The mechanism of ZnO nanorod growth by vapor phase transportation

ARTICLE IN PRESS Physica E 42 (2010) 2285–2288

Contents lists available at ScienceDirect

Physica E journal homepage: www.elsevier.com/locate/physe

The mechanism of ZnO nanorod growth by vapor phase transportation Seungjun Oh a, Mina Jung c, Jieun Koo a, Youngji Cho a, Sungkuk Choi b, Samnyung Yi a, Gyungsuk Kil b, Jiho Chang a,n a

Department of Applied Science, National Korea Maritime University, Busan, Republic of Korea Department of Electrical and Electronics Engineering, National Korea Maritime University, Busan, Republic of Korea c Institute for Materials Research, Tohoku University, Sendai, Japan b

a r t i c l e in fo

abstract

Article history: Received 22 January 2010 Received in revised form 3 May 2010 Accepted 5 May 2010 Available online 13 May 2010

We studied a kinetic model of the ZnO nanorod (NR) growth that occurs via vapor phase transportation (VPT). The mechanism of ZnO NR growth was discussed in terms of the length-to-diameter (L–D) relationship of NRs, and it was confirmed by varying both the type of catalyst and source material used. Both theoretical calculations and experimental data indicated that elongation of NRs is primarily determined by diffusion of the absorbed atoms (adatoms) within a nominal migration length on the surface, which is not considered in the simple vapor–liquid–solid model. & 2010 Elsevier B.V. All rights reserved.

Keywords: ZnO Nanorod Growth mechanism Diffusion-induced VLS

1. Introduction One-dimensional (1D) nanostructures, such as nanotubes, nanowires and nanobelts, have attracted considerable attention because of their interesting physical properties and various potential applications [1–5]. Among the 1D semiconductor nanostructures, ZnO nanorods (NR) are of particular interest for application to optoelectronic nanodevices due to their unique properties, such as a large exciton binding energy. Several methods have been employed to synthesize 1D ZnO nanostructures [6–9]. In particular, ZnO NRs are often grown via a catalystassisted vapor–liquid–solid (VLS) mechanism. During growth, the catalyst absorbs the vapor components, such as Zn (vapor) and ZnxO (x o1, vapor), to form a eutectic alloy. When the catalyst becomes supersaturated, crystallization and growth occur [10,11]. Therefore, the diameter of the catalyst in the liquid phase determines the diameter of the NR, and the amount of growth species supplied determines NR length. Consequently, there is a linear relationship between NR length and diameter because liquid catalysts with larger diameters absorb more of the growth species than catalysts with small diameters. However, this mechanism does not account for the role of adatoms that are outside the eutectic catalyst. Thus, the ability to control the diameter of the eutectic catalyst is essential for regulation of NR

n

Corresponding author. Fax: + 82 2 410 4783. E-mail address: [email protected] (J. Chang).

1386-9477/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.physe.2010.05.005

growth. However, there are many examples of previously reported results that cannot be explained by the simple VLS mechanism. For example, growth of ZnO NRs has been reported to occur via a self-catalyzed VLS mechanism [12]. The results of the aforementioned study suggest that during the initial stage of NR growth, either a Zn or ZnOx droplet formed and subsequently acted as a catalyst. A vapor-solidification (VS) mechanism has been reported as another potential mechanism by which ZnO NR growth occurs. The staircase-like morphology of the NRs, which consisted of terraces and steps, was explained by the layer-bylayer growth that occurred on the top of the NRs [13]. These earlier studies provide strong evidence that adatom kinetics have a significant effect on both length–diameter relationship and the morphology of NRs. Since the controlled synthesis of NRs requires a clear understanding of the growth mechanism, studies of the NR growth mechanism are necessary. The aim of the present study was to investigate the mechanism of ZnO NR growth via vapor phase transportation (VPT). The effects of both types of catalyst and the presence of impurities on the growth mechanism will be discussed in terms of the length– diameter (L–D) relationship of NRs.

2. Experimental ZnO NRs can be grown by means of various methods, including metal organic vapor phase epitaxy (MOCVD), molecular beam epitaxy (MBE), physical vapor deposition, hydrothermal

ARTICLE IN PRESS 2286

S. Oh et al. / Physica E 42 (2010) 2285–2288

Fig. 1. Schematic drawing of an experimental set-up.

processes, and so on. Among these methods, growth of ZnO NRs by vapor phase transportation (VPT) in a horizontal tube furnace is the simplest and most widely used method (Fig. 1). In the present study, ZnO and In-doped ZnO NRs were grown on AuGeand Ti-deposited Si(1 1 1) substrates using a horizontal reaction furnace. The composition x of Au1  xGex was 0.43; hence, the melting temperature was estimated to be 560 1C [14]. As shown in Fig. 1, the source materials, i.e., powder phase Zn (4 N, mesh¼50 mm) and In (5 N, mesh¼50 mm), were loaded into a quartz tray, which had two parts separated by 15 mm: the lower part of the tray was for loading the source material and the upper part was for substrate-loading. The evaporated vapor was supplied to the substrates through a via-hole in the upper part of the tray. When the temperature of the furnace reached the growth temperature (Tg), the quartz tray was inserted into the furnace. The ZnO and In-doped ZnO (ZnO:In) NRs were grown at 625 1C. After 30 min, the quartz tray was ejected and allowed to cool in air. Dry N2, which was used as the ambient gas, was allowed to flow through the quartz tube at a rate of 500 (ml/min) during growth. The surface morphology of each sample was observed using a Quanta 200 FEG scanning electron microscope (SEM) with a field emission gun. The In concentration in the ZnO:In NRs was determined by auger electron spectroscopy (AES) and energy dispersive spectroscopy (EDX) [15].

3. Results and discussion Fig. 2 shows SEM images of (a) ZnO NRs/Ti/Si(1 1 1) and (b) ZnO NRs/AuGe/Si(1 1 1); the insets show plain-view images of each sample. Despite the difference in melting temperature between the catalysts, we were able to grow well-aligned ZnO NRs with similar diameters. The lengths and diameters of the NRs were evaluated using the plain- and side-view SEM images, and the results are summarized in Fig. 3. NR length was not linearly correlated with NR diameter. Hence, we concluded that the conventional VLS model could not explain the length-to-diameter (L–D) dependence observed in our results. Furthermore, since Tg (625 1C) was considerably lower than the melting temperature of Ti (1668 1C), nucleation in the Ti-droplet was not expected. Note that VPT growth supports a long diffusion length, since reductions in kinetic energy on the substrate surface can be neglected. Therefore, the contribution of adatom diffusion to the overall growth of NRs should be carefully investigated. The diffusion-induced (DI) mechanism of NR growth has been thoroughly explained by Dubrovskii et al. [16], and it is worthwhile to consider the fundamentals of their explanation. According to the DI mechanism, the atoms that appear in the drop may come from both the source and the substrate surface due to diffusion. In the steady state, the normal growth rate of a NR dL/dt is given by [16]   pR2 dL VVs 2Crl  pR2 þ jL ð1Þ ¼ O dt O tl

Fig. 2. Cross-sectional SEM images of ZnO NRs on various catalysts, the insets are plain-view images of (a) ZnO NRs/Ti/Si(1 1 1) and (b) ZnO NRs/AuGe/Si(1 1 1).

Fig. 3. Experimental and theoretical L–D dependencies for (a) ZnO NRs on Ti/Si(1 1 1) and (b) ZnO NRs on AuGe/Si(1 1 1). The theoretical curves (solid lines) were obtained using Eq. (2).

ARTICLE IN PRESS S. Oh et al. / Physica E 42 (2010) 2285–2288

2287

here, L(t) is the NR length at time t measured from the surface of the epitaxial layer. R is the radius of NR, Vs is the growth rate of non-activated surface, V is the deposition rate, O is the volume per atom in the crystal, C is the volume concentration of alloy, rl is the interatomic distance in the liquid phase, tl is the mean lifetime of atoms in the liquid, and jL is the diffusion flux of adatoms towards the top of NR. The first term in Eq. (1) represents adsorption on the liquid surface, the second term stands for desorption and the third term describes the DI contribution to the growth rate. In many cases, this equation can be simplified as follows [17]:   D H ð2Þ L ¼ cþ D where H is the effective thickness (H¼Vt), Dn is characteristic diameter at which the DI effects become predominant, and c is constant related to adsorption and desorption rate. V is the deposition rate. The constant c is related to (e  g). The coefficient e  (V  Vs)/V is the relative difference between the V and Vs. The coefficient g accounts for desorption from the surface of the drop,

gffi

0:2 W

ð3Þ

where W is the rate of deposition in the mono-layers per seconds (ML/s). The constant c represents the difference between the adsorption–desorption VLS growth rate (1  g)V and the surface growth rate Vs ¼(1  e)V; this term can be either positive or negative. Eq. (2) contains the characteristic diameter Dn. In other words, it is the effective diffusion length of adatoms on the surface, which, in turn, determines the L–D dependence of NRs. For the ZnO NRs grown on Ti/Si(1 1 1), the effective thickness H and non-activated layer thickness Hs were measured in the sideview images of the NRs to be 420 and 300 nm, respectively. The deposition rate W, which was calculated from the total amount of ZnO deposited during growth, was 0.9 ML/s. For ZnO NRs on AuGe/Si(1 1 1), H was 1778 nm, Hs was 689 nm, and W was 3.8 ML/s. Using these values, the theoretical L–D dependence of each NR was plotted in Fig. 3. The best-fitting results for ZnO NRs on Ti/Si(1 1 1) and AuGe/Si(1 1 1) were obtained at Dn ¼250 and 180 nm, respectively. Thus, the adatom diffusion length for Ti/Si(1 1 1) is longer than that for AuGe/Si(1 1 1). The diffusion length on the surface (ls) is given by [18]

ls ¼ lexp½ðDGdep DGs Þ=2 kT

ð4Þ

where l is a constant that determines kinetic motion of adatom on the surface, kT is the Boltzmann constant times absolute temperature, and DGdep is the binding energy of the adatom to the surface. DGs is the activation energy for surface diffusion. Both the binding energy and the activation energy for surface diffusion influence the diffusion length. When one considers that DGs is mostly determined by Tg, a higher surface energy results in a longer diffusion length in the nucleation-limited growth regime. Also, the surface energy of Ti (STi) was higher than SSi and SAuGe at the growth temperature of 625 1C (STi  1900 ergs/cm2, SSi  800 ergs/cm2, SAuGe  730 ergs/cm2) [19,20]. Therefore, we attributed the longer diffusion length of the adatoms on the Ti/Si(1 1 1) surface to the increased surface energy. If this assumption is correct, incorporation of impurity atoms during NR growth would also affect the growth mechanism because the effective binding energy of the adatom to the surface would be altered by incorporation of impurities. Fig. 4 shows SEM images of In-doped ZnO NRs grown on AuGe/Si(1 1 1) with different In compositions; the insets show the plain-view images of each sample. Both the experimental and theoretical L–D dependencies are shown in Fig. 5. A reciprocal proportional relationship, i.e., L  (A/D) was observed in each

Fig. 4. Cross-sectional SEM images of In-doped ZnO NRs on AuGe-coated Si(1 1 1) substrates with different In compositions; the insets are plain-view images of (a) pure ZnO NRs, (b) ZnO:In0.27 NRs, and (c) ZnO:In0.33 NRs.

sample. The characteristic diameter Dn was obtained by curvefitting using Eq. (2). Typically, both the sticking coefficient and the migration length of an impurity atom would be considerably different from

ARTICLE IN PRESS 2288

S. Oh et al. / Physica E 42 (2010) 2285–2288

variation in the sticking coefficient of the adatoms. Curve-fitting of the experimental results indicated that the effective diffusion length (Dn) varied with both the type of catalyst and the adatom species.

Acknowledgements This research was supported by the MKE (The Ministry of Knowledge Economy), Korea, under the ITRC (Information Technology Research Center) support program supervised by the NIPA (National IT Industry Promotion Agency) (NIPA-2010C1090-1021-0015). Gyungsuk Kil and Jiho Chang were supported by Energy and Resource Technology R&D Program (2008-E-EL-02P-02-0-000-2008) of MKE. References

Fig. 5. Experimental and theoretical L-D dependencies for (a) pure ZnO NRs, (b) ZnO:In0.27 NRs, and (c) ZnO:In0.33 NRs. The theoretical curves (solid lines) were obtained using Eq. (2).

those of the host atoms. In has a considerably lower vapor pressure (  10  6 Torr at 625 1C) than Zn (  10 Torr at 625 1C). Hence, the nominal diffusion length of the adatoms was expected to vary, as was the sticking coefficient of the impurity atoms. Fig. 5 shows that effective diffusion length of the adatoms increased with the In content of the adatoms, as expected.

4. Summary and conclusions ZnO NRs were synthesized in a horizontal furnace via VPT. Both the type of catalyst and the growth species were varied to examine the growth mode. We found that the growth of ZnO NRs occurs via a diffusion-induced VLS mechanism. The growth mechanism was discussed and confirmed in terms of the L–D relationship of NRs, modification of the surface energy, and

[1] Y.J. Xing, Z.H. Xi, Z.Q. Xue, Appl. Phys. Lett. 83 (2003) 1689. [2] M.H. Huang, Y. Wu, H. Feick, N. Tran, E. Weber, P. Yang, Adv. Mater. 13 (2001) 113. [3] X.Y. Kong, Z.L. Wang, Appl. Phys. Lett. 84 (2004) 975. [4] W.L. Hughes, Z.L. Wang, J. Am. Chem. Soc. 126 (2004) 6703. [5] Z.L. Wang, X.Y. Kong, J.M. Zuo, Phys. Rev. Lett. 91 (2003) 185502. [6] J.J. Wu, S.C. Liu, Adv. Mater. 14 (2002) 215. [7] Y.C. Kong, D.P. Yu, B. Zhang, Appl. Phys. Lett. 78 (2001) 407. [8] C.X. Xu, X.W. Sun, B.J. Chen, Chem. Phys. Lett. 20 (2003) 1319. [9] C.X. Xu, X.W. Sung, B.J. Chen, J. Appl. Phys. 95 (2004) 661. [10] R.S. Wagner, W.C. Ellis, Appl. Phys. Lett. 4 (1964) 89. [11] E. McGlynn, J. Grabowska, K.K. Nanda, Surf. Coat. Tech. 20 (2005) 1093. [12] H. Tang, Z. Ye, L. Zhu, H. He, B. Zhao, M. Zhi, Z. Yang, Physica E 40 (2008) 507. [13] W.I. Park, D.H. Kim, S.-W. Jung, G.-C. Yi, Appl. Phys. Lett. 80 (2002) 4232. [14] H. Okamoto, T.B. Massalski, Bull. Alloy Phase Diagr. 5 (1984) 601. [15] M.N. Jung, J.E. Koo, S.J. Oh, B.W. Lee, W.J. Lee, S.H. Ha, Y.R. Cho, J.H. Chang, Appl. Phys. Lett. 94 (2009) 041906. [16] V.G. Dubrovskii, G.E. Cirlin, I.P. Soshnikov, A.A. Tonkikh, N.V. Sibirev, Y.B. Samsonenko, V.M. Ustinov, Phys. Rev. B 71 (2005) 205325. [17] V.G. Dubrovskii, I.P. Soshnikov, N.V. Sibirev, G.E. Cirlin, V.M. Ustinov, J. Cryst. Growth 289 (2006) 31. [18] K.N. Ti, J.W. Mayer, L.C. Feldman, Electronic Thin Film Science for Electrical Engineers and Materials Scientists, Macmillan Publishing Company, 1992. [19] L.E. Murr, Interfacial Phenomena in Metals and Alloys, Addison-Wesley, 1975. [20] Y.V. Nadich, V.M. Perevertailo, L.P. Obushchak, Poroshkovaya Metallurgiya 149 (5) (1975) 73.