Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164

Contents lists available at SciVerse ScienceDirect

Nuclear Instruments and Methods in Physics Research B


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

TEM observations of radiation damage in tungsten irradiated by 20 MeV W ions


ski a,, O.V. Ogorodnikova b, T. Pocin ski a, M. Andrzejczuk a, M. Rasin ski a, M. Mayer b, . Ciupin K.J. Kurzydowski a
a b

Warsaw University of Technology, ul. Woloska 141, PL-02507 Warsaw, Poland Max-Planck-Institut fr Plasmaphysik, EURATOM Association, Boltzmannstr. 2, D-85748 Garching, Germany

a r t i c l e

i n f o

a b s t r a c t
Polycrystalline, recrystallized W targets were subjected to implantation with 20 MeV W6+ ions in order to simulate radiation damage caused by fusion neutrons. Three samples with cumulative damage of 0.01, 0.1 and 0.89 dpa were produced. The near-surface zone of each sample has been analyzed by transmission electron microscopy (TEM). To this end, lamellae oriented perpendicularly to the targets implanted surface were milled out using focused ion beam (FIB). A reference lamella from non-irradiated, recrystallized W target was also prepared to estimate the damage introduced during FIB processing. TEM studies revealed a complex microstructure of the damaged zones as well as its evolution with cumulative damage level. The experimentally observed damage depth agrees very well with the one calculated using the Stopping and Range of Ions in Matter (SRIM) software. 2013 Elsevier B.V. All rights reserved.

Article history: Received 24 September 2012 Received in revised form 21 February 2013 Accepted 4 March 2013 Available online 29 March 2013 Keywords: Tungsten Radiation damage TEM observation

1. Introduction Tungsten (W) will be used in the high-ux region of the divertor in ITER and is a candidate material for plasma facing components in future fusion devices [1]. This is mainly due to its favorable physical properties under high heat and particle uxes. Its low sputtering yield can minimize impurity generation. Its good thermal properties, i.e. one of the highest melting point of all elements (3695 K) and high thermal conductivity (175 W m1 K1) [2] are also required for efcient plasma facing components. Since tritium is radioactive, safety requirements limit its in-vessel inventory to the total of 700 g [3]. If this level would be overcome, a clean-up of the vessel is necessary in order to reduce the radioactivity. According to the present knowledge, the hydrogen isotope retention in pure W is not a concern [39]. However, a production of neutron-like defects can result in a signicant increase of the deuterium retention in W as it was simulated by pre-irradiation of W by fast heavy ions [10]. Neutron (n) irradiation of materials leads to a signicant modication of their crystal structure due to the introduction and accumulation of radiation induced structural defects [11]. Sources of fusion neutrons for materials irradiation are still not available. Also, irradiation of materials in a fast nuclear reactor is not very practical due to the long time required for the accumulation of relevant damage levels and due to the different neutron en Corresponding author.
ski). E-mail address: lciupinski@gmail.com (. Ciupin 0168-583X/$ - see front matter 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.nimb.2013.03.022

ergy spectra, as compared to fusion neutrons. Therefore, in the present work, tungsten ions with a kinetic energy of 20 MeV were used to simulate displacement damage created by fast neutrons. This approach of mimicking neutron damage by fast ions has been recently adopted by many authors (e.g. [10,1216]). Despite some differences in primary knock-on atom energy spectra (>MeV from high energy ion-bombardment and <300 keV from 14 MeV neutrons) and a much higher rate of production of defects by ions compared to neutrons, ion irradiation is currently probably the only non-activating method of simulation of the fast neutron damage produced in fusion reactors. Within fusion research the effects of irradiation in tungsten are broadly investigated, as the process alters fuel retention [10,12,14 16] and mechanical properties [13,17]. These are of high importance for future fusion power plants. Equally high attention is paid to the microstructural evolution induced by ion implantation in an effort to understand the microstructure properties relationship and to elucidate the underlying mechanisms [18,19]. In our investigations, we have concentrated on the microstructural investigations of the damage proles observed at the cross-sections of the damaged zone. Although much research on TEM investigation of the radiation damage has been published, the available papers (e.g. [13,1820]) show microstructural defects in plan view with TEM images more-or-less parallel to the implanted material surface, whereas we are observing the defected zone at the cross sections normal to the surface. As the thickness of tungsten TEM lamella transparent to the electron beam is usually less than 100 nm, the plan-view images provide limited information on

160

ski et al. / Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164 . Ciupin

the defects contained in this thin layer at a given depth from the target surface. Moreover, often the exact location of this layer with regard to the target surface is difcult to dene due to specimen preparation procedure limitations. Our cross section images provide information on defects right from the target surface down to about 10 lm and allow evaluation of changes in defects distribution versus distance from the target surface. 2. Materials and methods Polycrystalline European ITER reference tungsten (WI) produced by Plansee AG [21] was used in our study. The WI was cut into 10 10 mm2 plates of the thickness of c.a. 0.5 mm and mechanically polished to a mirror like nish. Prior to ion implantation the plates were recrystallized at 2470 K for 10 min. The radiation damage was introduced by implantation with 20 MeV W6+ ions up to 0.01, 0.1 and 0.89 dpa in the damage maximum. This corresponds to the uences of 1.6 1016 W/m2, 1.6 1017 W/m2 and 1.4 1018 W/m2, respectively. The implantation was carried out at IPP Garching in a chamber connected to the 3 MV tandem accelerator [10]. The specimen holder was actively cooled during implantation. The background pressure in the chamber was better than 105 Pa. The ion current during the experiment was 10 30 nA/cm2. The temperature of the sample was measured by a thermocouple attached directly to the target and was kept around 290 K during the W ion bombardment. The damage prole was calculated using the program SRIM 2008.03 [22], full cascade option, with a threshold displacement energy of Ed = 90 eV. The damage prole in W irradiated by 20 MeV W6+ is shown in Fig. 1. The 20 MeV W ions create an inhomogeneous prole up to about 2.35 lm with a maximum at a depth of about 1.35 lm. The transmission electron microscopy (TEM) observations have been performed on lamellae cut as cross-sections perpendicular to the implanted sample surface. A focused ion beam system (FIB) Hitachi FB2100 with Ga+ ions beam and 40 kV accelerating voltage has been used for milling and extraction of TEM specimens. Extraction of the TEM lamella milled-out from the bulk of the material is realized using a microprobe inside the FIB chamber. This operation is called micro sampling [23]. Before ion milling the implanted surface of the tungsten plates has been covered with a protective tungsten layer using FIB-assisted chemical vapor deposition technique and tungsten hexacarbonyl W(CO)6 as processing gas. During FIB processing a sidewall damage of the milled lamella occurs [24]. In order to reduce this unwanted, additional damage that would ad-up to the damage induced by self-implantation in the TEM micrographs, low energy ion polishing has been employed. A LINDA Gentle Mill device with Ar ion beam operated with

Fig. 2. TEM lamella sample preparation and observations scheme showing the relative orientation of beams and samples.

an accelerating voltage in the range of 0.31 kV has been used for FIB damage attenuation. The same procedure was used for extraction of a lamella from an undamaged polycrystalline tungsten plate. This lamella was used as a reference sample for the evaluation of damage introduced during samples preparation. Finally, the TEM observations have been carried out on a scanning-transmission electron microscope Hitachi STEM HD2700 with an accelerating voltage of 200 kV. A scheme of cross-section micro sampling showing the relative orientation of the beams used during investigations is shown in Fig. 2. 3. Results A typical lamella is presented in Fig. 3. It is of uniform thickness, approximately 70 nm thick and the area transparent to the electrons is about 20 10 lm. Roughly one half of this observable lamella width is presented in Fig. 3. This lamella has been extracted from the sample with the damage level of 0.1 dpa. The top of the image corresponds to the top of the target surface. Several layers with varying contrast and morphology can be observed. These are, listed from the top: (a) two amorphous layers of tungsten deposited as protective coatings before ion milling, (b) a damaged layer due to 20 MeV W ions implantation, (c) an undamaged dee-

10

dpa

10

-1

10

-2

20 MeV W -> W Eth=90 eV


0 1 2 Tungsten depth, m 3
Fig. 3. Typical lamella extracted from ion-implanted tungsten sample damaged to 0.1 dpa with characteristic features: (a) two amorphous protective layers deposited prior FIB milling, (b) a damaged zone due to 20 MeV W ions implantation, (c) an undamaged deeper target region, and (d) material re-deposited during FIB milling.

6+

10

-3

Fig. 1. Damage prole calculated with SRIM for the 0.89 dpa damaged sample and Ed = 90 eV.

ski et al. / Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164 . Ciupin

161

Fig. 4. Overview of all analyzed materials (STEM bright eld images). Looking from the left: (a) undamaged target, (b) 0.01 dpa damage, (c) 0.1 dpa damage and (d) 0.89 dpa damage.

per layer of the target and (d) a bright layer with a columnar interface which is composed of material re-deposited during FIB milling. The top and bottom layers are inherent to the lamella extraction method and as such are not taken into account in further analyses. The lamella extracted from the reference, non-implanted target and the three damaged targets are juxtaposed in Fig. 4 in ascending damage level order (0, 0.01, 0.1 and 0.89 dpa) looking from the left. As the targets were recrystallized before implantation, coarse grain microstructures developed. Therefore, only one or two grains are usually visible in the cross sections as lamellae were smaller than the grains of the polycrystalline targets. Uniform contrast is observed for almost the whole depth of the undamaged sample except of the near surface region. The dotty

Fig. 5. Comparison of defect distributions in damaged zones (TEM bright eld images). Looking from the left: (a) 0.01 dpa, (b) 0.1 dpa and (c) 0.89 dpa.

contrast of the sample is due to the damage induced by the Ga ion beam of the FIB apparatus. This damage could not be removed even by the low-energy ion polishing employed after FIB milling, thus this unwanted feature is contributing to the damage images of all investigated specimens. Already at 0.01 dpa an additional damaged structure due to 20 MeV W6+ ions implantation is observable. This damaged structure extends from the surface down to roughly 2 lm in depth. Its morphology resembles that of FIB induced damage, however it can be noticed that it has seemingly a higher density of defects than the deeper, unaltered layer or the reference sample. At 0.1 dpa, the damaged structure is also easily observable and some differentiation of this region can be argued. Somewhere, in the middle of the altered region a coarsening of the defects is apparent. The deepest part of that damaged layer still preserves the dotty morphology observed earlier. Finally, the sample with the highest damage level (0.89 dpa) contains the most complex microstructure of all investigated ones. Further coarsening of defects is easily noticeable, especially in the middle of the damaged zone. The three damaged samples with higher magnication compared to Fig. 4 are shown in Fig. 5. The images are presenting the material volumes located at the depth range of approximately 0.5 1.5 lm from the surface. The coalescence and redistribution of defects are observed with increasing damage level. At 0.01 dpa the image is dominated by defects, less than 20 nm in size, uniformly distributed in the observed volume. At 0.1 dpa these defects begin to form chains and a web-like structure of defects emerges. At 0.89 dpa this web-like structure is already clearly visible. Moreover, we would argue that the material volumes inside that mesh have a much lower defect density, as we believe that the small defects visible inside those regions are due to FIB processing. The microstructure of the whole cross-section of the damaged zone of 0.89 dpa sample is shown in Fig. 6. The changes in the contrast of the defects with the distance from the sample surface are evident. The near surface area is characterized by a high density of tangled dislocations. Deeper, the tangled dislocations network grows, leaving larger areas with low defect density. Finally, at a distance of c.a. 2 lm from the surface, the damaged zone is composed of small dislocation loops densely distributed. Below this nal sub-layer small dislocation loops with lower areal density are observed that we attribute to FIB milling. The depth of implantation-induced damage in the sample with the highest level of dam-

162

ski et al. / Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164 . Ciupin

Fig. 8. STEM bright eld image of dislocation loops created by FIB milling.

Fig. 6. STEM bright eld damaged zone image of the 0.89 dpa damaged sample.

only noticeable difference is the size of the dislocation loops created, that of FIB origin being smaller.

4. Discussion Although the implantation depths calculated with SRIM and directly measured from TEM image are in a good agreement, the damage intensities from calculations do not match the images. SRIM predicts the damage density to peak at around 1.3 lm, whereas judging from the TEM image, obtained at 0.89 dpa, the areal density of defects is the lowest at this depth. This can be explained in terms of recombination and rearrangement of defects, which is observed with the increasing dose as mentioned earlier. There are two major driving forces inuencing the mobility of the structural defects, i.e., stress eld and temperature. As every dislocation loop is associated with a characteristic strain and stress eld, an increase of defect density will lead to the increase of the overall stress eld in the damaged material volume. That would explain the observed disagreement between calculated damage proles and observed in STEM as the defects in the highly damaged zone would have a higher tendency for recombination. However, the rate of defects migration in tungsten at room temperature, as discussed in [28], is extremely low. Therefore, for the defects to rearrange within the time-scale of our experiment, an increase of temperature seems indispensable. There are two potential sources of heat in our sample preparation procedure. The rst one is the W6+ ion irradiation. Although the whole implanted target is kept at room temperature, the local rise of temperature in a small volume directly under the ion beam cannot be excluded. The second one is FIB milling, which is basically also ion implantation. Park et al. [29] have shown that during a typical ion milling operation, the sample temperature may rise by 350 for materials with low thermal conductivity such as glass. They have also measured the temperature rise in a stainless steel sample to be about 150. Taking into account that tungsten has an approximately four times higher value of thermal conductivity than stainless steel, in our experiment the temperature rise due to ion-milling should not exceed several dozen degrees. According to [30], the temperatures between 370 and 720 K correspond to stage III, which is attributed to the migration of self-interstitial atoms and a temperature of 720 K, are the end of

Fig. 7. STEM bright eld image of dislocation loops in 0.01 dpa damaged sample.

age directly measured in TEM is equal to 2.3 lm (Fig. 6). This is in a very good agreement with the damage prole calculated with SRIM (Fig. 1) for the implantation of tungsten with 20 MeV W ions using the displacement energy of 90 eV [10]. The defects that are visible in the low damage level sample and form the dotty pattern mentioned above are dislocation loops. These are shown in Fig. 7 at high magnication for the 0.01 dpa specimen. Their size is c.a. 5 nm and it can be noticed that already at that damage level some agglomeration of the defects occurs. Formation of the dislocation loops of similar contrast in bright eld TEM images and characteristic size around 5 nm, has been observed by many authors [13,25,26] irrespective of the ions used for implantation. Also in our investigations the images of defects due to W ion implantation and FIB milling (that can be regarded as implantation of Ga ions) are very similar (compare Fig. 8). The

ski et al. / Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164 . Ciupin

163

stage III where vacancies became mobile. In [31] it was reported that from 570 to 770 K monovacancies migrate through the crystal lattice and either agglomerate with other vacancy-like defects to form larger defects or annihilate at defect sinks such as grain boundaries. Final recovery takes place only at temperatures higher than 1273 K. Consequently, several dozen degrees should not signicantly modify the ion-induced defects. Therefore, FIB preparation of TEM lamella cannot be responsible for the observed dislocation loop growth and coalescence. Thus, the rearrangement of defects should be attributed solely to the W ion implantation process. Kaoumi et al. [32] have studied the grain growth in nano-structured metals and alloys under irradiation. They have shown that the grain boundaries become mobile and resulting grain growth always occurs in the studied materials irrespective of the irradiation temperature. In this extensive study they have identied three temperature regimes: (1) a purely thermal regime where thermal effects dominate the grain boundary motion process, (2) a thermally assisted regime where thermal and irradiation effects combine to increase the rate of grain boundary mobility caused by either of these mechanisms and (3) a lowtemperature or nonthermal regime in which irradiation effects dominate the grain boundary mobility, and the kinetics do not depend on the irradiation temperature. Kaoumi et al. have also stated, that the transition temperature between the nonthermal and thermally assisted regimes is material dependent but can be related to their melting temperature (Tm) and occurs at a homologous temperature between 0.15 and 0.20 Tm. Further, they have developed a model for grain boundary migration in the nonthermal regime. According to this model the migration occurs by atomic jumps within the thermal spikes, due to incident high energy ions, hitting the grain boundary. The kinetics of the process is also biased by another driving force i.e. the local grain-boundary curvature. The migration of dislocation loops that we observe, leading to their growth and coalescence, may be triggered by the same mechanism since our samples have been irradiated at room temperature. This is certainly below 0.15 Tm (i.e., 550 K) for tungsten. The rearrangement of defects can be attributed to the strong local overheating of the region of dense collisions during W ions implantation. The kinetic energy of the atoms in the region of dense collisions can be recalculated with temperature using the basic equation E = 3/2NkBT. Therefore, the temperature is initially of the order of 10,000 K for MeV ions and this region is called a thermal spike [33,34]. The thermal spikes cool down to the ambient temperature in 1100 ps, so the temperature here does not correspond to thermodynamic equilibrium temperature. A curious feature of collision cascades is that the nal amount of damage produced after thermal spike may be much less than the number of atoms initially displaced in the thermal spike [35]. Typically, a heat spike is characterized by the formation of a transient underdense region in the center of the cascade, and an overdense region around it [35,36]. After the cascade, the overdense region becomes a region of interstitial defects, and the underdense region typically becomes a region of vacancies. It is also worth mentioning that in fusion devices tungsten elements will work at elevated temperatures above the 0.150.2 Tm limit, so the thermal regimes would prevail. Therefore, our future research will concern the samples irradiated at elevated temperatures. 5. Summary TEM investigations of radiation damage in self-irradiated tungsten have been performed. Samples with cumulative damage levels of 0.01, 0.1 and 0.89 dpa have been studied. The damaged zones have been analyzed in cross sections perpendicular to the implanted surface. This approach allowed the examination of changes

in the defect structure not only in relation to the radiation dose but also as function of the distance from the sample surface. It has been observed that at low damage level (0.01) the defected structure is composed of uniformly distributed dislocation loops of some 5 nm size. This structure further develops both with the radiation dose and distance from the implanted surface. At moderate damage (0.1 dpa) coarsening and entangling of defects occurs. This effect is more pronounced in the middle of the observed damaged zone depth. The defect structure in the sample damaged up to 0.89 dpa can be subdivided into three segments: (1) the near surface area down to about 0.4 lm characterized by a high density of tangled dislocations, (2) the intermediate area from about 0.4 lm down to about 1.9 lm with the tangled dislocations network growing/becoming coarser and leaving larger areas with low defect density and (3) the deep area at the depth of c.a. 2 lm, which is composed of small dislocation loops with a high density and a uniform distribution. The driving force of this observed defects rearrangement is related to the 20 MeV W6+ ion implantation and could be described by atomic jumps within thermal spikes. Acknowledgements The technical assistance of J. Dorner and M. Fueder during tungsten irradiation is gratefully acknowledged. This work, supported by the European Community, was carried out within the framework of the European Fusion Development Agreement. The views and opinions expressed herein do not necessarily reect those of the European Commission. The support by the Polish Ministry of Science and Education through Grant No. 2077/7.PR-EURATOM/2011/2 is also acknowledged. This work was partly supported by the Impuls- and Vernetzungs fund of the Helmholtz Society. References
[1] V. Philipps, J. Nucl. Mater. 415 (1) (2011) S2. [2] E. Lassner, W.-D. Schubert, Tungsten: Properties, Chemistry, Technology of the Element, Alloys, and Chemical Compounds, Kluwer Academic/Plenum Publishers, New York, 1999. p. 30. [3] J. Roth et al., J. Nucl. Mater. 390391 (2009) 1. [4] O.V. Ogorodnikova, J. Roth, M. Mayer, J. Nucl. Mater. 373 (2008) 254. [5] O.V. Ogorodnikova, T. Schwarz-Selinger, K. Sugiyama, T. Drbeck, W. Jacob, Phys. Scr. T 138 (2009) 014053. [6] Z. Tian, J.W. Davis, A.A. Haasz, J. Nucl. Mater. 399 (2010) 101. [7] V.Kh. Alimov, W.M. Shu, J. Roth, K. Sugiyama, S. Lindig, M. Balden, K. Isobe, T. Yamanishi, Phys. Scr. T 138 (2009) 014048. [8] R.A. Causey, C.L. Kunz, D.F. Cowgill, J. Nucl. Mater. 337339 (2005) 600. [9] M. Poon, A.A. Haasz, J.W. Davis, J. Nucl. Mater. 374 (2008) 390. [10] O.V. Ogorodnikova, B. Tyburska, V.Kh. Alimov, K. Ertl, J. Nucl. Mater. 415 (2011) S661. [11] P. Vladimirov, S. Bouffard, in: C.R. Physique 9 (2008) 303. [12] M.H.J. t Hoen, B. Tyburska-Pschel, K. Ertl, M. Mayer, J. Raap, A.W. Kleyn, P.A. Zeijlmans van Emmichoven, in: Nucl. Fusion 52 (2012) 023008. [13] D.E.J. Armstrong, X. Yi, E.A. Marquis, S.G. Roberts, J. Nucl. Mater. 432 (2013) 428. [14] M. Fukumoto, H. Kashiwagi, Y. Ohtsuka, Y. Ueda, M. Taniguchi, T. Inoue, K. Sakamoto, J. Yagyu, T. Arai, I. Takagi, T. Kawamura, J. Nucl. Mater. 390391 (2009) 572. [15] W.R. Wampler, R. Doerner, Nucl. Fusion 49 (2009) 115023. [16] Y. Oya, M. Shimada, M. Kobayashi, T. Oda, M. Hara, H. Watanabe, Y. Hatano, P. Calderoni, K. Okuno, Phys. Scr. T 145 (2011) 014050. [17] J.M. Steichen, J. Nucl. Mater. 60 (1) (1976) 13. [18] R. Sakamoto, T. Muroga, N. Yoshida, J. Nucl. Mater. 220222 (1995) 819. [19] H. Iwakiri, K. Yasunaga, K. Morishita, N. Yoshida, J. Nucl. Mater. 283287 (2000) 1134. [20] T. Matsuia, S. Mutob, T. Tanabe, J. Nucl. Mater. 283287 (2000) 1139. [21] ITER material assessment report 2001 G 74 MA 10 0107-11 W 0.2. [22] <http://srim.org>, SRIM. [23] <http://www.hht-eu.com/hht-eu/nte/FB-2100Brochure.pdf>. [24] J.P. McCaffrey, M.W. Phaneuf, L.D. Madsen, Ultramicroscopy 87 (2001) 97. [25] W. Jger, M. Wilkens, Phys. Status Solidi A 1 (1975) 89. [26] Y. Watanabe, H. Iwakiri, N. Yoshida, K. Morishita, A. Kohyama, Nucl. Instrum. Methods Phys. Res. B 255 (2007) 32.

164

ski et al. / Nuclear Instruments and Methods in Physics Research B 317 (2013) 159164 . Ciupin [34] C. Trautmann, S. Klaumnzer, H. Trinkaus, Phys. Rev. Lett. 85 (17) (2000) 3648. [35] R.S. Averback, T. Diaz de la Rubia, Displacement damage in irradiated metals and semiconductors, in: H. Ehrenfest, F. Spaepen (Eds.), Solid State Physics, vol. 51, Academic Press, 1998, p. 281. [36] F. Seitz, J.S. Koehler, Displacement of atoms during irradiation, in: F. Seitz, D. Turnbull (Eds.), Solid State Physics, vol. 2, Academic Press, 1956, p. 307.

[28] M.R. Gilbert, S.L. Dudarev, P.M. Derlet, D.G. Pettifor, J. Phys. Condens. Matter 20 (2008) 345214. [29] Y.M. Park, D.-S. Ko, K.-W. Yi, I. Petrov, Y.-W. Kim, Ultramicroscopy 107 (2007) 663. [30] L.K. Keys, J.P. Smith, J. Moteff, Phys. Rev. 176 (1968) 851. [31] A. Debelle, M.F. Barthe, T. Sauvage, J. Nucl. Mater. 376 (2008) 216221. [32] D. Kaoumi, A.T. Motta, R.C. Birtcher, J. Appl. Phys. 104 (2008) 73525. [33] A. Meftah et al., Phys. Rev. B 49 (18) (1994) 12457.

You might also like