Patterns of Ensemble Variation of the Optical Properties of ZnO

Jan 20, 2009 - Corresponding author. E-mail: [email protected]., †. Surface and Microanalysis Science Division. , ‡. Materials Reliability...
0 downloads 0 Views 3MB Size
J. Phys. Chem. C 2009, 113, 2277–2285

2277

Patterns of Ensemble Variation of the Optical Properties of ZnO Nanowires Grown with Copper and Gold Catalysts Susie Eustis,†,§ Lawrence H. Robins,‡ and Babak Nikoobakht*,† Surface and Microanalysis Science DiVision, National Institute of Standards and Technology, 100 Bureau DriVe, Mail Stop 8372, Gaithersburg, Maryland 20899, and Materials Reliability DiVision, National Institute of Standards and Technology, 325 Broadway, Mailcode 853.05, Boulder, Colorado 80305 ReceiVed: September 12, 2008; ReVised Manuscript ReceiVed: December 17, 2008

Understanding the source of variation of the optical properties within an ensemble of nanowires (NWs) is an important step toward fabrication of NWs with more uniform and better-controlled properties. This in turn will facilitate the development of complex architectures for device applications. Analysis of gold- and coppercatalyzed zinc oxide (ZnO) NWs grown in a high temperature tube furnace on a sapphire substrate shows that the structural and optical properties vary with NW growth sites on the substrate. Results show that there are systematic changes in the optical properties of the ZnO NWs from the center to the edge of the substrate, and also from “upstream” to “downstream” along the gas flow direction, implying changes in the availability of the source material and catalyst over the substrate surface during the growth. Photoluminescence microscopy has been used to map out these changes and unravel the growth patterns. We observe two distinct trends that are labeled the “edge” trend and the “gas flow” trend, respectively, and are linked to two growth phases. The first phase starts while the growth tube temperature and gas flow have not yet reached their final steady states. In this phase, due to the limited availability of the source material in gas phase, only metal nanodroplets that are located at the substrate edges have the advantage of early growth. While the edge NWs continue their growth, the second phase starts when the growth tube temperature and gas flow reach reasonable stabilities. Our results show that these two phases are more pronounced in the case of copper-catalyzed NWs. In the first phase, NWs have a better chance to grow at the substrate edges. In the second phase, copper diffuses downstream, causing an interesting variation in the optical properties of NWs. Numerous morphology examinations of NWs show that the variation in emission is related to the change in the ratio of the surface area to the bulk volume of the NWs within an ensemble of NWs. The patterns of the ZnO NW optical properties can be used as an indicator to follow the growth patterns of the NWs across the substrate. Introduction Zinc oxide (ZnO) nanowires (NWs) and related nanostructures have been extensively investigated in the past decade due to their interesting optical1-4 and electrical properties.5-7 These nanostructures are potential candidates for UV light sources, photodetectors,8 sensors,9 solar cells,10 field emission sources,5 and piezoelectric devices.6,11 In the vapor-liquid-solid (VLS) mechanism, NWs are grown from thin films of metal catalyst interacting with the vapor source material on a solid substrate at elevated temperatures.12,13 During the growth process, the initially solid metal catalyst forms molten nanoparticle droplets, and the source precursors flow downstream incorporating into the catalyst droplets on the substrate surface until supersaturation is reached. Supersaturation leads to precipitation of the source material to form NWs. The NWs continue to grow until either the source material is exhausted or the temperature is dropped. Many factors influence the properties of NWs grown by the VLS mechanism, including the supersaturation point,14,15 the temperature history of the sample,15-18 the vapor pressure of the source material,17-20 and the solid/solid13,15 and the liquid/solid13,15 solubility of the catalyst and growth material. * Corresponding author. E-mail: [email protected]. † Surface and Microanalysis Science Division. ‡ Materials Reliability Division. § Present Address: Directed Vapor Technologies International, Inc., 2 Boar’s Head Lane, Charlottesville, VA 22911.

10.1021/jp808129j

The optical properties of ZnO have been a source of interest due to the lasing capability of single and multiple NWs.1-4 The photoluminescence (PL) spectrum of ZnO typically displays two peaks. One peak, caused by Coulombically bound electron-hole pairs (excitons), is referred to as either the band gap emission or the excitonic emission. The large exciton binding energy of ∼60 meV in ZnO stabilizes the excitonic states against dissociation to free electron-hole pairs, even at room temperature, which makes ZnO an attractive light source. The other peak, called the deep trap emission, is in the visible and is due to impurities or defects, which possess electronic energy levels within the band gap of ZnO. Numerous parameters affect the optical properties of ZnO NWs. Increasing the growth temperature of ZnO NWs has been shown to increase the deep trap emission due to the increased rate of NW growth, which leads to more defects.16 Oxygen vacancies, zinc vacancies, surface defects, and transition metal and other chemical impurities are known to contribute to the deep trap emission of ZnO NWs.2,21-24 All of these factors will be influenced by the growth conditions, leading to the common use of the ratio between the intensity of the band gap emission and the intensity of the deep trap emission to characterize different syntheses of ZnO.2,23-25 This ratio is often used as a figure of merit for the crystalline quality of ZnO due to the defect state origin of the deep trap emission.

This article not subject to U.S. Copyright. Published 2009 by the American Chemical Society Published on Web 01/20/2009

2278 J. Phys. Chem. C, Vol. 113, No. 6, 2009 All of the ZnO NW samples in this study exhibited a deep trap emission peak in the green region near 2.48 eV. There have been numerous reports attributing the green deep trap emission to several different types of defect structures: singly ionized oxygen vacancies, antisite oxygen, oxygen vacancies, zinc vacancies, zinc interstitials, copper impurities, and surface defects.2,21-24 Green deep trap emission due to copper was first observed by Dingle21 with a characteristic fine structure at low temperature. Reshchikov et al.22 were able to distinguish five uniquely identifiable emission peaks in the green region from measurements at multiple temperatures. One of these green emission bands was assigned to the copper impurity and showed fine structure at low temperature due to phonon replicas with energy spacings of 32 and 72 meV.21,22 Copper incorporation in bulk ZnO has been shown to increase the deep trap emission.21-23,26,27 Kouklin8 has recently shown that doping ZnO with copper can increase the spectral sensitivity for photodetection applications. Wang et al.27 found that copper incorporation in bulk ZnO at mole fractions up to 2% (i.e., Zn0.98Cu0.02O) increases the band gap emission, while the deep trap emission remains relatively low in intensity. Other studies have shown that copper incorporation in ZnO NWs increases the band gap emission.28-30 Wang et al.27 also found that further increasing the copper concentration decreased the intensity of the band gap emission, while the intensity of the deep trap emission increased. Zhang et al.31 studied the physical properties of copper-doped nanoneedles and nanonails (which are similar in shape to the nanoneedles, but are topped with a flat head). The nanonails showed intense band gap emission, while the nanoneedles showed weak band gap emission and stronger deep trap emission than did the nanonails. These differences were attributed to oxygen availability during growth and surface defects.31 In a previous study, we reported that the growth of ZnO NWs on a copper support generated a high density of long ZnO NWs. The PL spectrum of these NWs showed an intense green deep trap emission with only a weak band gap emission.32 We attributed the green emission to a combination of copper incorporation and surface defects.32 The measured optical properties of an ensemble of NWs can be considered to be the average of the optical properties of individual NWs in the ensemble. The primary goals of this study are to characterize the spatial variation of the optical properties of ZnO NWs covering an area of several millimeters, and to better understand what factors influence this variation. To this end, the effect of the metal catalyst, either gold or copper, on the size (diameter and length), growth rate, and optical properties of the ZnO NWs is examined while maintaining the same growth conditions. Because of the sensitivity of ZnO PL to impurities, PL microscopy was chosen as the main tool for optical characterization of the ZnO NWs. Experimental Section Zinc oxide NWs were grown by chemical vapor deposition in a tube furnace at high temperatures. A 1:1 mixture (by mass) of ZnO powder and graphite was placed in the middle of a small quartz tube centered in the tube furnace. For most growth runs, a 4 mm × 7 mm a-plane sapphire substrate was used as the growth substrate after thermal evaporation of a thin layer of gold or copper using a circular mask to generate 160 µm diameter metal circles with a center-to-center distance of 300 µm arranged in a hexagonal pattern. The metal-coated sapphire substrate was placed downstream from the source material at the edge of the small tube. For most growth runs, the tube furnace was set to ramp up to 890 at 100 °C/min and held for

Eustis et al. 15 min (the furnace temperature was observed to initially reach ∼1000 °C, before slowly decreasing to the steady-state temperature of ∼900 °C, due to the high heating rate). The final set temperature and hold time were varied from the standard values of 890 °C and 15 min in some growth experiments, as discussed in more detail below. The temperature is the same throughout the growth area (0.5 °C. For all runs, the furnace was purged continuously with argon at 0.60 standard L/min for a space velocity of 0.50 min-1, which means that the reactor volume is replaced every 2 min. Cathodoluminescence (CL) studies were preformed on ZnO NWs grown on a copper transmission electron microscopy halfgrid substrate, and a gold-coated copper substrate, prepared by depositing a titanium coating followed by a gold coating on the copper half-grid via thermal evaporation. A photolithographic mask (rather than the circular mask) was used for the deposition of the Au to generated gold islands in the Au/sapphire sample used for CL studies (shown in Figure 1b). All substrates were then exposed to the conditions above for the growth of the ZnO NWs. ZnO NWs were excited by a 3.81 eV HeCd laser at room temperature (T ≈ 297 K), and PL was detected through a 40× microscope objective. (A more complete description of the PL setup is available in a previous publication.32) The PL emission spectra and images were obtained for each catalyst circle on the sapphire substrate and were linked to the scanning electron microscopy (SEM) images of the corresponding location. The spectra were corrected for the detector response before analysis. The ratio of the peak intensity of the band gap emission to the peak intensity of the deep trap emission was defined as the gapto-trap ratio. The gap-to-trap ratio was used in characterizing the PL of the ZnO NWs in the following discussion. SEM images of ZnO NWs obtained with an accelerating voltage of 5 kV were used to determine the diameter and the length of the ZnO NWs on the sapphire substrates. The samples were coated with carbon to reduce charging during SEM imaging. X-ray diffraction (XRD) was obtained from the samples using copper KR radiation and detected with a calibrated crossed-wire area detector. Multiple locations of numerous samples were measured to ensure the reproducibility of the trends shown in this Article. The data displayed herein were chosen to be representative of the trends observed in all of the samples. CL measurements were obtained at 12 K from samples mounted with graphite paste on the coldfinger of a liquid-He cold stage in a LaB6 filament SEM. The SEM was operated with an electron beam voltage of 4.9 kV, and a beam current of 5.8 nA. The emitted CL was collimated by mirrors located near the sample, transmitted through a fused-silica window in the SEM, and refocused onto the entrance slit of a 0.34 m spectrograph with a 600 line/mm grating and entrance slit width of 0.05 mm. The spectra were recorded by a computercontrolled, nitrogen-cooled CCD camera. The wavelength resolution of the spectrograph-CCD camera system was 0.3 nm, which is equivalent to an energy resolution of 2.3 × 10-4E2 eV, where E is photon energy (thus, the energy resolution was 2.7 meV at 3.36 eV, or 1.5 meV at 2.5 eV). Results and Discussion The room-temperature PL spectra for representative areas of ZnO NWs grown with gold and copper catalysts on sapphire are shown in Figure 1a. The two characteristic emissions are observed for each sample; the band edge emission is centered at 3.305 eV ( 0.008 eV for gold and 3.262 eV ( 0.009 eV for copper, and the deep trap emission due to defect states is around

Optical Properties of ZnO Nanowires

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2279

Figure 1. (a) Photoluminescence (PL) spectra of ZnO NWs grown on sapphire with gold and copper catalysts. (b) CL spectra of ZnO NWs on sapphire with a gold catalyst, copper grid, and copper grid coated with titanium and gold. (c) XRD spectra of ZnO NWs grown on sapphire with gold and copper catalysts. (d) Distribution of gap-to-trap ratio for gold-catalyzed ZnO NWs on sapphire. (e) Distribution of gap-to-trap ratio for copper-catalyzed ZnO NWs on sapphire.

2.54 eV for gold and 2.45 eV for copper. However, the band edge emission is much more intense than the deep trap emission for the gold-catalyzed sample, while the deep trap emission is much more intense than the band edge emission for the coppercatalyzed sample. To further investigate the difference of the emissions between copper- and gold-catalyzed ZnO NWs, CL of NWs is measured at 12 K. The CL emission spectrum from gold-catalyzed ZnO NWs on sapphire is shown in Figure 1b. The CL spectra of ZnO NWs grown on a “bare” copper grid and a copper grid coated with titanium and gold are also shown in this figure. (A sample from copper-catalyzed ZnO NWs on sapphire was unavailable for CL measurement.) The near band edge emission peak is observed at 3.358-3.360 eV for all three samples. Peaks in this energy range observed in low temperature PL or CL spectra of ZnO are usually attributed to “A excitons” bound to neutral donors,23 originating from low concentrations of donor impurities such as Al, Ga, or In. Phonon replicas of the free or donor-bound exciton are observed in the 3.1-3.34 eV range for all of the samples examined by CL. The deep trap emission of the ZnO NWs grown from the copper grid shows an abrupt high-energy cutoff at 2.863 eV. This is close to the zero phonon line of the green emission band, which is attributed to copper impurities by Reshchikov at 2.859 eV.22 In addition, phonon spacings of 72 and 27 meV are observed for the ZnO NWs grown on the copper grid, which are similar to the phonon spacings reported for the copper impurity band by Reshchikov.22 Thus, although a number of distinct sources of green emission have been observed in ZnO,2,21-24 the low temperature CL measurements of the ZnO NWs grown on a “bare” copper grid provide unambiguous identification of the copper impurity band.

The XRD spectra of the ZnO NWs grown on sapphire with gold and copper catalysts are shown in Figure 1c. The positions of the diffraction peaks correspond to the lattice spacings of Wurtzite structure ZnO as reported on card number 36-1451.33 The high intensity of the (0002) diffraction peak for both NW samples implies oriented growth along the [0001] direction. The similarity of the XRD spectra for the gold- and copper-catalyzed samples shows that the growth orientation and crystalline quality of these NWs is essentially the same for both catalysts. The XRD results also show that the lower PL gap-to-trap ratio in the copper-catalyzed sample is not due to a degradation of the crystal quality (as compared to the gold-catalyzed sample) induced by the catalyst material. To characterize the spatial variation of the optical properties of the NWs as a function of location on the substrate surface, as well as the differences between the optical properties of the copper- and gold-catalyzed NWs, the room-temperature PL emission spectrum is measured for each metal catalyst circle of ZnO NW (in samples that contain 160 µm diameter gold or copper circles on a sapphire substrate). The gap-to-trap ratio is then calculated from the PL spectra of each circle. The histograms in Figure 1d and e represent the distributions of the gap-to-trap ratio obtained for the gold- and copper-catalyzed samples, respectively. The large spatial variation of the PL, which is apparent from these histograms, shows the importance of collecting data from many locations on each sample to obtain a representative comparison between the optical properties of the gold- and copper-catalyzed NWs. The gap-to-trap ratio ranges from 11 to 81 for the gold-catalyzed NWs and from 0.027 to 39 for the copper-catalyzed NWs. The distribution of the gap-to-trap ratio for the copper-catalyzed NWs is broad and

2280 J. Phys. Chem. C, Vol. 113, No. 6, 2009

Eustis et al.

Figure 2. (a) SEM composite image of ZnO NWs grown on a sapphire substrate with a copper catalyst. (b) The gap-to-trap ratio of each spot area is depicted by the color of the spot. (c) The PL spectra of the areas circled above.

bimodal (with a peak at both low and high values); on the other hand, the distribution of the gap-to-trap ratio for the goldcatalyzed NWs contains a single peak and is less broad. Thus, the optical properties of the copper-catalyzed NWs show more variation, as a function of location on the substrate, than do the optical properties of the gold-catalyzed NWs. Growth of Copper-Catalyzed NWs. To effectively display how the optical properties of the copper-catalyzed ZnO NWs vary as a function of location on a sapphire substrate, a composite SEM image and a color-coded image of the varying gap-to-trap ratios are shown in Figure 2a and b. Two trends are observed in the spatial distribution of the gap-totrap ratio of such NWs, as shown in Figure 2b. The first trend, referred to as the edge trend, is that the NWs located near the edges of the substrate have a substantially higher gapto-trap ratio than do the NWs in the middle of the substrate. The second trend, referred to as the gas flow trend, is a

gradual decrease in the gap-to-trap ratio of the NWs from the leading to trailing edge of the sapphire substrate (along the gas flow direction), excluding the NWs located near the edges. The edge trend appears to be stronger than the gas flow trend, because all of the NW-catalyst circles located near the edges of the substrate have a higher gap-to-trap ratio than do any NW-catalyst circles located near the middle. The PL spectra at three different locations, which are marked by outlining in Figure 2a and b, are shown in Figure 2c. As can be seen in Figure 2c, the origin of the edge trend is a much larger intensity of the band gap emission for the NWs located near the edges of the substrate, as well as a somewhat lower intensity of the deep trap emission. In Figure 2c, the gas flow trend is demonstrated by an increase in the intensity of the deep trap emission, moving from the leading edge toward the trailing edge of the substrate. The intensity of the band

Optical Properties of ZnO Nanowires

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2281

Figure 3. (a-c) SEM images of ZnO NWs grown on a sapphire substrate with a copper catalyst in the areas shown in Figure 2. (d) Diameter distributions of the NWs from the areas selected. (e) Length distributions of the NW from the areas selected where the NW length increases toward the trailing edge of the sapphire substrate. (f) Plot of the maximum intensity of the band gap emission against the average diameter of the NWs.

2282 J. Phys. Chem. C, Vol. 113, No. 6, 2009

Eustis et al.

Figure 4. (a) SEM composite image of ZnO NWs grown on a sapphire substrate with a gold catalyst. (b) The gap-to-trap ratio of each spot area is depicted by the color of the spot. (c) The PL spectra of the areas circled above.

edge emission remains relatively constant in the middle of the substrate, as shown in the inset of Figure 2c. To further investigate the origin of the spatial variation of the gap-to-trap ratio in the copper-catalyzed ZnO NW sample (shown in Figure 2), the morphologies of the NWs from the areas marked by outlining (in Figure 2a) are examined by SEM. Representative images of the marked areas are shown in Figure 3a-c. The NWs near the leading edge of the substrate are thin, long, oriented, and less dense. ZnO also forms crawling structures on the substrate surface, which can be observed near the leading edge due to the separation of the NWs. Less oriented growth is observed in the NWs toward the middle and trailing edge of the substrate as shown in Figure 3a-c. The typical diameter and length of the ZnO NWs increase from the leading edge toward the middle of the sapphire substrate, as shown in Figure 3b and c. The typical length of the NWs shows a further increase from the middle toward the trailing edge, while the

typical diameter remains constant from the middle toward the trailing edge. The NW diameter and length distributions, obtained from analysis of the images in Figure 3a-c and other similar images of the marked areas, confirm and quantify these trends, as shown in Figure 3d and e, respectively. Figure 3f shows that as the diameter of the NWs decreases, the intensity of the band gap emission increases. To quantify the amount of material on the surface of NWs, the ratio of the surface area to the bulk volume is used, which is proportional to the inverse of the diameter of a nanowire. Thus, the Supporting Information shows that the band gap emission intensity increases as the ratio of the surface area to the bulk volume increases. We would have expected a decrease in the band edge emission and increase in the deep trap emission, if the variation of the band gap emission were due primarily to surface or near-surface recombination. This also means the NWs at the edge of the substrate experience a lower exposure to

Optical Properties of ZnO Nanowires

Figure 5. (a-c) SEM images of ZnO NWs grown on a sapphire substrate with a gold catalyst in the areas shown in Figure 4. (d) Diameter distributions of the NWs from the areas selected (differences are not significant). (e) Length distributions of the NW from the areas selected.

copper flow, which agrees with their stronger band edge emission. A weaker deep trap emission of NWs at the edges is another indication of less diffusion of copper to the overall body of a thin NW. The decrease in the intensity of the band edge emission as the NWs increase in diameter toward the trailing edge suggests that copper incorporates into the bulk volume of the NWs and generates nonradiative centers that quench the band edge emission. Thus, while surface defects do influence the intensity of the deep trap emission, they do not determine the intensity of the band edge emission. Growth of Gold-Catalyzed NWs. A composite SEM image is shown in Figure 4a of a similar sapphire substrate covered with gold-catalyzed ZnO NWs. The color-coded map of the gapto-trap ratio of the PL emission is shown in Figure 4b, where the ratio is less at the edge of the substrate. The average of the gap-to-trap ratio is 34.3 for the areas on the edge as compared to 43.8 for the rest of the sample. Thus, the gold-catalyzed NW sample shows an edge trend of smaller relative magnitude (less fractional change in the gap-to-trap ratio) with respect to the copper-catalyzed NW sample. The magnitude of the gap-totrap ratio emission of gold-catalyzed NWs is similar over the majority of the substrate (see the histogram in Figure 1d). Moving from the leading edge toward the trailing edge, the gapto-trap ratio increases, suggesting a gas flow trend. As shown in the Supporting Information, both the band gap and the deep trap emission increase from the leading to trailing edge; however, the band gap emission increases by a larger percentage, causing the gap-to-trap ratio to increase. The PL emission spectra of the three highlighted (outlined) areas indicated in Figure 4a and b are plotted in Figure 4c.

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2283 Examination of these spectra confirms that the band gap emission is dominant in all parts of the gold-catalyzed sample (as already indicated by the large values of the gap-to-trap ratio) with relatively weak deep trap emission (especially when compared to the copper-catalyzed sample). A large increase in the band gap intensity is observed on moving from the leading edge toward the trailing edge. The deep trap emission intensity remains relatively low at all locations, although some increase of the deep trap intensity is observed moving from the leading toward the trailing edge. SEM images of the gold-catalyzed ZnO NWs grown on sapphire, from the leading edge, the middle, and toward the trailing edge as highlighted in Figure 4, are shown in Figure 5a-c, respectively. The diameter and length distributions obtained for these areas are shown in Figure 5d and e, respectively. The size distributions in Figure 5d show that the average diameter of the NWs is close to 60 nm, with no systematic variation from the leading to the trailing edge. The SEM images in Figure 5a-c show that the longest NWs are toward the trailing edge of the sapphire substrate. These NWs are considerably shorter than the NWs observed in coppercatalyzed samples. These NWs are also shorter than many of the similar gold-catalyzed, but the trends are similar to those reported here (another sample is shown in the Supporting Information). Thus, both the copper-catalyzed and the goldcatalyzed NW samples show an increase in NW length toward the trailing edge of the sapphire substrate. Growth Time and Temperature Dependence. To investigate the significant difference in the length of NWs grown from gold and copper catalysts, the NW growth was carried out for different lengths of time and at different temperatures. In the initial set of time/temperature dependence experiments, the growth time (dwell time at 890 °C) was decreased from 15 to 5 min for both gold- and copper-catalyzed samples. For the goldcatalyzed samples, the length of the NWs decreased and the PL emission intensity decreased significantly with decreasing growth time. For the copper-catalyzed NWs, on the other hand, reducing the growth time from 15 to 5 min had almost no effect on the length and optical properties (PL spectrum) of the NWs. To further probe the effect of growth conditions on morphology and optical properties of the copper-catalyzed NWs, two sets of experiments were performed. First, the growth temperature was lowered from 890 to 790 °C and then to 700 °C, while the growth time was held constant at 15 min. The properties of the copper-catalyzed NWs grown at 790 °C were found to be similar to the properties of the NWs grown at 890 °C. Lowering the growth temperature further, to 700 °C, caused the PL emission intensity to decline and caused the length of the NWs to decrease significantly. The longest NWs in the copper-catalyzed sample grown at 700 °C were located near the leading edge, with lengths between 0.50 and 1.0 µm. The NWs decreased in length from the leading to the trailing edge, contrary to the trend observed in the NWs grown at higher temperatures. Despite low temperature growth in the case of copper, for gold nanodroplets, the lowest temperature at which the NW growth was observed was about 810 °C. Growth of NWs only from copper nanodroplets at 700 °C indicates that zinc and oxygen precursors are available in the tube furnace atmosphere; it also shows that gold nanodroplets are not active in crystallizing the ZnO at lower temperatures, even though the bulk phase diagrams of both metals with zinc show similar temperature behaviors.34,35 Therefore, when the temperature is raised to 890 °C, the observed differences between the gold- and copper-catalyzed NWs are due to the

2284 J. Phys. Chem. C, Vol. 113, No. 6, 2009 ability of the copper nanodroplets to start the growth process at lower temperatures. This would also increase the time over which the NWs experience conditions conducive to growth, explaining the longer growth of the copper-catalyzed NWs and the more pronounced trends observed. The higher band edge emission intensity of Cu-catalyzed NWs at the substrate edge also indicates the lower concentration of Cu in the gaseous phase in this area as compared to the central parts of the substrate. The gradual decline in the band edge emission toward the trailing edge indicates a stronger incorporation of copper and its migration to these regions (flow effect) (see Figure 2b). Therefore, we suggest the emergence of the edge effect and the flow effect in a serial fashion. At the early stage of the growth process, that is, 7-12 min after turning on the furnace, in the case of the Cu-catalyzed NWs, the Zn and O precursors approach the substrate with the highest concentration at the leading edge, thus resulting in formation of NWs in these areas. At this stage, the distribution of precursors, flow dynamic, and temperature have still not reached a steady state. As the tube furnace temperature reaches a more stable and steady condition, the gas flow containing Zn and O precursors steadily reaches the central and trailing areas of the substrate. It is at this stage where copper starts to migrate from the upper stream toward the trailing edge, causing the observed variation in the optical properties of NWs. Previously, Mensah et al.38 have also shown that the presence of metal catalyst toward the leading edge can influence the size and properties of the NWs, by its migration further downstream. Similar variations have been observed by varying the size and spacing of the catalyst droplet on the substrate surface.36,37 Comparison of the Results from the Gold- and CopperCatalyzed NWs. With the gold-catalyzed ZnO NWs, both the edge trend and the gas flow trend are observed, although the edge trend is weaker and the gas flow trend is reverse in direction with respect to the copper-catalyzed NWs. During the NW growth at 890 °C, in contrast to Cu-catalyzed NWs, gold nanodroplets do not result in longer ZnO NWs at the edges. This is explainable on the basis of the different activities of gold and copper nanodroplets observed in NW growth at 700 °C. Assuming the final growth temperature is at 890 °C, for copper nanodroplets, NW growth starts at about 700 °C, and by the time the temperature stabilizes at 890 °C (after 3-5 min), the NWs at the edges are well grown in length. For gold, the lowest observed growth temperature is about 810 °C, which is so close to the final growth temperature (890 °C) that it does not allow the appearance of an edge trend similar to the copper case. We think this is why the edge trend has a minimal effect on the morphology of the gold-catalyzed NWs. Once the temperature reaches 890 °C, a steady gas flow causes the appearance of longer NWs toward the trailing edge with a stronger band edge emission. While incorporation of copper leads to creation of surface trap states and nonradiative sites in the core of the NW, gold clearly introduces a lower number of nonradiative centers and perhaps more surface traps. Thus, the increase in the gap-to-trap ratio of gold-catalyzed NWs toward the trailing edge could mainly be due to a decrease in the surface to volume ratio of the NWs (as they grow longer). Although these results show that the variation in the optical properties could be a serious concern in the current state-ofthe-art growth of standing NWs, they are also promising, because we think the growth conditions can be tailored to minimize or maximize such variations. For instance, the gas flow trend and downstream diffusion of copper could be minimized by enhancing the substrate edge effect via, for

Eustis et al. example, designing a nonflat surface. This could result in copper NWs with reproducible and strong band edge emission. Alternatively, if copper doping and incorporation is of interest, the gas flow effect could be enhanced by inserting the sample inside the tube furnace when the tube has reached its equilibrium condition. Conclusions By comparing the sizes and the optical properties of ZnO NWs grown with copper and gold catalysts, the nonuniformity of the NWs due to different growth conditions was observed. The average of the gap-to-trap ratio of emission was 48.8 for gold-catalyzed NWs and 2.4 for the copper-catalyzed NWs. In the case of copper-catalyzed NWs, it was shown that due to the activity of copper nanodroplets, growth starts at lower temperature, thus resulting in the observation of two distinct effects of edge and gas flow. It was also concluded that the edge effect controls the morphology of NWs at the early stage of the growth and results in longer NWs at the substrate leading edge, while the gas flow effect takes the control of growth of NWs at a later time (the second phase), resulting in longer NWs at the substrate trailing edge. It was described that during the second phase of the growth, the copper diffuses to the vapor phase and is transported by the flow downstream, resulting in its incorporation into the NWs and a pronounced decrease in the gap-to-trap emission ratios. In the case of gold-catalyzed NWs, the temperature difference between the two phases of the growth was not as significant as that of the copper metal. This resulted in a less obvious edge effect and more dominant gas flow trend. A strong dependence of the band edge emission intensity on the ratio of the surface area to the volume of the NW was determined as responsible for the observed trend. The optical emission and specifically the gap-to-trap ratio are used in this setup to find areas of variability in the growth conditions as a visualization tool. This information can then be used to expand the areas of uniform growth by modification of the growth chamber shape, position, and gas flow. Using results of this study, it is possible to pave the way for prescribing conditions for growth of NWs with more homogeneous optical properties. It is also possible to develop platforms for quick and high throughput characterization, and quality control of NW ensembles grown on large areas. Supporting Information Available: Additional figures. This material is available free of charge via the Internet at http:// pubs.acs.org. References and Notes (1) Huang, M. H.; Mao, S.; Feick, H.; Yan, H.; Wu, Y.; Kind, H.; Weber, E.; Russo, R.; Yang, P. Science 2001, 292, 1897. (2) Djurisic, A. B.; Leung, Y. H. Small 2006, 2, 944. (3) Yang, P.; Yan, H.; Mao, S.; Russo, R.; Johnson, J.; Saykally, R.; Morris, N.; Pham, J.; He, R.; Choi, H. J. AdV. Funct. Mater. 2002, 12, 323. (4) Johnson, J. C.; Yan, H.; Schaller, R. D.; Haber, L. H.; Yang, P.; Saykally, R. J. J. Phys. Chem. B 2001, 105, 11387. (5) Lee, C. J.; Lee, T. J.; Lyu, S. C.; Zhang, Y.; Ruh, H.; Lee, H. J. Appl. Phys. Lett. 2002, 81, 3648. (6) Wang, X.; Jinhui Song, J. L.; Wang, Z. L. Science 2007, 316, 102. (7) Zhao, M. H.; Wang, Z. L.; Mao, S. X. Nano Lett. 2004, 4, 587. (8) Kouklin, N. AdV. Mater. 2008, 20, 2190. (9) Wan, Q.; Li, Q. H.; Chen, Y. J.; Wang, T. H.; He, X. L.; Li, J. P.; Lin, C. L. Appl. Phys. Lett. 2004, 84, 3654. (10) Baxter, J. B.; Aydil, E. S. Appl. Phys. Lett. 2005, 86, 053114/1-3. (11) Qin, Y.; Wang, X.; Wang, Z. L. Nature 2008, 451, 809. (12) Wagner, R. S.; Ellis, W. C. Appl. Phys. Lett. 1964, 4, 89. (13) Givargizov, E. I. J. Cryst. Growth 1975, 31, 20. (14) Roper, S. M.; Davis, S. H.; Norris, S. A.; Golovin, A. A.; Voorhees, P. W.; Weiss, M. J. Appl. Phys. 2007, 102, 034304.

Optical Properties of ZnO Nanowires (15) Adhikari, H.; Marshall, A. F.; Goldthorpe, I. A.; Chidsey, C. E. D.; McIntyre, P. C. ACS Nano 2007, 1, 415. (16) Khan, A.; Kordesch, M. E. Physica E 2005, 30, 51. (17) Kodambaka, S.; Tersoff, J.; Reuter, M. C.; Ross, F. M. Science 2007, 316, 729. (18) Dalal, S.; Baptista, D.; Teo, K.; Lacerda, R.; Jefferson, D.; Milne, W. Nanotechnology 2006, 17, 4811. (19) Song, J. H.; Wang, X. D.; Riedo, E.; Wang, Z. L. J. Phys. Chem. B 2005, 109, 9869. (20) Yoon, H.; Seo, K.; Moon, H.; Varadwaj, K. S. K.; In, J.; Kim, B. J. Phys. Chem. C 2008, 112, 9181. (21) Dingle, R. Phys. ReV. Lett. 1969, 23, 579. (22) Reshchikov, M. A.; Morkoc, H.; Nemeth, B.; Nause, J.; Xie, J.; Hertog, B.; Osinsky, A. Physica B 2007, 401-402, 358. (23) Ozgur, U.; Ya, I. A.; Liu, C.; Teke, A.; Reshchikov, M. A.; Dogan, S.; Avrutin, V.; Cho, S. J.; Morkoc, H. J. Appl. Phys. 2005, 98, 041301/ 1–103. (24) Djurisic, A. B.; Choy, W. C. H.; Roy, V. A. L.; Leung, Y. H.; Kwong, C. Y.; Cheah, K. W.; Rao, T. K. G.; Chan, W. K.; Lui, H. F.; Surya, C. AdV. Funct. Mater. 2004, 14, 856. (25) Fischer, A. M.; Srinivasan, S.; Garcia, R.; Ponce, F. A.; Guano, S. E.; Lello, B. C. D.; Moura, F. J.; Solorzano, I. G. Appl. Phys. Lett. 2007, 91, 121905. (26) Garces, N. Y.; Wang, L.; Bai, L.; Giles, N. C.; Halliburton, L. E.; Cantwell, G. Appl. Phys. Lett. 2002, 81, 622.

J. Phys. Chem. C, Vol. 113, No. 6, 2009 2285 (27) Wang, X. B.; Song, C.; Geng, K. W.; Zeng, F.; Pan, F. Appl. Surf. Sci. 2007, 253, 6905. (28) Li, S. Y.; Lin, P.; Lee, C. Y.; Tseng, T. Y. J. Appl. Phys. 2004, 95, 3711. (29) Yamamoto, K.; Nagasawa, K.; Ohmori, T. Physica E 2004, 24, 129. (30) Xu, C. X.; Sun, X. W.; Zhang, X. H.; Ke, L.; Chua, S. J. Nanotechnology 2004, 15, 856. (31) Zhang, Z.; Yi, J. B.; Ding, J.; Wong, L. M.; Seng, H. L.; Wang, S. J.; Tao, J. G.; Li, G. P.; Xing, G. Z.; Sum, T. C.; Alfred Huan, C. H.; Wu, T. J. Phys. Chem. C 2008, 112, 9579. (32) Eustis, S.; Meier, D. C.; Beversluis, M. R.; Nikoobakht, B. ACS Nano 2008, 2, 368. (33) Mc Murdie, H.; Morris, M.; Evans, E.; Paretzkin, B.; Wong-Ng, W.; Ettlinger, L.; Hubbard, C. Powder Diffr. 1986, 1, 76. (34) Miodownik, A. P., Subramanian, P. R., Chakrabarti, D. J., Laughlin, D. E., Eds. ASM International: Materials Park, OH, 1994; p 487. (35) Okamoto, H.; Massalski, T. B. 2nd ed.; ASM International: Metals Park, OH, 1990; p 456. (36) Borgstrom, M. T.; Immink, G.; Ketelaars, B.; Algra, R.; Bakkers, E. P. A. M. Nat. Nanotechnol. 2007, 2, 541. (37) Sun, X. H.; Lam, S.; Sham, T. K.; Heigl, F.; Jurgensen, A.; Wong, N. B. J. Phys. Chem. B 2005, 10, 3120. (38) Mensah, S. L.; Kayastha, V. K.; Yap, Y. K. J. Phys. Chem. C 2007, 111, 16092.

JP808129J