SnOx@β–SnO heterostructures and their enhanced photocurrent in photoelectrochemical cell

To improve the effective separation of the photogenerated electrons and holes, SnOx@β–SnO composites on titanium meshes were prepared by using a standard one-step hydrothermal procedure. As photoanodes in photoelectrochemical cell, the SnOx@β–SnO composites exhibited a photocurrent density of 150 μA cm−2 at 0.0 V versus Ag/AgCl under visible light illumination. The improved photocurrent can be explained as, the Sn2+ 5s impurity states in SnO2 have a band gap energy of 2.24 eV, SnOx@β–SnO composites construct a type-II heterojunction. And titanium meshes have a good conductivity, which is beneficial to transferring electron from the conduction band of SnO2 to external circuit. Furthermore, the SnOx@β–SnO heterostructures have a larger electrochemical specific surface area, and stronger light absorption from 350–800 nm.


Introduction
Photoelectrochemical cell has attracted many attentions due to its potential applications in the renewable energy. But the improvement of the energy conversion, and charge separation efficiency, as well as the catalyst stability is still a challenge, where people have made many efforts to deal with. Semiconductor heterostrucures were regarded as the suitable photcatalysts, as this structures can efficiently enhance the charge transfer and separation.
Tin oxides, with three ordinal types, SnO 2 and SnO, and Sn 3 O 4 can be applied in water splitting, supercapacitor [1] and phototransistor [2]. SnO 2 with a wide band gap has an excellent photocatalytic activity restricted only to UV light region. Because of the existence of defect in SnO 2 nanomaterials, we can prepare SnO and Sn 3 O 4 , which will be beneficial to the visible light absorption. Therefore SnO x /Sn 3 O 4 heterostructures were prepared to study the photocatalytic activity or gas sensitivity. For example, Sn 3 O 4 /SnO 2 heterostructures were used as gas sensor [3], SnO/Sn 3 O 4 heterostructures were used as photocatalyst [4]. And SnO x composites were used as active oxygen reduction reaction (ORR) catalysts to promote reaction kinetics [5]. And SnO/SnO 2 heterojunctions assembled from ultrathin nanosheets showed improved properties of gas sensors [6][7][8].
As the structure phases of the catalysts are tightly related to the energy conversion, and charge separation efficiency, it is important to study theirs effects on photocurrent. The phase changes of SnO 2 to Sn 3 O 4 /SnO 2 to SnO/Sn 3 O 4 were reported via changing precursor initial pH [9], but there is no discussion on the photocurrent, nor on SnO/SnO 2 related water splitting. Recently, Zaraska et al prepared porous anodic Sn 3 O 4 /SnO x films under different sets of annealing conditions and obtained a maximal photocurrent density of 30 μA cm −2 with a UV illumination light [10]. And Tian et al [11] prepared SnO and Sn 4+ -SnO nanosheets with a solvothermal method. The Sn 4+ -SnO nanosheets have a photocurrent density of 4 μA cm −2 , higher than pristine SnO nanosheets, because of the lower electron transfer resistance after doping Sn 4+ compared to pristine SnO. Usually we intend to believe that nanosheets have much more specific areas, which will be helpful for the semiconductor contacting to electrolytes. There are rarely studies of photocurrent on the microspheres composed with nanoparticles for SnO x .
In the paper, by using a standard one-step hydrothermal procedure with different ratios of Tin (II) chloride dehydrate [SnCl 2 ·2H 2 O] to sodium citrate dihydrate [Na 3 C 6 H 5 O 7 ·2H 2 O], we obtained two kinds of tin oxides microspheres (Sn-1, Sn-2) on titanium mesh. XRD patterns and Raman spectra show that sample Sn-1 is composed mainly with Sn 3 O 4 and a fraction of β-SnO, and sample Sn-2 was composed with SnO 2 and β-SnO nanoparticles. Then the two samples were used as photoanodes in PEC. The results show that Sn-2 has a photocurrent density of 150 μA cm −2 under on/off of 20 s at 0.0 V versus Ag/AgCl with illumination of 564±60 nm 300 mW cm −2 incident intensity, which is four times higher than that of Sn-1. The improved photocurrent density can be ascribed to the type-II energy band alignment between SnO 2 @β-SnO composites, larger electrochemical specific surface area and stronger light absorption from 500-800 nm.

Experimental section
Sample preparation SnO x nanomaterials were prepared by the standard one-step hydrothermal procedure.

Structural characterization
The morphologies and crystal phases were checked by a Hitachi S4800 filed-emission scanning electron microscopy (SEM) and a powder x-ray diffractometer (XRD, MiniFlex 600, Rigaku) with mono-chromatized Cu Kα radiation, respectively. High-resolution transmission electron microscopy (HRTEM, Tecnai G2 F30 S-TWIN) and selected area electron diffraction (SAED) images were obtained at the accelerating voltage of 300 kV. Raman measurements were conducted at room temperature on a Renishaw inVia Reflex Raman spectrometer with 532 nm lines. A 532 nm laser caused an incident power at the prepared nanomaterials of 3 mW, which used output power of 20 mW and objective lens magnified 50 times. Light absorption measurements were carried out using a UV-vis-NIR spectrophotometer (UV-vis, Cary-5000, Varian) with an integrating sphere. In addition, the chemical states of Sn and O elements were analyzed by x-ray photoelectron spectroscopy (XPS, Thermo ESCALAB250Xi) which was equipped with a standard monochromatic Al-kα source (hν=1486.6 eV).

PEC measurements
The electrochemical characteristics of as-prepared Sn-1, Sn-2 photoanodes were evaluated by photocurrent curves, cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS), and Mott-Schottky plots on the Zanner CIMPS electrochemical workstation (Germany) using a three-electrode cell, where SnO x photoanodes, a Pt wire and an Ag/AgCl were applied as working, counter and reference electrodes, respectively. A 300 W Xe lamp (CEL-HXF 300, Beijing Au-light, China) was employed as incident light source in 0.1 M Na 2 SO 4 electrolyte solution.

Results and discussion
The morphologies were investigated with SEM images, as shown in figures 1(a), (b). Figures 1(a), (b) is the SEM images of Sn-1 and Sn-2 microspheres on Ti mesh. The SEM images show that there are sparsely distributed microspheres with a size of 1 μm for Sn-1, as shown in figure 1(a). With increasing the precursor of Na 3 C 6 H 5 O 7 ·2H 2 O, the microspheres with a diameter of 1∼2 μm are densely covered on the surface of Ti mesh, as figure in 1(b). TEM and HRTEM characterization were carried out to investigate the crystal structures and microstructure feature. TEM image of a selected Sn-2 microsphere is displayed in figure 1(c). Those microspheres connect together, which will be beneficial to the carriers' transportation. Figure 1  Raman spectrum was carried out to explain the compositions, as shown in figure 2(a). It can be seen that there are two obvious peaks at 170 and 585 cm −1 for Sn-1. As the peak at around 170 cm −1 is the typical peak of the intrinsic phonon modes of Sn 3 O 4 [13,14], indicating that the dominant components of sample Sn-1 are Sn 3 O 4 nanoparticles. Same as for Sn-2, an obvious peak at ∼619 cm −1 corresponds to the A1g phonon modes of SnO 2 , a peak at ∼577 cm −1 is assigned to surface defects of the SnO 2 nanocrystals [15].
UV-vis−NIR absorption spectra were recorded to investigate the effect of the Sn vacancies on the optical/ electronic of the prepared samples, as shown in figure 2(b), where two samples exhibited a strong visible light absorption, but sample Sn-2 has a stronger absorption from 350 to 800 nm than sample Sn-1. There are two obvious absorbance edges between 400 and 800 nm. Samples Sn-1 and Sn-2 have a strong absorption edge at 501.5 (2.47 eV) and 541.07 nm (2.24 eV), respectively, with the same weak absorption edge at 739.4 nm  (1.68 eV), suggesting that it is not a homogeneous distribution of the defects in the prepared samples. For Sn-1, 2.47 eV corresponds to the energy of Sn 2+ 5s in Sn 3 O 4 , and 2.24 eV is to the energy of Sn 2+ 5s in SnO 2. Similar observation was reported, such as the values of 2.53 eV for defected SnO 2 [16], SnO (2.53 eV) [17] and SiO 2 /SnO 2 /SnO 2 :Sn 2+ nanomaterials [18]. As reported, self-doping of SnO 2−x nanocrystals with Sn 2+ red-shift their absorption to the visible region and simultaneously produces oxygen vacancies, which can effectively scavenge photogenerated holes and thus enable the color switching of redox dyes using visible light [19]. Here the reduction of band gaps for Sn 3 O 4 and SnO 2 can be ascribed to the oxygen vacancies. The second band gap at 1.68 eV is in agreement with the range of 1.7∼2.2 eV for the Sn 2+ -doped SnO 2 [20], and the calculation of 1.7 eV for β -SnO [21]. Figure 3(a) shows a survey XPS spectrum from the NPs, it indicates the presence of the following elements: Sn, O and adventitious C. No contaminants from the nanoparticle synthesis were detected on the sample surface. XPS spectrum of the Sn 3d core level regions is shown in figure 3(b). The double spectral lines of Sn 3d at the binding energy ∼487.1 and ∼495.5 eV for Sn-1; 486.9 and 495.3 eV for Sn-2, with a spin-orbit splitting of 8.4 eV were regarded as Sn 3d 5/2 and Sn 3d 3/2 [22]. Two Sn 3d 5/2 peaks can be well fitted with one single peak at 487.0 eV for Sn-1 and 486.88 for Sn-2 with a respective FWHM of 1.61 and 1.31 eV. These binding energies of Sn 3d 5/2 are higher than that reported for Sn 2+ (486.4-486.5 eV) and slightly lower than that reported for Sn 4+ (487.2-487.5 eV) [23]. It can thus be hypothesized that both samples might contain tin in both II and IV oxidation states, and the lower Sn 3d 5/2 binding energy of Sn-2 implies larger amount of Sn 2+ in SnO 2−x sample [24,25]. The asymmetric peak was observed in the O-1s region (in figure 3(c)), which can be deconvoluted into two components at binding energies of 530.45 and 532.18 eV for sample Sn-1, and 530.88, 531.74 and 533.67 eV for sample Sn-2, as displayed in figure 2(d) and table 1. The peak at ∼530 eV was associated to Sn 2+ and that ∼531.8 to SnO 2 (Sn 4+ ) [26]. The peak at 533.67 eV can be indexed to -OH groups, which was thought to stem from the hydrated water [27]. From the table 1, we find that the ratio of P1 to P2 intensity increases from 1.3 to 1.96, implying that sample Sn-2 has relative higher Sn 2+ than that of sample Sn-1, in agreement with the result of the bandgap redshift.
The measurement of the self-powered PEC-type detector was carried out using a continuous visible-light pulse with an on-off interval of 10 s (or 20 s) at different intensities under a Xe lamp with a wavelength of 564±60 nm, as shown in figure 4(a). The maximal photocurrent density of Sn-2 is over 200 μA cm −2 , about four times larger than that of sample Sn-1. This value is much larger than ever reported self-powered photodetectors (SPPDs) [28], 75 μA cm −2 in SnO x [29] and 60 μA cm −2 in Sn 2+ -SnO 2 composites [30]. To investigate the changes of the electrochemical specific surface area, the difference in current density variation plotted against scan rate is shown in figure 4(b). The linear slope, equivalent to twice of the doublelayer capacitance (C dl ), was used to represent the ECSA. We found that the slope of Sn-2 is about seven times larger than that of samples Sn-1, meaning that Sn-2 has quite larger electrochemical specific surface area, which will be beneficial to the improvement of photocurrent of Sn-2.
Electrochemical impedance spectroscopy (EIS) was also performed to measure the charge transfer resistance (R ct ) occurring at the interface of photoelectrode/electrolyte, as shown in figure 4(c). For each EIS spectrum, two distinct parts composing a semicircle in the high-frequency region and a straight slope in the low-frequency region are considered to be related to the charge transfer process and diffusion-limited process, respectively. The intercept at the real axis represents the resistance at interface between photoelectrode material and Ti mesh substrate (R s ) and the diameter of semicircle equals to the charge transfer resistances (R c ), respectively. It is obvious that the Sn-2 electrode possesses a lower R s and R c , consistent with the higher photocurrent of Sn-2, in figure 4(a). Mott-Schotty plots of Sn-1and Sn-2 electrodes were used to estimate the electron density, as shown in figure 4(d). Figure 4(d) is Mott−Schottky plots at fixed frequencies of 1 kHz on SnO x photoelectrode registered in 0.1 M Na 2 SO 4 electrolyte (pH 7), and a linear fitting of MS from −0.4 to 0.1 V was displayed in the inset of figure 4(d). The smaller slope of Sn-2 implies a larger carrier density. According to the Mott-Schottky equation, the capacitance of a semiconductor (C) is related to the applied potential (V) according to where E fb is the flat band potential and E, q, A, ε, e 0 and N d are the external applied potential, the elementary charge, the area of the metal contacts, the dielectric constant, the vacuum permittivity, and the carrier concentration, respectively. The optical static dielectric constant ε is equal to 7.1 [31] or 10 [32]. The flat potential E fb is obtained as −0.37, −0.16 V versus Ag/AgCl for Sn-1 and Sn-2, respectively. From equation (1), the charge carrier density N d can be calculated from the following figure 4(d), linear regions with positive slopes can be observed between −0.5 and 0.1 V, indicating an n-type semiconductor. From the inset of figure 4(d), it shows that the carrier density N d of Sn-2 is 1.36 times of Sn-1. Figure 5(a) shows the current density versus applied potential under dark and visible light irradiation, Sn-2 presented the better PEC property under illumination than Sn-1. The higher photocurrent density at different bias potentials for Sn-2 suggests more efficient charge carrier collection in SnO x nanoparticles under visible light illumination [33]. Figure 5(b) presented cyclic voltammograms recorded over a wide potential range, i.e., from −0.6 to 0.4 V versus AgCl. One anodic current peak at -0.5 V was observed for Sn-2, which can be attributed to Sn 2+ →Sn 4+ stepwise oxidation processes forming different Sn 2+ /Sn 4+ oxide/hydroxide species, consistent with the reported results. 34 Sn-2 has a larger dark current from −0.6 to −0.2 versus AgCl, meaning higher donor (electron) density as the charge transfer in the dark is dominated by the major carrier (electron). Comparing figures 5(c) and (d), one can find that sample Sn-2 has nearly seven times larger double-layer capacitance (C dl ) in comparison to Sn-1, indicating the fast transfer of carriers at the electrode/electrolyte interface and the efficient separation of electrons and holes for Sn-2 happen.
Using the previous band gap [34] of Sn 3 O 4 , we present the energy band structure of Sn-1 in figure 6(a), where the energy band structure of SnO is taken from [21]. by Wang et al. The energy band alignment of Sn-2 is displayed in figure 6(b). For Sn-2, the Sn 2+ -dopant in SnO 2 introduces more oxygen vacancies leading to the reduction of bandgap. And the electrons can excite from Sn 2+ 5s impurity state to the conduction band consisting of Sn 4+ 5s orbitals with an energy of 2.24 eV due to the introduction of Sn 2+ 5s impurity state in SnO 2. 20 Furthermore, under visible light illumination, electrons can be excited from the valence band to the conduction band of SnO, as the conduction band of SnO 2 is more positive to than that of SnO versus NHE, the photogenerated electrons can transfer from the conduction band of SnO to that of SnO 2 , and holes transfer from The photocurrent density decay for Sn-2 can be explained as the photocorrosion. Figure 6 shows that the conduction band energy of β-SnO is 0.6 V above NHE, in addition to the electrons tranfer from the conduction band energy of β-SnO to SnO 2 , Sn(II)→Sn(IV) oxidation processes will occur 27 , leading to the gradully decreased photocurrent density. As reported, photocorrosion existed in many semiconductor materials [35], which can be ameliorated with wide-band gap semiconductors capping them as protective layers [36][37][38].

Discussions and conclusions
From the discussions of HRTEM images, XRD patterns, Raman spectra and energy band gaps, we can conclude that the dominant content of sample Sn-1 is Sn 3 O 4 nanoparticles with a small fraction of SnO, while the main content of Sn-2 is SnO 2 nanoparticles with a fraction of SnO.
Because of the good conductivity of Ti net, more generated electrons can be collected by electrode and transfer to the working electrode (Pt/FTO) by the external circuit. The generated holes are driven from the valence band of SnO into the interface of SnO x nanoparticles /electrolyte and captured by the reduced form of the redox molecule (h + +OH − →OH • ). While for sample Sn-1, it is hard to realize the e-p separation. Furthermore, Sn-2 has a stronger visible absorption from 350 to 800 nm, the stronger light absorption can be ascribed to densely distributed microspheres covering on the surface of Ti mesh; Sn-2 has a larger electrochemical specific surface area, which is helpful to the ions' transfer between photoelectrode and electrolyte. Therefore, sample Sn-2 exhibits a maximal photocurrent density more than 200 μA cm −2 under on/ off of 20 s at 0.0 V versus Ag/AgCl with illumination of 564±60 nm 300 mW cm −2 incident intensity. This value is higher than previous reports of Sn-based nanostructures with a one-step procedure without any annealing treatments.
Defects in a semiconductor will induce impurity states, which generally lie within the semiconductor band gap. Those defect-induced localized states can selectively capture the approaching charge carriers, leading to spatial charge separation and thus an improved photocatalytic activity. Therefore it is an important subject to modulate defect behavior in semiconductors. Using a standard one-step hydrothermal procedure we obtained Sn 3 O 4 microspheres (Sn-1) and Sn 2+ self-doped SnO 2 microspheres (Sn-2) without any annealing treatments, and obtained a maximal photocurrent density more than 200 μA cm −2 . From the results, the photostability needs to be further improved. In the future work we will consider it through heterostructural growth.