NANO LETTERS
Growth of GaInAs/AlInAs Heterostructure Nanowires for Long-Wavelength Photon Emission
2008 Vol. 8, No. 11 3645-3650
Kouta Tateno,* Guoqiang Zhang, and Hidetoshi Nakano NTT Basic Research Laboratories, NTT Corporation, 3-1 Morinosato-Wakamiya, Atsugi, Kanagawa 243-0198, Japan Received June 5, 2008; Revised Manuscript Received September 1, 2008
ABSTRACT We investigated the growth of GaInAs/AlInAs heterostructure nanowires on InP(111)B and Si(111) substrates in a metalorganic vapor phase epitaxy reactor. Au colloids were used to deposit Au catalysts 20 and 40 nm in diameter on the substrate surfaces. We obtained vertical GaInAs and AlInAs nanowires on InP(111)B surfaces. The GaInAs nanowires capped with GaAs/AlInAs layers show room-temperature photoluminescence. The peak exhibits a blue-shift when the Ga content in the core GaInAs nanowire is increased. For the GaInAs/AlInAs heterostructure growth, it is possible to change the Ga content sharply but Al also exists in the GaInAs layer regions. We also found that the ratios of Ga and Al contents to In content tend to increase and the axial growth rate to decrease along the nanowire toward the top. We were also able to make vertical GaInAs nanowires on Si(111) surfaces after a short growth of GaP and InP.
The growth of nanowires by means of the vapor-liquid-solid (VLS) mechanism was discovered and proposed first by Wagner and Ellis in the 1960s.1 Later, after the encouraging studies on light-emitting diodes using nanowires by Hiruma et al. in the 1990s,2 many studies on the semiconductor nanowires toward functional devices have been reported. The VLS mechanism is considered as the base mechanism for freestanding nanowire growth from metal particles. It is similar to the mechanism of liquid-phase epitaxy but is driven catalytically in the nanoarea by nanosized metal particles such as Au. This feature is very useful from the viewpoint of reducing growth time and the consumption of both power and source materials for industrial fabrication. GaInAs semiconductor is suitable for 1.3 or 1.55 µm wavelength lasers for optical communications and also has the potential to emit photons in the 2 µm wavelength region where molecular spectroscopic and medical applications are expected.3 AlInAs can be used as a barrier material for quantum wells and as a cladding material for optical waveguides. By combining GaInAs and AlInAs, various optical devices become possible. The growth of GaInAs nanowires on Si is very promising for application to Si-based optoelectronic integrated circuits (OEICs). Silicon is the base material of integrated circuits for electronics: it is inexpensive, abundant, safe, and can be formed into large singlecrystal wafers as wide as 12 in. Vertical III-V compound semiconductor nanowires grown on Si(111) surfaces have been attracting interest for application to OEICs.4-6 In large* Corresponding author. E-mail:
[email protected]. 10.1021/nl801612p CCC: $40.75 Published on Web 10/14/2008
2008 American Chemical Society
area epitaxy, there are problems related to lattice and polarity mismatch and differences in thermal expansion coefficients,7 although the form of nanowire reduces these problems.8,9 The selective growth at the position of the predeposited catalysts is also a very favorable feature. Among the III-V nanowires on Si, GaP nanowires on Si(111) substrates exhibit quite orderly orientation in the [111]B direction compared with GaAs or InP nanowires because GaP’s small lattice mismatch of less than 0.4% relative to Si.4-6 III-V nanowires tend to grow in the [111]B direction so that a GaP(111)B face has to be formed on Si(111) for vertical nanowire growth. Effective techniques for vertical nanowire growth have been introduced;4-6 however, complete control of the polarity has not been achieved, which is still an important issue. As for heterostructures, several unique three-dimensional structures with bends or inserted spheres obtained using GaAs, AlAs, or InP starting from GaP nanowires on Si(111) have recently been demonstrated.10,11 VLS-grown nanowires themselves have several problems. Stacking faults tend to form in the nanowires, which cause localized potential fluctuations and rough side walls. The concentrations of impurities in nanowires are not apparent because measuring the concentrations under the 1% level is quite difficult. Therefore, investigating the basic characteristics of the various semiconductor nanowires is still very important. Kim et al. have reported GaInAs nanowires grown on GaAs(111)B substrates.12 They observed up to about 1.6 µm photoluminescence (PL) at 10 K and pointed out the problems of stacking faults and compositional variation along
Table 1. Growth Conditions for Nanowire Samples material nanowires capping layers alternating GaInAs/AlInAs nanowires
GaInAs AlInAs 2nd: GaAs 1st: AlInAs 3rd: AlInAs 2nd: 5 pairs of GaInAs/AlInAs 1st: AlInAs
pregrowth on Si substrate
2nd: InP 1st: GaP
source
flow rate of group III (µmol/min), flow rate of group V (µmol/min), growth temperature, growth time
TMGa + TMIn, AsH3 TMAl + TMIn, AsH3 TMGa/AsH3 TMAl + TMIn*, AsH3 *mole ratio: [TMGa]/[TMIn] ) 0.5/0.5 TMAl + TMIn**, AsH3
9.6, 2.0 × 102, 460 °C, 10 min 9.6, 2.0 × 102, 460 °C, 10 min 4.8, 1.6 × 103, 580 °C, 10 s 9.6, 1.6 × 103, 580 °C, 7 min
TMGa or TMAl + TMIn**, AsH3
9.6, 2.0 × 102, 460 °C, 5 s for GaInAs and 25 s for AlInAs 9.6, 2.0 × 102, 460 °C, 4 min
TMAl + TMIn**, AsH3 **mole ratio: [TMGa or TMAl] /[TMIn] ) 0.75/0.25 TMIn, PH3 TMGa, PH3
the nanowire. In this paper, we present several results for GaInAs and AlInAs nanowires grown on InP(111)B substrates. By using a capping growth technique, we were able to improve photon emission efficiency to obtain roomtemperature (RT) PL in the range of 1.2-2.0 µm wavelength. We also made multiheterostructures of GaInAs and AlInAs. In addition, we were able to grow vertical GaInAs nanowires on Si(111) substrates. Our nanowires, like Kim et al.’s, also show stacking faults and compositional variation, which are serious problems preventing application. The wire growth was carried out in a low-pressure (76 Torr) horizontal metalorganic vapor phase epitaxy (MOVPE) reactor.5,6,10,11 Trimethylgallium (TMGa), trimethylaluminum (TMAl), and trimethylindium (TMIn) were the group III sources. Phosphine (PH3) and arsine (AsH3) were the group V sources. The catalysts were Au particles (40 and 20 nm in diameter) obtained from Au colloids. The 20 nm Au colloids were used for the heterostructure nanowires and GaAs/AlInAs-capped nanowires. Details of the growth conditions for the samples described in this paper are summarized in Table 1. The samples were GaInAs or AlInAs nanowires, GaAs/AlInAs capped GaInAs nanowires, and
9.6, 2.0 × 102, 460 °C, 3.5 min
2.4, 4.5 × 102, 550 °C, 5 s 4.8, 4.5 × 102, 550 °C, 5 s
alternating GaInAs/AlInAs heterostructure nanowires. On Si(111) substrates, GaP and InP were shortly grown before the GaInAs nanowire growth. The structures of the nanowires were observed by scanning electron microscopy (SEM, Hitachi, S-5200, operated at 15 kV) and transmission electron microscopy (TEM, JEOL, JEM2100F, at 200 kV). Highangle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and energy dispersive X-ray spectrometry (EDS) analyses were also performed in the TEM chamber to evaluate the elemental distribution in the nanowires. Compositions of Ga, Al, As, and In were evaluated from X-ray peaks from Ga-K, Al-K, As-K, and In-L lines. For optical characterization, RT-PL (Nanometrics, RPM2000) was measured. The input light had 532 nm wavelength, and it illuminated about a 100 µmφ area of the sample. The excitation power was about 50 mW. The collected luminescence was fed to a spectrometer equipped with a silicon charged-coupled device (CCD) or InGaAsarray detectors. Figure 1 shows SEM images of GaInAs and AlInAs nanowires. Almost all the nanowires started to grow perpendicular to the surface in the [111]B direction. However,
Figure 1. SEM images of GaInAs and AlInAs nanowires. (1) Top views; (2) views 38° tilted from the normal direction. (a,b) GaInAs nanowires; (c,d) AlInAs nanowires; (a,c) 40 nm Au colloids, [TMGa or TMAl]/[TMIn] ) 1, 7 min of nanowire growth; (b,d) 20 nm Au colloids, [TMGa or TMAl]/[TMIn] ) 3, 10 min of nanowire growth. 3646
Nano Lett., Vol. 8, No. 11, 2008
To resolve it, we have to find some proper way to make single-crystalline nanowires.
Figure 2. High-resolution TEM images of GaInAs and AlInAs nanowires. Growth conditions: 20 nm Au colloids, [TMGa or TMAl]/[TMIn] ) 3 and 10 min of nanowire growth.
the AlInAs nanowires, which were grown using 20-nm Au particles, a higher TMGa flow rate, and longer growth time, show bent shapes as shown in Figure 1d1,d2. Thin nanowires are likely to be bent by some forces, so it is possible that they were bent by stress induced by the variation of composition. As described later, these ternary alloy nanowires showed compositional variation along the nanowire. The bending seems to have occurred because of the imbalance of the radial growth. This radial growth occurred along with the catalyst-assisted growth so that the nanowire structure became tapered as seen in Figure 1. The amount of source material coming from the surface of the substrates should not be equal in the horizontal direction because, for example, there are many structures around each nanowire, which would influence the local concentration and the flow of the reaction species. Thus, the difference in the thickness and the composition on each side wall is one possible reason for the nanowires bending. The shape of the GaInAs nanowires is hexagonal from the top views, although many triangular shapes are seen for AlInAs nanowires as shown in Figure 1c1. These shapes are related to the stable side facets, especially {111} facets.13 Stacking faults are also involved in the structures.13 However, at this stage, it is not clear what causes the two structures. Crystallographic structures are clearly different for the GaInAs and AlInAs nanowires as shown in Figure 2. The GaInAs nanowires are mainly zinc blende structures with many twins, and some wurtzite structures are seen. For the AlInAs nanowires, the base structure is wurtzite with some twins. It has been reported that AlAs shows wurtzite structures in GaAs/AlAs-alternated nanowires grown on GaP nanowires on Si(111) substrates, and InAs nanowires tend to have wurtzite structures.14,15 So it is considered that the combination of AlAs and InAs is involved in the formation of a stable wurtzite structure. A variation of the diameter of GaInAs is seen in Figure 2a. Zinc blende-type nanowires are likely to show a zigzag structure owing to {111}-related side facets, as has been reported by Johansson et al. for GaP,13 while wurtizite-type nanowires are likely to show smooth side walls, which has been reported by Mattila et al. for InP.16 The image in Figure 2a is the region where both structures exist so the variation of the diameter is conspicuous. This is an important issue. Nano Lett., Vol. 8, No. 11, 2008
Bare GaInAs nanowires did not show PL peaks at RT. The capping growth to form core-shell structure nanowires is effective for improving luminescence. Figure 3 shows SEM images and RT-PL spectra of the GaAs/AlInAs-capped GaInAs nanowires and the Ga content in the core GaInAs vs TMGa flow ratio in the group III gases evaluated from the PL peaks. First, as for the growth properties for the nanowires, it is seen that the density of the nanowires for the sample for R ) 9 was about 30% of that for R ) 1. The density of the predeposited Au particles is almost the same for all samples. In Figure 3a1, there are many nonactivated (or no-nanowire-grown) Au particles on the surface, so it is considered that the optimum growth condition for the nanowire growth is different for the different Ga-to-In ratio samples. The growth windows for nanowires are very narrow for general VLS-grown nanowires, so it is not strange that they show such variation. Because vertical nanowires were somehow obtained for all samples, we proceeded to investigate the PL properties. In Figure 3b, it is found that the emission wavelength decreased with increasing TMGa/TMIn mole ratio () R) for the GaInAs cores. From the PL peaks, the corresponding Ga content in the core GaInAs for each sample was evaluated simply by assuming the relaxed-lattice condition, which is plotted in Figure 3c in accordance with the ratio of the TMGa flow rate in the supplied group III gases. Notably, most of the photon emission comes from the part in the nanowire with the lower band gap; the emission at PL peaks does not correspond to the average Ga content when the Ga content varies along the nanowire. Nanowires for R ) 1 show a large curve at the top region as shown in Figure 3a4 and the emission peak is very broad, which indicates a wide compositional variation along the wires. For the evaluation of the compositional evaluation, PL measurement is not a proper approach. However, it is obvious that the Ga content at the local part that contributes main photon emission is almost in proportion to that of the gas phase. Although the compositional variation exists in the nanowires, it can be said that we can control the emission wavelength by varying the TMGa/TMIn mole ratio. An elemental analysis using electron beam is a useful method for evaluating the compositional variation in nanowires. We made five alternating GaInAs layers in an AlInAs nanowire and analyzed the compositional distribution. Figure 4 shows HAADF-STEM images and EDS data for Ga, Al, and In in the nanowires. The five GaInAs layers are seen in the EDS spectra. The amount of Ga sharply changed within 6 nm, which is estimated from Figure 4b2. However, Al was also detected in the Ga-supplied region, and the thickness and the interval of this region decreased along the axis toward the top. The bottom layer of the five (Al)GaInAs layers was Al0.08Ga0.24In0.68As and the top layer was Al0.10Ga0.42In0.48As. The Ga composition in the gas phase (the ratio of TMGa flow) was 0.75, while the Ga composition in the solid phase was very small, i.e., 0.24-0.42. The total nanowire length was about 5 µm. On GaAs(111)B, the estimated diffusion lengths of Ga and In species are 2 and 6 µm.12 The content 3647
Figure 3. (a) SEM images of GaAs/AlInAs capped GaInAs nanowires. The ratio of TMGa to TMIn flow rate () R) was varied in the core GaInAs nanowire growth. (b) RT-PL spectra obtained from the samples in (a). (c) Ga content in the GaInAs vs the ratio of TMGa flow rate in the gas phase. The values are estimated from the PL peaks of (b).
Figure 4. HAADF-STEM images and corresponding EDS data for five alternating GaInAs layers in an AlInAs nanowire. EDS data were obtained by line-scanning in the nanowire. The scanned distances are indicated in (a) and (b).
of Ga has to decrease when the In reaction species is dominant in the growth owing to the long diffusion length. We think In-rich AlGaInAs formed around the bottom of the nanowire by means of a layer-by layer radial growth mechanism, which results in the tapered structure. The amount of the radial growth for AlInAs is relatively large so that Al is also detected in the five GaInAs layer regions. Here, we can not say whether Al is actually incorporated at the core region because the electron beam influenced both 3648
the core region, which was grown from an Au alloy particle, and the shell region, which was grown without the assistance of Au. The axial growth rate decreased drastically at the third GaInAs layer. We speculate that this position is the diffusion limit of In reaction species from the surface and that the diffusion length is about 2.5 µm. The small In diffusion length may be related to the AlInAs surface. From the TEM images in Figure 4, bulging at the GaInAs segments is seen. It is considered that there are different growth rates between Nano Lett., Vol. 8, No. 11, 2008
from the PL peak of zinc blende structure.16 On the other hand, for the five GaInAs layers in AlInAs nanowires as shown in Figure 4, the band gap difference between the top and the bottom layer is estimated to be about 210 meV, which must be much larger than the difference in the energy between wurtzite and zinc blende structures. Therefore, compositional variation is the most critical problem for the emission control. If we can grow long nanowires far beyond the diffusion length of the reaction species on the surface, it seems that a region of constant composition would be obtainable. The point is how to keep the temperature and gas supply constant at the Au alloy particle region. Another solution would be to use very short nanowires as the height level of quantum dots. In this case, the point is how to control the ultra low gas flow rate.
Figure 5. SEM images and PL spectrum of GaAs/AlInAs capped GaInAs nanowires on shortly grown InP/GaP on Si (111) substrates. (b) PL spectrum of the nanowires grown on InP(111)B substrates is included for comparison.
GaInAs and AlInAs and also a drastic decrease of the amount of reaction species at the GaInAs segments that come from the bottom toward the tip. On Si(111) substrates, we tried to make vertical GaInAs nanowires with GaAs/AlInAs capping layers after a short growth of GaP and InP. We have recently reported the InP growth on GaP nanowires on Si(111) substrates.11 Straight InP nanowires could not be formed on GaP nanowires. However, InP blocks were likely formed on the tops or around the bottom of the GaP nanowires in the hightemperature condition.11 Considering this result, we thought that small blocks of InP on short GaP nanowires would be effective as base structures for vertical GaInAs nanowires, which means the InP(111)B would be formed below the Au particles. If the InP layer is within the critical thickness on GaP, the growth from GaP to GaInAs will proceed epitaxially with the Au alloy particles always on top. Conclusively, it is confirmed that the base InP/GaP nanostructures were ideally formed because successive GaInAs nanowires were vertically grown as shown in Figure 5a. Figure 5b shows the RT-PL spectrum of the capped GaInAs nanowires on Si(111) substrates and that of the nanowires on InP(111) substrates, which were grown using the same growth conditions except for the InP and GaP growth. The red-shift peak is seen in the Si case. It is possible that the diffusion properties of reaction species are different from that on InP surfaces. It also possible that the adsorption and desorption processes for reaction species are quite different on Si surfaces, where precise measurements are necessary. Finally, controllability of the emission wavelength is discussed. The In content of GaInAs and AlInAs nanowires varies along them, and the bottom of the nanowires is the In-rich region, which has a lower band gap energy so that most photon emission must occur there. For InP nanowires, the wurtzite structure shows a PL blue-shift of about 80 meV Nano Lett., Vol. 8, No. 11, 2008
In summary, we have investigated growth of GaInAs/ AlInAs heterostructure nanowires on InP(111)B and Si(111) substrates in a MOVPE reactor. Au colloids were used to deposit Au catalysts 20 and 40 nm in diameter on the surface. We obtained vertical GaInAs and AlInAs nanowires on InP(111)B surfaces. GaInAs nanowires capped with GaAs/ AlInAs layers show RT-PL in the range of 1.2-2.0 µm. The peak blue-shifts with increasing the Ga content in the core GaInAs nanowire. For the GaInAs/AlInAs heterostructure growth, it is possible to change Ga content sharply within 6 nm but Al also exists in the GaInAs layers, which is mainly due to the AlInAs radial growth. It was also found that the ratios of the Ga and Al contents to the In content tend to increase and the axial growth rate decreases along the nanowire toward the top. We were also able to make vertical GaInAs nanowires on Si(111) surfaces after the short growth of GaP and InP. Acknowledgment. We thank Drs. H Sanada, T. Tawara, and H. Gotoh for their fruitful discussions. We also thank Drs. H. Yamaguchi and Y. Tokura for their continuous encouragement throughout this work. References (1) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (2) Hiruma, K.; Yazawa, M.; Katsuyama, T.; Ogawa, K.; Haraguchi, K.; Koguchi, K.; Kakibayashi, H. J. Appl. Phys. 1995, 77, 447. (3) Sato, T.; Mitsuhara, M.; Watanabe, T.; Kondo, Y. Appl. Phys. Lett. 2005, 87, 211903. (4) Måartensson, T.; Svensson, C. P. T.; Wacaser, B. A.; Larsson, M. W.; Seifert, W.; Deppert, K.; Gustaffson, A.; Wallenberg, L. R.; Samuelson, L. Nano Lett. 2004, 4, 1987. (5) Tateno, K.; Hibino, H.; Gotoh, H.; Nakano, H. Appl. Phys. Lett. 2006, 89, 033114. (6) Zhang, G.; Tateno, K.; Sogawa, T.; Nakano, H. J. Appl. Phys. 2008, 103, 014301. (7) Yonezu, H.; Furukawa, Y.; Abe, H.; Yoshikawa, Y.; Moon, S.-Y.; Utsumi, A.; Yoshizumi, Y.; Wakahara, A.; Ohtani, M. Opt. Mater. 2005, 27, 799. (8) Glas, F. Phys. ReV. B 2006, 74, 121302(R) (9) Chuang, L. C.; Moewe, M.; Chase, C.; Kobayashi, N. P.; ChangHasnain, C.; Crankshaw, S. Appl. Phys. Lett. 2007, 90, 043115. (10) Tateno, K.; Zhang, G.; Sogawa, T.; Nakano, H. Jpn. J. Appl. Phys 2007, 46, L780. (11) Tateno, K.; Zhang, G.; Nakano, H. J. Cryst. Growth 2008, 310, 2966. (12) Kim, Y.; Joyce, H. J.; Gao, Q.; Tan, H. H.; Jagadish, C.; Paladugu, M.; Zou, J.; Suvorova, A. A. Nano Lett. 2006, 6, 599. 3649
(13) Johansson, J.; Karlsson, L. S.; Svensson, C. P. T.; Måartensson, T.; Wacaser, B. A.; Deppert, K.; Samuelson, L.; Seifert, W. Nat. Mater. 2006, 5, 574. (14) Zhang, G.; Tateno, K.; Sanada, H.; Nakano, H. Proceedings of the 19th International Conference on Indium Phosphide and Related Materials Conference, Matsue, Japan, 2007, WeB2-2.
3650
(15) Dick, K. A.; Deppert, K.; Samuelson, L.; Seifert, W. J. Cryst. Growth 2006, 297, 326. (16) Mattila, M.; Hakkarainen, T.; Mulot, M.; Lipsanen, H. Nanotechnology 2006, 17, 1580.
NL801612P
Nano Lett., Vol. 8, No. 11, 2008
