J. Phys. Chem. B 2005, 109, 2733-2738
2733
Controllable Synthesis and Optical Properties of Novel ZnO Cone Arrays via Vapor Transport at Low Temperature Xinhai Han,†,‡ Guanzhong Wang,*,†,‡ Jiansheng Jie,†,‡ Wallace C. H. Choy,§ Yi Luo,§ T. I. Yuk,§ and J. G. Hou† Hefei National Laboratory for Physical Sciences at Microscale, and Department of Physics, UniVersity of Science and Technology of China, Hefei 230026, China, and Department of Electrical and Electronic Engineering, UniVersity of Hong Kong, Pokfulam Road, Hong Kong, China ReceiVed: June 4, 2004; In Final Form: October 15, 2004
Novel ZnO cone arrays with controllable morphologies have been synthesized on silicon (100) substrates by thermal evaporation of metal Zn powder at a low temperature of 570 °C without a metal catalyst. Clear structure evolutions were observed using scanning electron microscopy: well-aligned ZnO nanocones, doublecones with growing head cones attached by stem cones, and cones with straight hexagonal pillar were obtained as the distance between the source and the substrates was increased. X-ray diffraction shows that all cone arrays grow along the c-axis. Raman and photoluminescence spectra reveal that the optical properties of the buffer layer between the ZnO cone arrays and the silicon substrates are better than those of the ZnO cone arrays due to high concentration of Zn in the heads of the ZnO cone arrays and higher growth temperature of the buffer layer. The growth of ZnO arrays reveals that the cone arrays are synthesized through a selfcatalyzed vapor-liquid-solid (VLS) process.
Introduction The synthesis of nanomaterials with controllable morphology, size, and chemical composition is very important to nanoscale science because the properties of the nanomaterials are usually sensitive to these aspects.1 Many nanomaterials, such as elemental semiconductors (Si2 and Ge3), nitrides (GaN4 and AlN5), carbides (SiC6 and TiC7), and oxide systems (CdO, SnO2, Ga2O3, In2O3,8 TiO2,9 and ZnO10) have been successfully synthesized. Among them, semiconductor ZnO, which has a wide direct band-gap of 3.37 eV at room temperature and a large exciton binding energy of 60 meV, has attracted much attention for practical applications such as transparent electrodes,11 gas sensors,12,13 acousto-optical devices,14 nanoscale lasers,15 and piezoelectric devices.16 There have been syntheses of ZnO nanowires,15 nanobelts,8,17 ZnO-Zn nanocables and ZnO nanotubes,18 nanocantilevers,19 and tetragonal-ZnO phases.20 The synthesis approaches include vapor-phase evaporation,15,17,19,21 solution-based methods,22,23 laser ablation,24 and so forth. In this paper, we report the controllable synthesis of several remarkable ZnO cone array structures with a clear structural evolution by a simple thermal evaporation of metal Zn powder. The optical properties of the ZnO cone arrays with different morphologies are also investigated. Experimental Section The synthesis of ZnO cone arrays with controllable morphologies was carried out in a conventional furnace with a horizontal alumina tube, as schematically shown in Figure 1. * Corresponding author. Tel.: +86-551-3600075. Fax: +86-5513601073. E-mail:
[email protected]. † Hefei National Laboratory for Physical Sciences at Microscale, University of Science and Technology of China. ‡ Department of Physics, University of Science and Technology of China. § University of Hong Kong.
Figure 1. Schematic illustration of the evaporation system for the synthesis of ZnO cone arrays.
Zinc powder and silicon (100) substrates placed in an alumina boat were inserted into a quartz tube and put close to the middle of the furnace. They were located at the same horizontal level. With the silicon substrates placed at the downstream side, the separation between the Zn powder and Si substrate was only 10 mm so that there was no significant temperature deviation between them. Before the evaporation, the reaction chamber was cleaned three times using an argon gas flow, to drive out any remaining oxygen. The reaction gases were directed into the reaction system through the mass flow controllers. The synthesis process was divided into two stages. The first stage was to lead argon into the reaction system with a flow rate of 50 sccm (standard cubic centimeter per minute) into the reaction system to initiate the experiment. In the second stage of synthesizing the novel ZnO cone arrays, air with a flow rate of 4 sccm was added into the argon when the furnace temperature reached 420 °C. The tube chamber pressure was kept at 1 × 103 Pa, and the system was maintained at 570 °C for 1 h. Finally, the furnace was cooled naturally to room temperature. We found that the morphology of the products on the substrates is very sensitive to its distance from the Zn source and can be definitely controlled by adjusting the position of the substrate relative to the source materials. The as-grown products were characterized and analyzed by X-ray diffraction (XRD) (MAC MXPAHF with Cu KR radiation), field-emission scanning electron microscopy (FE-SEM) (JEOL JSM-6700F), high-resolution transmission electron microscopy (HRTEM) (JEOL 2010), and confocal laser
10.1021/jp0475943 CCC: $30.25 © 2005 American Chemical Society Published on Web 01/21/2005
2734 J. Phys. Chem. B, Vol. 109, No. 7, 2005
Han et al.
Figure 2. Optical image and XRD patterns of an as-grown sample. (a) Optical image of the as-grown sample on a silicon substrate, showing two distinct colors on the surface. The eight regions, A-H, are labeled to investigate their morphologies and optical properties. The carrier gas flow direction is along from region A to region H. (b) and (c) are XRD patterns of the black and white regions of the sample, respectively. The overwhelming (002) peaks reveal a preferred orientation of the products in both regions.
micro-Raman spectrometer (JY LABRAM-HR Ar-ion laser with 514.5 emission lines). The photoluminescence (PL) measurements were carried out at room temperature, being excited by He-Cd (325 nm) laser. Results and Discussion Figure 2a shows an optical image of a sample deposited on a silicon substrate. The most interesting phenomenon is that the colors of the as-grown products gradually change from black to white as the distance between the silicon and the source increases. To investigate the reason of the color change, XRD and FE-SEM were employed to analyze the crystalline structure and morphology of the samples. Figure 2b and c shows the XRD patterns of the black and white regions, respectively. All of the diffraction peaks can be indexed to ZnO with a hexagonal structure. The overwhelming (002) peaks reveal that the preferred orientation of the products is along the c-axis. The peak signal from the silicon substrate in Figure 2b indicates that the products in the black regions are thinner than those in the white region, which is also confirmed by the SEM results described next. Figure 3 shows the SEM images of eight different morphologies of ZnO cone arrays along the carrier gas flow direction. All of the cone arrays exhibit prefer orientation perpendicular to the substrates and almost the entire surface of the substrates are covered with the arrays, even on the sides of the substrates, as shown in Figure 3a. Figure 3b shows the head diameter of ZnO nanocone arrays is smaller than 100 nm and the length of the nanocones is longer than 8 µm. Each individual nanocone consists of a well-faceted stem and head. The diameter of the nanocones gradually decreases from the root to form a hexagonal tip. The magnified image in the inset reveals a 6-fold symmetry for both the stems and the heads, indicating that they grow along the same direction, parallel to the c-axis. As the distance between the Zn source and the substrates increases, the morphology of the
Figure 3. SEM images of eight different morphologies of ZnO cone arrays along the carrier gas flow direction. (a-h) SEM images of ZnO cone arrays, corresponding to region A-H in Figure 2a, respectively. Clear structure evolutions are observed. The insets show the magnified view revealing the shapes of the heads of the cone arrays, and the scale bars in the insets are 500 nm.
structures is also changed in an interesting manner. Heads are grown on the top of the stems to form double-cone (the head
Synthesis and Properties of ZnO Cone Arrays
Figure 4. (a, b, c) Cross-sectional SEM images of the sample, corresponding to Figure 3b, e, and h, respectively.
and stem cones) arrays, as shown in Figure 3c-g. All doublecone arrays have very flat hexagonal faceted heads, and the diameters of the cones vary from 200 nm to 3 µm, depending on the distance to the Zn source. As the Zn vapor pressure decreases, the diameter of the head cones increases faster than that of the stem cones. As a result, the double-cone arrays with such peculiar morphology are formed. Once the head diameters of the double-cone arrays are large enough to almost cover the surface, straight hexagonal pillars are formed at the top of the head cones, as shown in Figure 3h. The cross-sectional images of Figure 3b, e, and h are shown in Figure 4a, b, and c, respectively. It should be noted that the thickness of the buffer layer between the substrates and the ZnO cone arrays decreases as the distance between the substrate and the source increases. Consequently, the morphology of the products on substrate can be definitely controlled by adjusting the position of the substrate from the source materials. To the best of our knowledge, the particular gradual changing morphologies in ZnO nanostructures shown here are being reported for the first time in the literature. The morphology and microstructure of the ZnO cone arrays at B, E, and H in Figure 2a are further investigated by TEM, as shown in Figure 5. Figure 5a is the low-magnification TEM image of one ZnO nanocone corresponding to Figure 3b, which indicates gradual decrease of the diameters of the nanocone from the root to form a tip. The select-area electron diffraction (SAED) pattern from the head of the nanocone is shown in the inset, which confirmed that the growth direction of the nanocone is along the c-axis. The high-resolution TEM (HRTEM) from the left upper section of the nanocone, indicated by a black arrow, is shown in Figure 5b. The lattice constant measured from the lattice fringes is 0.26 nm, which further confirms that the nanocone grows along the [0001] c-axis direction. The edges of the nanocone are clear, and no amorphous layer is observed on the surface. Figure 5c and d shows the other two typical morphologies of ZnO cones, double-cones and cones with straight hexagonal pillar corresponding to Figure 3e and h, respectively. The SAEDs in the insets are also from the heads of the ZnO cones and identical to that of the ZnO nanocone in Figure 5a, implying that all of the ZnO cones have the same growth direction, which is consistent with the XRD results. The
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2735 crystal habit planes of the ZnO cones in Figure 5d are depicted schematically in Figure 5e. The Raman spectra at room temperature of four different regions of the ZnO cone arrays at B, D, F, and H in Figure 2a are all shown in Figure 6. All of the Raman peaks from the ZnO cone arrays agree well with the published data for c-ZnO.25,26 The peaks with Raman shifts of 100 and 437 cm-1 are the ZnO nonpolar optical phonons (E2), while the peaks at 380 and 581 cm-1 are attributed to the transversal optical (TO) modes with A1 symmetry and the longitudinal optical (LO) modes, respectively. The second-order modes are located at 208, 332, and 1050-1200 cm-1. This exhibits that all ZnO cone arrays in general maintain the crystal structure of the bulk ZnO. The 521 cm-1 peak is the optical phonon mode excited from the Si (100) substrates, and its second-order mode is located at 900-1000 cm-1. It is interesting to note that there is a gradual enhanced fluorescence background in the Raman spectra as the cone heads grow larger. This implies that the ZnO cone arrays with larger heads are of a lower optical quality. This is not consistent with the conventional results.21 We have also investigated and compared the Raman spectrum of the buffer layer between the ZnO cone arrays and the substrate in Figure 4a with the Raman spectrum of region B, as is shown in the inset of Figure 6. Transversal optical (TO) modes with E1 symmetry at 407 cm-1 and with A1 symmetry at 380 cm-1 are weak but resolvable. The fluorescence background of the buffer layer is as low as that of the ZnO nanocone arrays of Figure 3b. This indicates that the optical quality of the buffer layer is better than that of the ZnO nanocone arrays in region B and other regions as well. In Raman measurement, the Ar ion laser (514.5 nm) can penetrate into the silicon substrate through the ZnO cone arrays and the buffer layer because of the wide band nature of ZnO. Therefore, the signal from the substrates can also be observed in the Raman spectra. As a consequence, the regions with a thicker buffer layer have lower fluorescence background in Raman spectra due to the better crystal quality of the buffer layer. Room-temperature PL spectra of regions B, D, F, and H in Figure 2a are shown in Figure 7. All regions exhibit UV and broad green emission peaks. The UV emission band can be explained by the near band-edge transition of the wide bandgap ZnO cone arrays, the recombination of free excitons through an exciton-exciton collision process.27 It has been suggested that the green band emission at 505 nm is mainly due to the presence of various intrinsic defects, such as oxygen vacancy,28 interstitial zinc,29 etc. The UV peak intensity decreases and the green peak intensity increases when the heads of the ZnO cone arrays increase. This is also in agreement with the results of the Raman spectra that the ZnO cone arrays with larger heads are of poorer optical quality. It should also be noted that the UV peak positions are red-shifted when the heads of the reverse cones grow larger. This may be due to the fact that the larger heads have more defects. Interestingly, two UV peaks (377 and 387 nm) are observed in the PL spectrum of region H. According to the trend of the red-shift, the peak at 387 nm should be homologous with the UV peaks from other regions. However, it is not clear what constitutes the peak at 377 nm, but we could probably attribute it to the near band-edge emission from the straight hexagonal pillar ZnO on the top of the cone arrays in region H. The PL spectrum of the buffer layer in Figure 4a is also shown in Figure 7. A more intense UV emission and a weaker visible band are observed, indicating a higher optical quality than that of the ZnO cone arrays. Contrary to the Raman results,
2736 J. Phys. Chem. B, Vol. 109, No. 7, 2005
Han et al.
Figure 5. (a, c, d) The low-magnification TEM images of three typical ZnO cones, nanocone, double-cone, and cones with straight hexagonal pillar, corresponding to Figure 3b, e, and h, respectively. The insets are the SAEDs from the heads of the ZnO cones. The HRTEM image from the left upper section of the nanocone in (a), indicated by the black arrow, is shown in (b). (e) is the schematic depiction showing the crystal habit planes of the ZnO cone with straight hexagonal pillar in Figure 5d.
most of the excitation source has been absorbed by the heads instead of penetrating though the heads to the buffer layer and substrate. This is particularly for the case of region G and region H where the heads are very large. Consequently, the products with larger heads have poorer optical properties when compared to the buffer layer. Both the Raman and the PL results suggest that the optical properties of the buffer layer between the ZnO cone arrays and the silicon substrate are better than the ZnO cone arrays. This is explained as follows. In previous reports, the vapor-liquid-solid (VLS) process is a conventional growth mechanism used for interpreting the growth of nanowires.2-4,15 The characteristic of the VLS growth is the existence of nanoparticles capped at the end of the 1D structure. Besides, metal catalyst is usually used to form the eutectic alloy in molten phase with the source material. The eutectic alloy is considered as the nucleation sites or seeds for the formation of the nanostructures. In our experiments, no metal catalyst is used. In other words, the conventional characteristics
Figure 6. Raman scattering spectra of regions B, D, F, and H in Figure 2a. The increasing fluorescence background from region B to H implies the decreasing optical quality. The inset is the Raman scattering spectrum of the buffer layer in Figure 4a, as compared to the Raman spectrum of region B.
Synthesis and Properties of ZnO Cone Arrays
J. Phys. Chem. B, Vol. 109, No. 7, 2005 2737
rlmin )
Figure 7. Room-temperature PL spectra of regions B, D, F, and H in Figure 2a and the buffer layer in Figure 4a.
2ΩlσlV kBT ln(PZn/P h Zn)
(1)
where Ωl is the volume of the average atom in the liquid, σlV is the liquid-vapor surface free energy, kB is Boltzmann’s constant, and T is the absolute temperature. PZn is the Zn partial pressure at the nucleating site on the substrate. P h Zn is the Zn vapor phase pressure in the thermal equilibrium coexistence with the liquid of composition in large quantity and having a flat h Zn in the different regions of the surface. Ωl, σlV, kB, T, and P substrates can be considered as equal because the silicon substrate is in the same temperature region as the zinc source. As the distance between the zinc source and the silicon substrates increases, the zinc vapor pressure PZn decreases. Therefore, this results in rlmin increasing, which induces the stem diameters of the ZnO cone arrays to grow larger from region A to region H. At the same time, the heads of the cone arrays grow larger and the buffer layer becomes thinner, probably indicating a competitive growth.21 Conclusion
Figure 8. SEM image reveals that only randomly oriented ZnO nanocones are obtained if air was added into the argon flow at the start of the experiment. The inset is the magnified image.
of VLS deposition do not apply to us. We have found that leading air into the reaction system when the temperature is raised to 420 °C is the critical factor to growing ZnO cone arrays. If air is added into the argon at the beginning of the experiment, only randomly oriented ZnO nanocones are obtained, as shown in Figure 8. However, when the flow rate of air is lowered to zero, there are almost no products formed on the substrates. Neither ZnO could form without oxygen in the reaction system nor zinc clusters could form when the temperature of the substrate is higher than the melting point of metalic zinc (419 °C).30 On the basis of the above results, we believe that a self-catalyst VLS mechanism31,32 plays an important role in the growth of ZnO cone arrays. The formation of ZnO cone arrays could involve a rather complex process described as follows: First, zinc clusters have been deposited on the substrate from the zinc source at a temperature below 420 °C. When the temperature exceeds 420 °C, the zinc clusters melt into droplets. The Zn droplets not only act as the reactant but also provide an energetically favored site for the absorption of O2. Continual feeding of Zn and O2 into the liquidized zinc droplets sustains the growth of the ZnO cone arrays. As a result, the heads of the cone arrays have abundant Zn, which leads to the poorer optical properties than the buffer layer in our experiments. The buffer layer grows at the higher temperature zone near the substrates than the ZnO cone arrays, which is also probably the reason that the buffer layer has better optical properties. Consequently, it is believed that the formation of the ZnO cone arrays can be explained by the self-catalyst VLS mechanism. The stem diameters of the ZnO cone arrays grow larger from region A to region H. This probably relates to the thermodynamic limit for the minimum radius of the metal-liquid clusters:10,33,34
In conclusion, ZnO cone arrays with controllable morphologies have been synthesized on Si (100) substrates by thermal evaporation of metal Zn powder at a low temperature of 570 °C without using a metal catalyst. All cone arrays are oriented along the c-axis. As the heads of the cones grow larger, gradual enhanced fluorescence backgrounds are observed in the Raman spectra, and UV emission decreases while green emission increases in PL spectra, indicating a gradual degradation of optical properties. The buffer layer between the substrates and the ZnO cone arrays has better optical quality than any of the cone arrays. The result suggested that the cone arrays have abundant metal zinc due to a self-catalyzed VLS process. Acknowledgment. This work was supported by the Natural Science Foundation of China (Grant Nos. 60376008, 50121202) and partially supported by the UDF of the University of Hong Kong. References and Notes (1) El-Sayed, M. Acc. Chem. Res. 2004, 37, 326. (2) Morales, A. M.; Lieber, C. M. Science 1998, 279, 208. (3) Wu, Y. Y.; Yang, P. D. Chem. Mater. 2000, 12, 605. (4) Duan, X. F.; Lieber, C. M. AdV. Mater. 2000, 12, 298. (5) Haber, J. A.; Gibbons, P. C.; Buhro, W. E. J. Am. Chem. Soc. 1997, 119, 5455. (6) Wang, Z. L.; Dai, Z. R.; Gao, R. P.; Bai, Z. G.; Gole, J. L. Appl. Phys. Lett. 2000, 77, 3349. (7) Dai, H. J.; Wong, E. W.; Lu, Y. Z.; Fan, S. S.; Lieber, C. M. Nature 1995, 375, 769. (8) Pan, Z. W.; Dai, Z. R.; Wang, Z. L. Science 2001, 291, 1947. (9) Lei, Y.; Zhang, L. D.; Fan, J. C. Chem. Phys. Lett. 2001, 338, 231. (10) Huang, M. H.; Wu, Y. Y.; Feick, H.; Tran, N.; Weber, E.; Yang, P. D. AdV. Mater. 2001, 13, 113. (11) Chopra, K. L.; Major, S.; Pandya, D. K. Thin Solid Films 1983, 102, 1. (12) Dayan, N. J.; Sainkar, S. R.; Karekar, R. N.; Aiyer, R. C. Thin Solid Films 1998, 325, 254. (13) Mitra, P.; Chatterjee, A. P.; Maiti, H. S. J. Mater. Sci. 1998, 9, 441. (14) Gorla, C. R.; Emanetoglu, N. W.; Liang, S.; Mayo, W. E.; Lu, Y.; Wraback, M.; Shen, H. J. Appl. Phys. 1999, 85, 2595. (15) Huang, M. H.; Mao, S.; Feick, H.; Yan, H. Q.; Wu, Y. Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. D. Science 2001, 292, 1897. (16) Minne, S. C.; Manalis, S. R.; Quate, C. F. Appl. Phys. Lett. 1995, 67, 3918. (17) Jie, J. S.; Wang, G. Z.; Han, X. H.; Yu, Q. X.; Liao, Y.; Li, G. P.; Hou, J. G. Chem. Phys. Lett. 2004, 387, 466. (18) Wu, J. J.; Liu, S. C.; Wu, C. T.; Chen, K. H.; Chen, L. C. Appl. Phys. Lett. 2002, 81, 1312.
2738 J. Phys. Chem. B, Vol. 109, No. 7, 2005 (19) Wang, Z. L.; Kong, X. Y.; Zuo, J. M. Phys. ReV. Lett. 2003, 91, 185502. (20) Wan, Q.; Yu, K.; Wang, T. H.; Lin, C. L. Appl. Phys. Lett. 2003, 83, 2253. (21) Jie, J. S.; Wang, G. Z.; Han, X. H.; Fang, J. P.; Xu, B.; Yu, Q. X.; Liao, Y.; Li, F. Q.; Hou, J. G. J. Cryst. Growth 2004, 267, 223. (22) Peterson, R. B.; Fields, C. L.; Gregg, B. A. Langmuir 2004, 20, 5114. (23) Wang, Z.; Qian, X. F.; Yin, J.; Zhu, Z. K. Langmuir 2004, 20, 3441. (24) Kawakami, M.; Hartanto, A. B.; Nakata, Y.; Okada, T. Jpn. J. Appl. Phys. 2003, 42, L33. (25) Damen, T. C.; Porto, S. P. S.; Tell, B. Phys. ReV. 1966, 142, 570. (26) Arguello, C. A.; Rousseau, D. L.; Porto, S. P. S. Phys. ReV. 1969, 181, 1351.
Han et al. (27) Kong, Y. C.; Yu, D. P.; Zhang, B.; Fang, W.; Feng, S. Q. Appl. Phys. Lett. 2001, 78, 407. (28) Vanheusden, K.; Seager, C. H.; Warren, W. L.; Tallant, D. R.; Voigt, J. A. Appl. Phys. Lett. 1996, 68, 403. (29) Bylander, E. G. J. Appl. Phys. 1978, 49, 1188. (30) Dean, J. A., Ed. Lange’s Handbook of Chemistry, 13th ed.; McGraw-Hill: New York, 1985. (31) Hu, J. Q.; Li, Q.; Wong, N. B.; Lee, C. S.; Lee, S. T. Chem. Mater. 2002, 14, 1216. (32) Tseng, Y. K.; Hsu, H. C.; Hsieh, W. F.; Liu, K. S.; Chen, I. C. J. Mater. Res. 2003, 18, 2837. (33) Tan, T. Y.; Li, N.; Go¨sele, U. Appl. Phys. Lett. 2003, 83, 1199. (34) Hu, J. T.; Odom, T. W.; Lieber, C. M. Acc. Chem. Res. 1999, 32, 435.
