Journal of Crystal Growth 378 (2013) 77–80
Contents lists available at SciVerse ScienceDirect
Journal of Crystal Growth journal homepage: www.elsevier.com/locate/jcrysgro
Problems in low-temperature grown polycrystalline InAs layers on glass and their relief by inserting GaSbAs buffer layers Y. Kajikawa n, T. Okuzako, Y. Matsui Department of Electric and Control Systems Engineering, Interdisciplinary Faculty of Science and Engineering, Shimane University, 1060 Nishi-Kawatsu, Matsue, Shimane 690-8504, Japan
a r t i c l e i n f o
abstract
Available online 3 January 2013
We investigated the thickness dependence of structural and electrical properties of polycrystalline InAs layers grown on glass substrates at a substrate temperature of 230 1C by molecular-beam deposition. Degradation in electron mobility with decreasing thickness is shown and is attributed not only to the decrease in crystallite size but also to the existence of a defective layer adjacent to the substrate. In order to investigate the effectiveness of GaSbAs as buffer layers on these problems, polycrystalline InAs layers of 0.1 mm thickness were gown with GaSbAs buffer layers of various thicknesses and alloy compositions inserted between the InAs layers and the substrates. It is shown that, in spite of the low substrate temperature of 300 1C and the thin thickness of 0.1 mm, the polycrystalline InAs film can exhibit an electron mobility as high as 600 cm2/(V s) by inserting a 0.5-mm thick GaSbAs buffer layer of an Sb composition around 0.6. & 2012 Elsevier B.V. All rights reserved.
Keywords: A3. Molecular beam epitaxy A3. Polycrystalline deposition B1. Arsenide B1. Antimonides B2. Semiconducting III–V materials
1. Introduction In recent years, numerous studies have been devoted for fabricating thin-film transistors (TFTs) not only on glass substrates but also on plastic substrates. In our previous study [1] on polycrystalline InAs (poly-InAs) films grown by molecular-beam deposition (MBD), it has been shown that poly-InAs is a promising n-type semiconductor for TFTs on plastic substrates, since it can be deposited even at substrate temperatures as low as below 250 1C, while deposited films of 1 mm thickness exhibit mobilities as high as above 600 cm2/(V s) in spite of the low substrate temperatures. However, it has been commonly observed for polycrystalline films of many semiconductors that mobility of carriers is apt to decrease with decreasing thickness, while the appropriate thickness of the active layer of TFTs is around 0.1 mm. It is therefore necessary to check whether the poly-InAs films exhibit sufficiently high mobilities even at such thin film thicknesses. In the present study, we first examine the thickness dependence of structural and electrical properties of poly-InAs single layers in Section 2. On the other hand, the insertion of an electrically-inactive buffer layer is commonly used to improve the quality of the active layer, not only in homoepitaxy but also in heteroepitaxy for field-effect transistors. In another study of ours [2], it has been found that single-phase poly-GaSbAs can be
n
Corresponding author. Tel./fax: þ 81 852 32 8903. E-mail address:
[email protected] (Y. Kajikawa).
0022-0248/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.jcrysgro.2012.12.054
deposited even on plastic substrates at a substrate temperature as low as 300 1C, whereas the poly-GaSb film contains the metallic Sb phase when deposited at such a low temperature. It has also been found there that the non-doped poly-GaSbxAs1 x films exhibit p-type conduction in the whole studied range of Sb composition, x, between 0.34 and 1. Furthermore, GaSbxAs1 x can be lattice-matched to InAs by adjusting the Sb composition, x, to 0.92. If x 0.92 the hetero-growth of InAs on poly-GaSbxAs1 x is expected to occur without any misfit dislocations at the InAs/ GaSbxAs1 x hetero-interface in each crystallite. Thus, p-type polyGaSbxAs1 x with x 0.92 is considered as an ideal buffer layer for n-type poly-InAs. We therefore investigate the feasibility of using the p-type poly-GaSbAs as a buffer layer for n-type poly-InAs active layer in Section 3.
2. Thickness dependence of poly-InAs single layers Non-doped poly-InAs films of various thicknesses were grown on glass substrates at a substrate temperature of 230 1C by MBD. The As/In beam equivalent pressure (BEP) ratio during the deposition was 12, and the deposition rate was 1.2 mm/h. Fig. 1(a) and (b) respectively show the volume concentration and the mobility of free electrons at 300 K in the poly-InAs films as a function of the film thickness. As can be seen in Fig. 1(a) and (b), the free-electron concentration does not so depend on the thickness, while the electron mobility decreases with decreasing thickness. The poly-InAs films thicker than 0.6 mm do exhibit high
78
Y. Kajikawa et al. / Journal of Crystal Growth 378 (2013) 77–80
intrinsic and extrinsic stacking-fault probabilities, respectively) can be estimated through a peak-shift analysis of XRD [6]. For the (111)-oriented crystallites of our InAs sample deposited at 230 1C, the net stacking-fault probability aSF ð1 1 1Þ , was estimated to be as high as 0.02. Such a high density of stacking faults may be an additional cause of the low electron mobilities in the very thin films of poly-InAs deposited at the low substrate temperatures. Furthermore, as shown in the inset of Fig. 1(a), when the sheet concentration of free electrons in the InAs films is plotted against the film thickness, the plotted data for our InAs samples with thicknesses of 0.05, 0.1, and 0.2 mm almost lie on a straight line, and the intersection of the extrapolated straight line with the abscissa indicates the existence of a depletion layer of about 0.016 mm in thickness adjacent to the substrate. This can explain the decrease in the apparent volume concentration of free electrons with decreasing thickness observed for the plotted data of the thinnest three samples in Fig. 1(a). A high density of defects generated in the early stage of the deposition, such as stacking faults, may be the cause of the interfacial depletion layer. In order to improve the electron mobility in low-temperature grown polyInAs thin films of thickness of the order of 0.1 mm, it is necessary not only to increase the crystallite size but also to eliminate the interfacial defective layer. For this purpose, we inserted the polyGaSbAs layer as the buffer layer between the poly-InAs layer and the substrate, as described in the next section.
3. Effects of poly-GaSbAs buffer layers
Fig. 1. (a) Volume concentration and (b) mobility of free electrons in the InAs single layer as a function of the thickness. The inset in (a) shows the sheet concentration of free electrons.
electron mobilities above 300 cm2/(V s), while thinner films than 0.1 mm exhibit only low electron mobilities below 60 cm2/(V s). The increase of mobility in polycrystalline semiconductor films with the film thickness has often been related to the increase in crystallite size with the film thickness [3]. Actually, the crystallite size estimated from the width of the (111) peak in the X-ray diffraction (XRD) pattern of our InAs samples was confirmed to increase from 13 to 26 nm when the film thickness was increased from 0.2 to 2.4 mm. Besides the small crystallite size, another cause can be speculated for the low mobilities in our thin InAs films. In each of the XRD patterns of these InAs films, additional weak diffraction can be noticed at a diffraction angle of 2y ¼24.01 other than the usual diffraction peaks due to cubic InAs crystal. This weak diffraction is indexable as the (1010) diffraction of the hexagonal wurtzite phase of InAs. Vlasov and Semiletov [4] observed the (1010) diffraction of the hexagonal phase in the electron diffraction patterns of the InAs films deposited at substrate temperatures below 400 1C, and attributed it to the formation of staking faults in the cubic crystals. Farukhi and Charlson [5] pointed out that the amount of the hexagonal region will be sufficient to be noticeable in XRD patterns if periodic occurrence of stacking faults is introduced in some manner. Actually, the shift and the asymmetry of several XRD peaks suggested large densities of stacking faults in our InAs samples. The net stacking-fault probability in ISF ESF ISF ESF (hkl)-oriented crystallites aSF ðhklÞ ¼ aðhklÞ aðhklÞ (aðhklÞ and aðhklÞ being
InAs layers of 0.1-mm thickness were deposited on glass substrates by MBD with GaSbxAs1 x (0 rx r1) as the buffer layers between the InAs layers and the substrates. The BEPs of In, Ga, and As were fixed at about 9 10 8, 5 10 7, and 2 10 6 Torr, respectively. These BEPs of In and Ga resulted in the growth rates of 0.2 and 1 mm/h for InAs and GaSbAs, respectively. The As/In BEP ratio of 22 used here is rather high for the InAs deposition to achieve high electron mobility [1], but was adopted so as to ensure the group-V stabilized surfaces during the deposition of the GaSbxAs1 x buffer layers even with x¼0. In the first series of deposition (Series A), the cell temperature of Sb was fixed so that the Sb composition, x, of the GaSbxAs1 x buffer layers was fixed at about 0.9, while the thickness of the buffer layers was changed from 0 to 1 mm by changing the deposition time. On the contrary, the deposition time was fixed at 0.5 h so that the thickness of the buffer layers was fixed at 0.5 mm, while the Sb composition, x, of the GaSbxAs1 x buffer layers was changed from 0 to nearly 1 by changing the BEP of Sb in the second series of deposition (Series B). The substrate temperature was kept at 300 1C through the deposition of the GaSbxAs1 x (0 rx o 1) buffer layers and the InAs top layers. On the other hand, for the deposition of the sample with a GaSb buffer layer of 0.5 mm, the substrate temperature was set to 450 1C for the deposition of the GaSb buffer layer, since the GaSb layer would contain the metallic Sb phase when deposited below 400 1C [2]. After completing the deposition of the GaSb layer, the substrate temperature was lowered to 300 1C for the deposition of 0.1-mm thick InAs. XRD measurements were performed for estimating the alloy composition of the GaSbAs buffer layers through the Nelson–Riley plots [2]. Fig. 2 shows the semi-log plots of the XRD patters of the samples in Series A with the buffer layers of different thicknesses. Arrows in the figure indicate diffraction angles for the reflections from various diffraction planes of cubic InAs with the lattice constant of the bulk value. In the XRD pattern of the sample with the 0.25-mm thick buffer layer, the diffraction peaks of the GaSbAs buffer layer cannot be resolved from those of the InAs layer,
Y. Kajikawa et al. / Journal of Crystal Growth 378 (2013) 77–80
Fig. 2. XRD patters of the InAs/GaSbAs samples with different thicknesses of the GaSbxAs1 x buffer layers with x around 0.9.
indicating that the GaSbAs is well lattice-matched to InAs. In the XRD patters for the samples with thicker buffer layers, on the other hand, more intense diffraction peaks of GaSbAs than those of InAs can be seen in the higher angle side of each diffraction peak of InAs. This indicates that the Sb compositions of the buffer layers of these samples are smaller than that of the sample with the 0.25-mm thick buffer layer, owing to the unintentional decrease of Sb flux in spite of the fixed temperature of the Sb cell. In the XRD pattern of the sample without the buffer layer, (1010) diffraction of the hexagonal InAs can be seen as a shoulder in the low-angle side of the (111) diffraction peak of cubic InAs. Similar weak diffraction can be seen also in the XRD pattern of the sample with the 0.25-mm thick buffer layer. Solely from this XRD pattern, it is hard to assign whether this diffraction is from GaSbAs or from InAs. However, for the samples of single layers of poly-GaSbAs deposited on glass at 300 1C, no trace of the hexagonal (1010) diffraction was observed in their XRD patterns even when their thicknesses were less than 0.3 mm. Furthermore, in the studies on nanowire growth on GaAs (111) surfaces [7,8], it has been reported that even small amount of Sb gives preferentially zincblende structure of GaSbxAs1 x, while InAs nanowires grown under similar conditions normally exhibit predominantly wurtzite crystal structure with frequent stacking faults. Therefore, the hexagonal (1010) diffraction appeared in the XRD patterns of the InAs/GaSbAs hetero-structure samples is speculated to be not from the GaSbAs buffer layer but from the InAs layer. The (1010) diffraction of the hexagonal phase of InAs was not observed for the samples with thicker buffer layers than 0.5 mm, as can be seen in Fig. 2. Gottschalk et al. [9] experimentally determined the staking fault energy of cubic III–V compounds and found that the staking fault energy decreases with the ionicity. GaSbAs has less ionicity and thus has a larger staking fault energy than InAs. It is, therefore, speculated that the generation of staking faults will be suppressed during the deposition of the GaSbAs buffer layer, which will result in suppressing the emergence of the hexagonal phase in the InAs over layer. Hall-effect measurements were performed at 77 and 300 K for investigating the electrical properties of the InAs top layers. It is assumed here that the GaSbxAs1 x (0 rx r1) buffer layers have little effect on the measurement results since they are p-type for x 40 or highly resistive for x¼0. Fig. 3(a) and (b) respectively show the concentration and the mobility of free electrons in the
79
InAs top layers of the samples in Series A as a function of the thickness, t, of the GaSbxAs1 x buffer layers whose Sb composition, x, was around 0.9. Fig. 4(a) and (b) show the same as Fig. 3(a) and (b) but for the samples in Series B as a function of the Sb composition, x, of the GaSbxAs1 x buffer layers whose thicknesses were all 0.5 mm. The lattice mismatch, e, of GaSbAs to InAs is represented on the up-side scale of Fig. 4(a) and (b). As can be seen in Fig. 3(a) and Fig. 4(a), the electron concentration in the InAs top layer is not so affected by either of the thickness or the Sb composition of the buffer layer. On the other hand, the electron mobility in the InAs top layer was affected by either of the thickness or the Sb composition of the buffer layer. It increased with t until t ¼0.5 mm and then saturated, as can be seen in Fig. 3(b). It reaches a maximum value of m ¼600 cm2/(V s) at x¼0.62 in the dependence on the Sb composition, x, as can be seen in Fig. 4(b). It should be noted that, at this composition, the lattice mismatch of GaSbAs to InAs is not zero but e ¼ 2.2%. For this lattice mismatch, the equilibrium critical layer thicknesses for the generation of misfit dislocations at the hetero-interface are estimated to be about 3.5 and 9 nm for the epitaxial growth on (100) and (111) surfaces, respectively, according to Ref. [10]. The thickness of 0.1 mm of the InAs layers of our samples is far beyond these equilibrium critical thicknesses. Since the crystallite size of InAs is considered to be as small as a few tens of nanometers for our samples, the amount of the crystallite grains which consist of both of InAs and GaSbAs portions is much less
Fig. 3. (a) Volume concentration and (b) mobility of free electrons in the InAs top layer of the InAs/GaSbAs samples as a function of the thickness of the GaSbxAs1 x buffer layer with x around 0.9.
80
Y. Kajikawa et al. / Journal of Crystal Growth 378 (2013) 77–80
than those composed of InAs only. Therefore, the electrical conduction in the InAs layer may be dominated by the crystallite grains composed of InAs only, while the crystallite grains which contain the InAs/GaSbAs hetero-interfaces may have less effect. It is therefore speculated that the lattice matching of the buffer layer to the top layer may not be so significant for improving the electron mobility in the top layer. In summary, it has been proved that, even at a low substrate temperature of 300 1C, the poly-InAs film of thickness as thin as 0.1 mm can exhibit an electron mobility as high as 600 cm2/(V s) by inserting the GaSbAs buffer layer of 0.5-mm thickness. References [1] M. Takushima, Y. Kajikawa, Y. Kuya, M. Shiba, K. Ohnishi, Japanese Journal of Applied Physics 47 (2008) 1469. [2] T. Okuzako, K. Okamura, Y. Matsui, K. Nakaya, Y. Kajikawa, Physica Status Solidi C 8 (2011) 266. [3] C.H. Ling, J.H. Fisher, J.C. Anderson, Thin Solid Films 14 (1972) 267. [4] V.A. Vlasov, S.A. Semiletov, Soviet Physics: Crystallography (1968) 761. [5] M.R. Farukhi, E.J. Charlson, Journal of Applied Physics 40 (1969) 5361. [6] Y. Kajikawa, Y. Iseki, Y. Matsui, Proceedings of the 23rd International Conference on Indium Phosphide and Related Materials (IPRM2011), Berlin (2011) 169. [7] D.L. Dheeraj, G. Patriarche, H. Zhou, T.B. Hoang, A.F. Moses, S. Grønsberg, A.T.J. van Helvoort, B.-O. Fimland, H. Weman, Nano Letters 8 (2008) 4459. [8] M. Ek, B.M. Borg, A.W. Dey, B. Ganjipour, C. Thelander, L.-E. Wernersson, K.A. Dick, Crystal Growth and Design 11 (2011) 4588. [9] H. Gottschalk, G. Patzer, H. Alexander, Physics Status Solidi A 45 (1978) 207. [10] T. Anan, K. Nishi, S. Sugou, Applied Physics Letters 60 (1992) 3159.
Fig. 4. (a) Volume concentration and (b) mobility of free electrons in the InAs top layer of the InAs/GaSbAs samples as a function of the Sb composition, x, of the GaSbxAs1 x buffer layer of 0.5 mm thickness.