# Fileset

[Epitaxial lateral overgrowth of α-Ga2O3 by halide vapor phase.pdf](https://mdr.nims.go.jp/filesets/edb014a1-255d-4ac3-b517-cbeaabe3e316/download)

## Creator

[Y. Oshima](https://orcid.org/0000-0001-8293-4891), K. Kawara, T. Shinohe, T. Hitora, M. Kasu, S. Fujita

## Rights

Copyright 2018 Author(s). This article is distributed under a Creative Commons Attribution (CC BY) License.[Creative Commons BY Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/)

## Other metadata

[Epitaxial lateral overgrowth of α-Ga2O3 by halide vapor phase epitaxy](https://mdr.nims.go.jp/datasets/9237280e-1e6b-42a4-871b-6717b63d93c1)

## Fulltext

Epitaxial lateral overgrowth of α-Ga2O3 by halide vapor phase epitaxyViewOnlineExportCitationCrossMarkRESEARCH ARTICLE |  DECEMBER 10 2018Epitaxial lateral overgrowth of α-Ga2O3 by halide vapor phaseepitaxy Special Collection: Wide Bandgap OxidesY. Oshima ; K. Kawara; T. Shinohe; T. Hitora; M. Kasu; S. FujitaAPL Mater. 7, 022503 (2019)https://doi.org/10.1063/1.5051058 29 January 2024 06:53:11https://pubs.aip.org/aip/apm/article/7/2/022503/1064099/Epitaxial-lateral-overgrowth-of-Ga2O3-by-halidehttps://pubs.aip.org/aip/apm/article/7/2/022503/1064099/Epitaxial-lateral-overgrowth-of-Ga2O3-by-halide?pdfCoverIconEvent=citehttps://pubs.aip.org/aip/apm/article/7/2/022503/1064099/Epitaxial-lateral-overgrowth-of-Ga2O3-by-halide?pdfCoverIconEvent=crossmarkhttps://pubs.aip.org/apm/collection/1052/Wide-Bandgap-Oxidesjavascript:;javascript:;javascript:;javascript:;javascript:;javascript:;javascript:;https://doi.org/10.1063/1.5051058https://servedbyadbutler.com/redirect.spark?MID=176720&plid=2288747&setID=592934&channelID=0&CID=840257&banID=521619172&PID=0&textadID=0&tc=1&scheduleID=2208988&adSize=1640x440&data_keys=%7B%22%22%3A%22%22%7D&matches=%5B%22inurl%3A%5C%2Fapm%22%5D&mt=1706511191112392&spr=1&referrer=http%3A%2F%2Fpubs.aip.org%2Faip%2Fapm%2Farticle-pdf%2Fdoi%2F10.1063%2F1.5051058%2F13146790%2F022503_1_online.pdf&hc=0103c09135cf9fa66220dd486d15a1b2a2cd9ef7&location=APL Materials ARTICLE scitation.org/journal/apmEpitaxial lateral overgrowth of α-Ga2O3 by halidevapor phase epitaxyCite as: APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058Submitted: 6 August 2018 • Accepted: 11 September 2018 •Published Online: 10 December 2018Y. Oshima,1,a) K. Kawara,2 T. Shinohe,2 T. Hitora,2 M. Kasu,3 and S. Fujita4AFFILIATIONS1Optical Single Crystals Group, National Institute for Materials Science, 1-1 Namiki, 305-0044 Tsukuba, Japan2FLOSFIA, Inc., Kyodai-Katsura Venture Plaza, 615-8245 Kyoto, Japan3Department of Electrical and Electronic Engineering, Saga University, 1 Honjo-machi, 840-8502 Saga, Japan4Department of Electronic Science and Engineering, Kyoto University, Katsura 615-8520, Kyoto, Japana)Author to whom correspondence should be addressed: OSHIMA.Yuichi@nims.go.jpABSTRACTWe demonstrate the epitaxial lateral overgrowth of α-Ga2O3 by halide vapor phase epitaxy. We prepared patterned SiO2 maskson a (0001) α-Ga2O3/sapphire template, and then α-Ga2O3 islands were regrown selectively on the mask windows. The islandsgrew vertically and laterally to coalesce with each other. Facet control of the α-Ga2O3 islands was achieved by controlling thegrowth temperature, and inclined facets developed by decreasing the temperature. Transmission electron microscopy revealedthat the crystal quality of the regrown α-Ga2O3 was improved owing to both the blocking of dislocations by the mask and thedislocation bending by the inclined facets.© 2018 Author(s). All article content, except where otherwise noted, is licensed under a Creative Commons Attribution (CC BY) license(http://creativecommons.org/licenses/by/4.0/). https://doi.org/10.1063/1.5051058Ga2O3 has been reported to possess five different poly-morphs; these are the α-, β-, δ-, ε-, and γ-phases.1 α-Ga2O3,the target material studied in the present work, is one ofthe metastable phases of Ga2O3 and crystalizes into thecorundum structure. α-Ga2O3 is a wide bandgap semicon-ductor, and the bandgap energy has been reported to be5.2–5.3 eV.2,3 A high breakdown voltage and low on-resistanceare expected because of the large bandgap energy, and there-fore, α-Ga2O3 is promising for power device applications.Actually, Schottky barrier diodes with very low on-resistancebeyond the SiC limit have already been demonstrated.4 Inaddition, corundum-structured α-Ir2O3 and α-(Rh, Ga)2O3have been shown to exhibit clear p-type conduction,5,6 andtherefore, bi-polar devices using the hetero-pn-junctions areexpected.In contrast to the case of thermodynamically stableβ-Ga2O3, freestanding α-Ga2O3 wafers cannot be producedthrough the melt-growth technique. Accordingly, α-Ga2O3films need to be grown by heteroepitaxy. The heteroepitaxyof α-Ga2O3 is possible on sapphire (α-Al2O3) substrates bymist-CVD2 and halide vapor phase epitaxy (HVPE).3 Althoughthe crystal structures of α-Ga2O3 and sapphire both havethe corundum structure, the lattice mismatch is large (∆a/a∼ 4.5%, ∆c/c ∼ 3.3%). Therefore, α-Ga2O3 epilayers usuallyinclude a high density of dislocations. Figure 1(a) shows across-sectional TEM image of a conventional α-Ga2O3 filmgrown on a (0001) sapphire substrate by HVPE. We can see ahigh density of dislocations propagating along the film growthdirection. The dislocation density was estimated to be on theorder of 1010 cm−2 from a plan-view TEM image of the samesample [Fig. 1(b)]. The crystal quality should be improved sincesuch crystal defects could deteriorate the performance ofα-Ga2O3 devices, although the influence of the defects has notyet been clarified.To improve the crystal quality of heteroepitaxial filmsgrown on highly mismatched substrates, such as GaN on sap-phire, the epitaxial lateral overgrowth (ELO) technique hasbeen shown to be effective.7–9 In this technique, epitaxialgrowth is performed on a seed layer (GaN on sapphire, forexample) with a periodically patterned mask on the surface.APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-1© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmhttps://doi.org/10.1063/1.5051058https://www.scitation.org/action/showCitFormats?type=show&doi=10.1063/1.5051058https://crossmark.crossref.org/dialog/?doi=10.1063/1.5051058&domain=aip.scitation.org&date_stamp=2018-December-10https://doi.org/10.1063/1.5051058mailto:OSHIMA.Yuichi@nims.go.jphttp://creativecommons.org/licenses/by/4.0/https://doi.org/10.1063/1.5051058APL Materials ARTICLE scitation.org/journal/apmFIG. 1. TEM images of a conventional (0001) α-Ga2O3 film grown by HVPE.(a) Cross-sectional image and (b) plan-view image.The dimension of the mask width and windows are typicallyof micro-meter size, and the epitaxial growth begins selec-tively on the windows to form isolated islands of the tar-get crystal. The islands then grow vertically and laterally andfinally coalesce with each other to form a flat film. In thisgrowth process, dislocations in the seed layer under the maskdo not propagate into the grown layer. Although dislocationsin the seed layer propagate into the grown layer throughthe windows, the dislocations bend toward the lateral direc-tion to minimize the elastic strain energy if the islands haveinclined facets. As a result, the density on the film surface isreduced dramatically. The ELO technique is essential to growhigh-quality GaN by heteroepitaxy, and the threading dislo-cation density reduces typically from 109 cm−2 to 106 cm−2 orless.7–9 The ELO of α-Ga2O3 has already been demonstrated.10In the demonstration, a stripe-patterned SiO2 mask with amask/window size of 2 µm/2 µm was formed directly on a(0001) sapphire substrate, and α-Ga2O3 was grown by mist-CVD. Although coalescence was not achieved, cross-sectionalTEM revealed that no dislocation propagated into the laterallygrown wing region on the mask.To reduce the dislocation density effectively by ELO, asmall mask fill factor, that is, a small window size and widewindow spacing, is desirable. However, a smaller fill factormask requires thicker growth for coalescence, and a fastgrowth rate is therefore preferable. From this point of view,we employed HVPE to grow α-Ga2O3 in this work. HVPE is atype of CVD technique that is characterized by a fast growthrate, and the HVPE of α-Ga2O3 has already been demon-strated.3 In the present work, we demonstrate the ELO ofα-Ga2O3 by HVPE.We used an HVPE-grown (0001) α-Ga2O3 template(approximately 3-µm-thick) on sapphire as a seed substrate.Periodic masks were formed on the template layer, andα-Ga2O3 was regrown by HVPE. The growth temperature ofα-Ga2O3 is typically as low as 500–600 ◦C, and the driv-ing force of HVPE growth is higher compared with thatFIG. 2. SEM images of α-Ga2O3 stripes grown on an (a) SiO2 mask and (b) TiO2mask (bird’s-eye view).of mist-CVD. Therefore, a key issue for successful ELO ofα-Ga2O3 would be the suppression of undesired nucleation onthe mask. From this point of view, we have investigated themask material, mask design, and HVPE growth conditions.We investigated SiO2, Ti, TiO2, and TiN as mask materi-als. SiO2 was deposited by RF sputtering. Ti was deposited byelectron beam evaporation. TiN and TiO2 were produced bynitridation and oxidation of the Ti layer, respectively.We employed a stripe-patterned mask or dot-patternedmask, which had circle-shaped windows arranged to form atriangular lattice pattern. The masks were fabricated by con-ventional photolithography using a maskless exposure sys-tem. The mask/window widths of the stripe-patterned maskwere 5 µm/5 µm. The diameter of the circle windows of thedot-patterned mask was 5 µm. The window spacing of thedot-patterned mask (the distance between mask edges of thenearest windows) was 5–20 µm. A 5-µm-wide dot-patternedmask was used unless otherwise mentioned.We used a home-made atmospheric horizontal quartzHVPE reactor for this study. GaCl and O2 were used as the pre-cursors. GaCl was synthesized upstream in the reactor by thechemical reaction between metal Ga (>99.999 99% pure) andHCl gas (>99.999% pure) at 570 ◦C. The GaCl and O2 were theninjected together with N2 carrier gas into the growth zonedownstream in the reactor to grow α-Ga2O3 on the substrate.The growth was carried out at 540 ◦C with partial pressures ofGaCl and O2 supply to be 1.25 × 10−1 kPa and 1.25 kPa, respec-tively, unless otherwise specified. The total gas flow rate wasfixed to be 8 slm. The growth rate for flat α-Ga2O3 films was12 µm/h under these growth conditions. We used α-Ga2O3templates with approximately 3.4 cm2 area, and the spa-cial growth rate variation on the wafer was typically ±1%or less.The morphologies of the grown crystals were observedby scanning electron microscopy (SEM). The crystal qualitywas evaluated by X-ray rocking curve (XRC) measurements.FIG. 3. SEM images of α-Ga2O3 islands grown on dot-patterned mask with window spacings of (a) 5 µm,(b) 10 µm, and (c) 20 µm (bird’s-eye view).APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-2© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmAPL Materials ARTICLE scitation.org/journal/apmFIG. 4. SEM images of α-Ga2O3 islands grown at nominal growth rates of(a) 7 µm/h and (b) 5 µm/h (bird’s-eye view).The behavior of the crystal defects was observed by transmis-sion electron microscopy (TEM).Figures 2(a) and 2(b) show the SEM images of the sam-ples grown on a SiO2 mask and a TiO2 mask, respectively. Inthe case of the SiO2 mask, α-Ga2O3 stripes grew selectivelythrough the windows, while poly crystal grains nucleated onthe TiO2 mask. This difference was probably not because of thedifference of the mask material but because whether the maskwas amorphous or crystalline. The results for Ti and TiN maskswere similar to those obtained for the TiO2 mask. We thereforeemployed an SiO2 mask in the following experiments.Figures 3(a)–3(c) show the SEM images of the samplesgrown for 40 min on the dot-patterned masks with differentwindow spacings of 5, 10, and 20 µm. The same recipe wasapplied to grow all three samples. When the spacing was 5 µm,we only observed a regular array of α-Ga2O3 islands [Fig. 3(a)].However, when the spacing was wider, extra grains of ε-Ga2O3(not α) nucleated around each α-Ga2O3 island [Figs. 3(b)and 3(c)]. In such selective area growth, most of precursorsare consumed only at window area. Therefore, the increasein the window spacing (i.e., the decrease in window density)leads to the increase in the effective precursor supply perwindow. Actually, the island size was larger on the larger spac-ing mask. Such an increase in the growth driving force wouldlead to the easier nucleation of the extra grains. Accordingly, adecrease in the precursor supply into the reactor should beeffective to suppress the undesired grains. To confirm this,we carried out the growth at slower growth rates. In thesegrowth experiments, only the partial pressure of GaCl sup-ply was decreased from 1.25 × 10−1 kPa to 6.3 × 10−2 kPa and3.8 × 10−2 kPa. As a result, the nominal growth rate (i.e., thegrowth rate for flat films) decreased from 12 µm/h to 7 µm/hand 5 µm/h, respectively. A 20-µm-wide dot-patterned maskwas used for the growth. Figures 4(a) and 4(b) show the SEMimages of the samples. It was found that the nucleation of theextra grains was markedly suppressed under slower growthconditions.Figures 5(a)–5(c) show the SEM images of the samplesgrown at 540, 500, and 460 ◦C. When the growth temperaturewas 540 ◦C, the island shape was a hexagonal pillar with a well-developed (0001) plane on the top and small inclined (101̄1)facets. When the temperature was decreased to 500 ◦C, the(0001) plane became unstable and (101̄1) facets developed well.When the temperature was further decreased to 460 ◦C, the(0001) plane disappeared and (101̄4) facets appeared instead.As a result, the island shape was dominated by inclined facets.Thus, it was found that the island shape can be controlled bythe growth temperature. This feature is useful to carry outfacet-controlled ELO.Figures 6(a)–6(d) show the SEM images of the samplesat different growth stages, which are indicated using nominalthickness (i.e., thickness for flat films). At the beginning of thegrowth, the island shape was similar to that of the circle win-dow [Fig. 6(a)]. Then a hexagonal crystal habit became clear[Fig. 6(b)]. The coalescence process began from the bottomFIG. 5. SEM images of α-Ga2O3 islands grown at (a)540 ◦C, (b) 500 ◦C, and (c) 460 ◦C (bird’s-eye view).FIG. 6. SEM images of α-Ga2O3 islandswith nominal thickness of (a) 0.5 µm,(b) 1.6 µm, (c) 8 µm, and (d) 12 µm(plan-view and bird’s-eye view).APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-3© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmAPL Materials ARTICLE scitation.org/journal/apmFIG. 7. An SEM image of ELO-grown α-Ga2O3 flat film (bird’s-eye view).part of the islands [Fig. 6(c)], and the valley between islandsbecame shallow by continuing the growth [Fig. 6(d)]. Finally, aflat α-Ga2O3 film was obtained (Fig. 7).Figure 8 shows XRC FWHMs of the 0006 and 101̄2 diffrac-tion peaks, measured in symmetric and skew-symmetricgeometry, respectively, as a function of the nominal thick-ness. The FWHM of the 0006 diffraction peak (tilt angle)reflects the tilting of the (0001) plane, while the FWHM ofthe 101̄2 diffraction peak (twist angle) reflects the twistingaround [0001]. It was found that both the tilt and twistangles decreased with increasing nominal thickness, proba-bly owing to the increase in the volume of the high-qualityarea.Figure 9 shows the XRC profiles of 0006 and 101̄2 diffrac-tions for the sample with a nominal thickness of 12 µm. Inthe case of GaN ELO, the XRC profile of the out-of-planediffraction sometimes show splitting reflecting the formationof small angle grain boundaries because of wing tilting dur-ing the lateral growth.11,12 However, no such splitting wasobserved in Fig. 9. This result indicated that wing tilting wasnot significant in the case of ELO-grown α-Ga2O3, althoughthe reason for this needs to be clarified in future work.To clarify the behavior of dislocations in the ELO-grownα-Ga2O3, we carried out cross-sectional TEM for a coa-lesced sample. The growth was carried out at 520 ◦C for 2 h.FIG. 8. XRC FWHMs of ELO-grown α-Ga2O3 as a function of nominal thickness.FIG. 9. XRC profiles of ELO-grown α-Ga2O3 with a nominal thickness of 12 µm.(101̄1) inclined facets develop well during the island growthprocess at this growth temperature. Figure 10(a) shows theplan-view SEM of the sample surface. The surface was stillbumpy, and the positions of the windows and coalescedboundaries could easily be identified. Figure 10(b) showsthe schematic of the cross section of the sample, and thedotted-line rectangle shows the observation area of the crosssectional TEM. Figure 10(c) shows the TEM image. In theα-Ga2O3 template layer under the mask, a high density ofdislocations was observed. The dislocations propagated intothe regrown layer through the window. Then the dislocationbending started from the vicinity of the mask edge towardthe window center successively as the growth proceeded.FIG. 10. (a) Plan-view SEM image of a coalesced α-Ga2O3 film. (b) Schematic ofthe cross section. (c) Cross-sectional TEM image of the film.APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-4© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmAPL Materials ARTICLE scitation.org/journal/apmFIG. 11. SEM and TEM images of α-Ga2O3 stripes withwell-developed (0001) plane. (a) Plan-view SEM image,(b) cross-sectional TEM image, and (c) plan-view TEMimage.As will be described later, such bending was not observedwhen the inclined facets were not well developed. The dis-location bending therefore probably arose from the inclinedfacets. We could still see the contrast of crystal defects in thetop part above the window, but the density was much lowerthan that in the template layer. The defect density shouldbe further decreased by controlling the island morphologyso that the (0001) plane disappears completely. It is possiblethat the residual defects include not only dislocations but alsoother types of crystal defects, such as staking faults. Dark-field observation is now ongoing to identify the types of thedefects. At the coalesced boundary just above the mask, wecan see dislocation contrasts, which indicated that the crystalorientations of the adjacent islands were not completely thesame. The number of dislocations at the boundary decreasedwith increasing thickness, and no dislocation was found at thetop part. The crystal quality should be further improved if wecarry out the ELO process twice with positioning the secondmask so that the first windows areas are covered. Note that thebehavior of dislocations described above might be dependenton the mask direction with respect to the crystal orientation.The mask direction issue is quite important to reduce the dis-location density effectively, and it should be addressed in thefuture work.To confirm that the dislocation bending described abovecan be attributed to the inclined facets, we carried out TEMof a sample which was grown so that the (0001) plane waswell developed by raising the growth temperature to 560 ◦C. A5-µm-wide stripe-patterned mask was used for the growth.Figure 11(a) shows the plan-view SEM of the sample. Eachα-Ga2O3 stripe was accompanied by (0001) ε-Ga2O3 on bothsides. Since lateral growth rate of (0001) ε-Ga2O3 is largerthan that of (0001) α-Ga2O3 under the growth conditionsused for this experiment, the ε-Ga2O3 parts were about tocoalesce with each other, while α-Ga2O3 parts were stillseparated. Figure 11(b) shows a cross-sectional TEM imageof the sample. No dislocation bending was found in con-trast to the case of the above-mentioned sample grownwith well-developed inclined facets. Figure 11(c) shows aplan-view TEM image of the sample. No dislocation was foundin the laterally grown areas (approximately 22 µm2 in total),and therefore, the dislocation density should be less than5 × 106 cm−2 in these areas. Note that it is not clear atpresent if this dislocation density is sufficiently low or notfor power device applications, because it is quite specific toeach semiconductor material, and is strongly dependent onthe device structure and drive conditions. Further work isrequired to clarify the influence of crystal defects in α-Ga2O3devices in order to make it clear how low the defect densityshould be.We have demonstrated the ELO of α-Ga2O3 by HVPEfor the first time. Selective area growth was achieved byusing SiO2 as a mask material. Although ε-Ga2O3 grainsnucleated around each α-Ga2O3 island when the windowspacing was wide, such extra grains were suppressed bydecreasing the precursor supply. It was found that the mor-phology of α-Ga2O3 islands was sensitive to the growthtemperature, and inclined facets developed well at low tem-peratures. XRC measurements showed that both tilt and twistangles decreased with increasing nominal growth thickness,which reflected the increase in the high-quality portion. Theout-of-plane XRC did not show peak splitting, which indi-cated no significant wing tilting and resulting small anglegrain boundaries, as can be observed in ELO-grown GaN.Cross-sectional TEM of a coalesced film visualized the dis-location bending, which should be caused by the inclinedfacets. Such dislocation bending is useful to reduce the dis-location density above the windows. Although some disloca-tions were observed just above the mask along the coalescedboundary, no dislocation was found in the vicinity of the sam-ple surface. These results show that the ELO technique byHVPE is promising to grow high-quality α-Ga2O3, which wouldlead to the realization of high-performance α-Ga2O3 powerdevices.Part of this work is based on the results obtained froma project commissioned by the New Energy and IndustrialTechnology Development Organization (NEDO).APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-5© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmAPL Materials ARTICLE scitation.org/journal/apmREFERENCES1R. Roy, V. G. Hill, and E. F. Osborn, J. Am. Chem. Soc. 74, 719(1952).2D. Shinohara and S. Fujita, Jpn. J. Appl. Phys., Part 1 47, 7311(2008).3Y. Oshima, E. G. Villora, and K. Shimamura, Appl. Phys. Express 8, 055501(2015).4M. Oda, R. Tokuda, H. Kambara, T. Tanikawa, T. Sasaki, and T. Hitora, Appl.Phys. Express 9, 021101 (2016).5K. Kaneko, S. Kan, T. Hitora, and S. Fujita, in The 64th JSAP Spring Meeting,16p-P8-19, Yokohama, Japan, 2017.6K. Kaneko, S. Fujita, and T. Hitora, Jpn. J. Appl. Phys., Part 2 57, 02CB18(2018).7A. Usui, H. Sunakawa, A. Sakai, and A. A. Yamaguchi, Jpn. J. Appl. Phys.,Part 2 36, L899 (1997).8Y. Oshima, T. Eri, M. Shibata, H. Sunakawa, K. Kobayashi, T. Ichihashi, andA. Usui, Jpn. J. Appl. Phys., Part 2 42, L1 (2003).9K. Motoki, T. Okahisa, N. Matsumoto, M. Matsushima, H. Kimura, H. Kasai,K. Takemoto, K. Uematsu, T. Hirano, M. Nakayama, S. Nakahata, M. Ueno,D. Hara, Y. Kumagai, A. Koukitu, and H. Seki, Jpn. J. Appl. Phys., Part 2 40,L140 (2001).10A. Takatsuka, M. Oda, K. Kaneko, S. Fujita, and T. Hitora, in The 62th JSAPSpring Meeting, 13a-P18-12, Hiratsuka, Japan, 2015.11P. Fini, C. Thompson, G. B. Stephenson, J. A. Eastman, M. V. Ramana Murty,O. Auciello, L. Zhao, S. P. Denbaars, and J. S. Speck, Appl. Phys. Lett. 76, 3893(2000).12A. Sakai, H. Sunakawa, and A. Usui, Appl. Phys. Lett. 73, 481 (1998).APL Mater. 7, 022503 (2019); doi: 10.1063/1.5051058 7, 022503-6© Author(s) 2018 29 January 2024 06:53:11https://scitation.org/journal/apmhttps://doi.org/10.1021/ja01123a039https://doi.org/10.1143/jjap.47.7311https://doi.org/10.7567/apex.8.055501https://doi.org/10.7567/apex.9.021101https://doi.org/10.7567/apex.9.021101https://doi.org/10.7567/jjap.57.02cb18https://doi.org/10.1143/jjap.36.l899https://doi.org/10.1143/jjap.36.l899https://doi.org/10.1143/jjap.42.l1https://doi.org/10.1143/jjap.40.l140https://doi.org/10.1063/1.126812https://doi.org/10.1063/1.121907