Variation of passivation behavior induced by sputtered energetic particles and thermal annealing for ITO/SiOx/Si system^*

Solar grade n-type Czochralski silicon [(100), phosphor doped, chemical polished, (1.5–3.0) ${\rm{\Omega }}\cdot$ cm, (150±3) ${\rm{\mu }}$ m, 3.0 cm $\times$ 3.0 cm] was cleaned by a standard RCA method.^[17] Silicon wafers were first dipped into 0.08-mol/L iodine, which is considered as the best passivation to silicon.^[13,14] Passivated samples were sequentially sonicated for 10 min in absolute ethanol, deionized water, removing iodine. An ultrathin ($\sim 2$ nm) SiO_x layer was formed by thermally oxidizing in an oxidation furnace for 10 min at 700 $^\circ$ C. The flow ratio of high purity nitrogen and oxygen was 4 L/min: 1 L/min. It was verified that the thin SiO_x layer was involved into four compositions: lots of SiO₂, very few of Si₂O, SiO, and Si₂O ${}_{3}$ (Si 2p peak in the XPS spectrum as shown in Fig. S1 in Appendix A). Subsequently, ITO films of about (80±3) nm were deposited by the radio frequency magnetron sputtering system. The ITO target used for sputtering purposes was a sintered ceramic mixture disk of 10-wt% SnO₂ and 90-wt% In₂O₃. The base pressure was 3 $\times$ 10 ${}^{-4}$ Pa for the sputtering system. The substrate temperature of 250 $^\circ$ C, a power of 100 W, an argon partial pressure of 1.0 Pa and an argon gas flow rate of 40 sccm were set during the ITO film deposition, respectively. The carrier concentration, electronic mobility, and resistivity of ITO film were $(3.79\pm 0.04)\times {10}^{20}$ cm ${}^{-3}$, ($39.4\pm 0.8)$ cm ${}^{2}\cdot$ V ${}^{-1}\cdot$ S ${}^{-1}$, $(4.18\pm 0.07)\times {10}^{-4}{\rm{\Omega }}\cdot$ cm, respectively, which were measured by Hall Effect measurement. The average transmittance of visible spectrum was more than 85% and the optical band gap was deduced to be 3.60 eV (as shown in Fig. S2 in Appendix A), which was measured by a UV-VIS spectrophotometer.

To compare the passivation effectiveness, sputtering damage and annealing effect, the samples were divided into two groups. (i) The first set of samples, which were successively cleaned by immersion in 37% concentrated hydrochloric acid, deionized water and 5% hydrofluoric acid, removing both the ITO film and the silicon oxide layer. Then, samples are secondary passivated by 0.08-mol/L iodine. For the comparison, ITO films were directly deposited on two-sides of the unoxidized silicon substrate, with the same process of magnetron sputtering and treatment conditions. (ii) The second set of samples, which were annealed at the temperatures of 100 $^\circ$ C–700 $^\circ$ C (step length of 100 $^\circ$ C) for 30 min in the transfer sample chamber of the magnetron sputtering apparatus. The working pressure was maintained at 1 $\times$ 10 ${}^{-3}$ Pa. By contrast, ITO films are directly deposited on two-sides of seven unoxidized silicon wafers, with the same process of magnetron sputtering and post-annealing.

The ${\tau }_{\mathrm{eff}}$ was measured using a μ-PCD instrument (Semilab WCT-2000) with two-dimensional image. A 200-ns laser pulsed at 904 nm was used for the gain of photon-generated carriers. The number of photons in one pulse was 1.20 $\times$ 10¹⁹. Each sample was scanned three times at 1 mm $\times$ 1 mm resolution. Silicon wafer treated with iodine was encapsulated in a plastic bag to prevent contamination prior to be contacted by the prober. The electronic states and the chemical bonding of ITO/SiO_x/c-Si samples were examined by XPS with depth profiling (Thermo Fisher Scientific ESCA-250Xi). The sample was etched by Ar ${}^{+}$ ion beam with 1 keV, and the etching step was 20 s. A monochromatic Al Kα excitation source $\sim 1486.60$ eV was used for the XPS measurements. The pass energy of 30.00 eV, scan step of 0.10 eV and test area of 500 ${\rm{\mu }}$ m $\times$ 500 ${\rm{\mu }}$ m were set for testing. The lattice array morphology of ITO film, SiO_x layer and SiO_x/c-Si interface were analyzed by a TEM (JEOL JEM-2010 F) operated at 200 kV.

In the ${\tau }_{\mathrm{eff}}$ measurements, an x–y two-dimensional coordinate system was taken for the fixed-points to monitor the changes of ${\tau }_{\mathrm{eff}}$ before and after depositing ITO film or annealing, so that the lateral non-uniformity of the sample could be avoided. The measuring principle was designed as: at the fixed-point beside the central region, each of ${\tau }_{\mathrm{eff}}$ was adopted to the average value, and the consistency of seven measurements in τ ${}_{\mathrm{eff}}$ before and after sputtering of ITO film was guaranteed. The average value of ${\tau }_{\mathrm{eff}}$ was taken from three-time measurements and a standard deviation was derived.

Generally, for the high pure silicon wafer, the minority carrier lifetime (${\tau }_{\mathrm{eff}})$ is substantially determined by both the bulk minority carrier lifetime (${\tau }_{\mathrm{bulk}})$ and the surface minority carrier lifetime (${\tau }_{\mathrm{surf}})$ as shown in the following equation^[13,18]

$$\begin{eqnarray}&&\left\{\begin{array}{l}\displaystyle \frac{1}{{\tau }_{\mathrm{eff}}}=\displaystyle \frac{1}{{\tau }_{\mathrm{bulk}}}+\displaystyle \frac{1}{({\tau }_{\mathrm{diff}}+{\tau }_{\mathrm{surf}})},\\ {\tau }_{\mathrm{diff}}=\displaystyle \frac{{d}^{2}}{({\pi }^{2}\times {D}_{n,p})},\\ {\tau }_{\mathrm{surf}}=\displaystyle \frac{d}{(2\times S)},\end{array}\right.\end{eqnarray} \tag{ 1 }$$

where ${\tau }_{\mathrm{diff}}$ is the diffusion time for minority carriers from the interior of the wafer to the surface, decided by the wafer thickness ($d)$ and the diffusion coefficient of minority carriers (${D}_{{\rm{n}},{\rm{p}}})$; ${\tau }_{\mathrm{surf}}$ is determined by wafer thickness ($d)$ and surface recombination velocity ($S)$. When the silicon surface is passivated by 0.08-mol/L iodine which is considered as the best passivation to silicon, S is very low, and the τ ${}_{\mathrm{eff}}$ primarily indicates the bulk characteristics of the wafer. When the silicon surface is not passivated, S is very high, so the ${\tau }_{\mathrm{eff}}$ will be dominant by the surface feature of the wafer.^[13,14]

Figure 1 shows the variation of minority carrier lifetime 2D distribution of the one silicon wafer under the different states. As shown in Figs. 1(b) and (c), for those passivated samples undergoing furnace oxidation, the ITO sputtering process leads to a loss in ${\tau }_{\mathrm{eff}}$ of over 90%, dropping to about 5 ${\rm{\mu }}$ s. For the collected data in Figs. 1(a),1(d), and 1(f), the ${\tau }_{\mathrm{eff}}$ is basically maintained steady when it is once, twice and thrice passivated by 0.08-mol/L iodine, and S is very low, indicating that the ${\tau }_{\mathrm{bulk}}$ remains unchanged before and after sputtering, and it is larger than 280 ${\rm{\mu }}$ s. By contrast, the surface passivation validity of the sample where the ITO film is directly deposited on two-sides of unoxidized silicon is also poor as shown in Fig. 1(e). As a result, the structural damage should be localized at the SiO_x/c-Si interface, most likely in the form of the deep defects and the Si dangling bonds (or interface states).^[19,20]

Zoom In Zoom Out Reset image size

In order to make clear what the new electronic states are within the intermediate region of ITO/c-Si, after sputtering deposition of ITO thin film, the chemical components in the region are determined by XPS in elemental inner core–electron peaks such as In 3d, Sn 3d, Si 2p, and O 1s, with depth profiling. The relative atomic percent of In, Sn, and Si across the interface of the sample (here ITO film is deposited on silicon after 700-$^\circ$ C/10-min thermal oxidation process) is shown in Fig. 2(a) by an area integral estimation. The depth distribution of the chemical components is divided into three parts with etching times: the front ITO film, the intermediate SiO_x layer, and Si substrate subsequently. At the etching time of 720 s, the measured Si 2p spectra as shown in Fig. 2(c) is dashed dots, while the Si 2p ${}_{3/2}$, Si 2p ${}_{1/2}$, SiO_x, and SiO₂ are the deconvolution, fitted by the likely Si-correlated compositions in the region. According to the surface model of silicon oxide proposed by Seah et al.,^[15] the Si dangling bond (or interface state) within the SiO_x and SiO_x/c-Si interface is induced by sub-oxide SiO_x, which increases interface recombination and decreases samples' ${\tau }_{\mathrm{eff}}$. However, for the samples without furnace oxidation in advance, just the ITO film being directly deposited on two-sides of silicon substrate, the relative atomic percent of In, Sn, and Si across the interface are given in Fig. 2(b). The depth distribution of chemical components is similar to the former case except a right shift of 40 s. The sub-oxide SiO_x is also observed in Fig. 2(d), corresponding to the SiO_x/c-Si interface. The different finding is that the existence of a broad SiO_x layer at the interface region between ITO and c-Si is also revealed by XPS signal, and SiO_x just come into being in the process of sputtering deposition of ITO film.^[21] On the other side, it is reasonable for the right shift of the middle position of the interface region taking place in Fig. 2(b) and a little decrease of the intensity of Si2p peaks in Fig. 2(d), because of the different processing for the two kinds of samples. Therefore, the Si dangling bond appeared in both kinds of samples is identically linked to the sputtering process.

Zoom In Zoom Out Reset image size

Figures 3(a) and 3(b) show the cross-section TEM images of the sample, corresponding to Figs. 2(a) and 2(b), respectively. It is apparent that an ultrathin SiO_x layer presents between the ITO layer and Si substrates in all samples. The thicknesses of SiO_x layers in Figs. 3(a) and 3(b) are separately found to be about 2.3 nm and 1.4 nm. Furthermore, we observed the amorphous phase of the SiO_x layer and "atomistic interleaving structure" at the SiO_x/c-Si interface, which may relate to the intermediate oxidation state of SiO_x and most of the Si dangling bonds.

Zoom In Zoom Out Reset image size

We now turn to the origin of the sputtering damage. During the sputtering deposition of ITO film, the average kinetic energy of energetic particles (ions, atoms, molecules, and atomic clusters) is about 10 eV.^[22] Oxygen atoms (O ${}^{+0})$, oxygen ions (O ${}^{-1,-2})$, indium atom (In ${}^{+0})$, indium ions (In ${}^{+3})$, tin atoms (Sn ${}^{+0})$, tin ion (Sn ${}^{+2,+4})$, and reflected argon atoms (Ar ${}^{+0})$, argon ions (Ar ${}^{+1,+2})$ (energies up to 150 eV) primarily bombarded the substrate in the processing of magnetron sputtering.^[6,9] The bond energies of Si–Si, Si(Si)–Si, Si(Si)₂–Si, Si–O, and Si(O)–O are 3.21, 4.19, 4.56, 8.29, and 4.71 eV, respectively, which are less than that of energetic particles.^[23] Therefore, the energetic particles will break the above-mentioned atomic bonds and alter the electron state of the SiO_x layer as well as the SiO_x/c-Si interface, increasing the Si dangling bonds and the interface recombination velocity, and reducing the ${\tau }_{\mathrm{eff}}$. In addition, ultraviolet (UV) radiation from glow discharge and particle collision, reach up to 7.8 eV (or higher), which can pass through the SiO_x layer, reaching to the SiO_x/c-Si interface and the interior of c-Si.^[6] UV photons can excite electrons, injecting from the conduction band of Si into the conduction band of SiO₂, activating the deep defects at the SiO_x/c-Si interface and generating additional interface states (or Si dangling bonds). The required energy for this effect is only 3.1 eV, which is less than the UV light energy.^[24] Therefore, UV irradiation can damage the SiO_x/c-Si interface, resulting in the attenuation of ${\tau }_{\mathrm{eff}}$. In summary, the Si dangling bonds induced by sputtering damage are consistent with above experiment results. The Si dangling bonds are created from particle bombardment and UV light irradiation.

Figure 4 shows the variation of ${\tau }_{\mathrm{eff}}$ dependent on the vacuum annealing temperatures, thermal oxidation passivation, and sputtering deposition of ITO on the polished silicon wafer. It is found that a large decrease of ${\tau }_{\mathrm{eff}}$ for the samples after a sputtering deposition of ITO films occurs, which means a deterioration of the normally thermal oxidation passivation. The ${\tau }_{\mathrm{eff}}$ of c-Si after thermal oxidation at 700 $^\circ$ C for 10 min is constantly kept at 107 ${\rm{\mu }}$ s, while the τ ${}_{\mathrm{eff}}$ is decreased to 5 ${\rm{\mu }}$ s upon subsequent sputtering deposition of ITO films. The likely interpretation is the worsening of the saturation of surface dangling bonds on c-Si, as the ${\tau }_{\mathrm{bulk}}$ may not be changed during the sputtering deposition of ITO films. However, the damage of SiO_x/c-Si interface states is partially repaired after vacuum annealing under 100 $^\circ$ C–400 $^\circ$ C for 30 min, improving the ${\tau }_{\mathrm{eff}}$ of the samples. The ${\tau }_{\mathrm{eff}}$ increases at first and then decreases with the temperature. Herein the ${\tau }_{\mathrm{eff}}$ is recovered to nearly 30 ${\rm{\mu }}$ s at 200 $^\circ$ C for 30 min, after that it is declined with the temperature until to about 3 ${\rm{\mu }}$ s after 500 $^\circ$ C. These changes of the ${\tau }_{\mathrm{eff}}$ can be verified by a two-dimensional mapping measurement as shown in Fig. 5. Figures 5(a), 5(b), and 5(c) are corresponding to the thermal oxidation, sputtering deposition of ITO, and vacuum annealing at 200 $^\circ$ C, respectively. It should be mentioned that all measurements of the ${\tau }_{\mathrm{eff}}$ for the different situation are taken at those fixed points according to the former established x–y frame of axes on the c-Si wafer.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Based on the definition and description of efficient minority carrier lifetime in Subsection 3.1, the lowering of the τ ${}_{\mathrm{eff}}$ either in the sputtering process of ITO films or in the vacuum annealing at higher temperatures must be correlated to the somehow variation of the electronic states on the surface of c-Si or inside the SiO_x or the interfacial region of SiO_x/c-Si, because the ${\tau }_{\mathrm{surf}}$ is very sensitive to the changes of the chemical components on the top several layers of SiO_x/c-Si. Thus, it is necessary to extract the microstructure information within the interfacial region between ITO and c-Si so as to gain essential cause to induce the lessening after the sputtering deposition of ITO film on Si. So far, the best choice and combination for the purpose of this characterization is to apply XPS for stoichiometry and TEM for phase structure determination from the surface, down to the interface and entering the bulk to the ITO/SiO_x/Si system. In Fig. 3 we have learned the physical phase and the geometrical structure of the interfacial materials through TEM. We are sure that the ITO film and SiO_x are polycrystalline and amorphous phases, respectively. In the following sections, we will investigate the chemical configuration and electronic states of the ITO/SiO_x/Si system, especially for the SiO ${}_{x}$ and SiO_x/Si intermediate matter.

From the point of view of the application of photovoltaic device, the last step of low temperature processing at about 200 $^\circ$ C is always used for the windows or electrode fabrication. The process is equivalent to the low temperature annealing treatment. Therefore, in the photoemission spectroscopy studies, we pay main attention on the annealing effect of the ${\tau }_{\mathrm{eff}}$ evolution associated with the new electronic structures and chemical states of the SiO_x/c-Si for those samples after ITO deposition by sputtering method and subsequent annealing. There are two kinds of samples selected for XPS with depth profiling. One is for the annealing at 200 $^\circ$ C, ${\tau }_{\mathrm{eff}}$ of about 30 ${\rm{\mu }}$ s. Another is for the annealing at 600 $^\circ$ C, ${\tau }_{\mathrm{eff}}$ of about 3 ${\rm{\mu }}$ s. They are two extreme conditions for the comparison, in order to reveal the change of the fraction of S–O bonds in the intermediate region. All the Si 2p electrons emission peaks are collected at the middle position of the depth profile of chemical components as shown in Fig. 2(a), which can well manifest the variation of the electronic structures and chemical states in the SiO_x/c-Si interface.

In detail, figure 6 shows XPS spectra for the vacuum annealing samples at 600 $^\circ$ C (${\tau }_{\mathrm{eff}}$ is about 3 ${\rm{\mu }}$ s) and 200 $^\circ$ C (${\tau }_{\mathrm{eff}}$ is about 30 ${\rm{\mu }}$ s), respectively. The typical binding energies for the different Si–O bonds stoichiometric amorphous SiO_x ($0\le x\le 2$) are listed in Table 1. Those XPS-Si 2p peak spectra with subtraction of the background noise in 3 and 30 ${\rm{\mu }}$ s statuses are shown in Fig. 6(a). It is apparent that there is a little difference for the peak shape of Si 2p, corresponding to SiO_x (the range of binding energy is 100 eV–102 eV) of the two samples, with this high resolution analysis (${\rm{\Delta }}{E}_{\mathrm{Bindingenergy}}\approx 0.45$ eV). In addition to the illustration, a set of normalized and partially enlarged spectra, corresponding to the SiO_x component, is obviously observed in Fig. 6(b), and an intensity increase of sub-dioxide SiO_x after 600-$^\circ$ C vacuum annealing has been obtained. To understand these variations on the electronic structures where they are closely correlated to the SiO_x fraction in the intermediate region of ITO-Si, the deconvolution of XPS spectra for 3 and 30 ${\rm{\mu }}$ s in ${\tau }_{\mathrm{eff}}$ are separately shown in Figs. 6(c) and 6(d). The Si 2p (SiO ${}_{x})$ peak of XPS spectra for 3 ${\rm{\mu }}$ s is stronger than that for 30 ${\rm{\mu }}$ s; oppositely, the Si 2p (SiO ${}_{2})$ peak of XPS spectra for 3 ${\rm{\mu }}$ s is weaker than that for 30 ${\rm{\mu }}$ s. This contrast implies an increase of SiO_x and a decrease of SiO ${}_{2}$ after 600-$^\circ$ C vacuum annealing. Secondarily, the distinct deconvolution of XPS spectra for 3 ${\rm{\mu }}$ s and 30 ${\rm{\mu }}$ s in ${\tau }_{\mathrm{eff}}$ are shown in Figs. 6(e) and 6(f), respectively. Several sub-dioxides such as Si₂O₃, SiO, and Si₂O exist in the SiO_x/c-Si interface for both samples. However, the distort Si–O bonding states in the SiO_x/c-Si interface have more of a fraction after 600-$^\circ$ C vacuum annealing. According to the gaseous unstable SiO forming principle proposed by Yanjun Wang et al.,^[25] under low oxygen partial pressure (below about 10 ${}^{-4}$ Pa during annealing) and high temperature (500 $^\circ$ C–700 $^\circ$ C) conditions, two chemical reactions occurred in the SiO_x/c-Si interface as follows:

$$\begin{eqnarray}&&\mathrm{Si}+{\mathrm{SiO}}_{2}\to 2\mathrm{SiO}\uparrow,\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&2\mathrm{Si}+{{\rm{O}}}_{2}\to 2\mathrm{SiO}\uparrow,\end{eqnarray} \tag{ 3 }$$

where SiO (gaseous unstable substances) is difficult to be oxidized into SiO₂ phase under low oxygen partial pressure, then SiO escapes from SiO₂, which will result in an oxygen vacancy in the SiO_x/c-Si interfacial region, so that the Si dangling bonds (or interface state density) was promoted and the severe attenuation in ${\tau }_{\mathrm{eff}}$ was induced.^[25]

Zoom In Zoom Out Reset image size

Table 1.  Binding energies for amorphous SiO_x ($0\le x\le 2$).^[15,26] (peak error $\le 0.10$ eV).

Peak name	Peak position/eV
Si 2p ${}_{3/2}$	99.20
Si 2p ${}_{1/2}$	99.80
Si2O	100.15
SiO	100.95
Si2O3	101.70
SiO2	103.00

From the literatures, the dissociation energies of Sn–O and In–O are separately 5.47 eV and 3.59 eV in the crystallized ITO.^[26] On the other hand, the unintentional oxide behavior of Si in the ITO/c-Si has been reported to grow when the sample is heat treated at high temperatures (heated at 785 $^\circ$ C for 33 min),^[27] but the oxygen atoms (or oxygen ions) come from the decomposition of ITO materials. He et al. found that the sub-dioxides oxidize into SiO₂ until annealing temperature exceeds 1000 K at enough oxygen ambient.^[28] However, as shown in Fig. 6(d), more SiO₂ and less SiO_x exist in the SiO_x/c-Si interface of our sample for 30 ${\rm{\mu }}$ s, indicating that additional oxygen atoms (or oxygen ions) involved in the reaction processes as Eqs. (2) and (3) under low temperature annealing (100 $^\circ$ C–400 $^\circ$ C). We did a comparison experiment to explore the source of oxygen atoms (or oxygen ions). The ITO films were directly deposited on two-faces of seven unoxidized silicon, and the samples were treated with the same processing of magnetron sputtering and post-annealing conditions. The unintentional oxide layer (seen in Fig. 3(b)) has a poor passivation effect on the Si layer (as shown in Fig. 1(e)). If these samples' τ ${}_{\mathrm{eff}}$ are not improved by low-temperature annealing, the additional oxygen atoms (or oxygen ions) are not derived from the ITO film. Figure 7 shows the varying curves of τ ${}_{\mathrm{eff}}$ with the temperature before and after vacuum annealing for 30 min (double-sided deposition of ITO by sputtering on a non-oxidized silicon wafer). The samples' ${\tau }_{\mathrm{eff}}$ are unchanged before and after low temperature annealing. By comparison with low temperature parts of the annealing curve in Fig. 4, they have different trends, indicating that oxygen atoms (or oxygen ions) are derived from the SiO_x/c-Si interface rather than from the ITO film (refer to the samples with thermally grown SiO ${}_{x})$. Nevertheless, the vacuum annealing under 500 $^\circ$ C–700 $^\circ$ C leads to a loss in ${\tau }_{\mathrm{eff}}$. It is also verified that the occurrence of chemical reaction is like Eqs. (2) and (3). The degree of attenuation in ${\tau }_{\mathrm{eff}}$ becomes weak at 700 $^\circ$ C, attributing to a small amount of oxygen atoms from ITO film diffused into the SiO_x/c-Si interface, which is consistent basically with the previous report.^[27]

Zoom In Zoom Out Reset image size

To summarize, on the one hand, the electronic structure of ultrathin silicon oxide passivation was inevitably deteriorated by energetic particles bombardment and UV light radiation during the sputtering deposition of ITO film on c-Si. Thus, the interface recombination rate of photovoltaic devices has to be increased, which is particularly prominent in the SIS and HIT devices. On the other hand, low-temperature annealing can improve samples' τ ${}_{\mathrm{eff}}$. The following factors are implied by the above discussion: the energetic particle breaks the Si–O bond of the SiO_x/c-Si interface through impact ionization, which damages the lattice structure of the oxide layer and generates defects (Si dangling bonds). Some oxygen atoms (or oxygen ions) will obtain certain energy and diffuse to the SiO_x/c-Si interface. These oxygen atoms (or oxygen ions) will relax if their energy is less than the reaction activation energy (2 eV) of an oxygen atom in the crystalline silicon.^[29] Moreover, excessive silicon ions of the SiO_x/c-Si interface have come into being in the intentional oxidation process.^[16] The relaxation oxygen atoms (or oxygen ions) will react with excessive silicon ions under low-temperature annealing for 30 min, partially repairing the SiO_x/c-Si interface damage and reducing the interface states. Therefore, the passivation quality of the SiO_x layer and samples' ${\tau }_{\mathrm{eff}}$ are improved. In consequence, from the formation of heterojuntion structure, establishment of built-in electric field, depositing of TCO films and preparation of external electrodes to the post-annealing process, which make it possible to prepare efficient silicon-based thin-film heterojunction solar cells in low temperature and achieve a good negative temperature coefficient ($-0.25$%/$^\circ )$.^[30]

A thin SiO_x layer was produced by thermally oxidizing in an oxidation furnace for 10 min at 700 $^\circ$ C. The flow ratio of high purity nitrogen and oxygen was 4 L/min:1 L/min. The thin SiO_x layer was made up of four compositions: plenty of SiO₂, very few of Si₂O, SiO, and Si₂O₃ (as shown in Fig. S1).

Zoom In Zoom Out Reset image size

The optical characteristics of ITO film were measured by UV-VIS spectrophotometer. The average transmittance of the visible spectrum is more than 85% and the optical band gap was extrapolated to be 3.60 eV (as shown in Fig. S2).

Zoom In Zoom Out Reset image size