Introduction

Being a wide direct band gap semiconductor, SnO2 is well known as one of the most promising functional nanostructure materials due to its wide energy gap (3.6 eV) and large binding energy (130 meV) at room temperature. SnO2 has been extensively used as solar cells, sensors, and optoelectronics devices due to its excellent optical and electric properties [1–7]. Therefore, various SnO2 nanostructures such as nanorods, nanowires, nanoflowers, and nanoarrays have been synthesized for the device [8–14]. Among these shapes, hierarchical nanostructure has attracted significant interest because of its widespread potential applications in many areas such as photodetectors and sensors [15–17]. The various morphologies of SnO2 hierarchical nanostructure are usually synthesized by the hydrothermal method, which is highly praised by many researchers for its low cost, environmental friendship, and convenient synthesis. The size and morphology of these SnO2 hierarchical nanostructures could be controlled by changing various reaction parameters, such as the reaction temperature, time, surfactant, and material source [18–26]. However, there are few literatures reported on synthesis SnO2 nanoflower-rod arrays (NFRAs) architecture. Furthermore, most of the mechanisms of SnO2 luminescence property mainly focus on the defects or band edge emission. In addition, the index of emission peaks both in the ultraviolet and visible emission region of SnO2 is still a challenge. Therefore, in this work, a hierarchical SnO2 NFRAs architecture was synthesized through a simple template-free hydrothermal process on the indium tin oxide (ITO) substrate. Moreover, a reasonable growth mechanism of the hierarchical SnO2 NFRAs architecture was proposed based on the morphology and structure characterizations. In addition, corresponding photoluminescence (PL) property was investigated with a first principles study. The luminous mechanism was demonstrated based on the density functional theory (DFT) calculation. This research of SnO2 material not only presents a way to synthesize the novel nanostructures but also provides theoretical reference of potential PL application.

Materials and methods

Materials synthesis

In a typical procedure, 0.3966 g SnCl4·5H2O (99 %) and 0.4667 g NaOH (99 %) were dissolved into deionized water, respectively, and the NaOH solution was added into SnCl4 solution drop by drop. After that, 28.0 mL solution in total was transferred into a 40.0 mL stainless steel Teflon-lined autoclave. Then, the cleanly ITO substrate was introduced into the precursor solution. The autoclave was kept in a bake oven at 210 °C for 24 h. Finally, the ITO substrate was purged in deionized water for several times and dried in air for further characterization.

Characterization

The structural characterization of the SnO2 NFRAs was done on ITO substrate at incidence angle of 2.0° by using X-ray diffraction (XRD, 6100, SHIMADZU) equipped with Cu Kα X-ray source operated at 40 kV and 30 mA. The scan rate of 6°/min and step size of 0.02° were used. The products were imaged using scanning electron microscope (SEM, Zeiss ΣIGMA/VP) at 3 kV with a working distance of ∼9 mm. Energy dispersive spectroscopy (EDS) was carried out at 15 kV under the SEM. Transmission electron microscope (TEM) studies were performed on JEM-3010 at operating voltage of 200 keV. The PL spectrum was recorded with a FluoroMax-4 spectrophotometer (Horiba Jobin-Yvon) with a Xe lamp and a 360 nm filter was used. Quantum yield was acquired on the same FluoroMax-4 spectrophotometer equipping with the integrating sphere. The Raman spectrum was acquired by using a Renishaw inVia micro-Raman spectrometer at room temperature, and a laser with 514.5 nm wavelength was used as the excitation light source. The ultraviolet–visible (UV–Vis) diffuse reflectance spectrum (DRS) is recorded with a UV–Vis-NIR spectrophotometer (Cary 4000, VARIAN, USA).

Results and discussion

The crystalline structure of the obtained product was checked by XRD as shown in Fig. 1. The diffraction peaks are in accordance with rutile of SnO2 (space group P42/mnm (136), JCPDS file No. 41-1445). The result shows that the pure SnO2 sample has been synthesized successfully.

Figure 1
figure 1

XRD pattern of the SnO2 NFRAs

Full size image

When ITO substrate was introduced into the precursor solution, SnO2 nanorod arrays could be readily grown on the substrate, decorated or covered by a layer of hierarchical SnO2 nanoflowers. As shown in Fig. 2, SnO2 nanoflowers are spread onto the SnO2 nanorod arrays and the nanoflowers and nanorod arrays can be observed synchronously in Fig. 2a. Figure 2b, d shows the top views and the cross section view of the SnO2 nanorod arrays with a diameter of 100 nm and length of 600 nm, which indicates that the SnO2 grown on the ITO substrate uniformly and compactly. In Fig. 2c, there are many nanoflowers, but the nanorod arrays are sightless, because the nanoflowers are plentiful and covered nanorod arrays. The TEM images of the hierarchical SnO2 NFRAs architecture are shown in Fig. 3. For measuring TEM, SnO2 NFRAs are scraped from the ITO substrate. The nanorod arrays are peeled off from the ITO substrate as shown in Fig. 3a, b. It is necessary to emphasize the fact that the SnO2 nanorod arrays epitaxially grow on the ITO substrate via the hydrothermal process. The HRTEM image from a single nanorod in Fig. 3c shows that the lattice spacing is 0.326 nm, which is corresponding to the interspacing of the (110) planes, indicating that the [001] direction is the preferential growth direction of SnO2. The fast Fourier transform (FFT) pattern in Fig. 3c (inset) also confirms the estimation that the growth was along the c axis, which was same with our previous report growth mode [27].

Figure 2
figure 2

Top and cross section views of SEM images of the hierarchical SnO2 NFRAs architecture on ITO substrate, top views (a–c) and cross section view (d)

Full size image
Figure 3
figure 3

TEM images of SnO2 nanorod arrays (a, b), HRTEM image (c) of SnO2 nanorod with corresponding fast Fourier transform (FFT) image (inset) taken from the black square

Full size image

In order to analyze the stoichiometric proportion of the comprised elements in as-prepared product, the EDS measurement has been carried out under the SEM. Figure 4 shows the EDS spectra of SnO2 nanorod arrays (Fig. 4a) and SnO2 nanoflowers (Fig. 4b). The values of the atomic ratio between O and Sn are 61.88:38.12 and 49.28:50.72 in SnO2 nanorod arrays and SnO2 nanoflowers, respectively, which means that there are some oxygen vacancy (V O) defects existing in both nanorod arrays and nanorod flowers architectures. The observed atomic ratio of Sn and O verifies the formation of nonstoichiometric phase of tin oxide. Therefore, the SnO2 NFRAs were annealed in the oxygen environment at 800 °C for 4 h. The EDS spectra of SnO2 nanorod arrays and SnO2 nanoflowers are shown in Fig. S1a and Fig. S1b, respectively, and corresponding atomic ratio between O and Sn is 69.16:30.84 and 67.25:32.75, which means the stoichiometric phase of tin oxide can be obtained.

Figure 4
figure 4

EDS spectra of SnO2 nanorod arrays (a) and SnO2 nanoflowers (b)

Full size image

Based on the above characterizations, a reasonable growth mechanism of the SnO2 NFRAs structures can be proposed. The major reaction equations during the reaction processes can be summarized as follows: [23, 28]

$$ {\text{Sn}}^{4 + } + 4{\text{OH}}^{ - } \to {\text{Sn}}\left( {\text{OH}} \right)_{ 4} $$
(1)
$$ {\text{Sn}}\left( {\text{OH}} \right)_{ 4}\,+\,2 {\text{OH}}^{ - } \to {\text{Sn}}\left( {\text{OH}} \right)_{ 6}^{{ 2 { - }}} \to {\text{SnO}}_{2} . $$
(2)

At the initial stage, while the NaOH solution was added into SnCl4 solution drop by drop, amounts of white precipitates Sn(OH)4 were produced (Eq. 1). Along with the drop process, the Sn(OH)4 was dissolved by the excess OH− to generate Sn(OH) 2−6 complex ions (Eq. 2), which was the basic cells of crystalline growth. There were large numbers of SnO2 seeds lying on the surface of the ITO substrate randomly, which offered an active sites of the SnO2 nanorod. Synthesizing SnO2 flowers usually required the formation of aggregated SnO2 nuclei in an initial homogeneous nucleation process. In the basic environment at room temperature, only a few basic cells could be hydrolyzed into SnO2 nanocrystals. However, during the hydrothermal condition, SnO2 basic cells hydrolyzation into nanocrystals could be accelerated, and then the SnO2 nanocrystals could be aggregated into SnO2 and oriented growth with the driving force of decreasing surface energy. In the reaction system, the nanorod arrays were grown on the ITO substrate and the nanoflowers formed in the solution at the same time. Following this, low formation energy of the heterogeneous crystal growth made few SnO2 nanocrystalline grow on the ITO substrate. Then, the ITO substrate served as seed layer and guided the nucleation of SnO2, which promoted the SnO2 nucleation and grown into nanorod arrays in the hydrothermal progress. Moreover, without the inducing effect of the ITO layer, SnO2 nanorod would also be produced in solution and tend to aggregate into a spherical morphology because the spherical materials reunited for high stability and low surface energy. The SnO2 nanoflowers deposited on the top of SnO2 nanorod arrays via natural precipitation [19].

For investigating the growth mechanism of the hierarchical SnO2 NFRAs, a series of experiments were carried out along with the reaction time, and corresponding SEM images of the hierarchical SnO2 NFRAs are shown in Fig. 5. Based on the morphology evolution process, the schematic illustration of the morphology evolution process can be established (Fig. 6). When the ITO was immersed in the precursor solution, the SnO2 growth units were formed in the precursor solution and grown on the substrate at the same time due to its heterogeneous nucleation. When the reaction time reaches at 15 min, the SnO2 growth units can flock together based on the seeds layer in the ITO and these nuclei also reunite to form the multitwin central nuclei under the synergistic interaction from Van der Waals force (Figs. 5a, 6a) [27]. The SnO2 nuclei further increasingly constitute the SnO2 grain with a diameter about 200 nm, and some perching sites protrude on the surface of the SnO2 grain in this stage (Figs. 5b, 6b). Following these perching sites, the bud can grow on the surface of the grain, which is made up of dozens of nanorod (Figs. 5c, 6c). Similarly, the bud is growing on the grain which is located at the ITO (Figs. 5c, 6c). The growth rate of the bud is attenuate while its length is on the verge of 600 nm (Figs. 5d, 6d), and some of the buds begin to bloom (Figs. 5e, 6d). As the reaction time goes on, the bud is in full bloom. The nanoflower and the nanorod arrays can be constructed at 24 h (Figs. 5f, 6e); the flowers deposit on the ITO substrate, and the complex hierarchical SnO2 nanoflower-rod arrays can be obtained. Cross section views of hierarchical SnO2 nanomaterials and ITO substrate in above reaction intervals are shown in Fig. S1, indicating a consistent evolution process. The microstructural information of the SnO2 homogeneous nucleation on ITO substrate directing SnO2 further growth into nanorod by increasing the hydrothermal reaction time. Therefore, the ITO is believed to serve as a seed layer at the stage of homogeneous nucleation.

Figure 5
figure 5

SEM images of hierarchical SnO2 NFRAs after different hydrothermal reaction times: a 15 min, b 30 min, c 2 h, d 8 h, e 16 h, f 24 h

Full size image
Figure 6
figure 6

Schematic illustration of the reasonable evolution processes of hierarchical SnO2 NFRAs

Full size image

The excitation and emission spectra of hierarchical SnO2 NFRAs are shown in Fig. 7a. The excitation spectrum is recorded at emission energy of 394 nm. The emission spectrum is recorded with the excitation wavelength of 337 nm in the spectral range of 350–650 nm, which is composed of an intensive UV-violet peak at 394 nm, and four shoulders at 376, 386, 415, and 437 nm (the quantum yield is 2.62 %). The sharp features of the emission spectrum should be caused by the sharp features of the excitation spectrum. Figure 7b exhibits the UV–Vis absorbance spectrum of hierarchical SnO2 NFRAs; the inset of (αhv)2 vs hv can be derived from the absorption data. As shown in Fig. 7b, the average band gap value of the SnO2 NFRAs is ~3.62 eV, which is slightly larger than that of bulk materials. In order to investigate the impact of hierarchical morphology on the PL property. The reaction time-dependent PL spectra of SnO2 nanomaterials are shown in Fig. 8a with an excitation wavelength of 337 nm. There is a weak emission peak of the ITO substrate centered at 368 nm. With the increase of reaction time, the intensity of the defect-related emission increases. When the reaction time reaches at 2 h, the SnO2 nanomaterials cover the ITO substrate completely (Fig. 5c), which results in the disappearance of the emission peak of ITO substrate. The intensity of the defect-related emission further increased until the hierarchical SnO2 NFRAs formed. The bud of SnO2 is in full bloom with the reaction time goes on (Fig. 5e), which enlarges the surface of SnO2 and results in the defect-related emission. The excitation wavelength-dependent emission spectra of hierarchical SnO2 NFRAs are shown in Fig. 8b. Interestingly, the position of emission peaks almost does not vary with the increasing of excitation wavelength; the intensity increases gradually as excitation spectrum increases from 320 to 340 nm, and the similar phenomenon of SnO2 nanomaterials has been reported [29, 30]. The Raman spectrum of the SnO2 NFRAs is displayed in the inset of Fig. 8c. The Raman peak locating at 628 cm−1 corresponding to the A1g modes, which is in good agreement with for the rutile bulk SnO2 [30, 31]. The Raman peak locating at 574 cm−1 corresponding to the AS modes, which indicated that there are V O in the SnO2 materials [30, 32].

Figure 7
figure 7

The emission spectrum of hierarchical SnO2 NFRAs under 337 nm excitation of Xe lamp and corresponding excitation spectrum at emission energy of 394 nm (a). The UV–Vis absorbance spectrum of hierarchical SnO2 NFRAs (b)

Full size image
Figure 8
figure 8

Reaction time-dependent PL spectra of SnO2 nanomaterials (a). The excitation wavelength-dependent emission spectra of hierarchical SnO2 NFRAs (b). Raman spectra of the SnO2 NFRAs (c). The diagram of relaxation process in photoexcited SnO2 (d, left) and the energy levels comparison of the V O with different charge states in band gap (d, right, the PL analysis results (blue) and the DFT calculated results (red)

Full size image

Generally, V O are known as the most common defects in SnO2, which usually play a role in radiative centers in luminescence processes on SnO2 grain boundaries and arouse to defect level emission [33–39]. In the previous discussions, the EDS and Raman results indicate that the V O defects are existed in the NFRAs. Therefore, we could conclude that V O are the main origin of the luminescence in comparison with the EDS and Raman results. The V O are the intrinsic defects in n-type SnO2, which can capture electrons and form ionized vacancies [38, 39]. There are three types of V O in SnO2 with different charge states namely V 0O , V +O , and V 2+O [34–37, 40]. The ionized vacancies in SnO2 can play a role as deep defect donors and form new energy levels, which further influences the PL properties of SnO2 [41, 42]. In fact, the V 0O state is a shallow donor near the conduction bands (CBs). Most of the V O state are V +O state, which are located under flat-band conditions. Furthermore, V 2+O state is V +O state combining with a hole. When the bands gap of SnO2 is smaller than the energy of the excitation photon, the electron can be excited to the CBs. Then, the V +O state receives a hole from the valence bands (VBs) and hence the V 2+O is created. Undeniably, the intensity of V 2+O states will increase. Accordingly, the photoexcited electrons can recombine with V +O or V 2+O state. More specifically, the peak near 437 nm (~2.84 eV), is attributed to V +O in the SnO2 nanorod arrays [33–35], i.e., the radiative transition of the V +O level to the CBs of SnO2. The 415 nm (~2.99 eV) peak can be explained by the V 0O [40], which is caused by the combination of the electron from V 0O level and a hole donated by V 2+O level. The 394 nm (~3.15 eV) emission peak is attributed to the combination of the electron from CBs and a hole from the V 2+O [35]. The 386 nm (~3.22 eV) emission peak only can be observed in our case, and it might attribute to the relaxation process of defects level to VBs. The 376 nm (~3.30 eV) emission peak is related to the impurity or defect concentration and not to the structure [43]. The energy levels of V 0O state, V +O state, V 2+O state center at 0.15, 2.84, and 3.15 eV in band gap, respectively.

It is known that different defects may cause different energy levels, which will be reflected on the spectrum (such as PL spectrum), and the spectrum can be measured directly in experiments [44–47]. The PL spectrum generally reflects the relaxation process in photoexcited materials. Furthermore, based on the first principles study, the mechanism of the PL spectrum can be explained more credibly. Additionally, the explanations of the luminous mechanism are accordant both in real materials and in first principles study [44, 45, 48–50]. Even so, there are few reports about the luminous mechanism in SnO2 based on the first principles study. Therefore, in order to further confirm the PL results, the exact energy level of V O, V +O , and V 2+O in SnO2 should be confirmed. Along this line of consideration, a DFT study is proceeded with the pristine and the defective SnO2 nanowire models. The schematic illustration of SnO2 nanowire model containing 75 atoms with a vacuum region up to 9.475 Å is shown in Fig. 9. All the calculations are computed by using the CASTEP software package in Materials Studios 5.0. The exchange correlation potential is described with the local-density approximation (LDA). The cutoff energy of plane wave is set as 400 eV. The Brillouin zone integration is approximated using k-point sampling scheme of Monkhorst–Pack and 1 × 1 × 8 k-point grids are used. The tolerance for self-consistent filed, energy, maximum force, maximum displacement, and maximum stress are set as 1.0 × 10−6 eV/atom, 1.0 × 10−5 eV/atom, 0.03 eV/Å, 1.0 × 10−3 Å, and 0.05 GPa, respectively.

Figure 9
figure 9

Schematic illustration of SnO2 nanowire model (gray, red, green color represent tin atom, oxygen atom, oxygen vacancy)

Full size image

In order to understand the relationship between defects structure and PL properties in SnO2 NFRAs, the ionization energy of V O with different types of charge states in SnO2 is calculated. The formation energy of an isolated neutral V O in SnO2 is defined by the equation as follows: [48, 51, 52]

$$ E^{f} \left[ {V_{\text{O}}^{q} } \right] = E_{\text{tot}} \left[ {V_{\text{O}}^{q} } \right] - E_{\text{tot}} \left[ {\text{pristine}} \right] - \sum {n_{x} \cdot \mu_{x} } + q\left( {E_{f} + E_{v} +\Delta V} \right), $$
(3)

where E tot[V qO ] is the total energy of the nanowire model containing one oxygen vacancy with q charge. E tot[pristine] is the total energy of the prefect nanowire model. The n x is the number of the defect atom and the μ x is chemical potential of the defect atom. The E f , E v , and ΔV are, respectively, refer to the Fermi level, the formation energy of the charged state, and the correction term. The ionization energy namely E A is computed by the following equation: [48]

$$ E^{f} \left[ {V_{\text{O}}^{*} } \right]\left( {E_{F} = E_{A} } \right) = E^{f} \left[ {V_{\text{O}}^{0} } \right]. $$
(4)

In the case of V O, the formation energy of the V 0O state centers at about 0.15 eV, and the ionization energy of the V +O state and V 2+O state center at about 2.75 and 2.98 eV, respectively. Compared with the energy level deduced in PL spectrum, the V 0O state, V +O state, and V 2+O state in band gap are 0.15, 2.84, and 3.15 eV respectively. The energy level diagram is shown in Fig. 8d and all the energy levels are respected to the VBs of prefect SnO2. The V 0O state is centered at about 0.15 eV both in real SnO2 NFRAs and the computation results, and the result is consistent with previous reports [53–55]. Furthermore, the computation results of the V +O and V 2+O states are centered all close to the real SnO2 NFRAs product. Inevitably, there are still have few deviations about 0.09 eV (V +O ) and 0.17 eV (V 2+O ), which is caused by the concentration of V O in the nanowires module. The concentration of V O is directly related to the total atoms in SnO2 nanowires module. More specifically, on the one hand, the small SnO2 nanowire model results in the high concentration of V O. The calculations with the 75 atoms nanowires model show that the ionization energy is converged well for this system size, indicating that there is a significant defect–defect interaction. On the other hand, the computational efficiency is directly affected by the system size of the SnO2 nanowire model. Consequently, a SnO2 nanowire model containing 75 atoms is chosen to confirm the PL results. Videlicet, the calculation of the V O in SnO2 can confirm the luminous mechanism in this report.

Conclusions

The SnO2 NFRAs are synthesized on the ITO substances through a template-free hydrothermal method. A reasonable growth mechanism of the SnO2 NFRAs architecture has been proposed. The ITO substrate plays a role of seed layer and guides the nucleation of SnO2 NFRAs. The PL property of SnO2 has been analyzed based on the PL spectrum, which are composed of an intensive UV-violet peak at 394 nm, and four shoulders at 376, 386, 415, and 437 nm. The three types of V O are regarded as luminescence mechanism, and the V 0O state, V +O state, and V 2+O state in band gap are 0.15, 2.84, and 3.15 eV, respectively. The SnO2 nanowire model containing 75 atoms is chosen to confirm the PL results. The DFT calculation discloses that the ionization energy of the V +O state and V 2+O state center at about 2.75 and 2.98 eV, respectively, which is in agreement with the PL spectrum.