- Email: [email protected]
Effect of tungsten crystallographic orientation on He-ion-induced surface morphology changes
Available online at www.sciencedirect.com
ScienceDirect Acta Materialia 62 (2014) 173–181 www.elsevier.com/locate/actamat
Effect of tungsten crystallographic orientation on He-ion-induced surface morphology changes q C.M. Parish a, H. Hijazi b, H.M. Meyer a, F.W. Meyer b,⇑ a
Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA b Physics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA Received 3 September 2013; received in revised form 17 September 2013; accepted 22 September 2013 Available online 18 October 2013
Abstract In order to study the early stages of nanofuzz growth in fusion-plasma-facing tungsten, mirror-polished high-purity tungsten was exposed to 80 eV helium at 1130 °C to a fluence of 4 1024 He m2. The previously smooth surface shows morphology changes, and grains form one of four qualitatively different morphologies: smooth, wavy, pyramidal or terraced/wide waves. Combining high-resolution scanning electron microscopy (SEM) observations to determine the morphology of each grain with quantitative measurement of the grain’s orientation via electron backscatter diffraction in SEM shows that the normal-direction crystallographic orientation of the underlying grain controls the growth morphology. Specifically, near-h0 0 1i || normal direction (ND) grains formed pyramids, near-h1 1 4i to h1 1 2i || ND grains formed wavy and stepped structures and near-h1 0 3i || ND grains remained smooth. Comparisons to control specimens indicate no changes to underlying bulk crystallographic texture, and possible explanations of the structure growth, particularly loop-punching, are discussed. Future developments to control tungsten texture via thermomechanical processing, ideally obtaining a sharp near-h1 0 3i || ND processing texture, may delay the formation of nanofuzz. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Tungsten; Electron backscatter diffraction; Ion irradiation; Helium; Dislocation loop punching
1. Introduction Because of its high melting point, excellent erosion resistance and material strength, tungsten is favored for wall materials for next-generation magnetic fusion devices. In deuterium-tritium (D–T) fusion operation, significant He ash will be formed at 3.5 MeV, and its interaction with tungsten wall materials is currently undergoing extensive evaluation. q
Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the US Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. ⇑ Corresponding author. E-mail address: [email protected] (F.W. Meyer).
Under certain conditions of wall temperature and He fluence, He ions were found to cause significant morphology changes in the W surface [1–5], even for impact energies below the physical sputtering threshold of 200–300 eV. While still not completely understood, such surface changes are thought to result from near-surface He trapping at intrinsic or extrinsic defect sites. Due to favorable energetics, the trapped He atoms form clusters, which result in progressively higher lattice distortions, which relax by dislocation loop punching, producing He-filled cavities (bubbles) of increasing size. These He bubbles can eventually burst at high sample temperatures where the tungsten has lower effective viscosity or yield strength, causing surface pinholes, and eventually evolving into an increasingly random nanostructuring of the surface that can ultimately lead to the production of nanofuzz [6–9] Orientation dependence on nanostructuring has been reported [10]. There is concern that such surface nanostructuring may adversely affect the
1359-6454/$36.00 Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actamat.2013.09.045
174
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
robustness of the W surface in the presence of the highpower fusion plasma due to increased W dust formation, as well as affect H and He retention. In addition, the effect of the bulk material damage due to 14.1 MeV neutrons also produced in the D–T fusion reactions on these near surface He–W interactions has yet to be definitely assessed. In this experiment, we exposed highly polished tungsten coupons to 80 eV helium ions of flux 1020 m2 s1 to a fluence of 4 1024 He m2 at 1130 °C (1400 K) to study the early stages of nanofuzz formation. Scanning electron microscopy–electron backscatter diffraction (SEM–EBSD) was used to correlate the changes in the surface morphology of the grains caused by the helium exposure to the underlying crystallographic orientation, revealing a strong and systematic effect.
4.0–4.5 nA, with a 19–20 mm working distance. The tungsten coupons were attached to the EBSD specimen holder with silver paint, rather than conductive tape, to minimize sample drift and to allow close correlation between SE images (acquired immediately before EBSD) and the EBSD data. Grain calculations are based upon the definition that a >15° pixel-to-pixel misorientation defines a grain boundary. Texture pole figures were calculated with 5.0° half-width, and order L = 34 harmonic series expansion, triclinic sample symmetry and without using any data cleaning methods [14] of the raw data unless explicitly noted below.
2. Experimental details
3.1. Irradiated areas
2.1. Ion irradiation
The irradiated specimen (Fig. 1a) showed significant surface morphology changes from the pre-irradiated, mirror-polished condition. Four broad morphology categories are identified, and in Fig. 1b these are marked (i)–(iv), and shown in more detail in Fig. 1c and d. In Fig. 1b, (i) denotes a smooth or nearly-smooth region, (ii) denotes a wavy morphology, (iii) denotes pyramidal morphology and (iv) denotes terraced morphology. Obviously, these are qualitative assessments and particular grains may show an intermediate or ambiguous morphology. Judgment was used to differentiate the types of morphology, and as will be seen below, this provided results that correlate well to underlying crystallography. A large-area texture scan of the irradiated area is shown in Fig. 2. Fig. 2a shows a normal direction (ND)-projected inverse pole figure (IPF) map, and Fig. 2b the calculated pole figures (PFs) and an ND IPF. PFs indicate the relative number of grains with a given orientation hh k li oriented in a particular sample direction, measured relative to a random polycrystal. A strong (8 random) h0 0 1i || ND grain texture is seen, also with h0 0 1i in-plane. Relatively few h1 0 1i || ND or h1 1 1i || ND-type grains are seen. The average effective grain diameter is 11 ± 7 lm, indicating a wide distribution. The sharp h0 0 1i texture is consistent with reports of rolling and recrystallization in tungsten [15–17]. It is important to note that EBSD data require significant tilting of the specimen away from the surface normal, 70° away from the normal electron-beam incidence angle in this case. Fig. 3 shows a small area imaged using the EBSD software (which compensates for dynamic focus and the 3:1 difference in pixel size in the Y and X scan directions) imaged at 70° tilt, and the same area imaged with the SEM software at 0° tilt. First, note that the images match well; second, note that the features’ third-dimension information, such as the relative heights and angles of the pyramids or waves, are greatly exaggerated in the EBSD-tilt-acquired image. Grains (i) and (ii) are of a wave morphology when imaged at 0°, but appear to be sharp pyramids at 70° tilt.
High-purity tungsten sheet 0.7 mm in thickness was fabricated by electrical discharge machining (EDM) into 13 13 mm coupons and polished to colloidal silica (0.05 lm) for a mirror polish. Tungsten substrates were examined via X-ray photoelectron spectroscopy (XPS) ascut and after cleaning with detergent and water rinse. Aside from W, carbon and oxygen were the primary surface species observed. Minimal argon-ion sputtering reduced the C from 30 at.% to 5 at.% and O from 20 at.% to 4 at.%, showing that these elements were adsorbed atmospheric contaminants. Several of the He-exposed substrates were examined and showed no additional surface contamination from the He processing. The irradiations were performed at the ORNL Multicharged Ion Research Facility (MIRF) [11] using a new high-current beam deceleration module installed in the beamline of our CAPRICE electron cyclotron resonance (ECR) ion source, which has been described elsewhere [12]. This deceleration module provides high-flux He ion beams at energies down to 50 eV to W targets floating at the deceleration potential. The W samples can be raised to temperatures exceeding 1300 °C by electron beam heating. A recently added beam profile measurement device consisting of a positively biased beam catcher inside an enclosure with a 1 mm2 aperture can be scanned across the incident He ion beam via a PC controlled, stepper motor driven, x–y–z manipulator to determine the twodimensional flux distribution of the incident beam [13]. 2.2. Electron microscopy Specimens were examined in a JEOL JSM-6500F fieldemission analytical SEM. High resolution secondary electron (SE) images were acquired at 5 keV with small beam current (<100 pA) and short working distance (4–7 mm). EBSD was performed using an EDAX-TSL Hikari high-speed camera. Beam conditions were 30 keV,
3. Results
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
175
Fig. 1. (a) SEM images of the helium-irradiated area. In (b), four morphologies are marked (i)–(iv): (i) smooth; (ii) wavy; (iii) pyramidal; (iv) wide waves or steps. (c) Detailed views of smooth, wavy and pyramid morphologies. (d) The terrace-like nature of the wide waves.
Fig. 2. (a) Normal direction-projected IPF map. (b) Texture pole figures for {0 0 1}, {1 0 1}, and {1 1 1}, as well as normal-direction (ND) IPF. Note strong h0 0 1i texture. IPF unit-triangle color key is the same for all figures.
Possible artifacts due to non-normal tilt angle such as the latter are important to keep in mind. Determination of morphology should be made from the 0° tilt data, while assignment of crystallographic orientation should be made from the 70° tilt EBSD data. At this point, a data cleaning step is applied: due to surface roughness, there is a narrow spread of orientations within many grains. Each pixel orientation is replaced with the average orientation of the grain, and then a dilation step removes spikes. Fig. 4 shows an SEM image (0° tilt) with identifiable surface features, and Fig. 4b the IPF
map. A white-bounded inset to the bottom-right corner of Fig. 4b shows the raw (uncleaned) data for comparison. Finally, this data was analyzed to determine if the crystallographic direction parallel to the ND correlated to the surface morphology. In Fig. 5a, the EBSD data is presented with grain boundaries superimposed as black lines and coloration of the grains according to the morphology, either smooth (red), wavy/stepped/terraced (green), or pyramidal (blue). A discrete IPF is plotted as Fig. 5b. First, the average orientation of each measured grain is plotted as a black circle with size proportional to the grain size. The
176
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
Fig. 3. A helium-exposed area, imaged at EBSD 70° tilt (left) and at 0° tilt, normal-electron-beam incidence (right). Note features such as (i) and (ii) are misrepresented at 70° tilt.
Fig. 4. (a) 0° tilt SEM image showing surface features. (b) Cleaned IPF EBSD map. IPF color scale as in Fig. 2. Inset at bottom-right shows uncleaned (raw) data.
Fig. 5. (a) EBSD data of Fig. 3 colored by SEM morphology. (b) Inverse pole figure || normal direction; black marks are average individual grain orientations, sized proportionally to grain size. Colored squares are the orientations of the grains colored in (a). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
grains are clustered around h0 0 1i, as expected from the texture calculations (Fig. 2). Second, the colored boxes are plotted for each grain according to the morphologyassigned color in Fig. 5a.
The results indicate that the pyramids cluster around h0 0 1i || ND, the wavy grains in the vicinity of h1 1 3i || ND and the smooth grains near h1 0 2i || ND. Almost no grains are in the h1 01 i–h1 1 1i region.
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
3.2. Thermally exposed, unirradiated specimen Helium unexposed areas subjected to the same high sample temperature were also examined. Fig. 6a shows the SEM image acquired with the EBSD scan, Fig. 6b shows a normal direction IPF scan (with data cleaning as in Fig. 4d) and Fig. 6c shows texture pole figures calculated for this area (calculated from raw, not cleaned, data). The texture is essentially the same as for the helium-exposed area, Fig. 2b. The IPF morphology is also essentially the same as the helium-exposed area, compare Fig. 6b to Fig. 3b. However, the SEM image shows smooth-polished grains with grooving at the boundaries, in contrast to the rough and well-developed morphology of the helium-exposed area. Thus, the texture and general grain structure are not influenced by the helium exposure, but the surface morphology is. 4. Discussion Comparison of the morphologies of the irradiated and thermally treated specimen indicates that the grooving at the grain boundaries is thermally induced, but the morphological changes to the mirror-polished surface are caused by the interaction of the helium beam with the hot tungsten surface. Although growth of surface facets has been previously observed with thermal annealing alone [18], those experiments were at higher temperature for much longer times than these experiments. Three explanations for the observed differences between differently oriented grains can be proposed: (1) ion channeling effects, (2) surface energy effects and (3) dislocation loop-punching effects. First, we consider ion channeling, the phenomenon in which ions incident on or near a low-index zone axis penetrate anomalously deeply into a crystal by traveling down the channels between atomic columns [19,20]. Channeling is dominant with higher-energy beams of heavier elements: for instance, in focused ion beam (FIB) exposures with
177
30 keV Ga+, significant changes in the texture of the surface grains are observed even at much lower fluences (1020 Ga m2) [21]. The current results – using much lower energy, much lighter ions and much higher fluence – are significantly different to the FIB-induced results of Michael [21]. Comparison of the textures of the as-polished starting material with the irradiated material indicates that there is no re-texturing of the near-surface grains induced by the ion exposure. The ion channel theory of Lindhard [22] and Onderdelinden [23] (L–O), as described by Kempshall et al. [24], and as applied to Ga-FIB problems by Michael [21] and Giannuzzi and Michael [25], allows the non-channeled fraction of ions, X uvw 0 , to be calculated as a function of the bombarding ion, bombarding ion energy, the target material and the incident ion direction into the crystal hu v wi. Also calculable is the critical angle for channeling, wuvw 0 , for the incident direction hu v wi, in which ions within an angle of wuvw from the perfect incidence hu v wi will 0 still channel strongly. For 80 eV helium incident onto body-centered cubic (bcc) tungsten, these are zero for any direction other than h1 1 1i. In the h1 1 1i case, the nonchanneled fraction X h111i 85%, so only 15% of the ions o would be channeled. (The h1 1 1i channels most strongly because it is the closest-packed direction in bcc.) Because almost no h1 1 1i grains were observed in either the irradiated or non-irradiated regions, channeling effects for this low-energy helium regime is discounted as a possible effect. Higher-energy bombardments, not studied here, may suffer significant channeling effects. Garrison and Kulcinski [26] used 30 keV He, and found that grains 4–8° away from h0 0 1i suffered the least helium induced morphology change; but grains closer to h0 0 1i than 4°, or near h1 0 1i or h1 1 1i, did suffer significant attack. The L–O model predicts nearly complete channeling down h0 0 1i,h1 0 1i and h1 1 1i for 30 keV (>90% each), but given the assumptions inherent to the L–O model, discrepancies are unsurprising. The Garrison and Kulcinski results are interesting and need further exploration.
Fig. 6. (a) SEM image of the unirradiated region. (b) ND IPF map of the EBSD data; colorscale is the same as Fig. 2a. (c) PFs and ND IPF texture; colorscale varies from 1 to 11.7 random.
178
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
Surface energy is another possible explanation. Different references disagree on absolute values and relative ratios of surface energy for tungsten planes, but in general {1 0 1} is considered the lowest energy plane and {1 1 1} almost as low energy. {0 0 1} and {1 1 2} are also particularly low energy. Surface energies chkl for a given plane and ratios khkl of surface energy khkl = chkl/c101 are shown in Table 1. SEM alone lacks the three-dimensional (3-D) information needed to determine the planes of the different facets that grew during these experiments, but strong inferences can be made from the available information. Fig. 7a shows the SEM image of the scanned area again (acquired at 70° tilt), and with three grains marked (i)–(iii) and with the grains’ unit cells superimposed, as measured via EBSD. Detailed views of these grains (acquired at 0° tilt) are given in Fig. 7b and c, and discrete pole figures calculated for each grain (i)–(iii) are in the bottom half of Fig. 7. The pyramidal-morphology grain (iii) is the simplest, so we consider it first. The grain is nearly h0 0 1i || ND oriented. The {1 0 1} and {1 1 1} PFs show the {1 0 1} planes lying 45° from the ND, with plane normals nearly vertical and horizontal in the reference frame, which is to say, toward the reference direction (RD) and transverse direction (TD) axes. The {1 1 1} planes are almost 60° off normal, and rotated 45° about the ND relative to the {1 0 1} plane normal. The SEM image (Fig. 7c) shows that most of the pyramidal faces have normals whose surface tracks are either horizontal or vertical. Along with the tetrahedral shape of the pyramids and the low value of {1 0 1} surface energy, the surfaces of most of the pyramids are almost certainly {1 0 1} planes. Again, SEM does not provide sufficient 3-D data to be completely sure, but the assignment of the pyramidal faces in grain (iii) as {1 0 1} planes is very likely. Grain (i), a wave morphology, is more difficult to assign plane faces to. The grain is measured to be [3 6 2 2] || ND, or approximately of type h1 2 7i || ND. Along the length of the
Table 1 Tungsten surface energies chkl (J m2) and ratios k of chkl to c101. {h k l}
Wang et al. [27]
Zhang et al. [28]
Kumar and Grenga [29]
c k
4.025 1.343
2.645 1.185
1.028
c k
2.998 1.000
2.232 1.000
1.000
{1 1 1}
c k
3.671 1.225
2.247 1.007
Not observed
{1 1 2}
c
3.461
2.299
Waves/ terraces
k
1.155
1.030
{1 0 2}
c k
3.696 1.233
2.595 1.163
{1 0 3}
c k
3.886 1.296
2.671 1.197
{0 0 1} {1 0 1}
Observed morphology Pyramids Not observed
1.030 Smooth Smooth 1.120
waves, the crystal direction is measured by EBSD to be approximately [6 8 3]. The waves are not straight, so multiple values of the in-plane length of the waves, or values in-plane perpendicular to the waves, are possible. From the PF (i), however, a h1 0 2i-type direction is approximately perpendicular to the lengths of the waves, and is marked in Fig. 7b. Neither {1 0 1} nor {1 1 1} planes line up precisely, but both are very close to the axis perpendicular to the waves and are possible facets. Given the fact that the waves are not straight and bifurcate, it is conceivable (but not proven) that both faces are present in different regions. Similarly, the wide-wave or step morphology of grain (ii) is complicated to assign faces to, because of bifurcations and shifts in the planes. However, the long steplengths may be of {0 0 1} type, and the shorter steps of {1 1 2} type, as marked on the pole figures. This is simply one possibility, however, and not definitive. A FIB liftout was performed to help confirm the above plane analyses. A “wide-wave” morphology grain was found (Fig. 8a) by imaging with the SEM column in a FIB/SEM instrument. A liftout was attached to a copper TEM-style sample carrier and polished to 5 keV Ga+, Fig. 8b (again imaged in FIB/SEM). The specimen was then transferred to the SEM–EBSD system and imaged and EBSD mapped. Five grains are present on the liftout’s face, and their unit cell orientations are marked (Fig. 8c). The wide-wave morphology grain is marked in red1, and that single red grain’s PFs for {0 0 1}, {1 0 1}, {1 1 1}, {0 1 2}, {1 1 2} and {1 0 3} are presented in Fig. 8d. The h0 0 1i direction in the center of the PF indicates that the length of the wide waves is h0 0 1i.h1 0 1i is very nearly parallel to the original surface normal. Because the liftout was cut perpendicular to the length of the waves, only planes with normals falling on the equator (outmost grid circle) of the PFs can be the surface faces of the waves. This eliminates {1 0 1}, {1 1 1} and {1 1 2} because none of those planes falls on the equator near the directions of the waves’ surface normal. Tilting the specimen’s polished face flat to the electron beam allows the angle of the facets relative to the original surface to be measured; the facets are 15–25° above the original surface (not shown; Fig. 8b is not normal to the electron beam due to FIB–SEM instrument hardware geometry). The grid lines in Fig. 8d are in 15° increments, so of the equator-lying planes, i.e., {0 0 1}, {1 0 1}, {1 0 2} and {1 0 3}, it can be seen that {1 0 2} matches most closely to these measured angles, although {1 0 3} is also within the margin of error. The lowest-energy surfaces, which would be expected from a pure surface-energy effect, are not necessarily the observed surfaces. It appears that both high-index arbitrary surface faces and low-index {0 0 1} and {1 0 1} faces are faceting with resultant changes in surface area and
1 For interpretation of color in Fig. 8, the reader is referred to the web version of this article.
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
179
Fig. 7. (a) SEM image with grain boundaries (black) and three grain unit cells superimposed, labelled (i)–(iii). (b and c) detailed views of the grains. The bottom row shows PFs for the grains (i)–(iii).
Fig. 8. (a) In-FIB/SEM image of a wide-wave grain. (b) Liftout imaged after FIB polishing. (c) EBSD mapped region, with five grains’ unit cells superimposed. (d) PFs of the wide-wave grain.
180
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
energy. {1 0 1} is considered the lowest-energy plane in tungsten, yet in the example of Fig. 8, for instance, a nearly-{1 0 1} pre-irradiated surface appears to facet into either {1 0 2} or {1 0 3} facets. Given that a faceted surface has a larger surface area (and thus larger integrated surface energy even at constant surface energy per unit area) than a smooth polished surface, and that the smooth-remaining grains (as noted in Fig. 5) are around the {1 0 4}–{1 0 3}– {1 0 2} region of the unit triangle, a simple explanation is hard to find. The third possibility, as suggested by Ohno et al. [10], is that dislocation loop punching caused by the growth of helium bubbles is driving the surface faceting. As more helium atoms accumulate in a cavity, the pressure increases above the equilibrium pressure. Donnelly [30] summarizes the conditions for different mechanisms to decrease the pressure, and for larger bubbles the dominant mechanism is the punching of a dislocation loop from the bubble to increase the size of the bubble. A number of references describe loop punching in detail [31–37]. Very briefly, as the helium content of a bubble increases it eventually becomes energetically more favorable for the pressure to decrease via bubble growth rather than for the bubble size to stay constant at increasing pressure. An interstitial dislocation loop – essentially a platelet of interstitial atoms – will be emitted from the bubble, which will then be able to glide under stress. Early transmission electron microscopy (TEM) work, for instance, showed arrays of dislocation loops pushed out from individual helium filled cavities [36,37] in molybdenum. Recent work by Ohno et al. [10] on 25–30 eV helium irradiated tungsten surfaces showed results similar to ours, and the growth of non-lowestenergy facets was attributed to the action of bubbles punching glissile dislocation loops on {1 0 1} slip planes (although {2 1 1}h 1 1 1i slip in bcc tungsten is suggested by Weertman and Weertman [38]). Similar to our work, Ohno et al. observed relatively little surface morphology change in the area around h1 0 3i specimen normal. However, one difference is that they observed wavy structures forming at h0 0 1i whereas we observed pyramidal growth. The FIB liftout of Fig. 8 was thinned further with lowenergy Ga+ and then polished with 900 eV Ar+ and imaged in an FEI TitanS 300 keV TEM/STEM. Fig. 9a shows a BF Fresnel-contrast TEM image, and bubbles are clearly visible to a depth of 200 nm. Fig. 9b shows a lattice image of the region arrowed in Fig. 9a, and the surface plane is very near (2 1 0), confirming the EBSD results of Fig. 8. Unfortunately, too much contrast arising from specimen damage (both bubble-related and Ga+/Ar+ related) is present to directly image dislocation loops. It is possible that future experiments, likely taking advantage of the aberration corrector on the Titan’s probe-forming optics, might allow direct observation of dislocation loops. Very recent work [39] using molecular dynamics (MD) has indicated that loops punched out of growing nearsurface helium bubbles in tungsten will be drawn to the surface by image forces and become adatom islands, which
Fig. 9. (a) Low-magnification TEM image of the surface of the grain analyzed in Fig. 8. Arrow marks the region of (b), a lattice image showing near-{1 0 2} faceting.
probably act as incipient nanofuzz. While those apparent morphologies appear to be different from our observations, for reasons not yet understood in detail, this MD work increases our confidence that loop-punching may be the root cause of our surface morphology changes. In short, no single explanation, neither surface energy nor loop punching, is precisely consistent with the observed surface features, indicating that a more thorough understanding must be sought. 5. Conclusions Sub-sputtering-threshold 80 eV helium exposure to 4 1024 He m2 at 1130 °C resulted in roughening and surface morphology changes to smooth-polished tungsten. Grain boundary grooving occurred, and four general grain morphologies were observed: smooth, pyramidal, wavy and terraced. Unexposed areas held at the same temperature showed grain boundary grooving but not intragranular surface roughening, indicating that the effect arose from helium exposure. Texture pole figures for unexposed and exposed areas were the same, indicating that new
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
grains are not forming and grain rotation did not occur. The morphology changes correlated with the underlying surface normal direction of the grain: pyramids formed around h0 0 1i || ND, grains remained smooth around h1 0 2i–h1 0 3i–h1 0 4i || ND, and waves and steps formed in the region h1 1 4i–h1 1 2i–h2 1 3i || ND. No correlations to ion channeling directions were observed, and the developing surface facets were not low-energy planes in tungsten. The facet faces formed appear to be of several types, including {1 0 1} and {1 0 2}. The growth of these faces is explained by neither surface energy nor loop punching arguments alone. We suggest that thermomechanical processing to enhance h1 0 2i–h1 0 3i || ND grain processing textures, while difficult, may be effective in delaying nanofuzz nucleation under bombardment by DT fusion helium ash of hot first wall W surfaces of magnetic fusion devices. Acknowledgements This research was sponsored by the LDRD Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US Department of Energy. HH was appointed through the ORNL Postdoctoral Research Associates Program administered jointly by Oak Ridge Institute of Science and Education (ORISE), Oak Ridge Associated Universities (ORAU) and Oak Ridge National Laboratory (ORNL). Work performed in part via Oak Ridge National Laboratory’s Shared Research Equipment (ShaRE) User Program, which is sponsored by the Office of Basic Energy Sciences, US Department of Energy. Thanks to Dr. Lucille Giannuzzi, L.A. Giannuzzi & Associates LLC, for discussions regarding ion channeling. Thanks to Dr. Maxim Gussev, ORNL, for critiquing the manuscript. References [1] Iwakiri H, Yasunaga K, Morishita K, Yoshida N. J Nucl Mater 2000;283:1134. [2] Minyou Y. Plasma Sci Tecnhol 2005;7:2828. [3] Nishijima D, Ye MY, Ohno N, Takamura S. J Nucl Mater 2003;313:97. [4] Baldwin MJ, Lynch TC, Doerner RP, Yu JH. J Nucl Mater 2011;415:S104. [5] Yamagiwa M, Kajita S, Ohno N, Takagi M, Yoshida N, Yoshihara R, et al. J Nucl Mater 2011;417:499.
181
[6] Baldwin MJ, Doerner RP. Nucl Fusion 2008;48:035001. [7] Baldwin MJ, Doerner RP. J Nucl Mater 2010;404:165. [8] Takamura S, Ohno N, Nishijima D, Kajita S. Plasma and fusion research, vol. 1; 2006. #051. [9] Meyer FW, Hijazi H, Krstic PS, Dadras J, Meyer III HM, Parish CM, Bannister ME. Phys Scripta; in press. [10] Ohno N, Hirahata Y, Yamagiwa M, Kajita S, Takagi M, Yoshida N, et al. J Nucl Mater 2013;438(Suppl.):S879. [11] Meyer FW, Bannister ME, Hale JW, Havener CC, Krause HF, Vane CR, et al. Recent activities at the ORNL multicharged ion research facility (MIRF). In: Proceedings of ECRIS2010. Grenoble, France; 2012. [12] Hijazi H, Meyer FW. Rev Sci Instrum 2013;84:033305. [13] Bannister ME, Hijazi H, Meyer III HM, Cianciolo V, Meyer FW. Submitted for publication. [14] Brewer LN, Michael JR. Microsc Today 2010;18:10. [15] Neges J, Ortner B, Leichtfried G, Stuwe HP. Mater Sci Eng A – Struct Mater Prop Microstruct Process 1995;196:129. [16] Dhers J. In: Chandra T, Torralba JM, Sakai T, editors. Thermec’2003. Zurich-Uetikon, Pts 1–5, vol. 426-4. Trans Tech Publications; 2003. p. 4455. [17] Brewer LN, Bewlay BP. Microsc Microanal 2002;8:696CD. [18] Zhang W, Sachenko P, Gladwell I. Acta Mater 2004;52:107. [19] Brandt W. Sci Am 1968;218:90. [20] Feldman LC, Mayer JW, Picraux ST. Materials analysis by ion channeling. New York: Academic Press; 1982. [21] Michael JR. Microsc Microanal 2011;17:386. [22] Lindhard J. Phys Lett 1964;12:126. [23] Onderdelinden D. Appl Phys Lett 1966;8:189. [24] Kempshall BW, Schwarz SM, Prenitzer BI, Giannuzzi LA, Irwin RB, Stevie FA. J Vac Sci Technol B 2001;19:749. [25] Giannuzzi LA, Michael JR. Microsc Microanal 2013;19:344. [26] Garrison LM, Kulcinski GL. Phys Scripta; in press. [27] Wang SG, Tian EK, Lung CW. J Phys Chem Solids 2000;61:1295. [28] Zhang JM, Ma F, Xu KW. Surf Interface Anal 2003;35:662. [29] Kumar R, Grenga HE. Surf Sci 1976;59:612. [30] Donnelly SE. Radiat Eff Defects Solids 1985;90:1. [31] Wolfer WG. Philos Mag A – Phys Condens Matter Struct Defects Mech Prop 1988;58:285. [32] Wolfer WG. Philos Mag A – Phys Condens Matter Struct Defects Mech Prop 1989;59:87. [33] Trinkaus H. Possible mechanisms limiting the pressure in inert-gas bubbles in metals. In: Donnelly SE, Evans JH, editors. Fundamental aspects of inert gases in solids, vol. 279. Plenum; 1991. p. 369. [34] Schroeder H, Fichtner PFP, Trinkaus H. Inert gas bubble coarsening mechanisms. New York: Plenum; 1991. [35] Evans JH. J Nucl Mater 1978;76–7:228. [36] Evans JH, Vanveen A, Caspers LM. Scripta Metall 1981;15:323. [37] Evans JH, Vanveen A, Caspers LM. Nature 1981;291:310. [38] Weertman J, Weertman JR. In: Chahn RW, editor. Physical metallurgy. Amsterdam: Elsevier; 1970. p. 921. [39] Sefta F, Hammond KD, Juslin N, Wirth BD. Nucl Fusion 2013;53:073015.
ScienceDirect Acta Materialia 62 (2014) 173–181 www.elsevier.com/locate/actamat
Effect of tungsten crystallographic orientation on He-ion-induced surface morphology changes q C.M. Parish a, H. Hijazi b, H.M. Meyer a, F.W. Meyer b,⇑ a
Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA b Physics Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA Received 3 September 2013; received in revised form 17 September 2013; accepted 22 September 2013 Available online 18 October 2013
Abstract In order to study the early stages of nanofuzz growth in fusion-plasma-facing tungsten, mirror-polished high-purity tungsten was exposed to 80 eV helium at 1130 °C to a fluence of 4 1024 He m2. The previously smooth surface shows morphology changes, and grains form one of four qualitatively different morphologies: smooth, wavy, pyramidal or terraced/wide waves. Combining high-resolution scanning electron microscopy (SEM) observations to determine the morphology of each grain with quantitative measurement of the grain’s orientation via electron backscatter diffraction in SEM shows that the normal-direction crystallographic orientation of the underlying grain controls the growth morphology. Specifically, near-h0 0 1i || normal direction (ND) grains formed pyramids, near-h1 1 4i to h1 1 2i || ND grains formed wavy and stepped structures and near-h1 0 3i || ND grains remained smooth. Comparisons to control specimens indicate no changes to underlying bulk crystallographic texture, and possible explanations of the structure growth, particularly loop-punching, are discussed. Future developments to control tungsten texture via thermomechanical processing, ideally obtaining a sharp near-h1 0 3i || ND processing texture, may delay the formation of nanofuzz. Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. Keywords: Tungsten; Electron backscatter diffraction; Ion irradiation; Helium; Dislocation loop punching
1. Introduction Because of its high melting point, excellent erosion resistance and material strength, tungsten is favored for wall materials for next-generation magnetic fusion devices. In deuterium-tritium (D–T) fusion operation, significant He ash will be formed at 3.5 MeV, and its interaction with tungsten wall materials is currently undergoing extensive evaluation. q
Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC05-00OR22725 with the US Department of Energy. The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. ⇑ Corresponding author. E-mail address: [email protected] (F.W. Meyer).
Under certain conditions of wall temperature and He fluence, He ions were found to cause significant morphology changes in the W surface [1–5], even for impact energies below the physical sputtering threshold of 200–300 eV. While still not completely understood, such surface changes are thought to result from near-surface He trapping at intrinsic or extrinsic defect sites. Due to favorable energetics, the trapped He atoms form clusters, which result in progressively higher lattice distortions, which relax by dislocation loop punching, producing He-filled cavities (bubbles) of increasing size. These He bubbles can eventually burst at high sample temperatures where the tungsten has lower effective viscosity or yield strength, causing surface pinholes, and eventually evolving into an increasingly random nanostructuring of the surface that can ultimately lead to the production of nanofuzz [6–9] Orientation dependence on nanostructuring has been reported [10]. There is concern that such surface nanostructuring may adversely affect the
1359-6454/$36.00 Ó 2013 Acta Materialia Inc. Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.actamat.2013.09.045
174
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
robustness of the W surface in the presence of the highpower fusion plasma due to increased W dust formation, as well as affect H and He retention. In addition, the effect of the bulk material damage due to 14.1 MeV neutrons also produced in the D–T fusion reactions on these near surface He–W interactions has yet to be definitely assessed. In this experiment, we exposed highly polished tungsten coupons to 80 eV helium ions of flux 1020 m2 s1 to a fluence of 4 1024 He m2 at 1130 °C (1400 K) to study the early stages of nanofuzz formation. Scanning electron microscopy–electron backscatter diffraction (SEM–EBSD) was used to correlate the changes in the surface morphology of the grains caused by the helium exposure to the underlying crystallographic orientation, revealing a strong and systematic effect.
4.0–4.5 nA, with a 19–20 mm working distance. The tungsten coupons were attached to the EBSD specimen holder with silver paint, rather than conductive tape, to minimize sample drift and to allow close correlation between SE images (acquired immediately before EBSD) and the EBSD data. Grain calculations are based upon the definition that a >15° pixel-to-pixel misorientation defines a grain boundary. Texture pole figures were calculated with 5.0° half-width, and order L = 34 harmonic series expansion, triclinic sample symmetry and without using any data cleaning methods [14] of the raw data unless explicitly noted below.
2. Experimental details
3.1. Irradiated areas
2.1. Ion irradiation
The irradiated specimen (Fig. 1a) showed significant surface morphology changes from the pre-irradiated, mirror-polished condition. Four broad morphology categories are identified, and in Fig. 1b these are marked (i)–(iv), and shown in more detail in Fig. 1c and d. In Fig. 1b, (i) denotes a smooth or nearly-smooth region, (ii) denotes a wavy morphology, (iii) denotes pyramidal morphology and (iv) denotes terraced morphology. Obviously, these are qualitative assessments and particular grains may show an intermediate or ambiguous morphology. Judgment was used to differentiate the types of morphology, and as will be seen below, this provided results that correlate well to underlying crystallography. A large-area texture scan of the irradiated area is shown in Fig. 2. Fig. 2a shows a normal direction (ND)-projected inverse pole figure (IPF) map, and Fig. 2b the calculated pole figures (PFs) and an ND IPF. PFs indicate the relative number of grains with a given orientation hh k li oriented in a particular sample direction, measured relative to a random polycrystal. A strong (8 random) h0 0 1i || ND grain texture is seen, also with h0 0 1i in-plane. Relatively few h1 0 1i || ND or h1 1 1i || ND-type grains are seen. The average effective grain diameter is 11 ± 7 lm, indicating a wide distribution. The sharp h0 0 1i texture is consistent with reports of rolling and recrystallization in tungsten [15–17]. It is important to note that EBSD data require significant tilting of the specimen away from the surface normal, 70° away from the normal electron-beam incidence angle in this case. Fig. 3 shows a small area imaged using the EBSD software (which compensates for dynamic focus and the 3:1 difference in pixel size in the Y and X scan directions) imaged at 70° tilt, and the same area imaged with the SEM software at 0° tilt. First, note that the images match well; second, note that the features’ third-dimension information, such as the relative heights and angles of the pyramids or waves, are greatly exaggerated in the EBSD-tilt-acquired image. Grains (i) and (ii) are of a wave morphology when imaged at 0°, but appear to be sharp pyramids at 70° tilt.
High-purity tungsten sheet 0.7 mm in thickness was fabricated by electrical discharge machining (EDM) into 13 13 mm coupons and polished to colloidal silica (0.05 lm) for a mirror polish. Tungsten substrates were examined via X-ray photoelectron spectroscopy (XPS) ascut and after cleaning with detergent and water rinse. Aside from W, carbon and oxygen were the primary surface species observed. Minimal argon-ion sputtering reduced the C from 30 at.% to 5 at.% and O from 20 at.% to 4 at.%, showing that these elements were adsorbed atmospheric contaminants. Several of the He-exposed substrates were examined and showed no additional surface contamination from the He processing. The irradiations were performed at the ORNL Multicharged Ion Research Facility (MIRF) [11] using a new high-current beam deceleration module installed in the beamline of our CAPRICE electron cyclotron resonance (ECR) ion source, which has been described elsewhere [12]. This deceleration module provides high-flux He ion beams at energies down to 50 eV to W targets floating at the deceleration potential. The W samples can be raised to temperatures exceeding 1300 °C by electron beam heating. A recently added beam profile measurement device consisting of a positively biased beam catcher inside an enclosure with a 1 mm2 aperture can be scanned across the incident He ion beam via a PC controlled, stepper motor driven, x–y–z manipulator to determine the twodimensional flux distribution of the incident beam [13]. 2.2. Electron microscopy Specimens were examined in a JEOL JSM-6500F fieldemission analytical SEM. High resolution secondary electron (SE) images were acquired at 5 keV with small beam current (<100 pA) and short working distance (4–7 mm). EBSD was performed using an EDAX-TSL Hikari high-speed camera. Beam conditions were 30 keV,
3. Results
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
175
Fig. 1. (a) SEM images of the helium-irradiated area. In (b), four morphologies are marked (i)–(iv): (i) smooth; (ii) wavy; (iii) pyramidal; (iv) wide waves or steps. (c) Detailed views of smooth, wavy and pyramid morphologies. (d) The terrace-like nature of the wide waves.
Fig. 2. (a) Normal direction-projected IPF map. (b) Texture pole figures for {0 0 1}, {1 0 1}, and {1 1 1}, as well as normal-direction (ND) IPF. Note strong h0 0 1i texture. IPF unit-triangle color key is the same for all figures.
Possible artifacts due to non-normal tilt angle such as the latter are important to keep in mind. Determination of morphology should be made from the 0° tilt data, while assignment of crystallographic orientation should be made from the 70° tilt EBSD data. At this point, a data cleaning step is applied: due to surface roughness, there is a narrow spread of orientations within many grains. Each pixel orientation is replaced with the average orientation of the grain, and then a dilation step removes spikes. Fig. 4 shows an SEM image (0° tilt) with identifiable surface features, and Fig. 4b the IPF
map. A white-bounded inset to the bottom-right corner of Fig. 4b shows the raw (uncleaned) data for comparison. Finally, this data was analyzed to determine if the crystallographic direction parallel to the ND correlated to the surface morphology. In Fig. 5a, the EBSD data is presented with grain boundaries superimposed as black lines and coloration of the grains according to the morphology, either smooth (red), wavy/stepped/terraced (green), or pyramidal (blue). A discrete IPF is plotted as Fig. 5b. First, the average orientation of each measured grain is plotted as a black circle with size proportional to the grain size. The
176
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
Fig. 3. A helium-exposed area, imaged at EBSD 70° tilt (left) and at 0° tilt, normal-electron-beam incidence (right). Note features such as (i) and (ii) are misrepresented at 70° tilt.
Fig. 4. (a) 0° tilt SEM image showing surface features. (b) Cleaned IPF EBSD map. IPF color scale as in Fig. 2. Inset at bottom-right shows uncleaned (raw) data.
Fig. 5. (a) EBSD data of Fig. 3 colored by SEM morphology. (b) Inverse pole figure || normal direction; black marks are average individual grain orientations, sized proportionally to grain size. Colored squares are the orientations of the grains colored in (a). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
grains are clustered around h0 0 1i, as expected from the texture calculations (Fig. 2). Second, the colored boxes are plotted for each grain according to the morphologyassigned color in Fig. 5a.
The results indicate that the pyramids cluster around h0 0 1i || ND, the wavy grains in the vicinity of h1 1 3i || ND and the smooth grains near h1 0 2i || ND. Almost no grains are in the h1 01 i–h1 1 1i region.
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
3.2. Thermally exposed, unirradiated specimen Helium unexposed areas subjected to the same high sample temperature were also examined. Fig. 6a shows the SEM image acquired with the EBSD scan, Fig. 6b shows a normal direction IPF scan (with data cleaning as in Fig. 4d) and Fig. 6c shows texture pole figures calculated for this area (calculated from raw, not cleaned, data). The texture is essentially the same as for the helium-exposed area, Fig. 2b. The IPF morphology is also essentially the same as the helium-exposed area, compare Fig. 6b to Fig. 3b. However, the SEM image shows smooth-polished grains with grooving at the boundaries, in contrast to the rough and well-developed morphology of the helium-exposed area. Thus, the texture and general grain structure are not influenced by the helium exposure, but the surface morphology is. 4. Discussion Comparison of the morphologies of the irradiated and thermally treated specimen indicates that the grooving at the grain boundaries is thermally induced, but the morphological changes to the mirror-polished surface are caused by the interaction of the helium beam with the hot tungsten surface. Although growth of surface facets has been previously observed with thermal annealing alone [18], those experiments were at higher temperature for much longer times than these experiments. Three explanations for the observed differences between differently oriented grains can be proposed: (1) ion channeling effects, (2) surface energy effects and (3) dislocation loop-punching effects. First, we consider ion channeling, the phenomenon in which ions incident on or near a low-index zone axis penetrate anomalously deeply into a crystal by traveling down the channels between atomic columns [19,20]. Channeling is dominant with higher-energy beams of heavier elements: for instance, in focused ion beam (FIB) exposures with
177
30 keV Ga+, significant changes in the texture of the surface grains are observed even at much lower fluences (1020 Ga m2) [21]. The current results – using much lower energy, much lighter ions and much higher fluence – are significantly different to the FIB-induced results of Michael [21]. Comparison of the textures of the as-polished starting material with the irradiated material indicates that there is no re-texturing of the near-surface grains induced by the ion exposure. The ion channel theory of Lindhard [22] and Onderdelinden [23] (L–O), as described by Kempshall et al. [24], and as applied to Ga-FIB problems by Michael [21] and Giannuzzi and Michael [25], allows the non-channeled fraction of ions, X uvw 0 , to be calculated as a function of the bombarding ion, bombarding ion energy, the target material and the incident ion direction into the crystal hu v wi. Also calculable is the critical angle for channeling, wuvw 0 , for the incident direction hu v wi, in which ions within an angle of wuvw from the perfect incidence hu v wi will 0 still channel strongly. For 80 eV helium incident onto body-centered cubic (bcc) tungsten, these are zero for any direction other than h1 1 1i. In the h1 1 1i case, the nonchanneled fraction X h111i 85%, so only 15% of the ions o would be channeled. (The h1 1 1i channels most strongly because it is the closest-packed direction in bcc.) Because almost no h1 1 1i grains were observed in either the irradiated or non-irradiated regions, channeling effects for this low-energy helium regime is discounted as a possible effect. Higher-energy bombardments, not studied here, may suffer significant channeling effects. Garrison and Kulcinski [26] used 30 keV He, and found that grains 4–8° away from h0 0 1i suffered the least helium induced morphology change; but grains closer to h0 0 1i than 4°, or near h1 0 1i or h1 1 1i, did suffer significant attack. The L–O model predicts nearly complete channeling down h0 0 1i,h1 0 1i and h1 1 1i for 30 keV (>90% each), but given the assumptions inherent to the L–O model, discrepancies are unsurprising. The Garrison and Kulcinski results are interesting and need further exploration.
Fig. 6. (a) SEM image of the unirradiated region. (b) ND IPF map of the EBSD data; colorscale is the same as Fig. 2a. (c) PFs and ND IPF texture; colorscale varies from 1 to 11.7 random.
178
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
Surface energy is another possible explanation. Different references disagree on absolute values and relative ratios of surface energy for tungsten planes, but in general {1 0 1} is considered the lowest energy plane and {1 1 1} almost as low energy. {0 0 1} and {1 1 2} are also particularly low energy. Surface energies chkl for a given plane and ratios khkl of surface energy khkl = chkl/c101 are shown in Table 1. SEM alone lacks the three-dimensional (3-D) information needed to determine the planes of the different facets that grew during these experiments, but strong inferences can be made from the available information. Fig. 7a shows the SEM image of the scanned area again (acquired at 70° tilt), and with three grains marked (i)–(iii) and with the grains’ unit cells superimposed, as measured via EBSD. Detailed views of these grains (acquired at 0° tilt) are given in Fig. 7b and c, and discrete pole figures calculated for each grain (i)–(iii) are in the bottom half of Fig. 7. The pyramidal-morphology grain (iii) is the simplest, so we consider it first. The grain is nearly h0 0 1i || ND oriented. The {1 0 1} and {1 1 1} PFs show the {1 0 1} planes lying 45° from the ND, with plane normals nearly vertical and horizontal in the reference frame, which is to say, toward the reference direction (RD) and transverse direction (TD) axes. The {1 1 1} planes are almost 60° off normal, and rotated 45° about the ND relative to the {1 0 1} plane normal. The SEM image (Fig. 7c) shows that most of the pyramidal faces have normals whose surface tracks are either horizontal or vertical. Along with the tetrahedral shape of the pyramids and the low value of {1 0 1} surface energy, the surfaces of most of the pyramids are almost certainly {1 0 1} planes. Again, SEM does not provide sufficient 3-D data to be completely sure, but the assignment of the pyramidal faces in grain (iii) as {1 0 1} planes is very likely. Grain (i), a wave morphology, is more difficult to assign plane faces to. The grain is measured to be [3 6 2 2] || ND, or approximately of type h1 2 7i || ND. Along the length of the
Table 1 Tungsten surface energies chkl (J m2) and ratios k of chkl to c101. {h k l}
Wang et al. [27]
Zhang et al. [28]
Kumar and Grenga [29]
c k
4.025 1.343
2.645 1.185
1.028
c k
2.998 1.000
2.232 1.000
1.000
{1 1 1}
c k
3.671 1.225
2.247 1.007
Not observed
{1 1 2}
c
3.461
2.299
Waves/ terraces
k
1.155
1.030
{1 0 2}
c k
3.696 1.233
2.595 1.163
{1 0 3}
c k
3.886 1.296
2.671 1.197
{0 0 1} {1 0 1}
Observed morphology Pyramids Not observed
1.030 Smooth Smooth 1.120
waves, the crystal direction is measured by EBSD to be approximately [6 8 3]. The waves are not straight, so multiple values of the in-plane length of the waves, or values in-plane perpendicular to the waves, are possible. From the PF (i), however, a h1 0 2i-type direction is approximately perpendicular to the lengths of the waves, and is marked in Fig. 7b. Neither {1 0 1} nor {1 1 1} planes line up precisely, but both are very close to the axis perpendicular to the waves and are possible facets. Given the fact that the waves are not straight and bifurcate, it is conceivable (but not proven) that both faces are present in different regions. Similarly, the wide-wave or step morphology of grain (ii) is complicated to assign faces to, because of bifurcations and shifts in the planes. However, the long steplengths may be of {0 0 1} type, and the shorter steps of {1 1 2} type, as marked on the pole figures. This is simply one possibility, however, and not definitive. A FIB liftout was performed to help confirm the above plane analyses. A “wide-wave” morphology grain was found (Fig. 8a) by imaging with the SEM column in a FIB/SEM instrument. A liftout was attached to a copper TEM-style sample carrier and polished to 5 keV Ga+, Fig. 8b (again imaged in FIB/SEM). The specimen was then transferred to the SEM–EBSD system and imaged and EBSD mapped. Five grains are present on the liftout’s face, and their unit cell orientations are marked (Fig. 8c). The wide-wave morphology grain is marked in red1, and that single red grain’s PFs for {0 0 1}, {1 0 1}, {1 1 1}, {0 1 2}, {1 1 2} and {1 0 3} are presented in Fig. 8d. The h0 0 1i direction in the center of the PF indicates that the length of the wide waves is h0 0 1i.h1 0 1i is very nearly parallel to the original surface normal. Because the liftout was cut perpendicular to the length of the waves, only planes with normals falling on the equator (outmost grid circle) of the PFs can be the surface faces of the waves. This eliminates {1 0 1}, {1 1 1} and {1 1 2} because none of those planes falls on the equator near the directions of the waves’ surface normal. Tilting the specimen’s polished face flat to the electron beam allows the angle of the facets relative to the original surface to be measured; the facets are 15–25° above the original surface (not shown; Fig. 8b is not normal to the electron beam due to FIB–SEM instrument hardware geometry). The grid lines in Fig. 8d are in 15° increments, so of the equator-lying planes, i.e., {0 0 1}, {1 0 1}, {1 0 2} and {1 0 3}, it can be seen that {1 0 2} matches most closely to these measured angles, although {1 0 3} is also within the margin of error. The lowest-energy surfaces, which would be expected from a pure surface-energy effect, are not necessarily the observed surfaces. It appears that both high-index arbitrary surface faces and low-index {0 0 1} and {1 0 1} faces are faceting with resultant changes in surface area and
1 For interpretation of color in Fig. 8, the reader is referred to the web version of this article.
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
179
Fig. 7. (a) SEM image with grain boundaries (black) and three grain unit cells superimposed, labelled (i)–(iii). (b and c) detailed views of the grains. The bottom row shows PFs for the grains (i)–(iii).
Fig. 8. (a) In-FIB/SEM image of a wide-wave grain. (b) Liftout imaged after FIB polishing. (c) EBSD mapped region, with five grains’ unit cells superimposed. (d) PFs of the wide-wave grain.
180
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
energy. {1 0 1} is considered the lowest-energy plane in tungsten, yet in the example of Fig. 8, for instance, a nearly-{1 0 1} pre-irradiated surface appears to facet into either {1 0 2} or {1 0 3} facets. Given that a faceted surface has a larger surface area (and thus larger integrated surface energy even at constant surface energy per unit area) than a smooth polished surface, and that the smooth-remaining grains (as noted in Fig. 5) are around the {1 0 4}–{1 0 3}– {1 0 2} region of the unit triangle, a simple explanation is hard to find. The third possibility, as suggested by Ohno et al. [10], is that dislocation loop punching caused by the growth of helium bubbles is driving the surface faceting. As more helium atoms accumulate in a cavity, the pressure increases above the equilibrium pressure. Donnelly [30] summarizes the conditions for different mechanisms to decrease the pressure, and for larger bubbles the dominant mechanism is the punching of a dislocation loop from the bubble to increase the size of the bubble. A number of references describe loop punching in detail [31–37]. Very briefly, as the helium content of a bubble increases it eventually becomes energetically more favorable for the pressure to decrease via bubble growth rather than for the bubble size to stay constant at increasing pressure. An interstitial dislocation loop – essentially a platelet of interstitial atoms – will be emitted from the bubble, which will then be able to glide under stress. Early transmission electron microscopy (TEM) work, for instance, showed arrays of dislocation loops pushed out from individual helium filled cavities [36,37] in molybdenum. Recent work by Ohno et al. [10] on 25–30 eV helium irradiated tungsten surfaces showed results similar to ours, and the growth of non-lowestenergy facets was attributed to the action of bubbles punching glissile dislocation loops on {1 0 1} slip planes (although {2 1 1}h 1 1 1i slip in bcc tungsten is suggested by Weertman and Weertman [38]). Similar to our work, Ohno et al. observed relatively little surface morphology change in the area around h1 0 3i specimen normal. However, one difference is that they observed wavy structures forming at h0 0 1i whereas we observed pyramidal growth. The FIB liftout of Fig. 8 was thinned further with lowenergy Ga+ and then polished with 900 eV Ar+ and imaged in an FEI TitanS 300 keV TEM/STEM. Fig. 9a shows a BF Fresnel-contrast TEM image, and bubbles are clearly visible to a depth of 200 nm. Fig. 9b shows a lattice image of the region arrowed in Fig. 9a, and the surface plane is very near (2 1 0), confirming the EBSD results of Fig. 8. Unfortunately, too much contrast arising from specimen damage (both bubble-related and Ga+/Ar+ related) is present to directly image dislocation loops. It is possible that future experiments, likely taking advantage of the aberration corrector on the Titan’s probe-forming optics, might allow direct observation of dislocation loops. Very recent work [39] using molecular dynamics (MD) has indicated that loops punched out of growing nearsurface helium bubbles in tungsten will be drawn to the surface by image forces and become adatom islands, which
Fig. 9. (a) Low-magnification TEM image of the surface of the grain analyzed in Fig. 8. Arrow marks the region of (b), a lattice image showing near-{1 0 2} faceting.
probably act as incipient nanofuzz. While those apparent morphologies appear to be different from our observations, for reasons not yet understood in detail, this MD work increases our confidence that loop-punching may be the root cause of our surface morphology changes. In short, no single explanation, neither surface energy nor loop punching, is precisely consistent with the observed surface features, indicating that a more thorough understanding must be sought. 5. Conclusions Sub-sputtering-threshold 80 eV helium exposure to 4 1024 He m2 at 1130 °C resulted in roughening and surface morphology changes to smooth-polished tungsten. Grain boundary grooving occurred, and four general grain morphologies were observed: smooth, pyramidal, wavy and terraced. Unexposed areas held at the same temperature showed grain boundary grooving but not intragranular surface roughening, indicating that the effect arose from helium exposure. Texture pole figures for unexposed and exposed areas were the same, indicating that new
C.M. Parish et al. / Acta Materialia 62 (2014) 173–181
grains are not forming and grain rotation did not occur. The morphology changes correlated with the underlying surface normal direction of the grain: pyramids formed around h0 0 1i || ND, grains remained smooth around h1 0 2i–h1 0 3i–h1 0 4i || ND, and waves and steps formed in the region h1 1 4i–h1 1 2i–h2 1 3i || ND. No correlations to ion channeling directions were observed, and the developing surface facets were not low-energy planes in tungsten. The facet faces formed appear to be of several types, including {1 0 1} and {1 0 2}. The growth of these faces is explained by neither surface energy nor loop punching arguments alone. We suggest that thermomechanical processing to enhance h1 0 2i–h1 0 3i || ND grain processing textures, while difficult, may be effective in delaying nanofuzz nucleation under bombardment by DT fusion helium ash of hot first wall W surfaces of magnetic fusion devices. Acknowledgements This research was sponsored by the LDRD Program of Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the US Department of Energy. HH was appointed through the ORNL Postdoctoral Research Associates Program administered jointly by Oak Ridge Institute of Science and Education (ORISE), Oak Ridge Associated Universities (ORAU) and Oak Ridge National Laboratory (ORNL). Work performed in part via Oak Ridge National Laboratory’s Shared Research Equipment (ShaRE) User Program, which is sponsored by the Office of Basic Energy Sciences, US Department of Energy. Thanks to Dr. Lucille Giannuzzi, L.A. Giannuzzi & Associates LLC, for discussions regarding ion channeling. Thanks to Dr. Maxim Gussev, ORNL, for critiquing the manuscript. References [1] Iwakiri H, Yasunaga K, Morishita K, Yoshida N. J Nucl Mater 2000;283:1134. [2] Minyou Y. Plasma Sci Tecnhol 2005;7:2828. [3] Nishijima D, Ye MY, Ohno N, Takamura S. J Nucl Mater 2003;313:97. [4] Baldwin MJ, Lynch TC, Doerner RP, Yu JH. J Nucl Mater 2011;415:S104. [5] Yamagiwa M, Kajita S, Ohno N, Takagi M, Yoshida N, Yoshihara R, et al. J Nucl Mater 2011;417:499.
181
[6] Baldwin MJ, Doerner RP. Nucl Fusion 2008;48:035001. [7] Baldwin MJ, Doerner RP. J Nucl Mater 2010;404:165. [8] Takamura S, Ohno N, Nishijima D, Kajita S. Plasma and fusion research, vol. 1; 2006. #051. [9] Meyer FW, Hijazi H, Krstic PS, Dadras J, Meyer III HM, Parish CM, Bannister ME. Phys Scripta; in press. [10] Ohno N, Hirahata Y, Yamagiwa M, Kajita S, Takagi M, Yoshida N, et al. J Nucl Mater 2013;438(Suppl.):S879. [11] Meyer FW, Bannister ME, Hale JW, Havener CC, Krause HF, Vane CR, et al. Recent activities at the ORNL multicharged ion research facility (MIRF). In: Proceedings of ECRIS2010. Grenoble, France; 2012. [12] Hijazi H, Meyer FW. Rev Sci Instrum 2013;84:033305. [13] Bannister ME, Hijazi H, Meyer III HM, Cianciolo V, Meyer FW. Submitted for publication. [14] Brewer LN, Michael JR. Microsc Today 2010;18:10. [15] Neges J, Ortner B, Leichtfried G, Stuwe HP. Mater Sci Eng A – Struct Mater Prop Microstruct Process 1995;196:129. [16] Dhers J. In: Chandra T, Torralba JM, Sakai T, editors. Thermec’2003. Zurich-Uetikon, Pts 1–5, vol. 426-4. Trans Tech Publications; 2003. p. 4455. [17] Brewer LN, Bewlay BP. Microsc Microanal 2002;8:696CD. [18] Zhang W, Sachenko P, Gladwell I. Acta Mater 2004;52:107. [19] Brandt W. Sci Am 1968;218:90. [20] Feldman LC, Mayer JW, Picraux ST. Materials analysis by ion channeling. New York: Academic Press; 1982. [21] Michael JR. Microsc Microanal 2011;17:386. [22] Lindhard J. Phys Lett 1964;12:126. [23] Onderdelinden D. Appl Phys Lett 1966;8:189. [24] Kempshall BW, Schwarz SM, Prenitzer BI, Giannuzzi LA, Irwin RB, Stevie FA. J Vac Sci Technol B 2001;19:749. [25] Giannuzzi LA, Michael JR. Microsc Microanal 2013;19:344. [26] Garrison LM, Kulcinski GL. Phys Scripta; in press. [27] Wang SG, Tian EK, Lung CW. J Phys Chem Solids 2000;61:1295. [28] Zhang JM, Ma F, Xu KW. Surf Interface Anal 2003;35:662. [29] Kumar R, Grenga HE. Surf Sci 1976;59:612. [30] Donnelly SE. Radiat Eff Defects Solids 1985;90:1. [31] Wolfer WG. Philos Mag A – Phys Condens Matter Struct Defects Mech Prop 1988;58:285. [32] Wolfer WG. Philos Mag A – Phys Condens Matter Struct Defects Mech Prop 1989;59:87. [33] Trinkaus H. Possible mechanisms limiting the pressure in inert-gas bubbles in metals. In: Donnelly SE, Evans JH, editors. Fundamental aspects of inert gases in solids, vol. 279. Plenum; 1991. p. 369. [34] Schroeder H, Fichtner PFP, Trinkaus H. Inert gas bubble coarsening mechanisms. New York: Plenum; 1991. [35] Evans JH. J Nucl Mater 1978;76–7:228. [36] Evans JH, Vanveen A, Caspers LM. Scripta Metall 1981;15:323. [37] Evans JH, Vanveen A, Caspers LM. Nature 1981;291:310. [38] Weertman J, Weertman JR. In: Chahn RW, editor. Physical metallurgy. Amsterdam: Elsevier; 1970. p. 921. [39] Sefta F, Hammond KD, Juslin N, Wirth BD. Nucl Fusion 2013;53:073015.