Growth temperature dependences of structural and electricalproperties of Ga2O3 epitaxial films grown on β-Ga2O3 (010)substrates by molecular beam epitaxy

Kohei Sasaki a,b,n, Masataka Higashiwaki b, Akito Kuramata a, Takekazu Masui c,Shigenobu Yamakoshi a

a Tamura Corporation, 2-3-1 Hirosedai, Sayama, Saitama 350-1328, Japanb National Institute of Information and Communications Technology, 4-2-1 Nukui-kitamachi, Koganei, Tokyo 184-8795, Japanc Koha Co., Ltd., 2-6-8 Kouyama, Nerima, Tokyo 176-0022, Japan

a r t i c l e i n f o

Article history:Received 25 October 2013Received in revised form24 January 2014Accepted 2 February 2014Communicated by: K.H. PloogAvailable online 8 February 2014

Keywords:A1. Crystal morphologyA1. DopingA3. Molecular beam epitaxyB2. Semiconducting gallium oxides

a b s t r a c t

We investigated the growth temperature dependence of the structural and electrical properties ofSn-doped Ga2O3 homoepitaxial films grown on single-crystal β-Ga2O3 (010) substrates by molecularbeam epitaxy. Ga2O3 films with an atomically smooth surface were obtained at growth temperatures of550–650 1C. On the other hand, a delay in the incorporation of Sn atoms in Ga2O3, which was probablydue to segregation, occurred in the initial stage of growth at higher than 600 1C. To ensure that Sn-dopedGa2O3 films with both high crystal quality and accurately controlled carrier density are obtained, theoptimum growth temperature should be set at 540–570 1C.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

β-Gallium oxide (β-Ga2O3) will be useful as a next-generationhigh-power device material because of its excellent materialproperties and ease of mass production. β-Ga2O3 has a large bandgap of 4.8–4.9 eV [1]. The breakdown electric field of β-Ga2O3 isexpected to be 8 MV/cm, as extrapolated from the relationbetween the band gaps and breakdown electric fields of othersemiconductors [2]. The electron concentration (n) in Ga2O3 can becontrolled over a wide range from 1016–1019 cm�3 by Si or Sndoping [3–6]. Moreover, semi-insulating substrates with resistiv-ities of about 1012 Ω cm can be fabricated by Mg or Fe doping [2,7].The electron mobility is estimated to be around 300 cm2/(V s) forn¼1015–1016 cm�3. Baliga's figure of merit for β-Ga2O3 is severaltimes larger than that of 4H–SiC or GaN [8]. Another importantfeature of β-Ga2O3 is that large single-crystal substrates can befabricated by using melt-growth methods under atmosphericpressure. In the past, β-Ga2O3 bulk crystals have been grown withthe floating zone (FZ) [3–5], Czochralski [9], and edge-definedfilm-fed growth (EFG) methods [10,11]. The melt-growth methods

enable us to produce large high-quality Ga2O3 substrates at a lowcost. This is a big advantage of β-Ga2O3 over SiC, GaN, anddiamond substrates. In fact, 2-inch-diameter single-crystalβ-Ga2O3 wafers have already been manufactured in our group byusing the EFG method.

Until now, Ga2O3 has been mainly studied for application tooptical devices such as transparent conductive substrates for InGaNLEDs and deep-UV photodetectors [12,13]. Power-device applica-tions require not only high-quality substrates but also the thickhomoepitaxial films, in which the carrier concentration is preciselycontrolled. There have been a few reports on epitaxial growth ofGa2O3 films on β-Ga2O3 (100) substrates by molecular beam epitaxy(MBE) [14,15]. These films had atomically flat surfaces because ofthe step-flow growth mode. However, there was a fatal problemwith the MBE growth on the (100) plane in that the growth ratewas only a few tens of nanometers per hour. To overcome thisproblem, we investigated the relation between the growth rate andthe Ga2O3 substrate orientation and succeeded in increasing thegrowth rate to several micrometers per hour by changing theorientation from (100) to (010) [6]. Furthermore, the use of Ga2O3

(010) substrates enabled us to precisely control the n-type con-ductivity of the MBE-grown Ga2O3 films by using the Sn dopingtechnique, which subsequently led to our building the Ga2O3

Schottky barrier diodes and transistors [2,16,17]. The devices

Contents lists available at ScienceDirect

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

Journal of Crystal Growth

http://dx.doi.org/10.1016/j.jcrysgro.2014.02.0020022-0248 & 2014 Elsevier B.V. All rights reserved.

n Corresponding author. Tel.: þ81 4 2900 0045; fax: þ81 4 2900 0059.E-mail address: kohei.sasaki@tamura-ss.co.jp (K. Sasaki).

Journal of Crystal Growth 392 (2014) 30–33

exhibited good enough performance for practical power-deviceapplications. However, there was still room for improvement inthe uniformity of the n-type dopant concentration in the MBE-grown Ga2O3 films, and an atomically smooth and flat surface and auniform doping profile are the keys to making further improve-ments to Ga2O3 electrical devices. In this work, therefore, westudied growth temperature (Tg) dependences of the structuraland electrical properties of Ga2O3 epitaxial films grown on β-Ga2O3

(010) substrates by MBE.

2. Experimental

We used Si-doped n-type β-Ga2O3 (010) substrates fabricatedfrom an FZ bulk crystal. It was cut into pieces along the (010)plane, and chemical mechanical polishing (CMP) was performedon the front side. The surface after the CMP process was atomicallyflat, with a root mean square (RMS) roughness of less than 0.2 nmin a 1�1 μm2 [18]. The Ga2O3 (010) substrates were about 10 mmin diameter and 650 μm in thickness. Prior to loading the sub-strates into an MBE growth chamber, they were cleaned withorganic solvent (acetone and methanol), acid [HF (46%) andH2SO4þH2O2], and ultrapure water, and then bonded to a Si waferwith In metal. The Tg was taken to be the temperature of the Siwafer measured through the Ga2O3 substrate by a pyrometer. Gaand Sn beam fluxes were supplied from 7 N Ga metal and 4 N SnO2

powder heated in normal Knudsen cells (K-cells). Note that the Snbeam flux would have become unstable if we had used Sn metal asa source because its surface oxidizes during growth. The Ga beamequivalent pressure was fixed at 2.1�10�4 Pa. The oxygen sourcewas an ozone (5%)–oxygen(95%) gas mixture. Note that Ga2O3 filmgrows very little when only pure oxygen gas is supplied. The flowrate of the ozone–oxygen gas mixture was 5 sccm. The growth ratewas 0.5–0.6 μm/h. The surface morphologies of the epitaxial filmswere evaluated by reflection high-energy electron diffraction(RHEED) and atomic force microscopy (AFM). Thicknesses andeffective donor concentrations (Nd�Na) of the Ga2O3 epitaxialfilms were estimated by making electrochemical capacitance–voltage (ECV) measurements using a dielectric constant of 10 forGa2O3 as a parameter [19].

3. Results and discussion

Fig. 1 shows RHEED patterns of unintentionally doped Ga2O3

epitaxial films grown at various Tg. The incidence direction of theelectron beam is perpendicular to the [001] direction. The growthtime was set at 1 h (corresponding film thickness of 0.5–0.6 μm).The RHEED pattern from the film grown at Tg¼500 1C was spotty.On the other hand, a clear streak pattern was observed for all filmsgrown at higher than Tg¼550 1C, indicating that the Tg rangegiving an atomically flat surface was comparatively wide.

Fig. 1. RHEED patterns of Ga2O3 epitaxial films grown at various Tg.

Fig. 2. Surface AFM images of Ga2O3 epitaxial films grown at various Tg.

K. Sasaki et al. / Journal of Crystal Growth 392 (2014) 30–33 31

Fig. 2 compares the surface morphologies of the unintention-ally doped Ga2O3 films grown at several Tg, as observed by AFM.The sample grown at 500 1C had a rough surface covered withsmall grains resulting from three-dimensional growth. In contrast,the samples grown at Tg¼550–650 1C had a smooth surface.Multi-atomic steps along the [100] direction that were due to stepbunching growth were observed in the samples grown atTg4700 1C, and the trend became worse with increasing Tg.Fig. 3 plots the surface RMS roughnesses of the Ga2O3 epitaxialfilms as a function of Tg (the roughness values were obtained fromthe AFM images). The surface RMS roughness was as small as0.4 nm for the sample grown at Tg¼600 1C.

Fig. 4 plots the Tg dependence of the growth rate of theunintentionally doped Ga2O3 epitaxial film. The growth ratedecreased a little above Tg¼700 1C. This behavior also occursduring GaAs and GaN MBE growth, and it is due to re-evaporation of Ga metal from the growth surface. During typicalGaAs MBE growth, Ga re-evaporation becomes significant atTg4700 1C; as a result, the growth rate at about Tg¼800 1C dropsto less than half compared with the value at Tgo700 1C [20]. Onthe other hand, during Ga2O3 MBE growth, the growth ratedecreased about 25%, even at Tg¼800 1C. Thus, it turned out thatthe Tg dependence of the growth rate is relatively small.

Fig. 5 shows X-ray diffraction (XRD) (020) rocking curve peaksfrom unintentionally doped Ga2O3 epitaxial films grown atTg¼500–800 1C, together with a peak from a Ga2O3 (010) FZsubstrate. The sample grown at Tg¼500 1C has a wider peak thanthose of the others. On the other hand, all the samples grown aboveTg¼550 1C show the same pattern as that from the substrate. Notethat the Ga2O3 substrates used in this study contained high-density

twin defects and low-angle grain boundaries because Ga2O3 sub-strates are in the developmental stage; as a result, the (020) rockingcurve peaks were spread into multiple peaks. From these results,we consider that fatal crystalline degradation of Ga2O3 epitaxialfilms can be avoided by conducting MBE growth at Tg4550 1C.

Fig. 6(a) shows depth profiles of the Nd�Na of Sn-dopedepitaxial films grown at Tg¼540, 570, and 600 1C, as estimatedfrom ECV measurements. The growth time was set at 30 min(corresponding to a film thickness of about 0.3 μm). A doping

Fig. 3. Surface RMS roughnesses of Ga2O3 epitaxial films as a function of Tg.

Fig. 4. Growth rates of Ga2O3 epitaxial films as a function of Tg.

Fig. 5. (020) XRD rocking curves from Ga2O3 epitaxial films grown at varioustemperatures and Ga2O3 (010) FZ substrate.

Fig. 6. (a) Nd�Na depth profiles of Sn-doped Ga2O3 epitaxial films grown at Tg¼540,570, and 600 1C. (b) Nd�Na in Ga2O3 epitaxial films grown at Tg¼540 and 570 1C as afunction of SnO2 K-cell temperature. The dashed line in Fig. 6(b) corresponds to theSnO2 vapor pressure.

K. Sasaki et al. / Journal of Crystal Growth 392 (2014) 30–3332

delay occurred in the initial stage of growth in the sample grownat Tg¼600 1C; this can be attributed to Sn atom segregation. Adelay in Sn incorporation into the film also occurs in GaAs MBEgrowth [21]. Therefore, it is likely due to an intrinsic property of Snatoms. Moreover, it is found that a doping delay can be reduced bydecreasing Tg to 570 1C or less. The Nd�Na of the films grown atTg¼540 and 570 1C are plotted as a function of the SnO2 K-celltemperature, together with the SnO2 vapor pressure (dashed line)in Fig. 6(b). The Nd�Na corresponded to the SnO2 vapor pressure,indicating that the donor concentration in the MBE-grown filmscould be precisely controlled with the SnO2 K-cell temperature.

4. Summary

In conclusion, we optimized the MBE growth temperature ofSn-doped β-Ga2O3 homoepitaxial films in order to control theirstructural and electrical properties. We obtained an atomicallysmooth and flat surface for the samples grown in the range ofTg¼550–650 1C. A doping delay occurred in the initial stage ofgrowth in the sample grown at Tg¼600 1C, and it turned out that itcan be reduced by decreasing Tg to 570 1C or less. These resultsindicate that high-quality Ga2O3 epitaxial films with a smoothsurface and uniform doping profile can be obtained in a narrow Tgrange of 550–570 1C.

Acknowledgments

Part of this work was supported by the New Energy andIndustrial Technology Development Organization (NEDO), Japan.

References

[1] H.H. Tippins, Phys. Rev. 140 (1965) A316.[2] M. Higashiwaki, K. Sasaki, A. Kuramata, T. Masui, S. Yamakoshi, Appl. Phys.

Lett. 100 (2012) 013504.[3] N. Ueda, H. Hosono, R. Waseda, H. Kawazoe, Appl. Phys. Lett. 70 (1997) 3561.[4] N. Suzuki, S. Ohira, M. Tanaka, T. Sugawara, K. Nakajima, T. Shishido, Phys.

Status Solidi C 4 (2007) 2310.[5] E.G. Víllora, K. Shimamura, Y. Yoshikawa, T. Ujiie, K. Aoki, Appl. Phys. Lett. 92

(2008) 202120.[6] K. Sasaki, A. Kuramata, T. Masui, E.G. Víllora, K. Shimamura, S. Yamakoshi,

Appl. Phys. Express 5 (2012) 035502.[7] K. Sasaki, M. Higashiwaki, A. Kuramata, T. Masui, S. Yamakoshi, Appl. Phys.

Express 6 (2013) 086502.[8] B. Jayant Baliga, IEEE Electron Device Lett. 10 (1989) 455.[9] Y. Tomm, P. Reiche, D. Klimm, T. Fukuda, J. Cryst. Growth 220 (2000) 510.[10] E.G. Villora, K. Shimamura, Y. Yoshikawa, K. Aoki, N. Ichinose, J. Cryst. Growth

270 (2004) 420.[11] H. Aida, K. Nishiguchi, H. Takeda, N. Aota, K. Sunakawa, Y. Yaguchi, Jpn. J. Appl.

Phys. 47 (2008) 8506.[12] K. Shimamura, E.G. Víllora, K. Domen, K. Yui, K. Aoki, N. Ichinose, Jpn. J. Appl.

Phys. 44 (2005) L7.[13] T. Oshima, T. Okuno, N. Arai, N. Suzuki, S. Ohira, S. Fujita, Appl. Phys. Express 1

(2008) 011202.[14] E.G. Víllora, K. Shimamura, K. Kitamura, K. Aoki, Appl. Phys. Lett. 88 (2006)

031105.[15] T. Oshima, N. Arai, N. Suzuki, S. Ohira, S. Fujita, Thin Solid Films 516 (2008)

5768.[16] K. Sasaki, M. Higashiwaki, A. Kuramata, T. Masui, S. Yamakoshi, J. Cryst. Growth

378 (2013) 591.[17] M. Higashiwaki, K. Sasaki, T. Kamimura, M.H. Wong, D. Krishnamurthy,

A. Kuramata, T. Masui, S. Yamakoshi, Appl. Phys. Lett. 103 (2013) 123511.[18] K. Sasaki, M. Higashiwaki, A. Kuramata, T. Masui, S. Yamakoshi, IEEE Electron

Device Lett. 34 (2013) 493.[19] M. Passlack, N.E.J. Hunt, E.F. Schubert, G.J. Zydzik, M. Hong, J.P. Mannaerts,

R.L. Opila, R.J. Fischer, Appl. Phys. Lett. 64 (1994) 2715.[20] T. Suzuki, I. Ichimura, T. Nishinaga, Jpn. J. Appl. Phys. 30 (1991) L1612.[21] A.Y. Cho, J. Appl. Phys 46 (1975) 1733.

K. Sasaki et al. / Journal of Crystal Growth 392 (2014) 30–33 33