Journal of Crystal Growth 195 (1998) 34—40 An X-ray standing wave study of ultrathin InAs films in GaAs(0 0 1) grown by atomic layer epitaxy J.A. Gupta , J.C. Woicik, S.P. Watkins *, K.E. Miyano, J.G. Pellegrino, E.D. Crozier Department of Physics, Simon Fraser University, Burnaby, BC, Canada V5A 1S6 National Institute of Standards and Technology, Gaithersburg, MD 20899, USA Department of Physics, Brooklyn College, New York, NY 11210, USA Abstract X-ray standing wave and X-ray diffraction measurements were used to determine the structure of nominal 1 monolayer and 1/2 monolayer InAs films buried in GaAs(0 0 1). The films were grown by atomic layer epitaxy using trimethylgallium, tertiarybutylarsine and trimethylindium. For the full monolayer sample the standing wave measurement shows that the indium atoms reside 1.577$0.014 A> above the GaAs(0 0 4) substrate planes. A calculation based on the macroscopic elastic theory suggests that this corresponds to a single In Ga As layer with x"0.794$0.068. The coherent fraction V \V of 0.766$0.051 indicates a reasonably abrupt interface, as confirmed by the In-excitonic photoluminescence full width at half maximum of 5.74$0.01 meV. The half monolayer sample is less strained, as expected, with the indium atoms at 1.502$0.030 A> above the substrate planes, corresponding to an In Ga As layer with x"0.446$0.145, and a V \V coherent fraction of 0.88$0.12. This study exemplifies the complimentary nature of XSW and XRD. 1998 Elsevier Science B.V. All rights reserved. PACS: 81.15Gh; 81.05.Ea; 68.65.#g; 61.10.Yh Keywords: InAs; GaAs; XSW; ALE; MOCVD; Diffraction 1. Introduction Advances in semiconductor growth techniques such as metalorganic chemical vapour deposition (MOCVD), the related technique of atomic layer * Corresponding author. Tel.: #1 604 291 5763; fax: #1 604 291 3592; e-mail: [email protected]. epitaxy (ALE), and molecular beam epitaxy (MBE) have allowed the growth and study of indium isoelectronic d-doping layers in GaAs(0 0 1). With In depositions of only fractions of a monolayer (ML), sharp and intense In-excitonic photoluminescence (PL) emission has been observed [1—5]. However, for these InAs layers buried beneath a GaAs capping layer, the determination of the In local structure is a non-trivial task. For 0022-0248/98/$ — see front matter 1998 Elsevier Science B.V. All rights reserved. PII: S 0 0 2 2 - 0 2 4 8 ( 9 8 ) 0 0 6 3 6 - 8 J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 submonolayers of InAs, there has been much work aimed at determining whether zero-dimensional or one-dimensional confinement occurs at InAs islands or wires [1—3]. There remains considerable interest in determining the effect of the InAs distribution on the optical properties. The synchrotron X-ray standing wave (XSW) technique [6] has previously been applied in the study of InAs monolayers embedded in GaAs grown by MBE. From the standing wave data it is possible to determine the location of the In atoms relative to the GaAs(0 0 4) planes, and to obtain a precise measurement of the InAs layer perfection (i.e. the d-like nature of the distribution). Results with MBE-grown films [7—10] have been consistent with predictions of the macroscopic elastic theory (MET) [11]. Here the lattice mismatch between InAs and GaAs (&7.2%) is accommodated by a tetragonal distortion of the InAs unit cell, resulting from pseudomorphic growth. In this study we have measured the strain in nominal 1 and 1/2 ML films of InAs in GaAs grown by the gas phase technique of ALE. X-ray standing waves were used to measure the In position relative to the underlying GaAs substrate unit cell, and the perfection of the layers. From the positional measurement and the MET we have deduced an In atomic distribution in the plane of each sample, which we find to be consistent with the results of X-ray diffraction (XRD) and PL measurements. 2. Experiment The samples were grown in a Thomas Swan vertical MOCVD reactor at 50 Torr. Semi-insulating vertical gradient freeze GaAs(0 0 1) epi-ready substrates were used with no additional cleaning. Following annealing under tertiarybutylarsine (TBAs) at 580°C for 300 s, a GaAs buffer layer of thickness 0.625 lm was grown by conventional MOCVD using triethylgallium (TEGa) and TBAs. ALE growth was performed at 390°C. GaAs barrier layers (of thickness &1.3 ML) above and below the InAs layer were grown by ALE in four 4-step cycles each consisting of a 3 s hydrogen purge, 8 s of trimethylgallium (TMGa), another 3 s hydrogen 35 purge, and finally 6 s of TBAs. The InAs layer was deposited in a single cycle as above, with a 3 or 1.5 s trimethylindium pulse, for the single and half monolayers, respectively. The InAs growth rate was based on XRD thickness measurements of InAs homoepitaxial films deposited by ALE [12]. After the top GaAs ALE layer, the temperature was raised to 500°C and a thicker GaAs layer was deposited by conventional MOCVD using TEGa and TBAs. This growth sequence was also employed in a previous PL study of samples with several InAs coverages [4]. In the present study three samples were produced. For the XRD and XSW measurements, two samples were grown with nominal In depositions of 1 and 1/2 ML, and GaAs caps (ALE#MOCVD) of approximately 100 A> . Actual cap thicknesses were determined by XRD, as will be discussed. A third sample was grown for PL characterization with 1 ML of InAs and a total GaAs cap thickness of 2550$25 A> , also determined by XRD. During each growth the sample surface was monitored by reflectance difference spectroscopy using the arrangement described elsewhere [13]. XSW measurements were made at the National Institute of Standards and Technology beamline X24A at the National Synchrotron Light Source, Brookhaven National Laboratory. The monochromator consisted of a pair of Si(2 2 0) crystals. Standing waves were produced within the GaAs crystals and modulated in phase by scanning in energy through the GaAs(0 0 4) back-reflection at normal incidence. Indium fluorescence yield data were obtained using a single-element Li-drifted Ge detector. Reflectivity measurements were made by recording the back-reflected photon intensity with a Ni mesh upstream from the sample. X-ray diffraction measurements were obtained in a standard h/2h geometry with a BEDE-D3 apparatus. An asymmetric double crystal channel-cut monochromator was used with Cu K X-rays of a wavelength j"1.54 A> . Rocking-curve measurements were obtained by scanning in angle, h, through the GaAs(0 0 4) symmetric reflection, near the Bragg angle, h , given by j"2d sin h , where 1 d is the spacing of the GaAs(0 0 4) planes. 1 Low-temperature photoluminescence measurements were performed at 4.2 K using the 514.5 nm 36 J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 line of an Ar-ion laser. Spectra were recorded using a double-grating 0.85 m spectrometer and GaAs photocathode photomultiplier tube. 3. Results and discussion In the XSW experiments, the X-ray reflectivity and resulting X-ray standing wave pattern are recorded simultaneously [6,14] as shown in Figs. 1 and 2. In each case the bottom curve depicts the reflectivity results obtained as the incident X-ray energy is scanned through the GaAs(0 0 4) backreflection condition. The standing wave pattern results from the superposition of the incident and back-reflected travelling waves and has the periodicity of the crystal lattice planes with diffraction vector, H. At the low energy side of the Bragg Fig. 1. Photon-energy dependence of the reflectivity (lower) and the In-L fluorescence yield (upper) near the GaAs(0 0 4) Bragg backreflection condition for the nominally 1 ML sample. The solid lines are the best fits to the data points. reflection the nodes of the standing wave lie on the substrate planes. As the phase changes in the region of the Bragg peak the wavefield moves continuously until the antinodes lie on the substrate planes at the high energy side. Indium fluorescence spectra, shown in the top curves of Figs. 1 and 2, are proportional to the absorption of the In atoms in the field of the standing wave. From this response the spatial distribution of the In atoms may be deduced. The indium fluorescence yield from the standing wave measurement is written [6] in terms of the coherent fraction, F&, and the coherent position, P&, which measures the position of the indium atoms relative to the substrate planes, normalized by the diffraction plane spacing, d . (For a single atom in bulk P&"1.) The yield, ½&, is ½&"½(1#R#2(RF& cos(a!2pP&)). (1) Fig. 2. Photon-energy dependence of the reflectivity (lower) and the In-L fluorescence yield (upper) near the GaAs(0 0 4) Bragg backreflection condition for the nominally 1/2 ML sample. The solid lines are the best fits to the data points. J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 Here R and a denote the reflectivity and phase as given by the dynamical theory of X-ray diffraction [6,14], and ½ is proportional to the beam flux which depends linearly on energy. Detailed discussions of the relationship between F&, P&, and Fourier analysis of the positional distributions have been published by other authors [6,15]. The coherent fraction provides a measure of the perfection of the deposited In layer. While an ideal, perfect interface would yield a coherent fraction of 1.0, corresponding to a perfect, d-like distribution of In atoms, thermal effects reduce the maximum value obtainable during an XSW measurement. Consideration of the Debye—Waller factor for the GaAs substrate suggests that for a structurally perfect layer the measured value would be approximately 0.9 [16]. In Figs. 1 and 2 the solid lines represent the fits to the experimental data. The reflectivity curves were fitted using the dynamical theory [6,14], and the fitting includes a Gaussian broadening function which accounts for the monochromator resolution. Using the reflectivity, phase and broadening determined from the reflectivity curves, the In-fluorescence spectra were fitted using Eq. (1) to determine the adjustable parameters F& and P&. The results are given in Table 1. Positions of the In atoms relative to the GaAs(0 0 4) planes with spacing d "1.4133 A> are then given by P&d and also 1 1 appear in Table 1. It is instructive to compare the results for the 1 ML sample with results obtained in similar samples produced by MBE. Table 1 shows the values of P& and F& which have been reported previously [7—10]. It should be noted that the coherent position is determined modulo d , so that a coherent 1 37 position of 0.116 is equivalent to 1.116 (allowing a more direct comparison with the other results) [6]. In Ref. [8], the XSW positional result was compared with the predictions of MET [11]. Taking account of the InAs lattice strains parallel, e , and , perpendicular, e , to the InAs/GaAs interface, , where C (2) e "!2 e , , , C the strained lattice constants may be determined. The unstrained InAs lattice constant is a " ' 6.0584 A> , while for pseudomorphic growth the coherency condition is a "a "5.6532 A> . Since , % e "(a !a )/a , (3) , , ' ' e "(a !a )/a , (4) , , ' ' the MET predicts that the In atoms should reside 1.6247 A> above the last As layer of the substrate. Comparing the positional result from our ALE monolayer with the MBE results, it appears that our sample is not as highly strained as the MBE samples. At the same time, the coherent fraction of 0.766$0.051 is among the best measured, but still somewhat lower than the ideal value of 0.9 which would be measured for a structurally perfect layer. Without the thermal effects, our result would be &0.851, which suggests that the quantum well interface in our sample is quite smooth; reduction from ideality likely reflects a combination of compositional variation at the interface, interface roughness, or a strain-induced spread in the In positions perpendicular to the plane. Processes such as In segregation have been observed for this Table 1 XSW parameters for ALE samples determined from the In L-fluorescence data compared with results reported for MBE-grown films. The In—As planar distances were calculated from the measured coherent positions 1 ML (Present work) 1/2 ML (Present work) 1 ML (MBE) [7] 1 ML (MBE) [8] 1 ML (MBE) [9] 1 ML In Ga As (MBE) [10] Coherent position Coherent fraction In—As planar distance (A> ) 1.116$0.010 1.063$0.021 1.17$0.02 0.16$0.02 1.154$0.01 0.053$0.02 0.766$0.051 0.88$0.12 0.58$0.07 0.73$0.1 0.43$0.03 0.53$0.1 1.577$0.014 1.502$0.030 1.65$0.03 1.64$0.03 1.63$0.01 1.488$0.028 38 J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 used to predict the concentration of the layer in a similar fashion to the XRD determination of alloy concentration by the measurement of lattice constant in thicker films [17]. Assume that the XSW position is correct, and that the In atoms are arranged in a single layer. Based on our experimental result this layer cannot be completely covered by In. If the layer actually consists of an alloyed distribution, In Ga As, V \V then we can solve for x based on the MET. From Vegard’s law, the unstrained lattice constant of the layer, a , is a "xa #(1!x)a . (5) ' % The elastic constants for the alloyed layer are given by the concentration-weighted values for InAs and GaAs [18], as C "xC'#(1!x)C% , C "xC'#(1!x)C% . From MET Fig. 3. Photoluminescence emission from a nominally 1 ML InAs film on GaAs(0 0 1), buried beneath a 2550$25 A> GaAs cap. system [13], although we cannot resolve details of such a process with an XSW measurement using only one set of substrate diffraction planes. For the 1/2 ML sample we have measured a coherent fraction of 0.88$0.12, which corresponds to &0.98 after accounting for the thermal effects. Fig. 3 shows a photoluminescence spectrum obtained at 4.2 K for a nominally 1 ML sample, with a 2550$25 A> cap as described previously. The peak location is found to be 1.459 eV, with a full width at half maximum of 5.74$0.01 meV. The location of the peak is close to the expected position for excitonic emission from an InAs monolayer [4], while the linewidth is as narrow as the best results reported elsewhere [2,4,5] and is indicative of satisfactory growth conditions. What is the significance of our positional measurement? XSW measurements of samples grown by MBE [7—10] support the macroscopic description of this interface. Our results may be (6) a "a (e #1), , , or (7) xC'#(1!x)C% (a a "a !2 !a ). , xC'#(1!x)C% % (8) The value of x can now be determined by substituting equation Eq. (5) into Eq. (8), and using the standing wave result a "P&a . Our measured , % values of a "6.309$0.057 A> and a "6.009$ , , 0.119 A> lead to x"0.794$0.068 and x"0.446$ 0.149, for the 1 and 1/2 ML films, respectively. If this same approach is applied to the results of Woicik et al. [10], in which XSW measurements were made for a single In Ga As layer with V \V a nominal x value of 0.48, it is found that the measured coherent position of 1.053$0.02 is indicative of an alloy composition x"0.378$0.142. This is quite reasonable, since the input Ga : In ratio and growth rates were determined from The values for InAs are C'"8.329;10 N m\, C'"4.526;10 N m\, and for GaAs C% "11.9; 10 N m\, C% "5.38;10 N m\. J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 Fig. 4. X-ray rocking curve data obtained near the GaAs(0 0 4) symmetric substrate reflection, for *h"h!h . Data for the 1 and 1/2 ML samples (upper and lower curves, respectively) are indicated by points. Solid lines indicate the results of dynamical simulations as described in the text. growths with much thicker layers. Because the indium and gallium shutters were open simultaneously during the MBE growth, the alloy-like distribution is to be expected. In contrast, for our ALE films no Ga-source molecules were incident on the substrate during the InAs layer growth. Fig. 4 shows the XRD data for the 1 and 1/2 ML samples. It is interesting to note that the Pendellösung fringes are clearly visible out to 7200 arcsec on either side of the main substrate reflection. This is somewhat remarkable for such thin layers of InAs capped by nominally 100 A> of GaAs. Highresolution XRD studies of thin InAs layers have been reported previously [2,7], in which cases the GaAs caps typically exceeded &275 A> . The XRD data were modeled using BEDE dynamical simulation software employing the Takagi—Taupin equations [19]. In the discussion of 39 the XSW data, we have conjectured that the In atoms reside in single In Ga As layers, and that V \V the group III layer has a distance a /4 from the last , As plane of the substrate. Thus the net displacement of the GaAs cap from the substrate is a /2. , The solid lines in Fig. 4 show the results of our simulations with the alloy compositions and layer thicknesses determined in this way. In addition, the GaAs cap thicknesses determined from the simulations are 97$3 A> for the 1 ML sample, and 85$3 A> for the 1/2 ML sample. Comparing the data and the simulations, it appears that our inferences based on the XSW results provide a good description of the XRD data. It is important to mention, however, that the XRD data could be interpreted differently. For instance, a reasonable simulation for the 1 ML sample rocking curve could alternatively have been obtained by supposing a single InAs layer and an associated substratecap displacement of approximately 3.0$0.3 A> , although this can not be related to a physical distribution of In atoms. Two factors may contribute to the lower than expected coverage in our monolayer sample. Without the benefit of reflection high-energy electron diffraction (RHEED) oscillations during the In deposition, the ALE process is at a disadvantage compared with MBE growth. Specifically, we rely upon ex situ XRD characterization to estimate the amount of In deposited. The precision in the In coverage determined in this way is not adequate to distinguish between a full monolayer and a 79.4% alloy layer, as illustrated here. Secondly, our growth process has employed the “flash-off” technique, first suggested by Brandt et al. for MBE growth of similar structures [20]. It is possible that raising the growth temperature after the GaAs cap ALE cycles may result in a greater-than-intended desorption of segregated In atoms, thus contributing to the lower In coverage. Indeed, a limited desorption is the goal of the flash-off step. Our interpretation of an alloyed distribution of In atoms, particularly in the 1/2 ML case, suggests that for submonolayer coverages the In atoms do not form monolayer islands, as proposed in the past [1—3]. Ideally, such islands would reside at the ideally strained InAs position (1.6247 A> ), although edge effects at InAs islands might lower 40 J.A. Gupta et al. / Journal of Crystal Growth 195 (1998) 34–40 the measured position. We must emphasize that our proposed structure does not constitute a unique and unambiguous interpretation of the data. It is well known that segregation of In atoms is a factor limiting the growth of perfect InAs layers [13], although our calculations suggest that the segregation process alone cannot explain the coherent positions we have measured in our samples [21]. 4. Conclusions We have performed the first X-ray standing wave study of ultrathin InAs layers buried in GaAs, deposited by ALE. A comparison between the results for our nominally 1 ML sample and previous studies with MBE samples suggests that our sample does not consist of a single, highly-strained In layer. Rather, a calculation based on the macroscopic elastic theory indicates that the In atoms are distributed within an In Ga As layer, with x" V \V 0.794$0.068. For the nominally 1/2 ML sample we find our XSW results to also be consistent with a single alloy layer, with x"0.446$0.149. Our simulations show that this interpretation provides an adequate description of the XRD data. For the 1 ML sample the coherent fraction of 0.766$0.051 is indicative of a fairly uniform In distribution, giving rise to the narrow PL emission observed. Factors such as segregation or terrace nucleation have not been included in the present analysis; with only one XSW measurement it is not strictly correct to infer any such multiple-location scenario. Finally, we have only attempted to measure the location of In atoms relative to the (0 0 4) GaAs planes. A more detailed, three-dimensional description of the In distribution is required and is the subject of ongoing experiments. Acknowledgements This work was supported by the Natural Sciences and Engineering Research Council of Canada. The National Synchrotron Light Source at Brook-haven National Laboratory, Upton, NY, is supported by the US Department of Energy under contract No. DE-AC02-76CH00016. References [1] M.V. Belousov, N.N. Ledentsov, M.V. Maximov, P.D. Wang, I.N. Yasievich, N.N. Faleev, I.A. Kozin, V.M. Ustinov, P.S. Kop’ev, C.M. Sotomayor Torres, Phys. Rev. B 51 (1995) 14 346. [2] M.I. Alonso, M. Ilg, K.H. Ploog, Phys. Rev. B 50 (1994) 1628. [3] W. Li, Z. Wang, J. Liang, B. Xu, Z. Zhu, Z. Yuan, J. Li, Appl. Phys. Lett. 67 (1995) 1874. [4] C.A. Tran, R.A. Arès, V.A. Karasyuk, S.P. Watkins, G. Letourneau, R. Leonelli, Phys. Rev. B 55 (1997) 4633. [5] Z.L. Yuan, Z.Y. Xu, B.Z. Zheng, J.Z. Xu, S.S. Li, W. Ge, Y. Wang, J. Wang, L.L. Chang, P.D. Wang, C.M. Sotomayor Torres, N.N. Ledentsov, Phys. Rev. B 54 (1996) 16 919. [6] A review article: J. Zegenhagen, Surf. Sci. Rep. 18 (1993) 199. [7] C. Giannini, L. Tapfer, S. Lagomarsino, J.C. Bouillard, A. Taccoen, B. Capelle, M. Ilg, O. Brandt, K.H. Ploog, Phys. Rev. B 48 (1993) 11 496. [8] J.C. Woicik, J.G. Pellegrino, S.H. Southworth, P.S. Shaw, B.A. Karlin, C.E. Bouldin, K.E. Miyano, Phys. Rev. B 52 (1995) R2281. [9] T.-L. Lee, Y. Qian, P.F. Lyman, J.C. Woicik, J.G. Pellegrino, M.J. Bedzyk, Physica B 221 (1996) 437. [10] J.C. Woicik, K.E. Miyano, J.G. Pellegrino, P.S. Shaw, S.H. Southworth, B.A. Karlin, Appl. Phys. Lett. 68 (1996) 3010. [11] J. Hornstra, W.J. Bartels, J. Crystal Growth 44 (1978) 513. [12] R.A. Arès, S.P. Watkins, C.A. Tran, J. Crystal Growth 170 (1997) 574. [13] R.A. Arès, C.A. Tran, S.P. Watkins, Appl. Phys. Lett. 67 (1995) 1576. [14] B.W. Batterman, H. Cole, Rev. Mod. Phys. 36 (1964) 681. [15] N. Hertel, G. Materlik, J. Zegenhagen, Z. Phys. B 58 (1985) 199. [16] R. Saravanan, S.K. Mohanlal, K.S. Chandrasekaran, Acta Crystallogr. Sect. A 48 (1992) 4. [17] J.C. Woicik, C.E. Bouldin, M.I. Bell, J.O. Cross, D.J. Tweet, B.D. Swanson, T.M. Zhang, L.B. Soerensen, C.A. King, J.L. Hoyt, P. Pianetta, J.F. Gibbons, Phys. Rev. B 43 (1991) 2419. [18] O. Madelung (Ed.), Semiconductors: Group IV Elements and III—V Compounds, Springer, Berlin, 1991. [19] S. Takagi, Acta Crystallogr. 15 (1962) 1311; S. Takagi, J. Phys. Soc. Japan 26 (1969) 1239; D. Taupin, Bull. Soc. Fr. Mineral. Crystallogr. 87 (1964) 469. [20] O. Brandt, K. Ploog, L. Tapfer, M. Hohenstein, R. Bierwolf, F. Phillip, Phys. Rev. B 45 (1992) 8443. [21] J.A. Gupta, J.C. Woicik, S.P. Watkins, E.D. Crozier, to be published.
© Copyright 2026 Paperzz