Introduction

The vapor–liquid–solid (VLS) crystal growth mechanism has been widely used to guide the growth of a wide range of nanowires1, such as Group IV (Si and Ge), III–V (GaAs, InAs, GaN and InP) and II–VI (CdS, CdSe, ZnS and ZnSe) semiconductors and oxides (ZnO, Ga2O3, SnO and SiO2) and the heterostructures2,3,4,5,6,7,8,9,10,11,12,13,14,15. A common morphological feature of VLS-grown nanowires is that each nanowire is terminated at one end by a catalyst nanoparticle with a comparable diameter. However, catalyst at the tip precludes the nanowires from a variety of applications including field emission, lasing, photovoltaic cells, etc. Though it is possible to remove the catalyst after growth, it also etches the nanowires8. Therefore, the requirement demands to synthesize catalyst free nanowires. Non-catalytic growth of several nanowires (Si, Ge, GaN, ZnO, Ga2O3, CdSe, B) and their heterostructures have been possible16,17,18,19,20,21,22,23,24,25,26. Nanowires produced by non-catalytic growth mechanism seem to best suit the nanowire technology development as there is no involvement of catalytic nanoparticles. It is believed that non-catalytic growth proceeds in two steps26: (1) nucleation of seed nanoparticles controlled by thermodynamic size limit and (2) subsequent one-dimensional anisotropic growth without catalyst.

The growth of the nanowires is a process governed by a competition between kinetics and thermodynamics. Theoretical descriptions on the catalytic as well as non-catalytic growth of nanowires are available in the literature26,27,28,29. The diameter of nanowires is mainly controlled by the size of the catalyst27,28,29, while the diameter of nanowires obtained by non-catalytic growth is governed by thermodynamic limit26. However, the length and hence, the aspect ratio of the nanowires, is expected to increase with growth duration. But the quest to improve aspect ratio is limited by various experimental shortcomings such as catalytic poisoning30, degradation of the precursors and growth in all possible directions31. Some efforts have been channelized in overcoming the above limitation such as by mixing NaCl with Zn to prevent degradation of precursors32.

Here, we report the non-catalytic synthesis of ZnO nanowires on soda lime glass (SLG) substrates and shown that the diameter of the nanowires is not only controlled by thermodynamically but also kinetically. It is shown that smaller diameter of the nanowires as compared to that set by the thermodynamic limit can be obtained. We have employed a technique wherein the source is protected from degradation ensuring the continuous supply of vapor. This is achieved by placing Si cap on Zn source. We have investigated the influence of temperature, growth duration and more importantly, the vapor flux. The results obtained are beyond the theoretical descriptions available in the literature. We have achieved ultralong ZnO nanowires (>300 μm) with huge aspect ratio (>104) and branched structures by controlling the growth both by thermodynamically and kinetically. Our results suggest the possibility of growing ZnO nanowires continuously without any apparent length limitation. Understanding the thermodynamic and kinetic controlled non-catalytic growth of nanowires is a great asset for the realization of longer and dense arrays of vertical nanowire for various applications. Furthermore, it is shown that the same source material can be used for several depositions of oxide nanowires.

Results

Fig. 1a–e shows the ZnO nanowires obtained by heating Zn powders with Si cap ( supplementary fig. 1 ) at 650°C for 5 min., 1, 2, 5, 10 and 50 h on SLG placed at 645°C. Optimum results have been obtained at a substrate temperature Tsub of 645°C (distance between the source and substrate is 5 mm). The length as a function of growth duration is shown in Fig. 1g . The length of the ZnO nanowires is found to increase almost linearly with time from 1 μm in 5 min, 12 μm in 1 h, 20 μm in 2 h, 112 μm in 10 h and 300 μm in 50 h. The increase in length with increasing in time in the case Zn source with Si cap is attributed to continuous supply of Zn vapors throughout the experiment. Since there is no catalyst used, the growth of nanowires is attributed to vapor-solid (VS) or self-catalytic mechanism. The morphology was fully consistent with the growth habit of the ZnO ideal crystal model33 in that the relative crystal growth velocity should follow the sequences: V(0001)>V(010)>V(000). The surface energy of (0001) surface is 2.0 J/m2, while that of (010) surface is 1.16 J/m2 34. Branched structures are obtained by this method are shown in Fig. 1h & i . The growth of branched structures was performed by keeping ZnO nanowires deposited on SLG at a distance of 5 mm from the source.

Figure 1
figure 1

ZnO nanowires on SLG synthesized at 650°C for (a) 5 min, (b) 1, (c) 2, (d) 5, (e) 10 (f) 50 h and (g) length as a function of growth duration for different temperatures.

The dashed lines in (g) are guide to the eye. Length of nanowires as obtained by chemical vapour deposition (CVD) is also plotted for comparison. (h & i) branched ZnO nanostructures. Regrowth of ZnO nanowires on already grown ZnO nanowires (grown at 650°C for 10 h) on a SLG substrate yield branched structures. ZnO nanowires grown on (j) MWCNT, (k) Si and (l) ITO.

Full size image

The nanowires can not only be grown on SLG but also on other substrates such as multiwalled carbon nanotubes (MWCNTs) bundles, indium tin oxide (ITO) and Si. Fig. 1j–l shows ZnO nanowires grown on MWCNT bundles, ITO and Si. The MWCNT growth was achieved by dispersing them on a SLG. The morphology obtained is uniform and similar to that obtained on SLG.

We have also investigated the growth of ZnO without the Si cap. The length of the nanowires ( supplementary fig. 2 ) is around 15 μm for 10 h deposition which is much smaller as compared to the length obtained with Si cap (length is 112 μm) but comparable with that obtained by using only zinc sources as reported in Ref. 24,25 where the length of the nanowires ranges from 2 μm to maximum of 15 μm. This clearly suggests that the length cannot be increased further without the protection of the Zn source. It may also be noted from Fig. 1g that growth rate is much higher (at least ten times) as compared to the non-catalytic growth of nanowires by chemical vapor deposition (CVD)21,22,23 and physical vapor transport technique without Si cap24,25. We also observed that the growth rate increases with the rate of gas flow ( supplementary fig. 3 ). The growth rate was increased from 12 to 20 μm/h when the gas flow rate was doubled.

Fig. 2a–c shows SEM images of ZnO nanowires obtained by heating Zn pellet at 550, 650 and 750°C for 1 h. The diameter of the ZnO nanowires reduced from 120 nm at 550°C to 46 nm at 750°C which is confirmed from the TEM images of the ZnO nanowires as shown in the insets. High resolution TEM (HRTEM) images taken from the side of the nanowires are shown in Fig. 2d–f which along with x-ray diffraction data ( supplementary figs 4 & 5 ) reveal the nanowire long axes are parallel to the [0002] direction. It may be noted that the density of nanowires increases and the diameter decreases as the temperature is increased, which is attributed to the increase in supersaturation of Zn vapours with increasing temperature. The minimum radius R* (the critical radius) can be predicted as it is given by35,36

Here, σ1 is the cluster surface energy (ZnO), σ12 is the cluster–substrate interface energy (ZnO–C), ν is the atomic volume, k is Boltzmann's constant, Tsub is the absolute temperature of the substrate, p is the pressure of the precursor at Tsub and is the equilibrium vapor pressure at Tsource (the set/source temperature). Both Tsource and Tsub at a distance of 5 mm were measured and provided in supplementary Table 1 . Based on the above equation and assuming that the Zn vapor condenses to form ZnO nanowires, the diameters at different temperatures were calculated and compared with the experimental results as shown in Fig. 2g . The theoretical results supported the decrease of diameter with increasing temperature but there is a quantitative difference. It may be noted that the diameter of nanowires increases with growth duration but very weakly as compared to the length. This corroborates the difference in the diameter and critical diameter as presented in Fig. 2g .

Figure 2
figure 2

(a–c) SEM and (d–f) HRTEM images of ZnO nanowires on SLG by heating at 550, 650 and 750°C for 1 h. TEM images of selected nanowires are shown in the insets. (g) Diameter (experimental) and critical diameter (theoretical) as a function of temperature. The red square represents the diameter of nanowires grown on a quartz substrate.

Full size image

According to Lifshitz-Slyozov-Wagner (LSW) theory37,38, the average diameter of the particles follows the relation where d is the average diameter at time t and d0 is the average initial diameter of the nanocrystals. The rate constant K is given by , where D is the diffusion constant at temperature T, Vm is the molar volume, γ is the surface energy and is the equilibrium concentration at a flat surface. For nanowire growth with negligible increase in the diameter, the length is expected to be linear with time. As the diameter is almost independent of time, LSW relation can clearly be rewritten as

where l represents the length of nanowires and α is the proportionality constant that depends on the diameter of the nanowires. Smaller is the diameter, larger is the increase in the length. Our experimental results ( Fig. 1g ) are consistent with the theoretical descriptions. Interestingly, the length is almost independent of temperature and follows a sacred relation with growth duration as evident from Fig. 1g . As discussed later, the growth of nanowires occurs through diffusion of adatoms onto the sidewalls. It has been argued for surface diffusion on sidewalls of nanowires that the length of the nanowires should be inversely proportional to the diameter39,

where c is constant. Accordingly, the length should increase with temperature as the diameter is decreasing. However, the increase in density along with the decrease in diameter with increase in the growth temperature results in the temperature independent length.

As the degree of supersaturation is increased, number of nuclei increases, subsequently leading to the reduction in the diameter of the nanowire. Saunders et al40 predicted that the degree of supersaturation of Zn first increases and then decreases as the temperature is increased during the formation of ZnO by vapour phase transport, thereby implying that it would be impossible to reduce the diameter of the nanowires with increasing temperature beyond certain limit. We have grown ZnO nanowires ( supplementary fig. 6 ) on a quartz substrate (as SLG cannot withstand the temperature) at 1123 K and the diameter is found to be ~80 nm as indicated in Fig. 2g by red square. This supports that the diameter of nanowires cannot be reduced beyond certain limit. Here we propose an alternate approach by which the diameter can be reduced further. We have shown that the diameter can be reduced by using more than one source to obtain more vapour flux without increasing the temperature.

SEM images shown in Fig. 3 are obtained by using one and three sources at 650°C. This indicates that the vapour flux is three times in the later case. TEM images of selected nanorods are shown in the insets. It is interesting to note that the diameter decreases as the flux is increased. The diameter of the ZnO nanorods is ~120 nm when single source is used while diameter reduces to ~40 nm when three sources is used as confirmed by TEM studies. Over all, aspect ratios greater than 104 can easily be achieved by this technique ( supplementary fig. 7 ).

Figure 3
figure 3

SEM images of ZnO nanowires on glass obtained by increasing the vapour flux (a) single source and (b) triple source with TEM images in the inset.

Schematic of (c) single and (d) triple source configurations.

Full size image

We believe that ZnO particles nucleates first on the substrate24,25,26,35. After reaching the thermodynamic size limit, nanowire growth proceeds in the [0001] direction ( supplementary fig. 8 ). Nanowire growth is primarily due to the adsorption of adatoms onto the sidewalls of nanowires and subsequent diffusion to the tips. This is consistent with the growth of long nanowires at long growth times. Branched structures are obtained when ZnO nanowires were grown on nanowires as shown in Fig. 1i . This also supports the adsorption of adatoms onto the sidewalls of nanowires and the formation of clusters followed by the growth of the nanowires.

There are two possible path ways for the growth of ZnO nanowires41. One is the formation of Zn nuclei followed by oxidation and growth, while the other one is the formation of ZnO nuclei followed by growth. As the oxidation of Zn is prevented, it is very likely that Zn vapors condense on ZnO in our case. However, as the source used in Ref. 24 is exposed to oxygen, the surface is expected to oxidize completely and ZnO vapor species are obtained. Variation of the degree of supersaturation with temperature is shown to be different for different pathways40. The different variation of diameter with temperature as discussed earlier can be due to the different degree of supersaturation. We would also like to point out that the vapors cannot be supplied continuously if the source is not protected from oxidation. The low growth rate reported in Ref. 24 is therefore lower as compared to ours. The source material cannot be changed as it may lead to the growth of branched structures without increasing the length of the nanowires.

The photoluminescence (PL) spectra of ZnO nanowires ( supplementary fig. 9 ) exhibit strong peaks at 380 nm in all cases suggesting the formation of high quality nanorods. There is also a weak broad peak around 530 nm which is attributed to oxygen deficiencies40,41,42. It may be noted that the PL peak at 530 nm is significant for ZnO nanowires synthesized at low temperature with one source for 1 h as compared to other nanowires. Raman spectra ( supplementary fig. 9 ) shows peaks at 333, 438, 480 and 560 cm−1 that match the position of the characteristic peaks of ZnO43. Fig. 4a & b show the PL imaging and corresponding PL spectra taken at different positions along the length of the nanowire. The strong peak around 380 nm is almost uniform along the length indicating that nanowire is optically uniform.

Figure 4
figure 4

(a) PL imaging (λexcitation = 355 nm and λdetection = 380 nm) (b) PL spectra at different positions of an isolated nanowire.

Full size image

The controlled growth of ZnO could be explained by using Si-Zn phase diagram ( Supplementary fig. 10 )44. The maximum solid solubility of Zn in Si is 0.0015 at%. When Zn pellet is heated while a part of it is in contact with Si cap as shown in Fig. 5a , the pellet gets oxidized except the surface in contact with Si. The optical photograph of the pellet recovered after the experiment is shown in Fig. 5b . Dark region inside red circle indicates the region of the pellet was in contact with the Si cap during the experiment and remains protected from oxidation. As the vapor pressure of Zn is much higher as compared to ZnO, Zn vapors escape through the Zn/Si interface and the higher pressure in that region prevent ingress of O2 and thus eliminates Zn oxidation in this region. The vapor is then transported to nearby regions and also onto the top of Si cap to form ZnO nanoneedles as shown in Fig. 5c . The solid solubility of Zn being small, Zn vapors do not react with Si and are not impeded from reaching the various substrates. Over all, the Si cap protects Zn in contact with it from oxidation thereby, acting as oxidation barrier, ensuring continuous vapor supply from Zn pellet leading to the controlled growth of ZnO. This is in contrast to the case of a “bare” Zn pellet24,25 whose surfaces oxidize and Zn supply consequently reduces over time. It may also be noted that the diameter is found to be reduced as the temperature is increased which is in contrast with that reported in Ref. 24 where the increase of diameter with increasing temperature has been reported.

Figure 5
figure 5

(a) Schematic of the formation ZnO nanowires on Si cap placed over Zn powder. (b) Photograph of the pellet recovered after the completion of experiment. Dark region (Zn) inside red circle indicates the region of the pellet was in contact with the Si substrate during the experiment. (c) SEM image of ZnO nanoneedles on Si placed on Zn pellet. Nanoneedles are obtained for a growth temperature of 650°C for 1 h.

Full size image

One of the challenges is the large area deposition. Fig. 6 shows the summary of the large area (75 × 15 mm2) deposition of ZnO nanowires on SLG. A SLG substrate (SLG-1) was placed on a support so that the height can be equal to that of Si cap placed on Zn pellet as schematically shown in the Fig. 6a . Another SLG substrate (SLG-2) is placed at a different height in a separate experiment. The optical photograph of ZnO grown at 650°C for 2 h on SLG-1 is shown in Fig. 6b which clearly reveals a uniform white deposition over length and breadth of the SLG substrate. SEM images of ZnO nanostructures taken from different location are shown in Fig. 6c–e which reveals needles-like morphology. It is interesting to note that the morphology and the dimensions of ZnO nanoneedles are identical over a large area of 60 × 15 mm2 and identical to that grown on Si cap. The morphology beyond 60 mm is different as shown in Fig. 6f . Similarly, nanorods-like morphology is obtained for ZnO grown on SLG-2 as shown in Fig. 6g–i . The morphology is identical over a large area of 60 × 15 mm2 and is different beyond 60 mm ( Fig. 6j ). These results indicate that the methodology adopted here is of a great promise for scale-up synthesis of ZnO nanowires. Over all, the diameter is increasing and the morphologies are different beyond 60 mm which may be due to difference in growth temperature.

Figure 6
figure 6

(a) Schematic for the large area synthesis of ZnO nanowires/nanoneedles.(b) Optical photographs of a bare SLG and a SLG with ZnO nanoneedles. The area of the SLG is ~75 × 15 mm2. (c–f) SEM images of ZnO nanoneedles taken from different region of SLG-1. (g–j) SEM images of nanowires grown on SLG-2. High magnification images are shown in the insets of (c & g). ZnO nanowires grown on SLG at 650°C for 2 h. (k) 1st, (l) 2nd, (m) 3rd, (n) 4th and (o) 5th subsequent deposition using the same source. (p) PL spectra of ZnO nanorods.

Full size image

Generally, the source material is used once for the oxide nanowires growth by vapour phase transport method. As the Si cap protects the source from the oxidation, we have investigated the reuse of source materials to grow nanowires. The substrates were replaced after each deposition without disturbing the source and Si cap. Fig. 6k–o show ZnO nanowires which reveal that the morphology of the ZnO nanowires is almost identical at least upto five subsequent depositions. The PL spectra as shown in Fig. 6p also reveal almost the same optical quality nanorods. The strong peak at 380 nm peak corresponds to the band-to-band transition and a weak broad peak around ~515 nm is due to the oxygen vacancy. Over all, the same source material can be used for several depositions of oxide nanostructured materials.

Discussion

We have developed a simple methodology to grow ultralong nanowires and branched structures with controllable aspect ratio without limitations in a large area. We demonstrate the significance and importance of local oxidation barrier on source for the growth of ultralong nanowires and more importantly for the first time, controlling the diameter of the nanowire kinetically without catalyst particle or seed layer. The insolubility of Zn in Si and the use of a Si cap on the Zn source to prevent oxidation is the key to grow ultralong nanowires. It has been shown that the aspect ratio can be controlled by thermodynamically (temperature) and kinetically (vapor flux). Over all, the length can be increased by increasing the deposition time and the diameter can be reduced by increasing the temperature and/or vapor flux. It is shown that smaller diameter of the nanowires as compared to that set by the thermodynamic limit can be obtained. The same source material can be used for several depositions of oxide nanostructured materials and suitable combination of materials can help to grow other oxide/sulfide nanostructures and oxide-sulfide branched structures in a controlled way.

For practical applications, selective growth of ZnO nanowires on pre-patterned substrates is required45,46,47,48,49,50,51. Our preliminary results indicate the possibility of nanowires growth on pre-patterned substrates ( supplementary fig. 11 ). Furthermore, the methodology described here along with that reported in Refs. 52 & 53 can be adopted to grow nanowires on a diverse range of pre-patterned substrates with or without catalysts.

Methods

A Si cap54 is placed over Zn pellets ( Supplementary Fig. 1 ) which prevents oxidation of Zn (which served as local oxidation barrier for Zn) and helps to grow longer nanowires without limitations. Zn pellets are made from 10 mg of Zn powders (S.D fine 99.95%) and along with Si cap is placed at the centre of the quartz boat. The whole assembly is then kept inside the alumina tube on a single zone furnace ( Supplementary Fig. 1 ). The growth substrates are placed at different distance of 5 mm from the source (Zn pellet with Si cap on top) and heated to various temperatures for different time duration under argon gas (99.99%) flow of 110–220 sccm (standards cubic centimetre per minute). The flow is controlled by a mass flow controller (MFC). The experiments are performed at atmospheric pressure and the outlet is connected to the exhaust. After the experiment, the product is analysed by Scanning electron microscopy (SEM) using SEM Quanta 200 and transmission electron microscopy (TEM) analysis using TEM technai G-20. Raman spectra were recorded using WITec alpha 200 and the photoluminescence (PL) spectra were recorded using a Jobin Yvon system with He-Cd laser with 325 excitation wavelength and WITec alpha 200 with Nd-YAG laser with 355 excitation wavelength.