Semiconductor nanowires have demonstrated unique

Transcription

1 pubs.acs.org/nanolett Understanding Self-Aligned Planar Growth of InAs Nanowires Yunlong Zi, Kyooho Jung, Dmitri Zakharov, and Chen Yang*,, Department of Physics and Department of Chemistry, Purdue University, West Lafayette, Indiana 47907, United States Birck Nanotechnology Center, Purdue University, West Lafayette, Indiana 47907, United States *S Supporting Information ABSTRACT: Semiconducting nanowires have attracted lots of attention because of their potential applications. Compared with free-standing nanowires, self-aligned planar nanowires grown epitaxially on the substrate have shown advantageous properties such as being twin defect free and ready for device fabrication, opening potentials for the large-scale device applications. Understanding of planar nanowire growth, which is essential for selective growth of planar vs freestanding wires, is still limited. In this paper, we reported different growth behaviors for self-aligned planar and free-standing InAs nanowires under identical growth conditions. We present a new model based on a revised Gibbs Thomson equation for the planar nanowires. Using this model, we predicted and successfully confirmed through experiments that higher arsenic vapor partial pressure promoted free-standing InAs nanowire growth. A smaller critical diameter for planar nanowire growth was predicted and achieved experimentally. Successful control and understanding of planar and free-standing nanowire growth established in our work opens up the potential of large-scale integration of self-aligned nanowires for practical device applications. KEYWORDS: Planar nanowire, InAs, Gibbs Thomson equation, growth model Semiconductor nanowires have demonstrated unique properties and various potential applications, such as nanowire transistors, 1 nanosensors, 2 nanowire photodetectors, 3 and self-powered nanowire devices. 4 The bottom-up growth approach has been successfully used to produce high-quality nanowires and nanowire heterostructures, enabling new device strategies for nanoelectronics. 5 Nanowires synthesized by various vapor deposition methods are typically free-standing on the growth substrates, 6 8 and the postgrowth processing required for large-scale device applications is challenging. In III V free-standing nanowires twin plane defects are frequently found, 9 which can decrease the performance of nanowire devices. Recently, planar nanowires grown epitaxially on the substrate were reported for InP nanowires on an InP (100) substrate. 10 A systematic study of GaAs planar nanowires shows that they are twin defect free with large yield self-alignment and transfer-printable for large-scale device integrations. 11 InAs planar nanowire growth was also reported on semi-insulating GaAs substrates 12 with the possibility of direct device fabrication on as-grown substrates. Potentially being structurally advantageous and ready for large-scale device fabrication, planar nanowires open up opportunities for fundamental studies and practical applications. However, the growth mechanism for planar nanowires has not been well-studied, which limits controlled growth of planar vs free-standing nanowires. Previously the Gibbs Thomson equation was successfully applied to several free-standing nanowire systems focusing on the thermodynamics of nanowire growth. 13,14 As stated in the Gibbs Thomson equation, elemental vapor partial pressure and diameter of the nanowires were confirmed as significant factors determining the supersaturations of catalysts which drive the nanowire growth directly. To understand the difference of planar and freestanding nanowire growth in thermodynamics, the Gibbs Thomson equation for planar nanowire growth is required to be developed. In this work, we have demonstrated the growth of InAs planar nanowires using a simple physical vapor deposition method. Growth behaviors of planar and free-standing InAs nanowires, specifically termination time under identical growth conditions, have been compared. We have developed a new model based on a revised Gibbs Thomson equation for semicylindrical planar nanowires. Results calculated based on the model predicted two distinguished aspects in planar nanowire growth in comparison with the growth of freestanding nanowires: planar nanowire growth is favored by a lower partial vapor pressure and leads to a smaller critical diameter of the nanowires synthesized. Both aspects were further confirmed by the experiments. A metal nanoparticle initiated physical vapor deposition method was used to grow InAs nanowires. 15 The precursor, InAs powder ( %, Alfa Aesar), was placed in the upstream zone of a multizone furnace, while (100) or (111)B InAs growth substrates were placed in the downstream zone. The temperatures of the precursor and the substrate were Received: March 21, 2013 Revised: April 25, 2013 Published: May 1, American Chemical Society 2786

2 Nano s Figure 1. SEM images of self-aligned planar InAs nanowires grown on InAs (100) substrates with (a) top view, (b) 45 tilted view, and (c) top view. Scale bars are (a) 200 nm, (b) 200 nm, and (c) 2 μm. Figure 2. (a c) Cross-sectional TEM images of planar InAs nanowires grown on (100) substrate. Inset in a shows fast Fourier transform taken from the nanowires body. Scale bars are (a) 50 nm, (b) 5 nm, and (c) 10 nm. (d) Schematic of gold catalyst, InAs nanowire, and substrate showing crystal directions and surfaces. (e) 3-Dimensional zinc blende unit cell of InAs. The (011) plane is highlighted in green. (f) (011) plane view of the InAs unit cell. controlled at 850 and 485 C, respectively. The growth pressure was 150 Torr, and H 2 was used as the carrier gas with a flow rate of 150 sccm unless otherwise specified. Prior to the growth, InAs substrates were cleaned by sonication in acetone, isopropanol, and ethanol, followed by a chemical etching using HCl:H 2 O (1:10) solution to remove the native oxide layer. Poly-L-lysine (Ted Pella) was used to enhance adhesion of gold particles to the substrate. 2, 5, and 40 nm gold colloid particles (BBI International) were dispersed on poly-l-lysine coated substrates to initiate the nanowire growth. Scanning electron microscopy (SEM) images show that selfaligned planar InAs nanowires were grown on InAs (100) substrates (Figure 1a and b) under the conditions described above. Nanowires were all grown in-plane with the substrate, and no nanowires were observed to be free-standing out of the substrate. Based on the cut of the InAs (100) substrate, nanowires were found to be self-aligned along [01 1] or [0 11] direction in parallel and distributed uniformly over the large areas up to the entire substrate (Figure 1c). Surface areas on the substrates without the nanowires grown were noticeably smooth, indicating that the InAs substrates do not serve directly as precursors for the growth under given conditions. Cross-sectional high-resolution transmission electron microscopy (HRTEM) images revealed that these planar nanowires were grown epitaxially on the (100) substrate (Figure 2 a and b). The fast Fourier transform (FFT) of the image taken from the nanowire body indicates that the nanowire growth direction is <0 11> with a zinc blende crystal structure and a lattice 2787

3 Nano s constant of 6.05 Å, in agreement with the bulk value of InAs. No defects, especially twin defects, were found in any of the 20 nanowires examined, which is consistent with the observation on planar GaAs nanowires reported. 11 A representative HRTEM image of the catalytic interface is shown in Figure 2c, in which the dark contrast indicates gold. We measured the angle between the catalyst nanowire interface and the catalyst substrate interface to be 125.5, which is consistent with the fact that the angle between [0 11] and [1 11] is expected to be ( = 35.5 ) (Figure 2d). Considering the orientation of the (100) substrate, this result, together with FFT shown in Figure 2a, confirmed that the interface plane between the gold particle and the nanowire is a (1 11) plane, more specifically, a (111)B plane. The result implies that, even though the nanowire axis is in [0 11] direction, the nanowire growth is still carried out by the growth of (111)B plane, which has the lowest surface energy among all InAs surfaces. 16 Previous InP, GaAs, and InAs planar nanowire growth have also been demonstrated to have (111)B planes as energy-favored growth planes. 10,11,17 Having noticed the (111)B plane as the growth front, we expected that epitaxial growth of InAs nanowires on InAs (111) B substrates would result in vertical InAs nanowires. Figure 3a and b shows vertical free-standing InAs nanowire grown epitaxially on (111)B substrate, and the growth conditions were identical with those on the (100) substrate. In addition, SEM images of samples with different growth time revealed that the lengths of the vertical InAs nanowires on InAs (111)B surface remain almost constant after the initial 5 min (Figure 3a and b), while the planar growth on InAs (100) surface continued until more than one hour (Figure 3c and d). Figure 3e shows average lengths of nanowires measured from SEM images as a function of growth time. These results demonstrate following key features: first, for planar nanowire growth on InAs (100) (red triangles), growth rate decreases after the initial 5 min, and growth was terminated at approximately 1 h; second, for freestanding nanowire growth on InAs (111)B (black triangles), the growth terminates approximately after 5 min, much earlier than the termination of the planar growth on InAs (100). It is known that for III V compounds, when temperature is above the congruent sublimation temperature T cs, the sublimation rate of the group V element is higher than that of group III element. 18 The temperature for InAs source in our experiments is 850 C, which is far higher than T cs of InAs 387 C, 18 resulting a much higher sublimation rate of arsenic than that of indium. After a few minutes of rapid arsenic sublimation, a solid InAs precursor turns into indium liquid with a small amount of arsenic dissolved. 19 Then the arsenic vapor will be released at a lower rate. As a result, arsenic vapor pressure in the growth chamber is initially high and then decreases after the first few minutes, 20 becoming a rate-limiting factor of the growth. The arsenic sublimation rate during the growth extracted from the weight of the arsenic left in the molten InAs precursor is a measure for arsenic vapor available in the growth chamber (see Supporting Information). Figure 3f confirmed that initially the average arsenic sublimation rate starts high and then decreases after a few minutes, eventually reaching to a minimal sublimation rate after approximately 40 min. This initial time period is consistent with the time period when we observed the termination of nanowire growth on (111)B substrate. The vapor pressure of arsenic is expected to continuously decrease. Eventually after 1 h of growth we observed the termination of planar growth on (100) substrate. The different termination Figure 3. (a d) SEM images with a 60 tilted view of InAs nanowires grown on InAs (100) and InAs (111)B substrates when varying the growth time. (a and b) Free-standing nanowire growth on InAs (111) B for 5 min and 60 min, respectively. (c and d) Planar nanowire growth on InAs (100) for 5 and 60 min, respectively. All scale bars are 500 nm. (e) Average length of InAs nanowires as a function of growth time. Black and red triangles represent growth on InAs (111)B and InAs (100), respectively. (f) The weight of the arsenic left in the molten InAs precursor after 3, 20, 40, and 100 min of source sublimation and the average sublimation rate for each time period estimated from the weight of the arsenic left. The weight of initial InAs precursor was 60 mg for all trials. time of growth on (100) and (111)B substrates discussed above indicates that the free-standing growth on the (111)B substrate requires a higher arsenic vapor pressure than that for planar growth on the (100) substrate. To understand the thermodynamics in nanowire growth, the Gibbs Thomson equation was utilized to predict feasibility of nanowire growth depending on the source vapor pressure and the targeted diameter. 14 In our work, the Gibbs Thomson equation was utilized to explain the difference in the termination time of free-standing and planar nanowire growth. For both free-standing and planar nanowires growth, under fixed total pressure and temperature, the Gibbs free energy change δg during the nanowire growth is given as: δg = δn( Δ μ ) + δsσ (1) where δn, δs, σ, and Δμ denote the number of arsenic atoms supplied by the vapors, the surface area increase, InAs nanowire average surface energy density, and the supersaturation over a catalyst surface with infinite radius of curvature, respectively

4 Nano s Figure 4. Schematic diagram illustrating (a) free-standing nanowire with a cylindrical shape and (b) planar nanowire with a semicylindrical shape. Arsenic rather than indium is focused here since arsenic vapor is the rate-limiting factor as discussed above. The supersaturation Δμ is the chemical potential difference between arsenic atoms in the vapor and solid phases in the bulk limit with infinite radius. 13 Δμ can be estimated as: Δ μ = p kt ln As p (2) due to its dependence on vapor pressure of arsenic p As. p is the vapor pressure of arsenic in bulk gold with infinite radius. 14,22 For free-standing nanowire growth on InAs (111)B substrate, in which nanowires can be considered as cylindrical structures 14 (Figure 4a), the changes in the number of atoms and surface area are proportional to the increase of nanowire length δl: δv πr δl δn = = Ω Ω δs = 2πδ r L 2 where Ω is atomic volume of arsenic in InAs material. Hence, the supersaturation in free-standing nanowires Δμ f is obtained from eq 1 as: δg σ σ Δ μ = =Δμ Ω p f 2 As 2 = kt ln Ω f δn r p r (3) Equation 3 is the well-known equation for the Gibbs Thomson effect, which is only valid for spherical and cylindrical systems. 13,21 For planar nanowires in InAs (100) substrate, the thickness of nanowires in cross-sectional TEM images (Figure 3a and b) was measured as nm, while the width of nanowires in top view SEM images (Figure 1a) was measured as nm. Thus it is reasonable to treat a planar nanowire as a semicylindrical structure (Figure 4b). For such semicylindrical geometry, the Gibbs Thomson effect equation developed previously (eq 3) fails. In this case, changes in number of atoms δn and surface areas δs are given as: δv (1/2) πr δl δn = = Ω Ω δs = πδ r L 2rδL 2 Note that the term 2rδL in δs denotes surface area decrease of InAs (100) substrate, which is the area of interface between the semicylindrical structure and the substrate. As a result, a revised equation of Gibbs Thomson effect for planar nanowire growth is obtained as: Δ μ p σ σ σ = Gp 2 =Δμ Ω 4 + Ω Ω Δ μ f N r πr r (4) To verify the accuracy of the presented model, the supersaturations for both free-standing and planar nanowire growth are estimated based on eqs 3 and 4 for the experimental conditions (see Supporting Information) and listed in Table 1. Table 1. Supersaturations for Free-Standing and Planar Nanowire Growth in Different Time Ranges growth time (min) free-standing Δμ f (ev) planar Δμ p (ev) For supersaturation calculations, arsenic vapor pressure p As was estimated from arsenic depletion rate extracted from Figure 3f. Pressure of arsenic in gold with infinite radius p was estimated by multiplying arsenic vapor pressure at 500 C 23 and arsenic solubility in gold. 24 A radius r of 20 nm was used. Arsenic atomic volume Ω in InAs and InAs surface energy density σ are both constants (see Supporting Information for details). As shown in Table 1, the sign of free-standing nanowire growth supersaturation Δμ f changes from positive to negative after the initial growth period (0 3 min), resulting in termination of growth. On the contrary, the sign of planar nanowire growth supersaturation Δμ p stays positive, and hence the growth continues. During the first three minutes, planar nanowire growth shows a larger positive value for Δμ p than that for free-standing growth, indicating that planar growth is preferred. The promotion in Δμ p is due to the additional term 4Ωσ/πr in eq 4, which is estimated at about ev and comparable with Δμ f. Vapor pressure of arsenic is expected to be further lower after one hour of planar nanowire growth, leading to a negative Δμ p and consequentially the termination of the nanowire growth. Based on this revised Gibbs Thomson effect equation, we speculate that, if vapor pressure of arsenic is increased, Δμ f can be enhanced and the free-standing nanowire growth might be promoted. Considering that a high H 2 flow rate not only results in low arsenic pressure (Supporting Information), but also quickly removes arsenic vapors from the growth chamber, a significantly lower H 2 flow (20 sccm) was used to achieve a higher arsenic vapor pressure. Free-standing nanowires were observed on (100) substrate (Figure 5a) with a variety of gold nanoparticle sizes from 5 to 40 nm with 20 sccm H 2 flow and other identical conditions. These free-standing nanowires were 2789

5 Nano s Figure 5. SEM images of free-standing nanowire growth on (a) InAs (100) substrate (tilt 85 view), (b) InAs (111)A substrate (top view), and (c) ultrathin planar nanowire growth on InAs (100) substrate (tilt 20 view). Insets in a and b show the corresponding orientations of both substrates. Both the nanowires are grown on an identical condition, with H 2 flow rate of 20 sccm. Sizes of Au nanoparticles used are (a) 40 nm, (b) 5 nm, and (c) 2 nm. Scale bars are (a) 500 nm, (b) 1 μm, and (c) 500 nm. The arrows in a and b highlight planar nanowires. self-aligned in the same direction with a tilted angle of about 35 to the (100) substrate, indicating the growth front is also (111)B. We also tested growth on an InAs (111)A substrate with 20 sccm H 2 flow and other identical conditions. We observed that the nanowires are in three directions with angle 60 apart, which is consistent with three <11 2> directions as the projections of <11 1> directions (which are <111>B directions) on (111)A substrate (Figure 5b). These growth directions again indicated that (111)B planes are the growth fronts on (111)A substrates. 10 A few planar nanowires were observed in Figure 5a and b as highlighted by the arrows. They were mostly shorter than free-standing nanowires. That could be because arsenic vapors are first consumed by free-standing nanowire growth before diffusing into the substrate surface. Additionally, the critical diameters for free-standing and planar nanowires grown under our pressure and temperature conditions can be calculated by setting Δμ f and Δμ p to 0 in eqs 3 and 4, respectively: 4σΩ df = 2r = kt ln( p / p ) As (5) d p 1.46σΩ = 2r = kt ln( p / p ) As (6) The critical diameters for free-standing and planar nanowires were calculated as 18.3 and 6.7 nm, respectively (see Supporting Information). This calculation indicates the possibility of achieving thinner planar nanowires compared to the free-standing case. Almost all of free-standing nanowires grown using 5 20 nm Au nanoparticles have diameters of larger than 18 nm, which is consistent with 18.3 nm as the freestanding nanowires critical diameter calculated here. By using 2 nm Au particles with some possible agglomeration and other identical conditions used for Figure 5a, planar nanowires with diameters mostly in range of nm were produced on (100) substrates (Figure 5c). For (111)B substrates, 2 nm Au nanoparticles lead to a few vertical nanowires thicker than 20 nm and no growth of nanowires with diameters below 18 nm. Such a difference in the critical diameter supports the model based on the Gibbs Thomson equations. In conclusion, we have demonstrated the InAs planar nanowire growth on InAs (100) substrates using a simple vapor deposition method. The growth is found to be epitaxial on the substrate with a catalyst nanowire interface of the (111)B plane and a wire axis direction of <0 11>. Compared with epitaxial growth of vertical InAs nanowire on (111)B substrates under the same growth condition, the planar InAs nanowire growth was found to be terminated at a later time and with a larger growth rate. We have derived Gibbs Thomson equations based on different geometries of free-standing and planar nanowires and gained further understanding of the difference between planar growth and free-standing growth. As predicted from the revised Gibb-Thomson equation, selfaligned free-standing nanowire growth was achieved on InAs (100) and (111)A substrates, under a significant lower H 2 flow rate. Planar nanowires grown were also predicted to have a smaller critical diameter with respect to free-standing nanowires, which was confirmed experimentally. Collectively, these experimental results supported the model, which provided insights on controlled growth of self-aligned planar nanowires. Current configuration where InAs nanowire growth is on InAs substrates is not suitable to be transformed directly to device fabrication. The expansion of the current work to a hetero substrate with a different bandgap is required for practical largescale device applications. ASSOCIATED CONTENT *S Supporting Information Sample preparation for cross-sectional TEM images, identification of Au catalyst-inas nanowire interface orientation, indium weight detection using the ICP MS method, estimation of arsenic partial pressure, and supersaturation and critical diameter calculation for free-standing and planar nanowire growth. This material is available free of charge via the Internet at AUTHOR INFORMATION Corresponding Author * yang@purdue.edu. Present Address K.J.: Department of Electrical and Computer Engineering, University of Illinois at Urbana Champaign, Urbana, IL 61801, United States. Author Contributions Y.Z. and K.J. contributed equally to this work. Notes The authors declare no competing financial interest. 2790

6 ACKNOWLEDGMENTS This work is supported by National Science Foundation DMR grant and Army Research Office grant EL. REFERENCES (1) Cui, Y.; Zhong, Z.; Wang, D.; Wang, W. U.; Lieber, C. M. Nano Lett. 2003, 3 (2), (2) Cui, Y.; Wei, Q.; Park, H.; Lieber, C. M. Science 2001, 293 (5533), (3) Kind, H.; Yan, H.; Messer, B.; Law, M.; Yang, P. Adv. Mater. 2002, 14 (2), (4) Xu, S.; Qin, Y.; Xu, C.; Wei, Y.; Yang, R.; Wang, Z. L. Nat. Nanotechnol. 2010, 5 (5), (5) Lu, W.; Lieber, C. M. Nat. Mater. 2007, 6 (11), (6) Tomioka, K.; Motohisa, J.; Hara, S.; Fukui, T. Nano Lett. 2008, 8 (10), (7) Zi, Y.; Zhao, Y.; Candebat, D.; Appenzeller, J.; Yang, C. ChemPhysChem 2012, 13 (10), (8) Dick, K. A.; Deppert, K.; Ma rtensson, T.; Mandl, B.; Samuelson, L.; Seifert, W. Nano Lett. 2005, 5 (4), (9) Dick, K. A.; Caroff, P.; Bolinsson, J.; Messing, M. E.; Johansson, J.; Deppert, K.; Wallenberg, L. R.; Samuelson, L. Semiconductor Sci. Technol. 2010, 25 (2), (10) Mattila, M.; Hakkarainen, T.; Jiang, H.; Kauppinen, E. I.; Lipsanen, H. Nanotechnology 2007, 18 (15), (11) Fortuna, S. A.; Wen, J.; Chun, I. S.; Li, X. Nano Lett. 2008, 8 (12), (12) Zhang, X.; Zou, J.; Paladugu, M.; Guo, Y.; Wang, Y.; Kim, Y.; Joyce, H. J.; Gao, Q.; Tan, H. H.; Jagadish, C. Small 2009, 5 (3), (13) Dayeh, S. A.; Picraux, S. T. Nano Lett. 2010, 10 (10), (14) Fro berg, L. E.; Seifert, W.; Johansson, J. Phys. Rev. B 2007, 76 (15), (15) Jiang, X.; Xiong, Q.; Nam, S.; Qian, F.; Li, Y.; Lieber, C. M. Nano Lett. 2007, 7 (10), (16) Moll, N.; Scheffler, M.; Pehlke, E. Phys. Rev. B 1998, 58 (8), (17) Fortuna, S. A.; Dowdy, R.; Xiuling, L. In Epitaxial III-V planar nanowires: Self-aligned, high-mobility and transfer-printable, Indium Phosphide & Related Materials (IPRM), 2010 International Conference on, May 31 June 4, 2010; pp 1 4. (18) Wood, C. E. C.; Singer, K.; Ohashi, T.; Dawson, L. R.; Noreika, A. J. J. Appl. Phys. 1983, 54 (5), (19) Pupp, C.; Murray, J. J.; Pottie, R. F. J. Chem. Thermodyn. 1974, 6 (2), (20) Hall, R. N. J. Electrochem. Soc. 1963, 110 (5), (21) Givargizov, E. I. Highly anisotropic crystals; D. Reidel Pub. Co.: Dordrecht, The Netherlands, (22) Smith, E. B. Basic chemical thermodynamics; Imperial College Press: London, (23) Nesmeianov, A. N. Vapor pressure of the chemical elements; Elsevier Pub. Co.: New York, (24) Park, H. D.; Gaillot, A.-C.; Prokes, S. M.; Cammarata, R. C. J. Cryst. Growth 2006, 296 (2), Nano s 2791