Surface morphology of GaSb grown on (1 1 1)B GaAs by molecular beam epitaxy

Surface morphology of GaSb grown on (1 1 1)B GaAs by molecular beam epitaxy

Journal of Crystal Growth 203 (1999) 297}301 Surface morphology of GaSb grown on (1 1 1)B GaAs by molecular beam epitaxy E. Hall *, H. Kroemer Mate...

616KB Sizes 2 Downloads 55 Views

Journal of Crystal Growth 203 (1999) 297}301

Surface morphology of GaSb grown on (1 1 1)B GaAs by molecular beam epitaxy E. Hall *, H. Kroemer Materials Department, University of California at Santa Barbara, Santa Barbara, CA 93106, USA ECE Department, University of California at Santa Barbara, Santa Barbara, CA 93106, USA Received 21 November 1998; accepted 1 March 1999 Communicated by A.Y. Cho

Abstract The growth by molecular beam epitaxy of GaSb on (1 1 1)B GaAs surfaces is examined. Both thin nucleation layers and thick bulk layers were grown using on-axis and misoriented substrates by varying growth temperature and antimony overpressure. The resulting surface structures were studied by atomic force microscopy, revealing for the best layers } grown on the misoriented substrates } a stepped morphology displaying a very regular period.  1999 Elsevier Science B.V. All rights reserved. PACS: 81.15.H; 61.16.C Keywords: Molecular beam epitaxy; GaSb

There has been extensive research into the epitaxial growth of GaSb and AlSb, which } along with InAs } forms a nearly lattice-matched system. These compounds are useful as the basis for both high-speed devices [1] and long-wavelength optics [2]. Similarly, despite the greater prevalence of the (1 0 0) surface of GaAs as a substrate for epitaxial growth, the (1 1 1) orientation has still received considerable attention since it has its own advantages. For example, the valence band structure of (1 1 1)GaAs has led to quantum well lasers with reduced threshold currents [3], and use of the

* Corresponding author. Fax: #1-805-893-8971. E-mail address: [email protected] (E. Hall)

piezo-electric e!ect } which becomes a maximum along this orientation [4] } has been used to introduce carriers into a quantum well without intentional doping in the barrier [5]. This orientation is even expected to reduce the undoped carrier density of GaSb grown by molecular beam epitaxy (MBE) [6]. Very little work, however, has been done examining the growth of the antimonide compounds on the (1 1 1) face, especially using GaAs substrates, which are often the substrate of default for such antimonide-based devices. Such growth requires good nucleation layers to accomodate the large lattice mismatch between (Al,Ga)Sb and GaAs (&7%) as well as high-quality, thick ('1 lm) buffer layers to attenuate the large density of dislocations that form at the highly mismatched interface.

0022-0248/99/$ - see front matter  1999 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 9 ) 0 0 1 3 2 - 3

298

E. Hall, H. Kroemer / Journal of Crystal Growth 203 (1999) 297}301

In the present paper we focus on the growth of GaSb on (1 1 1)B GaAs substrates, examining both the surface morphology of thin (&200 nm) nucleation layers and the development of this morphology as the thickness of these layers is extended to '1 lm. Both substrate temperature and antimony #ux were varied to study their e!ects on morphology. In addition, the unintentional background doping of the thick layers was measured to examine one of the possible advantages of this growth orientation. Samples were grown by MBE using a standard Varian GEN-II equipped with an EPI valved antimony cracker source to supply a Sb #ux [7]. Both  on-axis GaAs(1 1 1)B substrates and substrates nominally misoriented 1.53 towards [2 1 1 ] were used. Following thermal oxide desorption, approximately 200 nm of GaAs were grown to obtain a smooth starting surface for the subsequent antimonide layers. The substrate temperature and arsenic overpressure for these smoothing layers were set so that growth occurred at the high-temperature end of the (19;(19 RHEED reconstruction window [8], resulting in starting layers with a roughness of &1.5 nm (rms) as measured by atomic force microscopy (AFM) across a 2 lm; 2 lm area. The surface morphology of the subsequent GaSb layers was investigated with both Nomarski microscopy and AFM. Mobility and carrier concentration measurements were made using a standard van der Pauw geometry. Nucleation using on-axis wafers was plagued by macroscopic, pyramid-like features that gave the surface a grainy appearance. Such features were present across a substantial range of growth conditions but did appear, however, on any of the samples grown using the misoriented substrates, a result consistent with that of other authors [6,9]. The morphology results discussed in the remainder of this letter, therefore, were obtained using the misoriented substrates. An AFM micrograph of the best 200 nm nucleation layer is shown in Fig. 1a. This sample was 䉴 Fig. 1. AFM micrographs: (a) 8 lm;8 lm, (b) 2 lm;2 lm of 200 nm GaSb layer grown at optimal conditions on misoriented (1 1 1)B GaAs substrate. (c) Pro"le of the morphology along the dotted line.

E. Hall, H. Kroemer / Journal of Crystal Growth 203 (1999) 297}301

grown at a substrate temperature of ¹ "5003C 1 and a #ux ratio Sb /Ga"1.30R , where R is the    #ux ratio at the boundary between the Group III (Ga) and Group V (Sb) stabilizations [10]. The surface of this layer displays a fairly regular array of steps that form a &303 angle to the nominal miscut direction of the wafer, indicating a discrepancy in the wafer manufacturer's speci"cations. A closer view of this surface (Fig. 1b) reveals several interesting details. The angled slope of the steps, e.g. is mostly composed of short, jagged segments, which primarily have edges perpendicular to the [2 1 1 ] axis (the nominal miscut direction of the wafer), but also form edges perpendicular to both the [1 1 0] and [1 2 1] axes to accommodate the discrepancy in the miscut direction. The formation of [2 1 1 ] edges is consistent with the results of Yang et al. [11] the formation of the other edges, however, was not observed by those authors. In this closer view, there is also an indication of step bunching on this sample. Each of the `largea steps visible in the wider view corresponds to an elevation change of 3}5 GaSb bilayers (1 bilayer" a/(3+3.52 As ), and, within each of these large steps, several small steps } one bilayer high } can be seen. Although the single steps consist primarily of [2 1 1 ] edges, all three edges can be seen on both types of steps. Since the peak-to-valley height of the large steps is &2 nm, the roughness scale of this thin layer is less than that obtained by using thick GaSb bu!er layers on (1 0 0)GaAs which are characterized by &4 nm spiral growth mounds [12]. The growth window for this good nucleation layer is fairly narrow. A sample grown at ¹ " 1 4603C showed a more mottled appearance, consisting of short but rough steps (Fig. 2a). This layer was also marked by long, straight defects: possibly dislocations or stacking faults. A sample grown at ¹ "5403C showed an increase in the step 1 bunching, producing very long, smooth plateaus. However, the sharp cli!s that marked the edge of these plateaus resulted in a roughness that was signi"cantly worse than for the sample grown at the lower temperature (Fig. 2b). Similar bunching was found when the Sb/Ga ratio was lowered to Sb/Ga"1.05R (at ¹ "5003C), but increasing  1 the ratio to Sb/Ga"1.55R showed no e!ect on 

299

Fig. 2. AFM micrographs (2 lm;2 lm) of 200 nm GaSb grown at: (a) 4603C and (b) 5403C.

surface morphology compared to the Sb/Ga" 1.30R sample.  In order to study the evolution, with increasing layer thickness, of the surface morphology seen above, a 1 lm GaSb layer was grown at the optimal

300

E. Hall, H. Kroemer / Journal of Crystal Growth 203 (1999) 297}301

nucleation conditions previously de"ned. The wide-area AFM micrograph of this layer (Fig. 3a) shows a strikingly regular pattern as the step widths have reached an equilibrium value. Although a few straight defects } similar to those seen for lower temperature samples } appear near the bottom of the picture, the majority of this stepped surface is quite smooth, corresponding to a roughness of &0.8 nm (rms). Again, a closer view of this surface reveals several interesting details (Fig. 3b). The larger steps are seen to be composed } on average } of four singlebilayer steps of width &25 nm. Since each of the larger steps correspond to about a "ve-bilayer drop, we calculate a &1.43 miscut from the (1 1 1)B surface, which is consistent with the manufacturer's speci"cations. Subsequent growths using slight variations in temperature did not alter this equilibrium step width. The step edge character also shows a striking development. The [1 1 0] and [1 2 1] edges are now more prevalent on the large steps, accommodating most of the miscut discrepancy on these steps. The single bilayer steps within each of these larger steps, therefore, have mostly [2 1 1 ] edge character. The background doping of this thick layer was measured using a standard van der Pauw geometry. The layer was found to be p-type with a hole concentration p+1.2;10/cm at room temperature, slightly lower than similar layers grown in our laboratory on (1 0 0) surfaces (p+ 2.1;10/cm). In conclusion, we have demonstrated the growth of smooth, highly-regular GaSb bu!er layers on misoriented (1 1 1)B GaAs substrates. We feel these layers are a suitable template for the extension of antimonide-based devices to the (1 1 1)B surface, allowing the bene"ts of this orientation to be used in these devices. This work made use of the MRL Central Facilities supported by NSF under Award No. DMR 91-23048. 䉳 Fig. 3. AFM micrographs: (a) 8 lm;8 lm, (b) 2 lm;2 lm of 1 lm GaSb layer grown at optimal conditions. (c) Pro"le of the morphology along the dotted line.

E. Hall, H. Kroemer / Journal of Crystal Growth 203 (1999) 297}301

References [1] B. Brar, H. Kroemer, IEEE Electron Dev. Lett. 16 (1995) 548. [2] H.K. Choi, G.W. Turner, Z.L. Liau, Appl. Phys. Lett. 65 (1994) 2251. [3] T. Hayakawa, M. Kondo, T. Suyama, K. Takahashi, S. Yamamoto, T. Hijikata, Jpn. J. Appl. Phys. 26 (1987) L302. [4] D.L. Smith, C. Mailhiot, Rev. Mod. Phys. 62 (1990) 173. [5] E.S. Snow, B.V. Shanabrook, D.C. Gammon, Appl. Phys. Lett. 56 (1990) 758.

301

[6] K.F. Longenbach, W.I. Wang, Appl. Phys. Lett. 59 (1991) 1117. [7] E. Hall, H.-R. Blank, J.E. English, R. Naone, J.C. Champlain, H. Kroemer, J. Vac. Sci. Technol. B 16 (1998) 2660. [8] P. Chen, K.C. Rajkumar, A. Madhukar, J. Vac. Sci. Technol. B 9 (1991) 2312. [9] G.P. Schwartz, G.J. Gualtieri, W.A. Sunder, J. Crystal Growth 102 (1990) 147. [10] B. Hancock, H. Kroemer, J. Appl. Phys. 55 (1984) 4239. [11] K. Yang, L.J. Schowalter, B.K. Laurich, I.H. Campbell, D.L. Smith, J. Vac. Sci. Technol. B 11 (1993) 779. [12] B. Brar, D. Leonard, Appl. Phys. Lett. 66 (1995) 463.