Introduction

Vanadium pentoxide (V2O5) is the most stable oxide in the vanadium oxide system, which has an indirect bandgap of about 2.2–2.4 eV energy. Owing to its unique layered structure, V2O5 shows interesting optical and electronic properties, thereby attracting attention for the research and applications in diverse areas such as chemical sensing, multi-coloured photochromism, cathode material in batteries and catalysis (Legrouri 1993; Moshfegh 1991; Schoiswohl et al. 2004; Talledo et al. 2003; Fei et al. 2008; Hu and Zhong 2005; Stoyanov et al. 2006). Recently, research indicated that V2O5 has shown light emission in visible region; and the photoluminescence (PL) intensity can be improved by heat treatment (Wang et al. 2007). However, owing to the relatively weak PL intensity, V2O5 has not been considered for the light emission applications.

It has been observed that the combination of V2O5 with other transition metal oxides improves the electronic, optical and catalytic characteristics of the composite system (Zou et al.2009a, 2010a, b; Liu et al. 2007; Stoyanov et al. 2006). Similarly, vanadates such as Mg3(VO4)2, LiZn(VO4)2 and NaCaVO4 show good luminescence properties. Upon further doping with rare earth element ions such as Eu3+ ions, vanadate complexes such as Ba3V2O8 and Ca3Sr3(VO4)4 show enhanced fluorescence light emission (Chen et al. 2010; Choi et al. 2009). In fact, these compounds generally have VO 3−4 group in which V5+ ion is surrounded by four O2− ions in tetrahedral symmetry. Upon photoexcitation, the charge transfer from oxygen to vanadium ion is pronounced. Accordingly, efficient energy transfers from the vanadate ions to luminescent centres take place easily, making these compounds attractive candidates for luminescent applications.

As a typical vanadate, Zn3(VO4)2 has fascinating microstructure with high porosity. The lattice is assembled from layers of Zn octahedral connected by tetrahedral vanadate groups (Hoyos et al. 2001; Umemura et al. 2006). It has three polymorphs; α, β and γ___α-Zn3(VO4)2 is the stable room temperature phase, whereas β and γ are the non-quenchable high-temperature phases (Hng and Knowles 1999). Till now, few studies have reported the crystal structure, synthesis and characterization of Zn3(VO4)2. In a recent article, two methods have been described for the synthesis of Zn3(VO4)2: hydrothermal and citrate–gel combustion method (Pitale et al. 2012). In another study, Zn3(VO4)2 has been prepared by hydrothermal method from Zn3(OH)2V2O7.nH2O as the starting material.(Ni et al. 2010) These reports describe strong photoluminescence emission in the visible region from Zn3(VO4)2 compound. However, the origination of the visible light emission is still not fully understood, and the relationship between the microstructure and the optical property needs detailed investigation.

In this article, we report the preparation of Zn3(VO4)2 by means of magnetron sputtering method with two different deposition routes and subsequent heat treatment. The influencing factors, such as annealing temperature and deposition parameters, on the microstructure and luminescence properties of Zn3(VO4)2 have been systematically studied. Strong photoluminescence in visible region has been observed, which demonstrates that Zn3(VO4)2 compound can be a potential material for light emission applications.

Experimental details

Two types of samples, A and B, were prepared by DC magnetron sputtering on glass substrates in Ar atmosphere. The glass substrate was rinsed in alcohol, cleaned in ultrasonic bath for 10 min and then blown dried using hot air. After introduced into the sputter chamber, Ar plasma produced by RF was used to further clean the glass substrate surface for 1 h before deposition. Ar was used as the working gas. The background pressure of the sputter chamber was 2.67 × 10−4 Pa. The working gas pressure was 1.33 Pa with a flow rate of 1.67 × 10−7 m3/s. During the deposition, the substrates were rotated with the speed of 3 rpm.

Sample A was prepared by dual-target co-sputtering of pure ZnO (99.9 %) and V2O5 (99.9 %) at room temperature. In the sputtering chamber, ZnO and V2O5 targets were installed at 180o (parallel) to each other, and thus co-sputtering experiment can be conducted. The deposition time was 4 h with a DC current of 0.25 A. After deposition, the samples were annealed in a tube furnace at 500 and 550 °C in oxygen atmosphere with a flow rate of 3.3 × 10−6 m3/s. The crystallization and optical properties of V2O5 are known to be sensitive to annealing temperature. Therefore, temperature of furnace was carefully calibrated with thermocouple before annealing.

Sample B was prepared involving four steps: (1) deposition of V2O5 by sputter for 4 h with DC current of 0.35 A on glass substrate at room temperature; (2) subsequent annealing the V2O5/glass sample in oxygen atmosphere at 500 °C for 1 h to form the V2O5 crystals; (3) sputtered ZnO with DC current of 0.25 A onto the annealed V2O5/glass for 40 min at room temperature; and (4) after ZnO deposition, the ZnO/V2O5/glass samples were again annealed at 500 °C in oxygen atmosphere for 1 h at the same flow rate as for sample A.

Crystal structures of the obtained samples were investigated by X-ray Diffraction (XRD) with Cu Kα radiation (Bruker D2). Microstructures and compositions of the products were examined by means of Philips XL-30S scanning electron microscope (SEM) equipped with energy-dispersive x-ray spectroscopy (EDS). For EDS analysis, beam line with spot size 4 and 20 kV voltage has been used which gives penetration depth of ~1 μm. The room temperature photoluminescence (PL) test was performed by a He–Cd laser (λ = 325 nm) to investigate the light emission properties of the samples.

Results and discussion

To study the crystalline structure and phase changes in sample A before and after heat treatment, XRD has been performed as shown in Fig. 1. The sample without heat treatment shows no diffraction peaks in the pattern, indicating an amorphous structure (Fig. 1a). It is consistent with the previous report that the pure V2O5 prepared by sputter at room temperature always shows amorphous structure at room temperature (Zou et al. 2009a, b). The obtained XRD result indicates that the simultaneous sputtering from both targets does not favour the ZnO crystallization, which is perhaps due to the disruption of crystal structure of ZnO by the unbalanced stoichiometry of V2O5 and ZnO. Another possible explanation is ascribed to the presence of high content of vanadium in the film which leads to the distortion of ZnO crystal lattice through incorporation of V atoms at the Zn points (Wang et al. 2010).

Fig. 1
figure 1

XRD pattern of sample A: (a) without heat treatment; (b, c) annealed at 500 and 550 °C in oxygen atmosphere

Full size image

Upon heating to 500 °C, diffraction peaks indexed as ZnO and α-Zn3(VO4)2 are noted (Fig. 1b), implying that weak crystallization process has started because of the surface diffusion process. At the annealing temperature of 550 °C, three major phases appeared in the XRD pattern, α-Zn3(VO4)2, V5O9 and ZnO (Fig. 1c). α-Zn3(VO4)2 is a low temperature polymorph, which has the orthorhombic crystalline structure with a = 0.6088, b = 1.1498 and c = 0.8280 nm according to JCPDS No. 73–1300.

Microstructure and morphology of the sample A before and after heat treatments at 500 and 550 °C have been investigated by SEM as shown in Fig. 2. The total thickness of the film measured by the SEM cross section is ~950 nm. It is observed that the film without heat treatment consist of nanoparticles separated by void regions (Fig. 2a). Changes in surface morphology are noted by annealing the sample at different temperatures. At 500 °C, the amorphous nanoparticles started to agglomerate and form small crystals as viewed in Fig. 2b. The cross section shows thick and dense film after heat treatment as shown in the inset of Fig. 2b. After an annealing at 550 °C, some big crystals appeared on the surface (Fig. 2c). The underneath bush-like structures can be seen through the voids between these crystals.

Fig. 2
figure 2

SEM images of sample A: a without heat treatment; b, c annealed at 500 and 550 °C in oxygen atmosphere and d EDS pattern of sample A annealed at 500 °C

Full size image

To further investigate the composition of the sample annealed at 500 °C, EDS was performed. Vanadium and zinc elements are clearly observed from the spectrum as shown in Fig. 2d. The Si and Au peaks come from the glass substrate and coating of sample respectively. The atomic ratio of Zn:V:O is 50:15:35, while the atomic ratio of Zn:V:O should be 23:15:62 according to the molecular formula of Zn3(VO4)2. Thus, the EDS result shows the presence of Zn interstitials or oxygen vacancies, if we assume that all ZnO and V2O5 have been transformed to a compound Zn3(VO4)2. However, XRD results indicate the presence of ZnO and V5O9 in the annealed sample at 550 °C. This confirms that the ZnO and V2O5 are not fully mingled at temperature ≤550 °C. By combining the EDS and XRD results, two possible explanations for incomplete formation of Zn3(VO4)2 are proposed: first, the annealing temperature is not high enough to provide the required activation energy for the formation of Zn3(VO4)2 as mentioned in the ZnO–V2O5 phase diagram (Kurzawa et al. 2001). The second reason lies on the fact that Zn, V and O atoms are not stoichiometrically distributed in the film as the sample is prepared by the dual target co-sputtering, thus the atomic ratio of Zn and O atoms does not match with the composition of Zn3(VO4)2. Zn is found in excess whereas O is deficient as indicated by EDS result. Thus, it is understandable that Zn interstitials have combined with O atoms to form ZnO, whereas V and O atoms in combination formed low oxidation state vanadium oxide, i.e., V5O9 in the samples annealed at 550 °C. As a result, the V:O ratio has been reduced from 2.5 to 1.8 during the phase change from V2O5 to V5O9 because of the deficiency of O atoms in the film.

The XRD pattern of pure V2O5/glass sample after heat treatment at 500–550 °C in oxygen atmosphere give rise to two diffraction peaks associated with strong β-V2O5 with the preferred (200) orientation and γ-V2O5 as observed in Fig. 3a–b. The intensity of both β-V2O5 and γ-V2O5 phases increases with increasing temperature, which indicates the temperature sensitive crystallization. α-V2O5 has not been observed at any stage as reported previously for pure V2O5 annealed at 550 °C (Zou et al. 2009b).

Fig. 3
figure 3

XRD pattern of: (a, b) Pure V2O5 annealed at 500 and 550 °C; (c) sample B without heat treatment and (d) annealed at 500 °C in oxygen atmosphere. Inset shows the detailed XRD pattern of the same samples in the range 2θ = 15–50o

Full size image

To investigate the crystal structure and phases formed in the V2O5–ZnO binary system, XRD was performed on sample B before and after heat treatment. Fig. 3c shows the XRD pattern of the sample without heat treatment after ZnO thin layer deposition. Two major peaks have been noted in the pattern; β-V2O5 at 12.2o and ZnO at 34.4o indicating that both V2O5 and ZnO maintains the separate phase and requires high energy to form zinc vanadium oxide compounds. The sputtering process at room temperature does not provide this energy, thus the post-deposition heat treatment at critical temperature is required to provide enough kinetic energy for the solid phase reaction. It is clear that after heat treatment at 500 °C in oxygen atmosphere, sample B shows α-Zn3(VO4)2 as major phase along with two other peaks from ZnO and γ-V2O5, representing that ZnO and V2O5 are not fully reacted to form Zn3(VO4)2 at this temperature as shown in Fig. 3d.

The SEM images of sample B before and after heat treatment are shown in Fig. 4a–b. The sample without heat treatment after ZnO thin layer deposition shows two types of structures: microrods with the length of several microns, and small grains. These microrods are assumed to be β-V2O5 according to the XRD pattern in Fig. 3a, which is consistent with the previous report that pure V2O5 on glass transforms to nanorods after annealing at 500 °C (Zou et al. 2009a, b). The small grains which are dispersed between the microrods might be ZnO crystals. The inset of Fig. 4b shows the cross section of the sample after heat treatment at 500 °C. It is to be noted that V2O5 microrods have disappeared, and the sample is quite dense, homogenous and smooth with low porosity. It can be speculated that V2O5 microrods are mingled with ZnO grains to form zinc vanadium oxide crystals as confirmed by XRD (Fig. 3d). By comparing the SEM images of samples A and B after 500 °C heat treatment, it can be observed that the surface morphologies seem to be similar. The EDS spectrum of annealed sample B is shown in Fig. 4c. The Zn:V:O atomic ratio is 25:24:51 which agrees quite well with the atomic ratio of Zn3(VO4)2. Compared with sample A, sample B contains less oxygen vacancies and Zn interstitials.

Fig. 4
figure 4

SEM image of sample B: a without heat treatment; b, c SEM image and EDS spectra of the sample annealed at 500 °C in oxygen atmosphere

Full size image

He–Cd laser with the wavelength of 325 nm was used to investigate the photoluminescence (PL) properties of the samples annealed at different temperatures. Light emission could not be observed for sample A without heat treatment. Emission bands in the UV and visible region can be clearly seen from sample A after heat treatment. The PL intensity in UV and visible region increases with increasing annealing temperature from 500 to 550 °C. The UV emission band is observed at 384 nm, which corresponds to the bandgap energy of 3.23 eV for ZnO. Red shift has been noted in the UV emission band because of the deep level transition in ZnO. It is suggested that the V ions have been partially incorporated into the ZnO lattice to form some defects, resulting in the decreased bandgap. This is the “so-called” Brustein–Moss effect induced by the external atom doping (Krithiga and Chandrasekaran 2009). The PL in the visible region centred at 528 nm shows some vibrational fine spectra, which can be attributed to the emission bands of Zn3(VO4)2.

Weak PL has been observed from pure V2O5 sample annealed at 550 °C in the visible region centred at 538 nm. It corresponds to 2.34 eV, which matches the bandgap of V2O5. The un-annealed sample B shows weak photoemission in the UV region which corresponds to ZnO light emission. PL in visible region has not been observed which shows that light emission from V2O5 has been suppressed by ZnO thin layer. After an annealing at 500 °C, sample B showed the highest PL emission at the same centre as sample A, i.e., at 528 nm. By comparing the XRD and PL results, we can speculate that both samples have same luminescent centre (Fig. 5).

Fig. 5
figure 5

PL spectra measured at room temperature (a) sample B without heat treatment; (b) pure V2O5 annealed at 550 °C; (c, d) sample A annealed at 500 and 550 °C and (e) sample B annealed at 500 °C, all in oxygen atmosphere

Full size image

The PL intensity of annealed sample B is double compared with sample A. Usually lattice defects such as oxygen vacancies, Zn interstitials, oxygen or Zn antisites are considered to be the sources for transitions in the visible region for ZnO. On the other hand, Zn interstitials have been considered to be highly mobile, and they can easily move to other defects to form the non-radiative defect complexes which cause the decrease in PL intensity (Muller et al. 2008). The deficiency of these defects in sample B is one of the reasons for intensive light emission in the visible region.

As the bandgap of V2O5 also lies in the same range as of Zn3(VO4)2, but the un-annealed sample B has not shown any PL in the visible region, even though crystalline V2O5 exists. Therefore, we can speculate from the results obtained that the strong PL emission in annealed sample B is due to Zn3(VO4)2. In this case, coupling mechanism for photoemission is not involved between the V2O5 and ZnO as reported earlier for V2O5/ZnO bilayer composite (Zou et al. 2010a, b). It indicates that coupling mechanism or resonance effect is effective only when two particles of different metal oxides come in close contact to each other to form hetrostructures. For the current situation, the un-annealed sample B does not form hetrostructures, while the annealed sample forms a new phase Zn3(VO4)2 because of the solid solution of ZnO–V2O5.

For V2O5 doped ZnO system, the strong green band emission has been observed, whereas UV emission has become weaker than pure ZnO (Kim et al. 2005). However, for the higher concentration of V2O5 and at a critical annealing temperature, V2O5 becomes soluble in ZnO, and new phase such as Zn3(VO4)2 appears apart from ZnO and V2O5. High concentration of V2O5 introduces secondary phases in the ZnO–V2O5 system and gives rise to photoemission corresponding to the new phase.

β-V2O5 has tetragonal or monoclinic crystal structure with lattice parameters a = 0.71140, b = 0.35718 and c = 0.62846 nm. The characteristic building unit of β-V2O5 consists of edge sharing of four octahedra or forming a quadruple unit. Each octrahedra consist of off-centred vanadium atom coordinated by six oxygen atoms. V2O5 forms layered structure with the V–O double bond, i.e., vanadyl bond of the shortest bond length of 0.1583 nm in case of β phase (Filonenko et al. 2004). Like some other transition metal oxides, V2O5 forms the electronic states by hybridization of V3d-O2p states. The conduction band mainly arises from the V3d bands, and it can be divided into two sub-bands: one is a broad band located at higher energy region, whereas the other one is narrow split-off band below the broad band separated by additional gaps of ~0.35 and ~0.45 eV. The valance and split-off conduction band are separated by the indirect optical bandgap of ~2.2 eV (Khyzhun et al. 2005; Zhang and Henrich 1994).

Several articles reported the visible light emission of vanadium oxide supported on different metal oxides upon excitation (Anpo 1980; Patterson et al. 1991; Iwamoto et al. 1983; Garcia et al. 2000). It has been found that V = O vanadyl groups are the most active sites for the PL emission; the position of the emission band depends on the carriers and contents in various supported vanadium oxides (Iwamoto et al. 1983). The PL of V2O5 anchored on SiO2 support indicates that phosphorescence takes place because of charge transfer process on the surface vanadyl groups. It involves an electron transfer from O2− to V5+ ions, which results in the formation of pairs of hole centres (O−) and trapped electrons (V4+) as well as a reverse radiative decay by disappearance of hole–electron pair (Patterson et al. 1991). Frank–Condon analysis indicates that the inter-nuclear equilibrium distance between the vanadium and oxygen ions during the charge transfer process in excited state is larger by 0.012 nm than that in its ground state. This change in the inter-nuclear distance between the two states allows the transitions to a number of excited vibrational levels according to Frank–Condon principle. The PL spectrum obtained from V2O5 supported on porous vycor glass (PVG) further confirms that the energy band separation in the vibrational fine structure corresponds to the vibration energy of the double bond in the surface vanadyl groups, which becomes weak on excitation (Anpo 1980).

In another study, the solid–state reaction has been investigated between zeolite and V2O5 (Zhang et al. 1998). The PL with the strong peak at 500 nm has been reported with the vibrational fine structure similar to that of the four-fold tetrahedrally coordinated V5+ species, which have V = O vanadyl groups and were highly dispersed on SiO2.

The crystal structure of Zn3(VO4)2 consists of octahedral Zn ions connected by vanadate groups. The VO43− anion group is tetrahedral and four O2− ions are bounded by covalent bonds to the central V5+ metal ion. Luminescence phenomenon reported for other vanadate complexes such as Mg3(VO4)2, LiZnVO4, and NaCaVO4 also involves one-electron charge transfer process from the oxygen 2p orbital to the 3d orbital of the V5+ ion. Considering the electronic structure of VO 3−4 ion in Td symmetry, the bluish-green luminescence of the vanadate group in Ba3V2O8 has been observed due to both 3T2 → 1A1 transition for λmax = 490 nm and the 3T1 → 1A1 transition for λmax = 525 nm (Park and Mho 2007). Red emission has been reported by the Eu3+-activated Ca3Sr3(VO4)4 complex at 618 nm because of non-radiative transfer of absorbed photons by VO43− groups inside the host matrix to the luminescent centres like Eu3+ (Choi et al. 2009). According to the above reports and the results obtained here, we can deduce that the origin of high PL in the Zn3(VO4)2 is due to the charge transfer transitions of VO43− group, and it can act as an efficient self-activated phosphor material.

Conclusions

Two types of samples have been prepared by DC magnetron sputtering of ZnO and V2O5 targets and subsequent annealing in oxygen atmosphere. Sample A was prepared by dual-target co-sputtering of ZnO and V2O5, whereas sample B was formed by deposition of V2O5 and ZnO thin layers with separate, independent steps. SEM and XRD results have shown the formation of crystalline α-Zn3(VO4)2 on post heat treatment at 500 °C. The un-annealed samples do not show Zn3(VO4)2 compound, which confirms that V2O5 and ZnO can only form a solid solution at a critical annealing temperature. The atomic ratio of Zn:V:O is in good agreement with the molecular formula of Zn3(VO4)2 in the sample B, while the sample A has shown more Zn interstitials and oxygen vacancies because of the nonuniform distribution of Zn, V and O atoms from the dual-target co-sputtering.

PL measurements at room temperature exhibit a weak UV emission band from ZnO and strong emission band in the visible region from Zn3(VO4)2. It is observed that the presence of Zn interstitials and oxygen vacancies decreases the photoemission in the visible region. Based on the charge transfer and Frank–Condon principle, we propose that the (VO4)3− ions should be the luminescent centres for the visible light emission from Zn3(VO4)2. By optimizing the deposition parameters, PL intensity is expected to be improved further, which demonstrates that the Zn3(VO4)2 compound should be a promising material for light emission applications.