar) at temperatures ranging from 200 to 1200 °C for different annealing times. Annealing in high vacuum was carried out at IPNL with a Pekly ETF 30-50/15-S tubular furnace using a silica tube. The annealing time ranged up to about 300 h for temperatures ranging from 500 to 1200 °C with a heating ramp of 5 °C min1 and without any programmed cooling ramp. Heat treatments in inert gas were carried out in a pure argon 5.01 flow at about 1 bar (0.2–2 L min1 depending on the annealing setup), either at Ampère laboratory (Institut National des Sciences Appliquées – INSA of Lyon, France) using a Jipelec JetFirst rapid thermal annealing (RTA) furnace for flash treatments (2–3 min) with a heating ramp ranging from 10 to 30 °C s1 and a cooling ramp of a few minutes, or at IPNL using a Thermolyne 21100 tubular furnace for relatively short annealing durations (below 24 h) and temperatures up to 1000 °C with heating and cooling ramps of a few minutes, as well as with a Nabertherm RHTH 120-150/18 tubular furnace for heat treatments at 1200 °C with heating and cooling ramps of 5 °C min1. Duplicate experiments (that will not be shown here) performed either in argon or in vacuum showed that the thermal evolution of the implanted D did not depend on the type of inert atmosphere nor the heating and cooling ramps. 2.4. Graphite structure analysis
309
3. Results and discussion 3.1. Dependence of D behavior on implantation depth: role of surface proximity Let us remind that the main D thermal migration process, already observed in SLA2 nuclear graphite implanted at 70 keV, is the release [3]. It starts around 500 °C and is completed around 1200 °C. Moreover, first results obtained after 24 h annealing on 70 keV (Rp ffi 0.75 lm) and 390 keV (Rp ffi 3.2 lm) implanted SLA2 samples had shown that the release was more effective when the D profile was closer to the surface. Three release steps had been proposed, corresponding to three distinct regimes [4]: an almost instantaneous release regime corresponding to the detrapping of D located near the free surfaces, a transient regime attributed to the thermal detrapping of D from low energy sites and its diffusion along crystallites and coke grain edges, the limiting factor being the diffusion process, and a saturation regime corresponding to the detrapping of interstitial D located at high energy sites inside the crystallites and chemically bound to the graphene layer boundaries. In the present paper, a systematic study has been carried out for three implantation depths and increasing annealing times from 1 to almost 300 h. Results are displayed in Figs. 2 and 3. Fig. 2, presenting the D release as a function of annealing temperature, shows that, for a given annealing time, a deeper implantation depth shifts the release curve towards higher temperatures. In other words, at a given temperature, the release decreases with the implantation depth. However, the 1.7 lm curves are closer to the 3.2 lm ones below 700 °C, whereas for temperatures above 900 °C, they are closer to the 0.75 lm curves. This contrasted behavior is probably related to the progressive detrapping of tightly bound deuterium above 900 °C for which the release is facilitated due to the proximity of surface (1.7 lm) compared to the deeper located one (3.2 lm). Fig. 3a presents the D release, for 96 h annealing, as a function of the implantation depth for various annealing temperatures. The data have been obtained at different incident particle energies: 900 keV for the 0.75 lm implantation depth, 1150 keV for 1.7 lm and 1600 keV for 3.2 lm. This figure shows that at 900 °C, the release becomes nearly linear as a function of the implantation depth reaching an almost constant value at 1200 °C whatever the depth. Fig. 3b, showing the release for various annealing times and at two temperatures (1200 and 600 °C), puts forward that a 600 °C annealing favors the release of D located close to the surface whereas a 1200 °C annealing is more effective for detrapping deeper located D.
Raman microspectrometry experiments were carried out on both nuclear graphite and HOPG samples at ENS (Ecole Normale Supérieure, Paris, France) or CEMHTI (Conditions Extrêmes et Matériaux : Haute Température et Irradiation, Orléans, France), using a Renishaw inVia Reflex spectrometer in order to follow the structural modifications induced by the different treatments. Ar laser source (514.5 nm wavelength – 2.41 eV energy) focused on a 1 lm2 spot through a Leica microscope using 50 objective), while the Raman-scattered light was collected by a CCD camera. Moreover, given that sample heating due to the laser spot might affect the measurements, we used a low incident power (<1 mW) [15]. Since the penetration depth of a Raman laser into graphite is limited to a hundred nanometers [16], the analyses were limited to the samples implanted at the lowest energy of 70 keV in order to probe zones as close as possible to the most disordered ones.
1
Minimum purity of 99.999% guaranteed by the supplier (Linde Gas).
Fig. 2. Fitted release curves and (experimental data for the sample annealed for 48 h) corresponding to the D release as a function of temperature for the SLA2 graphite implanted respectively at 70, 200 and 390 keV and annealed for 24, 48 and 96 h.
310
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
Fig. 3. D releases as a function of the implantation depth for (a) different temperatures and an annealing duration of 96 h and (b) for annealing temperatures of 600 and 1200 °C and annealing times varying from 1 h to 288 h. The dashed lines allow guiding the eye. The data have been obtained at different incident particle energies: 900 keV for the 0.75 lm implantation depth, 1150 keV for 1.7 lm and 1600 keV for 3.2 lm.
Fig. 4. Fitted release curves corresponding to the D release as a function of temperature for SLA2 and HOPG graphite implanted respectively at 70, 200 and 390 keV and annealed for 24, 48 and 96 h.
Fig. 5. HOPG Raman spectra of virgin, as-implanted at 70 keV and implanted and annealed samples in the temperature range 500–1000 °C during 300 h. The spectra have been normalized with respect to the G band.
3.2. Effect of the graphite structure on the implanted D behavior Let us now compare the results obtained in SLA2 nuclear graphite in the previous section to those obtained in HOPG for which
Fig. 6. SLA2 Raman spectra of virgin, as-implanted at 70 keV and implanted and annealed samples in the temperature range 500–1000 °C during 300 h. The spectra have been normalized with respect to the G band.
the structure is much more homogeneous at a larger scale. The comparison of the D release in both materials as a function of temperature is given in Fig. 4. Whatever the annealing time and implanted depth, the release curves of the 0.65 and 2.8 lm implantation depths in HOPG are shifted towards high temperatures compared to the SLA2 implanted samples. This means that the release in HOPG is always lower than that in SLA2. Indeed, in HOPG, during implantation, D is mainly trapped into interstitial high energy sites, which means that more energy will be necessary to displace it from these sites than for SLA2 graphite, for which a great part of D should be trapped at low energy sites located along crystallites and coke grain edges [4]. At the same time, graphene planes are also fractured by the D+ ion beam, which gives rise to release short-cuts. However, the impact of these short-cuts on the D release remains less important compared to those induced by the high inter and intragranular-porosity of the nuclear graphite. In order to go further in the understanding of these results, and especially to follow the increase of the structural disorder due to the ion implantation, Raman analysis was performed. Raman spectra are displayed in Figs. 5 and 6 respectively for HOPG and SLA2 samples implanted at the lowest energy (see Table 1 which displays the implantation data for SLA2 and HOPG graphite). These figures present the normalized intensity of the symmetry-allowed G band and the defect-induced D1 and D2 bands as a function of the Raman shift. Raman spectra clearly show that the virgin nuclear graphite is already more disordered than HOPG since its spectrum exhibits a D1
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311 Table 1 Implantation data (energy, Rp, maximum D concentration and dpa) for SLA2 and HOPG graphite. Implantation energy (keV) Rp in SLA2 (lm) Rp in HOPG (lm) [D] at the Rp (at.%) for both graphites [dpa] at the Rd for both graphites
70 0.7 0.6 3.4 0.3
200 1.7 1.4 2.9 0.3
390 3.2 2.8 2.5 0.2
band which does not exist on the HOPG one. In both types of graphite, implantation results in the appearance of significant defect bands compared to the virgin samples. Temperature annealing favors the structure reordering without reaching the structural states of the virgin samples even at 1000 °C. 4. Conclusion Finally, the D(3He,p)4He nuclear reaction is an appropriate probe to study the D thermal release in graphite up to an implantation depth of a few microns. It allows confirming the crucial role of the graphite structure on this release and drawing interesting conclusions for management of nuclear graphite. Indeed, D is significantly more easily released in highly porous and less graphitized graphite compared to highly graphitized and lamellar model graphite. Raman microspectroscopy results also put forward the important role of temperature on the reordering of the structure in both types of graphite. Thus, by extrapolating our results to the behavior of T during reactor operation, it is likely that (i) at around 500 °C, temperature close to the maximum operating UNGG temperature, a major part of the T located close to free surfaces has been released, (ii) T should be less released in the coke grains for which the graphitization level is higher than in the binder which is less graphitized and more porous. This has been well evidenced by Le Guillou et al. [4] using nuclear microprobe analysis. Considering irradiated graphite waste purification options, the thermal process should therefore be more efficient for the most disordered graphite. Moreover, a fine crushing as well as temperatures higher than 1200–1300 °C should allow getting rid of a major part of T after dozens of annealing hours.
311
Acknowledgments This work was performed in the frame of an Andra Ph.D. grant and financially supported by Andra, EDF and the European project CARBOWASTE (Seventh EURATOM Framework Programme). The authors are grateful to A. Perrat-Mabilon (IPNL) and Y. Le Gall (ICube) as well as to R. Ingren and B. Marchand (University of Helsinki) for D ion implantations. They also would like to thank F. Laariedh (INSA) for RTA treatments and the accelerator staff of IPNL for its helpful support during ion beam experiments. References [1] M.-C. Perrin, B. Poncet, M.-T. Pascal, Livre blanc du Tritium, Le Tritium dans les déchets des réacteurs d’EDF en cours de démantèlement, ASN, 2010, p. 120 (Chapitre 2). [2] Andra, Etude des scénarios de gestion à long terme des déchets de faible activité massique à vie longue, Andra technical report, FRPADPG.12.0020/A 2012, p. 37. [3] M. Le Guillou, N. Toulhoat, Y. Pipon, N. Moncoffre, N. Bérerd, A. Perrat-Mabilon, R. Rapegno, Nucl. Instr. Meth. B 332 (2014) 90. [4] M. Le Guillou, N. Toulhoat, Y. Pipon, N. Moncoffre, H. Khodja, J. Nucl. Mater. 461 (2015) 72–77. [5] J.-P. Bonal, J.-C. Robin, Les Réacteurs Nucléaires à Caloporteur gaz, Un Matériau Fascinant: Le Graphite, CEA/DEN, CEA Saclay et Groupe Moniteur (Editions du Moniteur), Paris, 2006, p. 27. [6] C.-E. Vaudey, N. Toulhoat, N. Moncoffre, N. Bérerd, L. Raimbault, P. Sainsot, J.-N. Rouzaud, A. Perrat-Mabilon, J. Nucl. Mater. 395 (2009) 62. [7] J.F. Ziegler, J.P. Biersack, M.D. Ziegler, SRIM, The Stopping and Range of Ions in Matter, SRIM Co, 2008. [8] T. Hayashi, K. Sugyiama, K. Krieger, M. Mayer, V.K. Alimov, T. Tanabe, K. Masaki, N. Miya, J. Nucl. Mater. 363 (2007) 904. [9] I. Takagi, K. Yoshida, K. Shin, K. Higashi, Nucl. Instr. Meth. Phys. Res. B 84 (1994) 393. [10] H. Khodja, C. Brosset, N. Bernier, Nucl. Instr. Meth. Phys. Res. B 266 (2008) 1425. [11] V.K. Alimov, M. Mayer, J. Roth, Nucl. Instr. Meth. Phys. Res. B 234 (2005) 90. [12] M. Mayer, Nucl. Instr. Meth. B 266 (2008) 1852. [13] M. Mayer. Tech. Rep. IPP 9/113, Max-Planck-Institut fur Plasmaphysik, Garching, 1997. [14] M. Mayer, E. Gauthier, K. Sugiyama, U. von Toussaint, Nuclear Instruments and Methods in Physics Research B: Beam Interactions with Materials and Atoms, 267, No. 3, 2009, pp. 506–512. [15] N.J. Everall, J. Lumsdon, D.J. Christopher, Carbon 29 (1991) 133–137. [16] M.R. Ammar, J.N. Rouzaud, C.E. Vaudey, N. Toulhoat, N. Moncoffre, Carbon 48 (2010) 1244–1251.
Contents lists available at ScienceDirect
Nuclear Instruments and Methods in Physics Research B journal homepage: www.elsevier.com/locate/nimb
Thermal migration of deuterium implanted in graphite: Influence of free surface proximity and structure M. Le Guillou a, N. Moncoffre a,⇑, N. Toulhoat a,b, Y. Pipon a,c, M.R. Ammar d, J.N. Rouzaud e, D. Deldicque e a
Institut de Physique Nucléaire de Lyon, CNRS/IN2P3 UMR 5822, Université Claude Bernard Lyon 1, Université de Lyon, F-69622 Villeurbanne cedex, France CEA/DEN – Centre de Saclay, F-91191 Gif-sur-Yvette cedex, France c Institut Universitaire Technologique, Université Claude Bernard Lyon 1, Université de Lyon, F-69622 Villeurbanne cedex, France d CNRS, CEMHTI UPR3079, Université Orléans, CS90055, F-45071 Orléans cedex 2, France e Laboratoire de Géologie de l’Ecole Normale Supérieure, Paris, UMR CNRS ENS 8538, F-75231 Paris cedex 5, France b
a r t i c l e
i n f o
Article history: Received 30 June 2015 Received in revised form 12 November 2015 Accepted 17 November 2015 Available online 22 December 2015 Keywords: Nuclear Reaction Analysis Graphite Deuterium Temperature
a b s t r a c t This paper is a contribution to the study of the behavior of activation products produced in irradiated nuclear graphite, graphite being the moderator of the first French generation of CO2 cooled nuclear fission reactors. This paper is focused on the thermal release of Tritium, a major contributor to the initial activity, taking into account the role of the free surfaces (open pores and graphite surface). Two kinds of graphite were compared. On one hand, Highly Oriented Pyrolitic Graphite (HOPG), a model well graphitized graphite, and on the other hand, SLA2, a porous less graphitized nuclear graphite. Deuterium ion implantation at three different energies 70, 200 and 390 keV allows simulating the presence of Tritium at three different depths, corresponding respectively to projected ranges Rp of 0.75, 1.7 and 3.2 lm. The D isotopic tracing is performed thanks to the D(3He,p)4He nuclear reaction. The graphite structure is studied by Raman microspectrometry. Thermal annealing is performed in the temperature range 200–1200 °C up to 300 h annealing time. As observed in a previous study, the results show that the D release occurs according to three kinetic regimes: a rapid permeation through open pores, a transient regime corresponding to detrapping and diffusion of D located at low energy sites correlated to the edges of crystallites and finally a saturation regime attributed to detrapping of interstitial D located at high energy sites inside the crystallites. Below 600 °C, D release is negligible whatever the implantation depth and the graphite type. The present paper clearly puts forward that above 600 °C, the D release decreases at deeper implantation depths and strongly depends on the graphite structure. In HOPG where high energy sites are more abundant, the D release is less dependent on the surface proximity compared to SLA2. In SLA2, in which the low energy sites prevail, the D release curves are clearly shifted towards lower temperatures when D is located close to free surfaces. Extrapolating our data to Tritium mobility in irradiated graphite, we show that thermal selective extraction of T would be all the more so efficient as the graphite structure is more disordered, which means in the most irradiated and damaged graphite zones in the reactor. Ó 2015 Elsevier B.V. All rights reserved.
1. Introduction Graphite has been used as neutron moderator in many types of nuclear reactors due to its ability to slow down fast neutrons without capturing them. Whatever the reactor design, the irradiated graphite waste management has to be faced sooner or later regarding the production of long lived radioactive species or radionuclides that might be dose-determining for disposal such as 14C, 36 Cl or 3H. Thus, all over the world, around 250,000 tons of ⇑ Corresponding author. E-mail address: [email protected] (N. Moncoffre). http://dx.doi.org/10.1016/j.nimb.2015.11.034 0168-583X/Ó 2015 Elsevier B.V. All rights reserved.
irradiated graphite have been produced through commercial and military nuclear power operation. Many of these reactors are now being decommissioned. Whatever the management options, a particular attention should be paid to Tritium. This radionuclide that was mainly produced through thermal neutron activation 6 Li(nth,a)3H of 6Li impurities [1], has a short life time (around 12 years) but it is a major contributor to the initial radioactive dose [2]. In particular, T release might impact safety of the dismantling operation stage as well as the radioactive waste package management at the operational stages of disposal. Indeed, according to its location and speciation in the irradiated graphite waste, it might be potentially released through isotopic exchange with water or
308
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
water vapor. Thus, in order to anticipate eventual T release, its radioactive inventory, location and speciation in the irradiated graphite should be previously assessed and its behavior during dismantling and waste management foreseen. The protocol used in this work is based on the use of D implanted graphite samples in order to simulate the presence of T, displaced from its original structural site through recoil during reactor operation. NRA (Nuclear Reaction Analysis) is particularly well adapted to the analysis of D. The implanted D profiles are analyzed using the D(3He,p)4He nuclear reaction. First results on the effects of temperature on the behavior of D in nuclear graphite have already been reported in previous papers [3,4]. The aim was to elucidate the behavior of T during reactor operation and to explore options of thermal decontamination in inert gas or oxidizing conditions. The results show that the D release is strongly correlated to proximity of the free surfaces and also to the nature and relative abundance of the different D trapping sites. Thus, higher temperatures are necessary to detrap D located at high energy sites inside the crystallites compared to D located at the coke grain edges or in the porous binder. It is important to remind that nuclear graphite has a polycrystalline structure and contains micrometer sized coke grains that are blended with coal tar pitch acting as a binder. The grains are formed by several more or less oriented crystallites with a size of a few hundreds nm and each crystallite is formed by a triperiodic stacking of graphene planes. The structural organization and the graphitization level of nuclear graphite are thus strongly heterogeneous at different scales [5,6].Therefore, in order to get more insight into the D release mechanisms, we decided to investigate also the role of graphite structure using Raman microspectrometry. For this purpose, two kinds of graphite have been considered: (i) Virgin nuclear graphite provided by EDF (Electricité De France) with an average porosity of 32%; (ii) Highly Ordered Pyrolytic Graphite (HOPG) Grade SPI-1graphite whose micrometer sized crystallites are all oriented parallel to the HOPG platelet, giving it a perfect lamellar structure; this HOPG is highly crystalline and not porous. Then, in order to study the role of surface proximity on the D release, we implanted both kinds of graphite samples respectively at three different implantation depths. 2. Experimental 2.1. Sample preparation and deuterium ion implantation The virgin nuclear graphite is issued from the UNGG (Uranium Naturel-Graphite-Gaz) CO2 cooled reactor Saint-Laurent A2 (EDF, SLA2, St-Laurent-des-Eaux, France), currently under dismantling. The samples are cut to a size of about 9 5 4 mm3 with a diamond saw and one face is polished down to the micrometer with diamond paste. HOPG is obtained from SPI Supplies (West Chester, US) through Neyco SA (Paris, France). We received the HOPG samples as 1 mm thick 10 10 mm2 plates. The plates are cut with a diamond saw into approximately 10 5 1 mm3 plates. Both nuclear graphite and HOPG samples were annealed at 1000 °C– 1200 °C for 8 h either in high vacuum (P ffi 107 mbar) in order to desorb most of the gaseous impurities and also in order to anneal at least partially the defects induced by polishing (for the nuclear graphite samples). The graphite samples are implanted with D+ ions at room temperature (RT) under vacuum either using the 400 kV ion implanter IMIO400 of the Institut de Physique Nucléaire of Lyon (IPNL, France) or the 200 kV ion implanter EATON 200MC of ICube laboratory of the University of Strasbourg, France or the 500 kV ion implanter KIIA of the University of Helsinki, Finland. Deuterium implantation was carried out at three different energies, 70, 200 and 390 keV. A density of 2.2 g cm3 was taken for HOPG samples (density of
perfect graphite) whereas for SLA2, we have assumed a density of 1.92 g cm3 which will be discussed in the next section. Hence, the respective projected ranges (Rp) deduced from SRIM 2011 [7] are respectively of 0.75, 1.7 and 3.2 lm for SLA2 and 0.65, 1.45 and 2.8 for HOPG. Since the D implantation profiles were subsequently analyzed by NRA with a detection limit around 1 at.% in our experimental conditions, D+ ions were implanted at a fluence of 5 1016 D+ cm2. On the basis of SRIM calculation, the D concentrations at Rp are respectively of 3.4, 2.9 and 2.5 at.% and the number of displacements per atom (dpa) at the projected damage range (Rd), using the Kinchin–Pease calculation mode, are 0.3, 0.28 and 0.25 for both types of graphite. It can be noticed that the implanted D concentrations are higher than those of T produced in graphite moderator during UNGG reactor operation which range from some ppb to several tenths of ppm. In spite of the relatively high D+ fluence, the maximum amount of displaced atoms generated by the implantation process in the graphite matrix remains quite low, around 0.3 dpa. 2.2. Analysis of the implanted deuterium The D profiles were analyzed using the D(3He,p)4He nuclear reaction with an incident 3He beam. This reaction has already been used in particular in the context of studies devoted to fusion reactors [8–10].The analyses were carried out with a millimetric beam at the 4 MV Van de Graaff facility of IPNL. The reaction produces 13 MeV protons and 2 MeV 4He+ particles. The incident beam was normal to the sample surface and only protons were detected at 155° using a 2000 lm thick Si surface barrier detector with a solid angle of 12 msr. A 23 lm thick MylarÒ screen was placed in front of the detector to stop the emitted 4He+ particles as well as the backscattered 3He+ particles. Experimental details can be found in Refs. [3,4]. Depending on the increasing implantation depths, the D profiles were analyzed using increasing 3He+energies of 900, 1150 and 1600 keV. These energy values allow adjusting the maximum of the D(3He,p)4He reaction cross section rNRA located around 620 keV at a detection angle of 135° [11], as close as possible to the calculated D projected ranges. As mentioned previously, a density of 2.2 g cm3 was assumed for HOPG due to its highly ordered state and its perfect lamellar structure whereas for the SLA2 samples, a lower density of 1.92 g cm3 was chosen due to its porous structure. Fig. 1 represents the D distribution profiles for both graphite grades as a function of depth for the implantation energy of 390 keV. This figure shows the implantation profiles
Fig. 1. D distribution profiles in graphite as a function of depth for an implantation energy of 390 keV: experimental (black triangles) and fitted SIMNRA profile (black dotted line) and SRIM calculated profiles for respective densities of 1.92 g cm3 (green dashed dotted line) and 2.2 g cm3 (red dashed dotted line). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
obtained by SRIM for both sample densities as well as the experimental data and the SIMNRA fits for the SLA2 sample (1.92 density). Using the RESOLNRA code [12] of SIMNRA [13], we calculated the D(3He,p)4He nuclear reaction depth resolution values for the three incident particle energies (900, 1150 and 1600 keV) for both sample densities at the sample surface and at the respective projected ranges Rp. Calculations show that, within 10%, these values are similar for both densities. The following mean values have been obtained: (i) at 900 keV:270 nm (surface) and 190 nm (Rp), (ii) at 1150 keV:370 nm (surface) and 185 nm (Rp) (iii) at 1600 keV:600 nm (surface) and 180 nm (Rp). Considering our experimental conditions (in particular a detector resolution of 18 keV), these depth resolutions are in rather good agreement with those obtained by Mayer et al. in carbon samples [14]. 2.3. Heat treatments Heat treatments of graphite samples in inert atmosphere were performed either in high vacuum (P ffi 107 mbar) or in inert gas flow (Ar) at temperatures ranging from 200 to 1200 °C for different annealing times. Annealing in high vacuum was carried out at IPNL with a Pekly ETF 30-50/15-S tubular furnace using a silica tube. The annealing time ranged up to about 300 h for temperatures ranging from 500 to 1200 °C with a heating ramp of 5 °C min1 and without any programmed cooling ramp. Heat treatments in inert gas were carried out in a pure argon 5.01 flow at about 1 bar (0.2–2 L min1 depending on the annealing setup), either at Ampère laboratory (Institut National des Sciences Appliquées – INSA of Lyon, France) using a Jipelec JetFirst rapid thermal annealing (RTA) furnace for flash treatments (2–3 min) with a heating ramp ranging from 10 to 30 °C s1 and a cooling ramp of a few minutes, or at IPNL using a Thermolyne 21100 tubular furnace for relatively short annealing durations (below 24 h) and temperatures up to 1000 °C with heating and cooling ramps of a few minutes, as well as with a Nabertherm RHTH 120-150/18 tubular furnace for heat treatments at 1200 °C with heating and cooling ramps of 5 °C min1. Duplicate experiments (that will not be shown here) performed either in argon or in vacuum showed that the thermal evolution of the implanted D did not depend on the type of inert atmosphere nor the heating and cooling ramps. 2.4. Graphite structure analysis
309
3. Results and discussion 3.1. Dependence of D behavior on implantation depth: role of surface proximity Let us remind that the main D thermal migration process, already observed in SLA2 nuclear graphite implanted at 70 keV, is the release [3]. It starts around 500 °C and is completed around 1200 °C. Moreover, first results obtained after 24 h annealing on 70 keV (Rp ffi 0.75 lm) and 390 keV (Rp ffi 3.2 lm) implanted SLA2 samples had shown that the release was more effective when the D profile was closer to the surface. Three release steps had been proposed, corresponding to three distinct regimes [4]: an almost instantaneous release regime corresponding to the detrapping of D located near the free surfaces, a transient regime attributed to the thermal detrapping of D from low energy sites and its diffusion along crystallites and coke grain edges, the limiting factor being the diffusion process, and a saturation regime corresponding to the detrapping of interstitial D located at high energy sites inside the crystallites and chemically bound to the graphene layer boundaries. In the present paper, a systematic study has been carried out for three implantation depths and increasing annealing times from 1 to almost 300 h. Results are displayed in Figs. 2 and 3. Fig. 2, presenting the D release as a function of annealing temperature, shows that, for a given annealing time, a deeper implantation depth shifts the release curve towards higher temperatures. In other words, at a given temperature, the release decreases with the implantation depth. However, the 1.7 lm curves are closer to the 3.2 lm ones below 700 °C, whereas for temperatures above 900 °C, they are closer to the 0.75 lm curves. This contrasted behavior is probably related to the progressive detrapping of tightly bound deuterium above 900 °C for which the release is facilitated due to the proximity of surface (1.7 lm) compared to the deeper located one (3.2 lm). Fig. 3a presents the D release, for 96 h annealing, as a function of the implantation depth for various annealing temperatures. The data have been obtained at different incident particle energies: 900 keV for the 0.75 lm implantation depth, 1150 keV for 1.7 lm and 1600 keV for 3.2 lm. This figure shows that at 900 °C, the release becomes nearly linear as a function of the implantation depth reaching an almost constant value at 1200 °C whatever the depth. Fig. 3b, showing the release for various annealing times and at two temperatures (1200 and 600 °C), puts forward that a 600 °C annealing favors the release of D located close to the surface whereas a 1200 °C annealing is more effective for detrapping deeper located D.
Raman microspectrometry experiments were carried out on both nuclear graphite and HOPG samples at ENS (Ecole Normale Supérieure, Paris, France) or CEMHTI (Conditions Extrêmes et Matériaux : Haute Température et Irradiation, Orléans, France), using a Renishaw inVia Reflex spectrometer in order to follow the structural modifications induced by the different treatments. Ar laser source (514.5 nm wavelength – 2.41 eV energy) focused on a 1 lm2 spot through a Leica microscope using 50 objective), while the Raman-scattered light was collected by a CCD camera. Moreover, given that sample heating due to the laser spot might affect the measurements, we used a low incident power (<1 mW) [15]. Since the penetration depth of a Raman laser into graphite is limited to a hundred nanometers [16], the analyses were limited to the samples implanted at the lowest energy of 70 keV in order to probe zones as close as possible to the most disordered ones.
1
Minimum purity of 99.999% guaranteed by the supplier (Linde Gas).
Fig. 2. Fitted release curves and (experimental data for the sample annealed for 48 h) corresponding to the D release as a function of temperature for the SLA2 graphite implanted respectively at 70, 200 and 390 keV and annealed for 24, 48 and 96 h.
310
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311
Fig. 3. D releases as a function of the implantation depth for (a) different temperatures and an annealing duration of 96 h and (b) for annealing temperatures of 600 and 1200 °C and annealing times varying from 1 h to 288 h. The dashed lines allow guiding the eye. The data have been obtained at different incident particle energies: 900 keV for the 0.75 lm implantation depth, 1150 keV for 1.7 lm and 1600 keV for 3.2 lm.
Fig. 4. Fitted release curves corresponding to the D release as a function of temperature for SLA2 and HOPG graphite implanted respectively at 70, 200 and 390 keV and annealed for 24, 48 and 96 h.
Fig. 5. HOPG Raman spectra of virgin, as-implanted at 70 keV and implanted and annealed samples in the temperature range 500–1000 °C during 300 h. The spectra have been normalized with respect to the G band.
3.2. Effect of the graphite structure on the implanted D behavior Let us now compare the results obtained in SLA2 nuclear graphite in the previous section to those obtained in HOPG for which
Fig. 6. SLA2 Raman spectra of virgin, as-implanted at 70 keV and implanted and annealed samples in the temperature range 500–1000 °C during 300 h. The spectra have been normalized with respect to the G band.
the structure is much more homogeneous at a larger scale. The comparison of the D release in both materials as a function of temperature is given in Fig. 4. Whatever the annealing time and implanted depth, the release curves of the 0.65 and 2.8 lm implantation depths in HOPG are shifted towards high temperatures compared to the SLA2 implanted samples. This means that the release in HOPG is always lower than that in SLA2. Indeed, in HOPG, during implantation, D is mainly trapped into interstitial high energy sites, which means that more energy will be necessary to displace it from these sites than for SLA2 graphite, for which a great part of D should be trapped at low energy sites located along crystallites and coke grain edges [4]. At the same time, graphene planes are also fractured by the D+ ion beam, which gives rise to release short-cuts. However, the impact of these short-cuts on the D release remains less important compared to those induced by the high inter and intragranular-porosity of the nuclear graphite. In order to go further in the understanding of these results, and especially to follow the increase of the structural disorder due to the ion implantation, Raman analysis was performed. Raman spectra are displayed in Figs. 5 and 6 respectively for HOPG and SLA2 samples implanted at the lowest energy (see Table 1 which displays the implantation data for SLA2 and HOPG graphite). These figures present the normalized intensity of the symmetry-allowed G band and the defect-induced D1 and D2 bands as a function of the Raman shift. Raman spectra clearly show that the virgin nuclear graphite is already more disordered than HOPG since its spectrum exhibits a D1
M. Le Guillou et al. / Nuclear Instruments and Methods in Physics Research B 371 (2016) 307–311 Table 1 Implantation data (energy, Rp, maximum D concentration and dpa) for SLA2 and HOPG graphite. Implantation energy (keV) Rp in SLA2 (lm) Rp in HOPG (lm) [D] at the Rp (at.%) for both graphites [dpa] at the Rd for both graphites
70 0.7 0.6 3.4 0.3
200 1.7 1.4 2.9 0.3
390 3.2 2.8 2.5 0.2
band which does not exist on the HOPG one. In both types of graphite, implantation results in the appearance of significant defect bands compared to the virgin samples. Temperature annealing favors the structure reordering without reaching the structural states of the virgin samples even at 1000 °C. 4. Conclusion Finally, the D(3He,p)4He nuclear reaction is an appropriate probe to study the D thermal release in graphite up to an implantation depth of a few microns. It allows confirming the crucial role of the graphite structure on this release and drawing interesting conclusions for management of nuclear graphite. Indeed, D is significantly more easily released in highly porous and less graphitized graphite compared to highly graphitized and lamellar model graphite. Raman microspectroscopy results also put forward the important role of temperature on the reordering of the structure in both types of graphite. Thus, by extrapolating our results to the behavior of T during reactor operation, it is likely that (i) at around 500 °C, temperature close to the maximum operating UNGG temperature, a major part of the T located close to free surfaces has been released, (ii) T should be less released in the coke grains for which the graphitization level is higher than in the binder which is less graphitized and more porous. This has been well evidenced by Le Guillou et al. [4] using nuclear microprobe analysis. Considering irradiated graphite waste purification options, the thermal process should therefore be more efficient for the most disordered graphite. Moreover, a fine crushing as well as temperatures higher than 1200–1300 °C should allow getting rid of a major part of T after dozens of annealing hours.
311
Acknowledgments This work was performed in the frame of an Andra Ph.D. grant and financially supported by Andra, EDF and the European project CARBOWASTE (Seventh EURATOM Framework Programme). The authors are grateful to A. Perrat-Mabilon (IPNL) and Y. Le Gall (ICube) as well as to R. Ingren and B. Marchand (University of Helsinki) for D ion implantations. They also would like to thank F. Laariedh (INSA) for RTA treatments and the accelerator staff of IPNL for its helpful support during ion beam experiments. References [1] M.-C. Perrin, B. Poncet, M.-T. Pascal, Livre blanc du Tritium, Le Tritium dans les déchets des réacteurs d’EDF en cours de démantèlement, ASN, 2010, p. 120 (Chapitre 2). [2] Andra, Etude des scénarios de gestion à long terme des déchets de faible activité massique à vie longue, Andra technical report, FRPADPG.12.0020/A 2012, p. 37. [3] M. Le Guillou, N. Toulhoat, Y. Pipon, N. Moncoffre, N. Bérerd, A. Perrat-Mabilon, R. Rapegno, Nucl. Instr. Meth. B 332 (2014) 90. [4] M. Le Guillou, N. Toulhoat, Y. Pipon, N. Moncoffre, H. Khodja, J. Nucl. Mater. 461 (2015) 72–77. [5] J.-P. Bonal, J.-C. Robin, Les Réacteurs Nucléaires à Caloporteur gaz, Un Matériau Fascinant: Le Graphite, CEA/DEN, CEA Saclay et Groupe Moniteur (Editions du Moniteur), Paris, 2006, p. 27. [6] C.-E. Vaudey, N. Toulhoat, N. Moncoffre, N. Bérerd, L. Raimbault, P. Sainsot, J.-N. Rouzaud, A. Perrat-Mabilon, J. Nucl. Mater. 395 (2009) 62. [7] J.F. Ziegler, J.P. Biersack, M.D. Ziegler, SRIM, The Stopping and Range of Ions in Matter, SRIM Co, 2008. [8] T. Hayashi, K. Sugyiama, K. Krieger, M. Mayer, V.K. Alimov, T. Tanabe, K. Masaki, N. Miya, J. Nucl. Mater. 363 (2007) 904. [9] I. Takagi, K. Yoshida, K. Shin, K. Higashi, Nucl. Instr. Meth. Phys. Res. B 84 (1994) 393. [10] H. Khodja, C. Brosset, N. Bernier, Nucl. Instr. Meth. Phys. Res. B 266 (2008) 1425. [11] V.K. Alimov, M. Mayer, J. Roth, Nucl. Instr. Meth. Phys. Res. B 234 (2005) 90. [12] M. Mayer, Nucl. Instr. Meth. B 266 (2008) 1852. [13] M. Mayer. Tech. Rep. IPP 9/113, Max-Planck-Institut fur Plasmaphysik, Garching, 1997. [14] M. Mayer, E. Gauthier, K. Sugiyama, U. von Toussaint, Nuclear Instruments and Methods in Physics Research B: Beam Interactions with Materials and Atoms, 267, No. 3, 2009, pp. 506–512. [15] N.J. Everall, J. Lumsdon, D.J. Christopher, Carbon 29 (1991) 133–137. [16] M.R. Ammar, J.N. Rouzaud, C.E. Vaudey, N. Toulhoat, N. Moncoffre, Carbon 48 (2010) 1244–1251.