Investigation of Nucleation Mechanism and Tapering Observed in ZnO Nanowire Growth by Carbothermal Reduction Technique
© Kar et al. 2010
Received: 20 June 2010
Accepted: 5 August 2010
Published: 19 August 2010
ZnO nanowire nucleation mechanism and initial stages of nanowire growth using the carbothermal reduction technique are studied confirming the involvement of the catalyst at the tip in the growth process. Role of the Au catalyst is further confirmed when the tapering observed in the nanowires can be explained by the change in the shape of the catalyst causing a variation of the contact area at the liquid–solid interface of the nanowires. The rate of decrease in nanowire diameter with length on the average is found to be 0.36 nm/s and this rate is larger near the base. Variation in the ZnO nanowire diameter with length is further explained on the basis of the rate at which Zn atoms are supplied as well as the droplet stability at the high flow rates and temperature. Further, saw-tooth faceting is noticed in tapered nanowires, and the formation is analyzed crystallographically.
KeywordsZnO Nanowires Microscopy
Interest in nanowires continues to grow fueled by applications in electronics, optoelectronics, sensors, piezoelectric and thermoelectric devices, and energy storage . In spite of considerable advances in growth and application development of nanowires, the various proposed growth mechanisms are still controversial and subject to immense discussion. For example, it is well documented that the diameter of nanowires grown via the vapor–liquid–solid (VLS) mechanism is determined by the size of the droplet. This is true but it does not necessarily imply that the diameter of nanowires is constant along its axis. It has been recently reported that the dynamic reshaping of the catalyst particles during the nanowire growth determines the length and shape of the nanowires . Also, just as in elemental semiconductors , there is a general consensus that even for oxide nanowires the whole molten alloy particle, referred to as the catalyst, rises above the surface of the substrate and rides at the tip of the nanowire during the growth process. The objective of this paper is twofold. First, we investigate the initial stages of nucleation, oxidation of Zn atoms, and growth of ZnO nanowires. Secondly, we investigate the formation of tapered nanowires from the growth kinetics point of view. Compared to elemental and III–V nanowires, growth behavior of semiconducting oxide nanowires, and in particular ZnO is not well understood . ZnO has been proven to be quite a complex and interesting material with a variety of structures such as nanowires, nanobelts, and tetrapods . Each of these structures can be formed by different growth mechanisms under widely different thermodynamic conditions. The recent surge in applications of ZnO nanowires as a piezoelectric material  for energy harvesting has led to the present investigation. The dependence of nanowire diameter on the amount of generated piezoelectricity requires clarification of the role of gold at the nanowire tip in controlling the shape and diameter of the nanowire .
The source consists of zinc oxide (ZnO) metal basis of 99.999% purity mixed with graphite in a weight ratio of 1:1 to carry out a carbothermal reduction process. A 1" diameter quartz tube was inserted inside an isothermal furnace, and the source mixture was kept in a quartz boat inside this tubular reactor. Gold colloids were used as the catalyst for different experiments. The substrate with the Au catalyst was placed downstream from the quartz boat located at the center of the heating zone. One end of the quartz tube was connected to a mass flow controller, which controls the flow rate of the carrier gas, argon, and the other end was connected to the exhaust. Nanowires were grown within a temperature window of 900–980°C, whereas the carrier gas flow rate was varied between 100–160 sccm.
Results and Discussion
The oxidation process of the Zn leading to the formation of ZnO contributing to the nanowire growth is a critical step. In the present case, it is likely that ZnO nanowires originate from the oxidation of the Zn atoms within the Au–Zn alloy particle since it is commonly known that the activity of metals can be increased upon alloying. It has been previously reported that the oxidation of alloys such as Zn–Ag and Zn–Cu results in the formation of ZnO crystalline precipitates . In a similar manner, it is believed ZnO will precipitate in the form of ZnO nanowires due to the oxidation of Au–Zn particles. There have also been recent claims stating that there is no involvement of a liquid or solid catalyst at the tip of ZnO nanowires in the growth process involving thermal evaporation . The SEM images in Figures 1d, e and 2d here show a distinct catalyst at the nanowire tip that serves to control the diameter of the nanowire.
We propose here that Zn atoms are the main species contributing to crystal growth and not ZnO. The gold particles alloy with Zn and the oxide nanowires grow with the assistance of the liquid catalyst particles at the wire tip where the Zn atoms are oxidized, as discussed previously. Since the ZnO crystal nucleates at the solid–liquid interface, the nanowire diameter is determined and controlled by the size of the gold catalyst droplet at the tip, which is a feature common in VLS process [1, 9]. We also point out contradicting reports  wherein nanowire branches with substantially different diameters compared to the catalyst particle have been seen; a growth mechanism different from VLS was suggested, and ZnO atoms were believed to be the source of nanowire growth instead of Zn atoms.
which means that the droplet approaches a larger solid angle of spherical section. Here, σ l , σ ls , and σ s are the surface tension of the droplet surface, the liquid–solid interface, and the solid, respectively. The droplet at the tip of the nanowire has a contact area of radius R and τ is the line tension. The increase in the contact angle β causes a decrease in the contact area and a decrease in the radius R of the droplet. Consequently, the radius of the nanowire R should be smaller than the initial radius r' of the contact area of the nuclei on the Au–Zn droplet. Thus, the nanowire diameter is largest at the base and decreases as the length of the wire increases, with the diameter being directly controlled by the shrinking radius of the contact area R at the droplet tip. The average rate of decrease in the nanowire diameter with time (tapering rate) for needle-like nanowires here is estimated as 0.36 nm/s and plotted in Figure 4d. Also, plotted is the growth rate of the nanowires, and the average rate throughout the process is estimated as 3.9 nm/s. However, the growth rate is relatively slower at the onset of growth and speeds up with time. The decrease in diameter of the nanowire with length of both needle-like nanowires and the ones with just tapered bases are plotted in Figure 4c. The rate of decrease in diameter is much larger close to the base and then decreases as the length increases. This can be explained by the fact that the rate of increase of angle α with length is large close to the base of the nanowire and decreases gradually. Thus, one reason for the decrease in nanowire diameter with length, i.e. tapering, is the reduction in the size of the catalyst at the tip as growth progresses.
Nanowires of two different morphologies are recalled here: in one case, we have the tapering just restricted to the base, as seen in Figure 4b, and another where we see a continuous decrease in nanowire diameter with length, as in Figure 3b–e; both morphologies can be explained on the basis of the rate at which the Zn atoms are deposited/delivered to the substrate if the temperature is kept constant. The concentration of the Zn adatoms at the substrate will increase as the Ar carrier gas flow rates are increased. Due to the concentration gradient between the substrate surface and the nanowire, diffusion of the adatoms becomes prominent at flow rates >140 sccm. Excess growth species are available at the base of the ZnO nanowire, where the mobility of Zn atoms diffusing from the substrate surface to the nanowire tip is impeded and allows radial tapering of the base . The base diameter thus increases steadily with an increase in flow rates. This gives rise to tapered, as in Figure 4b, but not needle-shaped nanowires, as in Figure 3. Further, it is to be noticed that the tapered segment of the nanowire in Figure 4b is very small. This is probably due to the limitation of adatom diffusion via the nanowire sidewall, since the upward adatom mobility via the nanowire sidewalls decreases at high temperatures . However, as the flow rates are increased further, the tapered nanowires give way to needle-like nanowires, obtained in this case at a flow rate of 160 sccm as shown in Figure 3b–e where there is a constant decrease in diameter with length. Such observations have been reported previously in the growth of InAs nanowires . The tapering also indicates that adatom surface diffusion from the substrate up the nanowire sidewalls forms a path for the growth species reaching the alloy droplet other than the direct impingement of the atoms on the droplet, as has been reported previously . A theoretical study  investigating the shape of the nanowires on the basis of the contact angle β of the liquid droplet at the nanowire tip found surprisingly that tapered nanowire growth (∂ > 0) is more likely for a wide range of contact angles β compared to nanowires with uniform diameters.
Oversupply of source vapor as well as interplay of surface energies of the wire and liquid droplet was also reported to cause the droplet to be unstable, leading to oscillations in resulting Si NW structures . This is seen here in the case of the tapered ZnO wire in Figure 5d, and such oscillations of the droplet resulting in faceting in the nanowire sidewalls is discussed below in detail. To our knowledge, this is the first observation of periodic saw-tooth faceting in ZnO nanowires; the observed faceting is periodic, where the period P of the facets is about 19.2 nm on average and the height H is about 4 nm. Incidentally, such faceting is observed only in tapered nanowires and not in straight ones as seen in Figure 5d, indicating such faceting might have some relation to the tapering mechanism observed in our case.
One mechanism that could lead to such faceting is surface reconstruction due to charge stabilization in polar compounds. ZnO crystal is formed by alternating stacks of oppositely charged O2- and Zn2+ planes parallel to the surface. If the resulting dipole moment perpendicular to the surface is nonzero, stabilization of such a surface is accomplished by a rearrangement of surface charges or by introducing compensating charges into the outermost surface planes . This could lead a significant modification of the surface geometric structure and stoichiometry. The stabilization mechanism for the Zn-terminated face of ZnO has been experimentally investigated by a variety of techniques, and various mechanisms have been proposed for the reduction in the surface charge density to yield a stable Zn termination. Recent experimental studies  combined with theoretical calculations  suggest that the Zn-terminated surface can be stabilized by a reconstruction involving triangular surface structures . However, due to the unstable nature of the catalyst at the wire tip, it is believed here that faceting due to droplet oscillation is the dominant mechanism that causes the surface reconstructions. The observations seen here can be explained based on a thermodynamic model used earlier to explain similar faceting in Si nanowires . The allowed facets correspond to a wire that is widening or narrowing as it grows. The wire widens as the droplet is stretched thinner and contact angle β (Figure 4a) decreases, which generates an inward force favoring the introduction of the narrowing facet. Conversely, the narrowing of the wire leads to the droplet applying an increasing outward force on the wire, favoring an introduction of the widening facet. This oscillatory mechanism leads to the periodic faceting seen here.
In this investigation, using the carbothermal reduction technique, ZnO nanowires are found to grow from nuclei on the molten Au–Zn clusters. A tapering of diameter along the axis is observed with the largest diameter at the base. This may be due to the change in the contact angle β of the catalyst droplet at the nanowire tip causing a change in the contact area R at liquid–solid interface. The growth rate of the needle-like nanowires is found to be 3.9 nm/s and the tapering rate is established to be 0.36 nm/s on the average. Finally, the rate of addition of Zn atoms is found to control the tapering where the tapering limited to the nanowire base gives way to a continuous taper along the nanowire length at higher deposition rates. The tapered wires are also found to have unstable droplets at their tips, which are believed to be a cause of tapering.
A.K would like to thank Dr. Shadi Dayeh and Dr. Ranadeep Bhowmick for helpful discussions and suggestions. We thank Professor Sreeram Vaddiraju for his insightful comments and help with the manuscript. Part of this work was supported by the WCU-ITCE Program at Postech funded by the Korea Science and Engineering Foundation from the Ministry of Education, Science, and Technology.
- Meyyappan M, Sunkara MK: Inorganic Nanowires: Applications, Properties and Characterization. (Boca Raton, FL: CRC press); 2010. see Chapter 9 for Oxide nanowires see Chapter 9 for Oxide nanowiresGoogle Scholar
- Hannon JB, Kodambaka S, Ross FM, Tromp RM: Nature Lett. 2006, 440: 69. 10.1038/nature04574View ArticleGoogle Scholar
- Wu Y, Yang P: J Am Chem Soc. 2001, 123: 3165. 10.1021/ja0059084View ArticleGoogle Scholar
- Wang Z, Ma XY, Song JW, Yao JH: Nano-Micro Lett. 2009, 1: 45.View ArticleGoogle Scholar
- Wang ZL, Song J: Science. 2006, 312: 242. 10.1126/science.1124005View ArticleGoogle Scholar
- Stan G, Ciobanu CV, Parthangal PM, Cook RF: Nano Lett. 2007, 7: 3691. 10.1021/nl071986eView ArticleGoogle Scholar
- Vellinga WP, DeHosson JTM: Acta Mater. 1997, 45: 933. 10.1016/S1359-6454(96)00252-2View ArticleGoogle Scholar
- Kim DS, Gosele U, Zacharias M: J Cryst Growth. 2009, 311: 3216. 10.1016/j.jcrysgro.2009.03.026View ArticleGoogle Scholar
- Givargizov EI: J Cryst Growth. 1975, 31: 20. 10.1016/0022-0248(75)90105-0View ArticleGoogle Scholar
- Kim DS, Scholz R, Gosele U, Zacharias M: Small. 2008, 4: 1615. 10.1002/smll.200800060View ArticleGoogle Scholar
- Chang YC, Chen LJ: J Phys Chem C. 2007, 111: 1268. 10.1021/jp066786qView ArticleGoogle Scholar
- Schmidt V, Senz S, Gosele U: Appl Phys A. 2005, 80: 445. 10.1007/s00339-004-3092-1View ArticleGoogle Scholar
- Dayeh SA, Yu ET, Wang D: Nano Lett. 2007, 7: 2486. 10.1021/nl0712668View ArticleGoogle Scholar
- Reep DJ, Ghandhi SK: J Electrochem Soc. 1984, 131: 2697. 10.1149/1.2115386View ArticleGoogle Scholar
- Nebol'sin VA, Shchetinin AA: Inorg Mater.39(203):899.Google Scholar
- Mohammad SN: Nano Lett. 2008, 8: 1532. 10.1021/nl072974wView ArticleGoogle Scholar
- Ross FM, Tersoff J, Reuter MC: Phys Rev Lett. 2005, 95: 146104. 10.1103/PhysRevLett.95.146104View ArticleGoogle Scholar
- Tasker PW: J Phys C. 1979, 12: 4977. 10.1088/0022-3719/12/22/036View ArticleGoogle Scholar
- Diebold U, Koplitz LV, Dulub O: Appl Surf Sci. 2004, 237: 336.View ArticleGoogle Scholar
- Dulub O, Diebold U, Kresse G: Phys Rev Lett. 2003, 90: 016102. 10.1103/PhysRevLett.90.016102View ArticleGoogle Scholar
- Ostendorf F, Torbrugge S, Reichling M: Phys Rev B. 2008, 77: 041405R. 10.1103/PhysRevB.77.041405View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.