Structural Transition in Indium Phosphide Nanowires - American

Apr 13, 2010 - Sappoo 060-0814, Japan, ‡ JST-PRESTO, Japan Science and Technology ... Hokkaido University, North 13 West 9, Sappoo 060-8628, Japan...
5 downloads 0 Views 3MB Size
pubs.acs.org/NanoLett

Structural Transition in Indium Phosphide Nanowires Yusuke Kitauchi,† Yasunori Kobayashi,† Katsuhiro Tomioka,‡,§ Shinjiro Hara,†,§ Kenji Hiruma,†,§ Takashi Fukui,†,§ and Junichi Motohisa*,† †

Graduate School of Information Science and Technology, Hokkaido University, North 14 West 9, Sappoo 060-0814, Japan, ‡ JST-PRESTO, Japan Science and Technology Agency, Kawaguchi 332-0012, Japan, and § Research Center for Integrated Quantum Electronics, Hokkaido University, North 13 West 9, Sappoo 060-8628, Japan ABSTRACT We study the catalyst-free growth of InP nanowires using selective-area metalorganic vapor phase epitaxy (SA-MOVPE) and show that they undergo transition of crystal structures depending on the growth conditions. InP nanowires were grown on InP substrates where the mask for the template of the growth was defined. The nanowires were grown only in the opening region of the mask. It was found that uniform array of InP nanowires with hexagonal cross section and with negligible tapering were grown under two distinctive growth conditions. The nanowires grown in two different growth conditions were found to exhibit different crystal structures. It was also found that the orientation and size of hexagon were different, suggesting that the difference of the growth behavior. A model for the transition of crystal structure is presented based on the atomic arrangements and termination of InP surfaces. Photoluminescence measurement revealed that the transition took place for nanowires with diameters up to 1 µm. KEYWORDS Nanowire, selective-area metalorganic vapor phase epitaxy, crystal structure, photoluminescence

R

ecently, semiconductor nanowires have attracted much attention because of their unique electrical and optical properties and application for various nanoscale devices.1-4 In particular, indium phosphide (InP)-based nanowires are promising for high-speed electron, optoelectronic, and photovoltaic devices because of their superior material property and possibility to develop various kinds of heterostructures. To date, InP-based nanowire devices including field effect transistors,5 photodetectors,6 lightemitting devices,5,7 waveguides,8 and solar cells9 have been demonstrated, and studies on heterostructures including quatum dots7,10,11 or core-shell structures12 have been reported. One of the key issues is to form high-quality nanowires free from crystal defects and unintentional impurities, which could be nonradiative recombination centers or scattering centers of carriers. In particular, control of the crystal structure and elimination of twin defects and/or stacking faults are important since it has been shown both theoretically13 and experimentally14,15 that the bandgap energy of InP is different by about 80-90 meV, between zincblende (ZB) and wurtzite (WZ) crystal structures and that, if they are mixed together, they exhibit type-II superlattice structures. Thus, a comprehensive understanding of the growth mechanism is required to control both structure and electronic states. We study the growth of InP nanowires by catalyst-free selective-area (SA) epitaxy. We show that the growth condition controls the growth mode, morphology, shape, and crystal structures of nanowires in selective-area metalorganic vapor phase epitaxy (SA-MOVPE).

Most InP nanowires have been fabricated using the catalyst-assisted vapor-liquid-solid (VLS) growth mechanism. The crystal structure of InP nanowires has been discussed rather controversially. For instance, Duan and Lieber16 and Bhunia et al.17 report that the crystal structures are zincblende (ZB), whereas Mattila et al. report that they are wurtzite (WZ) structures, which is different from that in bulk crystals.18 A recent study has revealed that the crystal structure depends on the growth temperature,19 and more recently, ZB becomes a stable phase when Zn is introduced as a dopant during growth.20 These phenomena have been discussed based on the stability of ZB and WZ crystals or on the formation energy of nuclei. For the former, the stability is determined by the bulk cohesive energy and the surface energy of the nanowire sidewalls.21 Since the sum of these energies depends on the size of the nanowires, there exist a critical radius of nanowires, at which the transition of the crystal structure from ZB to WZ occurs, and it is predicted to be larger for III-V semiconductors with larger ionicity. On the contrary, the latter shows that nuclei are formed near the vapor/liquid/solid interfaces, and the neucli of WZ-type stacking are energetically favorable.22 It should be noted that the size of the nanowires discussed in these studies is in the order of or less than 50 nm. On the other hand, we have been reporting on the catalyst-free growth of III-V semiconductor nanowires by SA-MOVPE.23-25 In this method, a SiO2 mask with a periodic circular opening is formed on the III-V semiconductor substrate as a template for growth, and at appropriate growth conditions, growth starts from the opening of the mask where the bare surface of the substrate is exposed. Position-controlled growth of nanowires is achieved because

Received for review: 01/6/2010 Published on Web: 04/13/2010 © 2010 American Chemical Society

1699

DOI: 10.1021/nl1000407 | Nano Lett. 2010, 10, 1699–1703

than in (i). In other growth conditions, we obtained either a hexagonal rod with a small aspect ratio, tetrahedral structure, or tapered nanowires, and the uniformity of the nanowire arrays was deteriorated (see the Supporting Information S1). Comparing the two SEM images, we can see differences between the two types of nanowires. First, the diameter is not the same, although growth started with the same mask opening size. The larger diameter in Figure 1b indicates that growth takes place in the lateral direction as well as the parallel direction of the nanowires, whereas the nanowire diameter in Figure 1a is the same as the opening size, d0. Therefore, the growth modes of the two types of nanowires are different. Second, a rotation of the hexagon was observed in their cross section, as shown in the insets. Because tapering is negligible in both cases, the surface of the sidewalls is flat in the atomic scale. This is also confirmed by transmission electron microscopy (TEM) images. Hence, facets vertical to the surface are formed as the sidewalls, and they are parallel to the [2¯11] and [1¯10] direction of InP (111)A surfaces for nanowires (i) and (ii), respectively. The difference in the growth mode becomes more evident if one sees the dependence of size and height of nanowires on the pitch, a, of the mask hole (Figure 1c). The diameter, d, of nanowires (i) did not change from the mask diameter, d0, but its height decreased with a. The tendency was quite similar to our previous results on SA-MOVPE growth of GaAs nanowires.24 On the contrary, both size and height increased as a increased for nanowires (ii). The most striking difference is the crystal structures. Figure 2 summarizes the results of TEM analysis of the two types of nanowires. The incident direction of the electron beam is parallel to the 〈11¯0〉 direction of the substrate. A high-resolution TEM image and a selected area electron diffraction pattern (SAED) indicate that nanowires (i) were WZ. This is consistent with our previous results and reasonable because they were grown under similar Tg and V/III ratio conditions.26 Statistic analysis proposed by Tomioka et al.25 shows that 92% of the atomic-layer stacking order follows that of WZ and the average thickness of a WZ segment is 25 ML. Meanwhile, nanowires (ii) were ZB. In addition, the contrast of the high resolution TEM image and the streak of electron diffraction pattern indicate the high density of stacking faults and twinning dislocations. The stacking faults appear in every 2.3 monolayers on average (see the Supporting Information S2). The difference in the growth mode can be explained by the diffusion of growth species and precursors in the gas phase and on the surface. Consider the case where reevaporation of growth species is negligible and the supply of group V species is sufficient. This situation is quite normal for MOVPE. In this case, the volume of nanowires is determined by the diffusion of group III species or their precursor, that is, how much they collect In atoms via the gas phase and surfaces. If the supply of source materials per unit area is constant, materials to be incorporated into a single nano-

FIGURE 1. Summary of SA-MOVPE-grown InP nanowires obtained in two different conditions: SEM image of InP NW array grown at: (a) growth condition (i); growth temperature 660 °C, V/III ratio 18. (b) growth condition (ii); growth temperature 600 °C, V/III ratio 55. The pitch, a, of nanowire array is 1 µm, and the initial opening diameter, d0, of mask is 100 nm. Each inset shows top view of nanowire, showing difference in size and orientation of hexagon. (c) average diameter, d, and (d) height, h, of two types of nanowires as function of a. d is almost same as d0 and independent of nanowires grown in condition (i) but increases for those grown in condition (ii).

of the confinement of the growth area by the SiO2 mask. For InP, we use trimethylindium (TMIn) and tertiallbutylphophine (TBP) as source materials and InP (111)A as a substrate and have reported that the crystal structure of nanowires is WZ,26 which is different from GaAs23 or InAs nanowires.25 The aforementioned theories alone are not applicable to the formation of WZ nanowires in SA-MOVPE, due to the fact that the size of the nanowires is much larger than the predicted critical size and due to the absence of the liquid phase. What we describe here is that the surface energy is one of the key issues for understanding the transition between WZ and ZB. We also show that a wider variety of structures can be controlled by the growth parameters including growth temperature, material supply ratio, mask opening size, and pitch of the nanowire array. As a result of a series of nanowire growth, we found two kinds of distinct growth conditions, in which highly uniform arrays of hexagonal nanowires without tapering were formed by SA-MOVPE. In Figure 1a,b, we show scanning electron microscopy (SEM) images of nanowires grown in different conditions, (i) and (ii), respectively. Hereafter, we call them nanowires (i) and (ii), respectively. The initial diameter, d0, of the mask pattern was 100 nm and the hole pitch, a, of the mask was 0.5 µm for both cases. For nanowires (i), the growth temperature Tg and ratio of partial pressure of the supply gas, [TBP]/[TMIn] (V/III ratio), were set to be 660 °C and 18, respectively. This is a standard condition for growing InP nanowires on InP (111)A substrates as stated in our previous reports. On the other hand, nanowires (ii) were grown at 600 °C and at a V/III ratio of 55. Note that the growth temperature is lower and V/III ratio is higher in (ii) © 2010 American Chemical Society

1700

DOI: 10.1021/nl1000407 | Nano Lett. 2010, 10, 1699-–1703

FIGURE 2. TEM image of two types of InP nanowires. (a) nanowires (i). (b) nanowires (ii). Scale bar: 10 nm. Upper and middle insets show selected area electron diffraction (SAED) and its simulated pattern, respectively, for each figure. Electron beam projection is parallel to 〈1¯10〉 direction of substrate. Both image and SAED demonstrate nanowires (i) have WZ crystal structure and very few stacking faults or twinning dislocations. Comparison with simulation also reveals that electron beam projection is parallel to 〈112¯0〉 of WZ. On the other hand, nanowires (ii) exhibit high-density of stacking faults, as one can see from streak pattern in SAED as well as TEM image. Simulated SAED pattern of nanowires (ii) represents superposition of two directions, Rand L. Lower inset shows magnification of each image, and arrows indicate positions of stacking faults. Scale bar: 2 nm.

wire increases with decreasing nanowire density. Therefore, the volume of the nanowire becomes larger with decreasing a, corresponding to nanowires (ii). Note also that growth of the {11¯0}-oriented surfaces of ZB crystals requires sufficient amount of both In and P atoms for nucleation, and generally enhanced for such low temperature and high group V material supply. Thus, growth on {11¯0} sidewall facets is noticeable for nanowires (ii), resulting in nanowires with a larger diameter than the mask opening size. On the other hand, under condition (i), that is in higher Tg and lower V/III ratio, the effect of re-evaporation becomes stronger than in condition (ii), and the effective density of In and P atoms decreases on the side and top surfaces of the nanowires. This results in the suppression of the growth on the sidewall in the lateral direction as well as in the vertical direction. Furthermore, once In atoms (or their precursors) evaporate from sidewalls of one nanowire, they will not be reincorporated into the adjacent nanowires and contribute to growth if they are apart from each other. Conversely, if a is short enough, In atoms can be readsorbed, then diffused to the top and contribute to axial growth. For this reason, lateral growth is restricted and the axial growth rate is enhanced © 2010 American Chemical Society

with a higher density of nanowires, resulting in longer and thinner nanowires. This kind of “synergetic” growth mode has already been reported in GaAs nanowires both in SAMOVPE24 and VLS growth.27 For the former, the catalytic decomposition of monomethylgallium (MMGa) around the catalyst is suggested, but it is essentially the same mechanism described here. Next, we describe one of a possible model for structural transition (Figure 3). To start with, we point out that the termination of the surface with constituent atoms depends on the growth conditions. It is reported experimentally that a surface reconstruction of the InP (111)A surface is either (2 × 2) or (3 × 3)R30° structures.28 The former appears at P deficient conditions, and its atomic arrangements are such that the surface is In-terminated and with missing In atoms in the center of the (2 × 2) unit cell. This surface reconstruction is supported by theoretical calculation.29 The latter, on the other hand, appears when more P is present on the surface, and it is suggested that P trimers are formed both in experiment and theory. Keeping this information in mind, let us recall that the difference between WZ and ZB is in the relative position of 1701

DOI: 10.1021/nl1000407 | Nano Lett. 2010, 10, 1699-–1703

FIGURE 3. Atomistic model for structural transition in InP nanowires. (a) side and top view of atom arrangements in WZ crystal. (b) side and top view of atom arrangements in ZB crystal. A and B represent adsorption sites of In for WZ and ZB stacking, respectively. If surface is terminated with In and desorption of P is significant, additional In atoms are needed for growth to proceed in [111]A direction, and binding of In at site A is stronger than in site B due to larger Coulomb interaction through second nearest pair of In and P (enclosed with purple dotted circles, and indicated with arrows). As a result, WZ stacking is favored. On the other hand, difference between site A and B is less important if surface is terminated with P, as schematically shown in b.

the [111]B direction of ZB, exhibiting {11¯0} facets.23,25 This is because conditions of high growth temperature and low V/III ratio are required for appropriate growth of nanowires in the [111]B direction, and binding of group V atoms on {11¯0} surfaces are less strong in such As-deficient conditions (in other words, {11¯0} surface is stable), resulting in a slower growth rate on {110} than on (111)B. Formation of other possible sidewall facets, for instance, {112¯}, has not been confirmed. Although the difference is not so large for nanowires (ii) and allows growth in both axial and radial directions, the situation is similar and the growth rate is the slowest on {11¯0}. One the other hand, among possible orientations as sidewall facets, that is, {112¯0} or {11¯00} is s reported to be more stable because the density of the surface dangling bonds are smaller in WZ21,30 (see the Supporting Information T3). Thus, it exhibits lower growth rate because it has the smallest density of surface bonds. Formation of {11¯00} facets in other WZ nanowires have been reported.31 Our discussion is mostly based on the thermodynamics at the surface including surface termination. Yet to be discussed is how much kinetics associated with the reaction and nucleation of growth species affects the formation of ZB and WZ crystal structures. The effect of the sidewall surface, which is crucial in terms of surface energy and the surface diffusion of In, should also be taken into account. In particular, it is expected that the crystal structure become ZB nearly independent of growth conditions in the limit of large diameter or on planar layers. To further investigate this point, we show the results of photoluminescence (PL) measurements for nanowires (i) and their dependence on nanowire diameter d in Figure 4. d was changed from 100 nm to

the second nearest pair of groups III and V atoms. For instance, in WZ, group III (V) atoms sit atop the secondnearest group V (III) atoms (site A), as shown in Figure 3. This gives rise to enhanced Coulomb interaction in the second-nearest pairs as compared to site B of ZB stacking. For growth on (111)A to proceed on the In-terminated surface, P atoms should be adsorbed first. However, bonding of In an P is thought to be weak since it has one single bond toward the surface, and the P atoms could be desorbed easily, particularly at a P-deficient situation as in condition (i). Thus, simultaneous In adsorption is required to help P incorporation into the crystal, and the binding is stronger if In atoms sit on site A. Hence, WZ stacking is favored in nanowires (i). In contrast, under condition (ii), at lower Tg and a higher V/III ratio, supply of P is sufficient, thus (111)A surface is terminated more with P. This hinders the difference between sites A and B, resulting in no preference for ZB and WZ stacking. Consequently, nanowires (ii) become mixed and exhibit a high density of stacking faults. As we mentioned earlier, the direction of the hexagon is different for the two types of nanowires, indicating that the crystallographic orientation of the sidewall facets is different. Analysis of SAED patterns and their comparison with the simulation (Figure 2) concluded that the facets are {11¯00} of the surface of WZ and {11¯0} of ZB in nanowires (i) and (ii), respectively. The orientation coincides with the substrate for ZB and is reasonable for WZ (see the Supporting Information S2 and S3). The difference can also be explained by the difference in the crystal structure as follows. Generally speaking, the surface of the lowest growth rate appears as facets. SA-MOVPE-grown GaAs and InAs nanowires grow in © 2010 American Chemical Society

1702

DOI: 10.1021/nl1000407 | Nano Lett. 2010, 10, 1699-–1703

This material is available free of charge via the Internet at http://pubs.acs.org. REFERENCES AND NOTES (1) (2) (3) (4) (5) (6) (7)

(8)

FIGURE 4. PL spectra of nanowires (i) and their dependence on diameter. PL of nanowires (i) shows a blue-shift of about 90 meV, as compared to the InP layer grown on planar substrate and nanowires (ii), indicating formation of a WZ crystal structure up to diameter of 1 µm. Reference spectra from planar layers grown in conditions (i) and (ii) are also shown. Dotted vertical line indicates the low temperature bandgap energy of bluk InP.

(9) (10) (11)

1 µm. For comparison, the PL spectrum of nanowires (ii) and those of planar layer grown on (111)A InP substrate are also shown. The planar layers were simultaneously grown with nanowires (i) or (ii). As we mentioned earlier, the WZ InP has been shown to have a larger band gap as compared to ZB InP both theoretically and experimentally. For all series of nanowires (i), a large blue-shift was observed as compared to nanowires (ii) and planar epi-layers, which is consistent with the formation of the WZ crystal phase. It should be noted that even nanowires with d ) 1 µm show a comparable blue-shift as thinner nanowires. This indicates that WZ is formed in most of the regions with a very slight mixture of the ZB crystal phase. Furthermore, PL from planar layers grown in condition (i) shows a blue shift as compared to bulk InP. This suggests the formation of WZ crystal phase even in (111)A planar layers. Therefore, the structural transition is mostly dependent on the growth conditions in InP and the effect is more pronounced in case of selective area growth of nanowires. In summary, we have shown that two types of InP nanowires can be grown in selective area-MOVPE, which exhibit different crystal structures, shapes, and growth modes, by controlling the growth condition. A model for the transition of the growth mode and crystal structures was described based on the state of termination of the InP (111)A surface.

(12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) (26) (27)

Acknowledgment. The work is partly financially supported by a Grant-in-Aid for Scientific Research, supported by Ministry of Education, Science, Sports, and Culture, Japan.

(28) (29) (30) (31)

Supporting Information Available. Additional experimental procedures including additional figures and a table.

© 2010 American Chemical Society

1703

Xia, Y.; Yang, P.; Sun, Y.; Wu, Y.; Mayers, B.; Gates, B.; Yin, Y.; Kim, F.; Yan, H. Adv. Mater. 2003, 15, 353–389. Xiang, J.; Lu, W.; Hu, Y.; Wu, Y.; Yan, H.; Lieber, C. M. Nature 2006, 441, 489–493. Ng, H.; Han, J.; Yamada, T.; Nguyen, P. Nano Lett. 2004, 4, 1247– 1252. Rehnstedt, C.; Mårtensson, T.; Thelander, C.; Samuelson, L.; Wernersson, L.-E. IEEE Trans. Electron Devices 2008, 55, 3030– 3036. Duan, X.; Huang, Y.; Cui, Y.; Wang, J.; Lieber, C. M. Nature 2001, 409, 66–69. Wang, J.; Gudiksen, M. S.; Duan, X.; Cui, Y.; Lieber, C. M. Science 2001, 293, 1455–1457. Minot, E. D.; Kelkensberg, F.; van Kouwen, M.; van Dam, A. J.; Kouwenhoven, L. P.; Zwiller, V.; Borgstro¨m, M. T.; Wunnicke, O.; Verheijen, M. A.; Bakkers, E. P. A. M. Nano Lett. 2007, 7, 367– 371. Ding, Y.; Motohisa, J.; Hua, B.; Hara, S.; Fukui, T. Nano Lett. 2007, 7, 3598–3602. Goto, H.; Nosaki, K.; Tomioka, K.; Hara, S.; Hiruma, K.; Motohisa, J.; Fukui, T. Appl. Phys. Exp. 2009, 2, No. 035004. Fuhrer, A.; Fro¨berg, L. E.; Pedersen, J. N.; Larsson, M. W.; Wacker, A.; Pistol, M.-E.; Samuelson, L. Nano Lett 2007, 243–246. Zwiller, V.; Akopian, N.; van Weert, M.; van Kouwen, M.; Perinetti, U.; Kouwenhoven, L. P.; Algra, R.; Rivas, J. G.; Bakkers, E.; Patriarche, G.; Liu, L.; Harmand, J.-C.; Kobayashi, Y.; Motohisa, J. C. R. Physique 2008, 9, 804–815. Mohan, P.; Motohisa, J.; Fukui, T. Appl. Phys. Lett. 2006, 88, 133105. Murayama, M.; Nakayama, T. Phys. Rev. 1994, B 49, 4710–4724. Mishra, A.; Titova, L. V.; Hoang, T. B.; Jackson, H. E.; Smith, L. M.; Yarrison-Rice, J. M.; Kim, Y.; Joyce, H. J.; Gao, Q.; Tan, H. H.; Jagadish, C. Appl. Phys. Lett. 2007, 91, 263104. Kobayashi, Y.; Fukui, M.; Motohisa, J.; Fukui, T. Physica E 2008, 40, 2204–2206. Duan, X.; Lieber, C. M. Adv. Mater. 2000, 12, 298–302. Bhunia, S.; Kawamura, T.; Watanabe, Y.; Fujikawa, S.; Tokushima, K. Appl. Phys. Lett. 2003, 83, 3371–3373. Mattila, M.; Hakkarainen, T.; Lipsanen, H.; Jiang, H.; Kauppinen, E. I. Appl. Phys. Lett. 2006, 89, No. 063119. Mattila, M.; Hakkarainen, T.; Mulot, M.; Lipsanen, H. Nanotechnology 2006, 17, 1580–1583. Algra, R. E.; Verheijen, M. A.; Borgstro¨m, M. T.; Feiner, L.; Immink, G.; van Enckevort, W. J. P.; Vlieg, E.; Bakkers, E. P. A. M. Nature 2008, 456, 369–372. Akiyama, T.; Sano, K.; Nakamura, K.; Ito, T. Jpn. J. Appl. Phys. 2006, 45, L275–L278. Glas, F.; Harmand, J.-C.; Patriarche, G. Phys. Rev. Lett. 2007, 99, 146101. Motohisa, J.; Noborisaka, J.; Takeda, J.; Inari, M.; Fukui, T. J. Cryst. Growth 2004, 272, 180–185. Noborisaka, J.; Motohisa, J.; Fukui, T. Appl. Phys. Lett. 2005, 86, 213102. Tomioka, K.; Motohisa, J.; Hara, S.; Fukui, Jpn. J. Appl. Phys. 2007, 46, L1102–L1104. Mohan, P.; Motohisa, J.; Fukui, T. Nanotechnology 2005, 16, 2903– 2907. Borgstro¨m, M. T.; Immink, G.; Ketelaars, B.; Algra, R.; Bakkers, E. P. A. M. Nat. Nanotechnol. 2007, 2, 541–544. Li, C. H.; Sun, Y.; Law, D. C.; Visbeck, S. B.; Hicks, R. F. Phys. Rev. 2003, B 68, No. 085320. Akiyama, T.; Kondo, T.; Tatematsu, H.; Nakamura, K.; Ito, T. Phys. Rev. B 2008, 78, 205318. Akiyama, T.; Nakamura, K.; Ito, T. Phys. Rev. B 2006, 73, 235308. Sekiguchi, H.; Kishino, K.; Kikuchi, A. Appl. Phys. Exp. 2008, 1, 124002.

DOI: 10.1021/nl1000407 | Nano Lett. 2010, 10, 1699-–1703