# Fileset

[Maskless_230605_clean.pdf](https://mdr.nims.go.jp/filesets/c44aeb04-3ad2-4f5d-919f-68263b41a43e/download)

## Creator

[Yuichi Oshima](https://orcid.org/0000-0001-8293-4891), Hiroyuki Ando, Takashi Shinohe

## Rights

© 2023 The Japan Society of Applied Physics<br>
This is an author-created, un-copyedited version of an article accepted for publication/published
in Applied Physics Express. IOP Publishing Ltd is not responsible for any errors or omissions in this version of the manuscript or
any version derived from it. The Version of Record is available online at https://doi.org/10.35848/1882-0786/acddca .[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Reduction of dislocation density in α-Ga2O3 epilayers via rapid growth at low temperatures by halide vapor phase epitaxy](https://mdr.nims.go.jp/datasets/89b0a845-c6a6-4ca9-a1c2-eb11ab66a858)

## Fulltext

Title of Paper Goes Here:  Template for APEX (Mar. 2022) 1 Reduction of dislocation density in -Ga2O3 epilayers via rapid growth at low temperatures by halide vapor phase epitaxy Yuichi Oshima1*, Hiroyuki Ando2, and Takashi Shinohe2 1Research Center for Electronic and Optical Materials, National Institute for Materials Science, Tsukuba, Ibaraki 305-0044, Japan 2FLOSFIA Inc., Kyodai-Katsura Venture Plaza, 615-8245 Kyoto, Japan E-mail: OSHIMA.Yuichi@nims.go.jp  We demonstrate that the dislocation density in -Ga2O3 epilayers is markedly reduced via rapid growth at low temperatures by halide vapor-phase epitaxy. An -Ga2O3 epilayer grown on (0001) sapphire at a high growth rate of 34 µm/h and a low temperature of 463 °C exhibited a dislocation density of 4  108 cm−2, which was approximately 1/100 of that in a conventional film. It is likely that the three-dimensional surface morphology developed during the growth enhanced the bending of the dislocations to increase the probability of pair annihilation. The combination of this technique with thick film growth and epitaxial lateral overgrowth resulted in a further low dislocation density of 1.1  107 cm−2.      Template for APEX (Mar. 2022) 2 Corundum-structured -Ga2O3 is a meta-stable phase of -Ga2O3.1) The thermal stability of -Ga2O3 depends on the film thickness and strain, and the transition temperature to -Ga2O3 is reported to be 500–650 °C.2,3) -Ga2O3 is an ultra-wide-bandgap semiconductor with an energy gap, Eg, of 5.2–5.3 eV,4,5) which is the widest among Ga2O3 polymorphs. Accordingly, a large critical field can be expected. In addition, -Ga2O3 forms solid solutions with other corundum-structured oxides. For example, it is possible to make -(AlxGa1−x)2O3 solid solutions without compositional limitation,6) in contrast with -(AlxGa1−x)2O3. Therefore, -Ga2O3 should allow a higher freedom of band engineering than -Ga2O3. Furthermore, -Ga2O3 can form a hetero-pn junction with isomorphic p-type oxides, such as -(GaxIr1−x)2O3, which virtually lattice matches to -Ga2O3.7,8) These features make -Ga2O3 an attractive candidate for power-device applications.9-12)  Despite these advantages, a major drawback of -Ga2O3 is the absence of melt-grown high-quality native substrates, in contrast to the case of -Ga2O3. Accordingly, -Ga2O3 films need to be grown on foreign substrates. Isomorphic sapphire is the most frequently used substrates4,5,13-15) because large-scale wafers are available at a reasonable price. The dislocation density in a heteroepitaxial -Ga2O3 film is typically of the order of 1010 cm−2 because of lattice mismatch if no measure is taken to reduce the dislocation density. For example, Kaneko et al. reported that the dislocation density in their mist-CVD-grown -Ga2O3 film was as high as 7  1010 cm−2.16) Such a high density of dislocations should degrade the electrical properties of -Ga2O3. Indeed, Son et al. experimentally confirmed that -Ga2O3 films with higher dislocation density exhibited lower electron mobility.17) According to theoretical prediction by Takane et al., the influence of dislocations on the electron mobility in -Ga2O3 should be negligible when the density is 107–108 cm−2 or less.18) Accordingly, these values should be a target for the reduction of the dislocation density.  Epitaxial lateral overgrowth (ELO) is one of the most frequently reported techniques to decrease the dislocation density in -Ga2O3.19-23) In the ELO technique, -Ga2O3 islands are selectively grown on the windows of a dielectric mask pattern formed on a seed -Ga2O3   Template for APEX (Mar. 2022) 3 layer (or directly on a sapphire substrate). Each -Ga2O3 island grows vertically and laterally, and the neighboring islands coalesce to form a compact film.21,23) The dislocation density on the laterally grown areas on the mask is typically as low as 107 cm−2 because dislocations in the seed layer do not propagate into those areas. However, the dislocation density on the window areas is still as high as that in the seed layer because the dislocations in the seed layer directly propagate into those areas.23) Thus, a limitation of ELO is that such low-quality areas remain. Although double-ELO process was reported to be effective to reduce the dislocation density on the window areas, the process was not cost-effective since the process required two rounds of photolithography and epitaxial growth.22)  The use of a super-lattice buffer layer was reported as a maskless method to uniformly reduce the dislocation density.24) In this technique, a mist-CVD-grown quasi-graded -(AlxGa1−x)2O3 super lattice buffer layer was inserted between the -Ga2O3 top layer and the sapphire substrate. As a result, the dislocation density decreased to the order of 108 cm−2. However, this technique requires a growth apparatus equipped with an additional gas line to supply the Al precursor, and the growth process is complicated.  In this paper, we demonstrate a maskless method to reduce the dislocation density in an -Ga2O3 epilayer via rapid growth at low temperatures by halide-vapor-phase epitaxy (HVPE). This technique uniformly reduces the dislocation density from 1010 to 108 cm−2 or less without using extra materials or a complicated process unlike the cases of the ELO or super-lattice buffer layers. One of the most advantageous features of HVPE is the rapid growth, with a very high growth rate of -Ga2O3 of over 100 µm/h having been reported.5,25) We therefore believe this technique is promising to grow thick drift layers for high-performance -Ga2O3-based power devices.  In the present work, we grew the -Ga2O3 epilayers in a lab-made HVPE reactor under atmospheric pressure using O2 and GaCly (y = 1, 3) as the precursors. GaCl was produced by chemical reaction of HCl and metal Ga at 570 °C upstream in the reactor. Additional HCl gas was also supplied to enable rapid growth by suppressing the parasitic gas-phase reaction owing to the conversion of part or all of the GaCl to GaCl3.25) The HCl partial pressure needs   Template for APEX (Mar. 2022) 4 to be increased with increasing the GaCl supply for rapid growth. N2 was used as the carrier gas. A (0001) patterned sapphire substrate (PSS) was used21) because the growth of an -Ga2O3 epilayer with a thickness of approximately 3 µm or more on a flat sapphire substrate (FSS) result in cracking or peeling off of the film.26) We note that the use of a PSS is not essential in our method from the viewpoint of crystal-quality improvement, and similar results can also be obtained on an FSS.  We investigated the effect of the growth temperature and growth rate on the dislocation density. The -Ga2O3 epilayers were prepared by three-step growth, as described in Table I. Step 1 was to nucleate -Ga2O3 on sapphire at our standard growth temperature of 520 °C with a relatively low growth rate. If the temperature in step 1 is too low, the growth results in an amorphous layer. Then, the growth was interrupted to change the temperature for steps 2 and 3. The duration of the interruption was set according to the temperature difference. Upon restarting the growth after the interruption, the HCl supply needed to be started in prior to the growth precursors to avoid gas-phase parasitic reaction. The HCl pre-flow should also work to etch the film surface grown at step 1. Our aim in step 2 was to perform the pre-flow with a low HCl partial pressure to suppress the etching as possible. If step 2 was skipped, the dislocation density in the subsequent layer grown at step 3 should increase due to the surface residuals produced as a result of rapid etching in the pre-flow process with a high HCl partial pressure, which is necessary for rapid growth. The layer formed in step 3 was the target of our investigation.  Table I. Growth conditions for the three-step growth. P denotes the partial pressure of the gas supply.  Step 1 Interruption Step 2 Step 3 P(GaCl) [kPa] 0.125 0 0.125 0.125 – 1.00 P(O2) [kPa] 1.25 1.25 1.25 1.25 – 2.50 P(HCl) [kPa] 0.188 0 0.188 0.188 – 1.25 Temperature [°C] 520 - 463 – 520 Growth rate [µm/h] 14 - 8.3 8.3 – 80 Thickness [µm] 0.5 - 0.5 10 – 200  After step 3, the sample was etched by HCl gas in the same HVPE reactor to visualize the dislocations as etch pits.23) The dislocation density was measured as the etch pit density (EPD), which was determined by counting the number of etch pits on four successive surface   Template for APEX (Mar. 2022) 5 areas of 8 µm  12 µm (approximately 384 µm2 in total) under SEM. Bright-field plan-view transmission electron microscopy (TEM) was also used to characterize a sample. Three successive surface areas of 2.2 µm  2.5 µm (approximately 16.5 µm2 in total) of a sample were observed by TEM to calculate the dislocation density. The surface morphology was observed by SEM. Bright-field cross-sectional TEM was used to investigate the behavior of the dislocations.  Figure 1(a) shows a plan-view SEM image of an -Ga2O3 film with an approximately 10-µm-thick third layer grown at a high growth rate of 34 µm/h and a low temperature of 463 °C. The surface exhibited a rough morphology, which is characteristic of rapid growth at low-temperature. The surface was HCl-gas-etched, and etch pits were visible as black dots. The etch pits were sufficiently isolated to determine the density. The EPD was 4  108 cm−2, which was approximately 1/100 of that in conventional -Ga2O3 films. Figure 1(b) shows a plan-view TEM image of the same -Ga2O3 film. The dislocation density was as low as 6  108 cm−2. Thus, the remarkable reduction in the dislocation density was confirmed although the TEM value was slightly higher than the EPD. We note that the EPD value should be more accurate than the TEM result because a much larger area was observed, as explained in the experimental details.  Fig. 1. (a) Plan-view SEM image and (b) plan-view TEM image of an -Ga2O3 epilayer grown at a growth rate of 34 µm/h and a temperature of 463 °C.  Figure 2(a) is a cross-sectional TEM image of the sample shown in Fig. 1, including the substrate boundary to the top surface. A steep decrease in the dislocation density was observed at a thickness of a few micrometers. The magnified image (Fig. 2(b)) revealed that the steep decrease took place in step 3. Bending of dislocations and dislocation loops were also observed. Figure 2(c–e) exhibits the surface morphologies at different growth stages. At   Template for APEX (Mar. 2022) 6 the end of step 1 (Fig. 2(e)), the surface was smooth except for a bump on the PSS. At the end of step 2 (Fig. 2(d)), surface roughening was observed. During step 3 (Fig. 2(c)), three-dimensional (3D) growth was further promoted. The 3D growth probably occurred because of the destabilization of the (0001) plane at low temperature19) triggered by the rapid growth, whereby surface migration of the growth species should be suppressed. In general, a dislocation line tends to be perpendicular to the free surface to minimize the elastic strain energy. Accordingly, a dislocation line bends to follow the surface when the surface angle changes. Such bending was reported for ELO-grown -Ga2O3 and GaN.19,27,28) The bending mechanism in the present work should be similar. It is likely that the bending increased the probability of pair annihilation, and resulted in the reduction in the dislocation density.  Fig. 2. (a) Cross-sectional TEM image of the -Ga2O3 epilayer shown in Fig. 1. (b) Magnified image of (a). (c, d) Bird’s-eye-view SEM image of the surface at each growth step.  To clarify the effect of growth temperature on the dislocation density, -Ga2O3 epilayers were grown at different temperatures with a fixed precursor supply in step 3. GaCl, O2, and HCl were supplied with partial pressures of 0.75, 2.5, and 1.25 kPa, respectively. The growth rate under these growth conditions was 52 µm/h at 463 °C. We note that the growth rate increased with increasing temperature, reaching 74 µm/h at 520 °C. This is probably because the growth was in the transient regime of mass-transport limited and reaction-limited growth. The growth time was set such that the thickness of the third layer was approximately 10 µm. Figure 3(a) shows the EPD as a function of growth temperature. Surface SEM images of selected samples are also shown. The EPD monotonically decreased with decreasing temperature. We note that the EPD of a sample grown at the standard temperature of 520 °C could not be calculated because the dislocation density was too high. The increase in the surface roughness at low temperature supports the reduction mechanism described above.   Template for APEX (Mar. 2022) 7  To clarify the effect of growth rate on the dislocation density, -Ga2O3 epilayers were grown at different growth rates at a fixed low temperature of 478 °C in step 3. The film thickness of the third layer was approximately 10 µm. Figure 3(b) shows the EPD as a function of growth rate. Surface SEM images of selected samples are also shown. The EPD was countable even at the lowest growth rate of 8.3 µm/h, at which it was 1.6  109 cm−2. The EPD further decreased at 37 µm/h but gradually increased at higher growth rates. The samples with smaller EPD exhibited rougher surfaces as expected. We note that the surface of the sample with a growth rate of 80 µm/h was smoother than that with a growth rate of 37 µm/h sample. We speculate that this was because of the increase in the effective growth temperature owing to the heat of reaction of GaCl and HCl to produce GaCl3.    Fig. 3. (a) Dislocation density in -Ga2O3 epilayers as a function of growth temperature. Plan-view SEM images of the selected samples are also shown. (b) Dislocation density in -Ga2O3 epilayers as a function of growth rate. Plan-view surface SEM images of the selected samples are also shown.  For GaN, the dislocation density is known to decrease during thick film growth even if the surface is smooth.29,30) This is because dislocations with Burgers vector of opposite signs gradually approach each other owing to the attractive force and finally react to form a dislocation loop. In the present work, thick film growth was combined with rapid growth at a low temperature to further decrease the dislocation density. In step 3 of the growth, GaCl, O2, and HCl were supplied with partial pressures of 1.00, 2.5, and 1.25 kPa, respectively, and the growth temperature was 463 °C. The growth rate was 70 µm/h under these growth   Template for APEX (Mar. 2022) 8 conditions. Figure 4 (a–c) shows SEM images of the etched surface of the thick -Ga2O3 epilayers. The EPD decreased with increasing thickness, reaching 1.5  108 cm−2 at 200 µm (Fig. 4(e)).   ELO was combined with thick film growth to further decrease the dislocation density. -Ga2O3 was grown at 463 °C on a 3-m-thick (0001) -Ga2O3 template with a striped TiOx mask pattern along [ 11̅00  . The widths of the mask and window were 5 and 5 µm, respectively. The details of the process are described elsewhere.21,23) The three-step growth was performed using the same gas-supply conditions as those for the samples shown in Fig. 4(a–c). The film thickness was 140 µm excluding the template layer. Figure 4(d) shows a plan-view SEM image of the etched sample. In contrast with the conventional ELO sample, high-dislocation density areas did not exist, although the surface still exhibited a rough morphology that originated from stripe coalescence. The EPD was as low as 1.1  107 cm−2. We attribute the good uniformity to the reduction of dislocation density on the windows via the rapid growth at low temperature, and the dispersion of the dislocations (including those newly formed at the coalesced boundaries) during the thick film growth after coalescence owing to the repulsive interaction between dislocations with Burgers vectors of the same sign. We note that the dislocation density on the window regions was much lower than those of the thick films with similar thickness. This is probably because the faceting was clearer as shown in Fig. 5. V-shaped facets were observed on the top of the stripes with high density, although the reason for such morphological difference from the plain films is under investigation. It is also likely that the V-shaped facets swept the surface like waves as the growth proceeds14) to further increase the possibility of dislocation bending.   Template for APEX (Mar. 2022) 9  Fig. 4. (a–c) Plan-view SEM images of -Ga2O3 epilayers with thicknesses of 35, 100, and 200 µm, respectively. (d) Plan-view SEM images of an -Ga2O3 epilayer prepared using the ELO technique with a thickness of 140 µm. (e) Dislocation density in the samples in (a–d) as a function of film thickness.   Fig. 5. Bird’s-eye-view SEM images of -Ga2O3 stripes grown using the same recipe with that for the sample shown in Fig. 1.   In summary, we demonstrated that rapid growth at low temperatures by HVPE effectively reduces the dislocation density in -Ga2O3 epilayers without leaving high-dislocation-density areas, in contrast with conventional ELO. We attribute the reduction of the dislocation density to the enhancement of pair annihilation by the bending triggered by the 3D growth. The dislocation density of an -Ga2O3 epilayer grown by this method was 4  108 cm−2, which was approximately 1/100 of that in a film grown without measures against dislocations. The combination of this method with ELO and thick film growth further reduced the dislocation density to 1.1  107 cm−2, to which the carrier scattering should be negligible even for a very low carrier concentration of 1  1015 cm−2.18)   Template for APEX (Mar. 2022) 10  Acknowledgement Part of this work was supported by the Innovative Science and Technology Initiative for Security (JPJ004596), ATLA, Japan.    References 1) R. Roy, V. G. Hill, and E. F. Osborn, J. Am. Chem. Soc. 74, 719 (1952). 2) S. Lee, Y. Ito, K. Kaneko, and S. Fujita, Jpn. J. Appl. Phys. 54, 030301 (2015). 3) R. Jinno, K. Kaneko, and S. Fujita, AIP Advances 10, 115013 (2020). 4) D. Shinohara and S. Fujita, Jpn. J. Appl. Phys. 47, 7311-7313 (2008). 5) Y. Oshima, E. G. Villora, and K. Shimamura, Appl. Phys. Express 8, 055501 (2015). 6) S. Fujita and K. Kaneko, J. Cryst. Growth 401, 588-592 (2014). 7) K. Kaneko, S. Fujita, and T. Hitora, Jpn. J. Appl. Phys. 57, 02CB18 (2018). 8) K. Kaneko, Y. Masuda, S. Kan, I. Takahashi, Y. Kato, T. Shinohe, and S. Fujita, Appl. Phys. Lett. 118, 102104 (2021). 9) M. Oda, R. Tokuda, H. Kambara, T. Tanikawa, T. Sasaki, and T. Hitora, Appl. Phys. Express 9, 021101 (2016). 10) News release from FLOSFIA and Kyoto university, July 13, 2018 (https://flosfia.com/struct/wp-content/uploads/79cd9d2dfa54a771f642e008cc4f9cb0.pdf). 11) Y. Jeong, J. Park, M. Yeom, I. Kang, J. Yang, H. Kim, D. Jeon, and G. Yoo, Appl. Phys. Express 15, 074001 (2022). 12) T. Shinohe, Proceedings of the 2022 International Power Electronics Conference (IPEC-Himeji 2022- ECCE Asia), Himeji, Japan, page 627-631 (2022). 13) Z. Cheng, M. Hanke, P. Vogt, O. Bierwagen, and A. Trampert, Appl. Phys. Lett. 111, 162104 (2017). 14) Y. Oshima, S. Yagyu and T. Shinohe, J. Appl. Phys. 130, 175304 (2021). 15) A. F. M. A. U. Bhuiyan, Z. Feng, H.-L. Huang, L. Meng, J. Hwang, and H. Zhao, APL Mater. 9, 101109 (2021). 16) K. Kaneko, S. Fujita, and T. Hitora, Jpn. J. Appl. Phys. 57 02CB18 (2018). 17) H. Son, Y. Choi, J. Park, B. Ryu, and D. Jeon, ECS Journal of Solid State Science and Technology 9, 055005 (2020). 18) H. Takane, H. Izumi, H. Hojo, T. Wakamatsu, K. Tanaka and K. Kaneko, J. Mater. Res.   Template for APEX (Mar. 2022) 11 38, 2645 (2023). 19) Y. Oshima, K. Kawara, T. Shinohe, T. Hitora, M. Kasu, and S. Fujita, APL Mater. 7, 022503 (2019). 20) R. Jinno, N. Yoshimura, K. Kaneko, and S. Fujita, Jpn. J. Appl. Phys. 58 120912 (2019). 21) Y. Oshima, K. Kawara, T. Oshima, M. Okigawa, and T. Shinohe, Jpn. J. Appl. Phys. 59 025512 (2020). 22) K. Kawara, Y. Oshima, M. Okigawa, and T. Shinohe, Appl. Phys. Express 13, 075507 (2020). 23) Y. Oshima, S. Yagyu, T. Shinohe, J. Cryst. Growth 576, 126387 (2021). 24) R. Jinno, T. Uchida, K. Kaneko, and S. Fujita, Appl. Phys. Express 9, 071101 (2016). 25) Y. Oshima, K. Kawara, T. Oshima, M. Okigawa, and T. Shinohe, Semicond. Sci. Technol. 35, 055022 (2020). 26) M. Oda, K. Kaneko, S. Fujita, and T. Hitora, Jpn. J. Appl. Phys. 55, 1202B4 (2016). 27) A. Usui, H. Sunakawa, A. Sakai and A. Yamaguchi, Jpn. J. Appl. Phys. 36, L899 (1997). 28) A. Sakai, H. Sunakawa and A. Usui, Appl. Phys. Lett. 73, 481 (1998). 29) K. Fujito, S. Kubo, H. Nagaoka, T. Mochizuki, H. Namita, S. Nagao, J. Cryst. Growth 311, 3011-3014 (2009). 30) H. Fujikura, T. Konno, T. Suzuki, T. Kitamura, T. Fujimoto, and T. Yoshida, Jpn. J. Appl. Phys. 57, 065502 (2018).