We report on the effect of sputtering deposition of indium tin oxide (ITO) as the transparent conductive oxide layer on the passivation performance of hydrogenated amorphous silicon/crystalline silicon heterojunctions. The influence of sputtering damage on passivation performance is studied by varying the ITO layer thickness from 0 nm to 80 nm. The passivation performance decreases considerably up to 10 nm and increases gradually from 20 nm to 80 nm, indicating that damage and recovery stages are present during the sputtering process. We focus on the injection energy as the cause of the recovery phenomenon. To optimize the passivation performance by intentionally enhancing the effect of the recovery stage while minimizing the initial damage at the heterointerface, we develop a two-step sputtering process in which the radiofrequency power is changed from 50 W to 100 W during deposition to prepare ITO double layers. Two step sputtering gives the lower damage, and the properties of ITO double layers are better than those of ITO single layers. These results show that two-step sputtering can realize greater a-Si:H passivation. Furthermore, better optical properties are obtained in ITO double layers compared with conventional ITO single layers. Therefore, modulating the radiofrequency power during ITO deposition can offer higher conversion efficiency.

Silicon heterojunction (SHJ) solar cells with hydrogenated amorphous silicon (a-Si:H) exhibit high power conversion efficiency.1–3 In an SHJ solar cell, undoped a-Si:H is used as a passivation layer to suppress the recombination of photogenerated carriers at the crystalline silicon (c-Si) surface, thereby allowing a high open-circuit voltage (VOC).4,5 The hydrogenation of dangling bonds (DBs) at a-Si:H/c-Si heterointerfaces is regarded as a dominant mechanism of the passivation effect of a-Si:H.6–11 In general, a transparent conductive oxide (TCO) layer is essential between the electrode and the doped a-Si:H layer because the conductivity of the a-Si:H layer is insufficient to deliver the photogenerated carriers to the external electrodes. A representative TCO material is indium tin oxide (ITO). Sputtering is widely used to deposit ITO, but consequent degradation of the passivation and cell performance is unavoidable (often called “sputtering damage”). Indeed, the performance of SHJ cells is degraded due to the generation of DBs during ITO deposition by sputtering.12–19 To avoid the sputtering damage, other methods for ITO deposition have been reported such as thermal evaporation and atomic layer deposition.20–22 However, another approach (e.g., post-plasma treatment) is necessary for realizing the great electrical and optical properties.21 Since sputtering is one of the simplest method to fabricate ITO films with great properties, a lower damage process that causes minimal sputtering damage is attractive for enhancing the conversion efficiency and further improving solar cells.

Sputtering damage is attributed to ion bombardment (IB) and light-induced degradation by the Staebler–Wronski effect in a-Si:H. Demaurex et al. studied sputtering damage by using glass and quartz shields to separate the damage into that of IB and vacuum ultraviolet (VUV) illumination generated by the plasma.12 Konishi and Ohdaira showed that sputtering damage depends strongly on deposition power.14 Furthermore, the DBs generated by sputtering can be re-terminated by high deposition temperature and post-annealing. Although annealing during and after deposition can help us to terminate the DBs and suppress the sputtering damage, the temperature in all processes of SHJ solar cells should be below 200 °C to prevent the a-Si:H layer from crystalizing, which is well known to degrade passivation performance, leading to low Voc.18,19 However, Huang et al. reported that better optical and electrical properties of ITO films could be obtained by using higher deposition power and temperature.23 Thus, it is crucial to establish a low-damage process as well as excellent optical and electrical properties under limited deposition power and temperature to improve the performance of SHJ solar cells.

In this paper, we study the effect of sputtering damage on passivation performance by varying the ITO layer thickness from 0 nm to 80 nm. The passivation performance decreased considerably up to 10 nm and increased gradually from 20 nm to 80 nm, which indicates that damage and recovery stages are present during sputtering. Based on these results for the passivation performance, we developed two-step deposition by changing the radiofrequency (RF) power from 50 W to 100 W during sputtering to minimize the influence of the initial damage and make effective use of the effect in the recovery stage. The electrical properties of the ITO double layers are identical to those of an ITO single layer, and then, the optical properties of the former are superior to those of the latter. Moreover, two-step deposition offers superior passivation performance compared with conventional one-step deposition.

Double-sided mirror-polished Czochralski (Cz)-grown n-type c-Si (100) wafers were used in all the experiments. The resistivity and wafer thickness were 3.0 Ω cm and ∼200 µm, respectively. These wafers were cleaned with Semicoclean-23 (Furuuchi Chemicals) and dipped in a 2.5% HF solution for 1 min to remove the native silicon oxide. The substrates were immersed in a 1.5% H2O2 solution for 30 s to form ultra-thin silicon oxide to prevent crystallization of the a-Si:H layer during annealing. Intrinsic a-Si:H (i-a-Si:H) and n-type a-Si:H (n-a-Si:H) were deposited on the rear side, while i- and p-type a-Si:H (p-a-Si:H) were deposited on the front side by plasma-enhanced chemical vapor deposition (PECVD) with a frequency of 27.12 MHz (CME-200J; ULVAC, Inc.). SiH4, H2, B2H6, and PH3 were used for a-Si:H deposition as the process gas. The deposition conditions of the a-Si:H layers are summarized in Table I.

TABLE I.

Deposition conditions of i-, n-, and p-a-Si:H prepared by PECVD.

DepositionLayerFlow rateFlow rateFlow rateFlow rate
temperaturethicknessof SiH4of H2of 2% PH3of 1% B2H6PressurePower density
Layer( °C)(nm)(SCCM)(SCCM)(SCCM)(SCCM)(Pa)(mW/cm2)
p-a-Si:H 260 40 15 25 32.5 
i-a-si:H (front) 260 40 25 32.5 
i-a-si:H (rear) 275 40 25 32.5 
n-a-Si:H 275 40 380 20 50 32.5 
DepositionLayerFlow rateFlow rateFlow rateFlow rate
temperaturethicknessof SiH4of H2of 2% PH3of 1% B2H6PressurePower density
Layer( °C)(nm)(SCCM)(SCCM)(SCCM)(SCCM)(Pa)(mW/cm2)
p-a-Si:H 260 40 15 25 32.5 
i-a-si:H (front) 260 40 25 32.5 
i-a-si:H (rear) 275 40 25 32.5 
n-a-Si:H 275 40 380 20 50 32.5 

After depositing the a-Si:H layers, an ITO layer was deposited by RF sputtering of an In2O3:SnO2 target (Sn 8 wt. %) with Ar sputtering gas, the purity of which was 99.999% (5 N). In all depositions, the Ar gas flow rate, the base pressure, the total process pressure, and the deposition temperature were 50 SCCM, 4.0 × 10−4 Pa, 0.2 Pa, and room temperature, respectively. To prepare the ITO single layer, ITO was deposited on the rear side at an RF power of 50 W [hereinafter referred to as low-power ITO (LP-ITO)]. The LP-ITO layer thickness dLP-ITO was varied from 10 nm to 80 nm. To prepare the ITO double layers, a second ITO layer was deposited on the initial LP-ITO layer at an RF power of 100 W [hereinafter referred to as high-power ITO (HP-ITO)]. The HP-ITO layer thickness dHP-ITO was varied from 40 nm to 70 nm, and the total layer thickness (i.e., dtotal = dLP-ITO + dHP-ITO) was 80 nm (supplementary material, Fig. S1). After ITO deposition, all samples were annealed in air at 200 °C for 15 min to cure the sputtering damage.

As shown in detail in Fig. 1, we fabricated four samples with different structures. For samples A and B, we investigated the lifetime change during ITO deposition until 80 nm thickness, and for samples C and D, we studied the effect of HP-ITO deposition on passivation performance and took samples between LP-ITO and HP-ITO depositions from the process chamber and measured the effective carrier lifetime τeff by quasi-steady-state photo conductance (QSSPC). To investigate the light-induced degradation in samples B and D, we covered them with glass substrates (EAGLE XG, Corning) during sputtering to eliminate the IB effect of the sputtering damage, but the glass substrates were sufficiently transparent to transmit the plasma illumination.

FIG. 1.

Schematic structures of samples using ITO single (samples A and B) or double layers (sample C and D). Glass substrates were mounted on the a-Si:H/c-Si heterojunctions during ITO deposition to eliminate ion bombardment (samples B and D).

FIG. 1.

Schematic structures of samples using ITO single (samples A and B) or double layers (sample C and D). Glass substrates were mounted on the a-Si:H/c-Si heterojunctions during ITO deposition to eliminate ion bombardment (samples B and D).

Close modal

We measured τeff by QSSPC (WCT-120 TS; Sinton Instruments) in generalized 1/1, 1/64, and transient modes.24,25 The layer thickness of the ITO films on glass substrates was determined by spectroscopic ellipsometry (SE; M-2000DI-Nug; J.A. Woollam Co.). In the SE analysis, we used the Cauchy, Tauc–Lorentz, Lorentz, Gauss, and Drude models for the dielectric functions of the samples based on Ref. 26. To characterize the electrical and optical properties of the ITO films on glass substrates, we used a Hall-effect measurement system (HMS-3000; ECOPIA), a spectrophotometer (V-570 UV/VIS/NIR; JASCO), and SE analysis. To investigate hydrogen depth profiles and hydrogen concentration (CH) in the samples, resonant 1H(15N,αγ)12C nuclear reaction analysis (NRA), which can detect H atoms in the near-surface region with nanometer scale depth resolution, was employed. The details of NRA are given elsewhere.27,28 For the Hall measurements, silver electrodes in the Van der Pauw configuration were deposited on the ITO films by vacuum evaporation, the layer thickness of all the ITO films was 80 ± 5 nm, and the sample size was 8 mm2.

Figure 2 shows the τeff of the LP-ITO deposited structure (a) without and (b) with glass covers as a function of the minority carrier density (MCD). The results for the samples with dLP-ITO = 0 (bare glass), 20 nm, and 80 nm are plotted in Fig. 2. The schematics of the lifetime measurements are illustrated as samples A and B in Fig. 1. τeff without glass covers was remarkably decreased compared to that with glass covers. These results mean that IB is the dominant degradation mechanism of sputtering damage, which is in good agreement with the previous work.12,29 Furthermore, we can discuss about the cause of carrier recombination by the figure of MCD vs τeff. It is known that the carrier recombination at lower and higher MCD regions is mainly caused by Shockley–Read–Hall (SRH) recombination and Auger recombination, respectively.30,31 At the lower MCD region in Fig. 2(a), the higher τeff is observed for the samples with 80-nm-thick ITO in comparison with the sample with 20-nm-thick ITO. This improved τeff at lower MCD is probably attributed to suppressed SRH recombination, suggesting that the number of DBs was decreased by ITO deposition.

FIG. 2.

(a) Injection-dependent effective carrier lifetime of the samples A and B after LP-ITO deposition (a) without and (b) with glass covers.

FIG. 2.

(a) Injection-dependent effective carrier lifetime of the samples A and B after LP-ITO deposition (a) without and (b) with glass covers.

Close modal

Figure 3 shows τeff at the MCD of 1.0 × 1015 cm−3 as a function of dLP-ITO. The average value of τeff at the MCD of 1.0 × 1015 cm−3 was around 1.77 ms before LP-ITO sputtering and was observed to decrease after LP-ITO deposition. Furthermore, τeff for the samples with and without glass covers shows a minimum at around 10 nm. After initial degradation, a slight increase in τeff with dLP-ITO can be seen. Therefore, the sputtering process seems to be separated into damage and recovery stages. A similar behavior of τeff up to 80 nm was observed in the previous work,29 and light soaking has been reported to increase the τeff of structures containing a-Si:H.4,32 Note that because all the depositions were carried out at room temperature, there is no influence of heat during sputtering.

FIG. 3.

Ratio of effective carrier lifetime τeff after LP-ITO deposition to that before LP-ITO deposition as a function of layer thickness of LP-ITO. Values of τeff at a minority carrier density of 1.0 × 1015 cm−3 are used. Triangles and circles represent the samples with and without glass covers, respectively.

FIG. 3.

Ratio of effective carrier lifetime τeff after LP-ITO deposition to that before LP-ITO deposition as a function of layer thickness of LP-ITO. Values of τeff at a minority carrier density of 1.0 × 1015 cm−3 are used. Triangles and circles represent the samples with and without glass covers, respectively.

Close modal

Figure 4 shows the transmittance spectra of LP-ITO layers of various thicknesses on glass substrates. The plasma light of pure Ar gas comprises mainly VUV with wavelengths of 100 nm, 160 nm, 220 nm, and 320 nm.33 The values of the transmittance at 220 nm and 320 nm as measured by SE are given in Table II. The transmittance at 220 nm ranged from 0.070 to 0.084, whereas that at 320 nm decreased gradually with the ITO thickness. Consequently, the photon flux density induced by the Ar plasma to the samples tends to decrease with the thickness of an ITO layer. From these results, the increase in τeff in the later sputtering stage may be caused by (i) changing the band diagram by ITO deposition and (ii) the re-termination of DBs by hydrogenation. However, the band diagram for an a-Si:H/c-Si heterointerface does not change much according to simulations using AFORS-HET (supplementary material, Fig. S2 and Table S1).34 These results mean that the increase in τeff in the recovery stage is not due to a changed band diagram and enhanced carrier selectivity during sputtering but rather due to re-termination of the DBs at the a-Si:H/c-Si heterointerface. Therefore, we consider that the degradation of the passivation performance was due to the de-hydrogenation of DBs and that the sputtering damage was partially cured by the re-hydrogenation of interfacial DBs.

FIG. 4.

Transmittance spectra of LP-ITO films on glass. The layer thickness of LP-ITO is 0 (bare glass), 10 nm, 20 nm, 40 nm, and 80 nm.

FIG. 4.

Transmittance spectra of LP-ITO films on glass. The layer thickness of LP-ITO is 0 (bare glass), 10 nm, 20 nm, 40 nm, and 80 nm.

Close modal
TABLE II.

Transmittance of LP-ITO with layer thickness dLP-ITO.

dLP-ITO (nm)0 (bare glass)10204080
T at 220 nm (%) 0.087 0.079 0.071 0.084 0.081 
T at 320 nm (%) 66.14 60.84 49.25 44.41 34.67 
dLP-ITO (nm)0 (bare glass)10204080
T at 220 nm (%) 0.087 0.079 0.071 0.084 0.081 
T at 320 nm (%) 66.14 60.84 49.25 44.41 34.67 

Note again that the passivation effect of a-Si:H is believed to be caused by the hydrogenation of DBs.6–11 In the damage stage, IB and plasma light induce de-hydrogenation, resulting in the generation of DBs in the a-Si:H layer and at the a-Si:H/c-Si heterointerface. After depositing 10-nm-thick ITO, the deposited ITO layer reduces the absorption of VUV in a-Si:H and protects the a-Si:H layer from IB. In the recovery stage, re-hydrogenation of DBs may be induced by the energy released from decelerating the incident ions and by the absorption of VUV. A similar consideration has been proposed previously.35 Furthermore, we also consider that the defect gradient from the interface to the surface causes H diffusion in the a-Si:H layer. It has been stated that hydrogen atoms can move easily from one site to another due to weakened Si–H and Si–Si bonds.35,36

NRA was used to analyze the hydrogen distributions in samples A and B before sputtering and after 10- and 80-nm-thick ITO deposition, as shown in Fig. 5, where the a-Si:H/c-Si heterointerface positions are indicated by vertical lines near 6.40 MeV of 15N ion energy. Furthermore, CH at the a-Si:H/c-Si interface and integral hydrogen concentration (IH) in the a-Si:H layer are summarized in Table III. CH and IH mean the hydrogen concentration at the specific energy and the total hydrogen content at the specific region, respectively. IH are given from the following equation:

IH=E1E2CHEdE,

where E1 and E2 are the energies corresponding to the a-Si:H surface and a-Si:H/c-Si interface. The slightly smaller depth of the interface in Figs. 5(b) and 5(c) is probably due to slight fluctuation of deposition rate during the fabrication of the a-Si:H layers and removal of ultrathin silicon oxide by HF treatment (supplementary material, Fig. S3). Regardless of whether or not glass covers were used, the total hydrogen content in the a-Si layer of all plasma-exposed samples is identical [Figs. 5(b)–5(f) and Table III). The very slight reduction of the H content in these samples compared to the condition before sputtering [Fig. 5(a)] may be attributed to a small loss of H due to VUV-induced desorption of H from the a-Si layer.37 More importantly, the NRA H profiles reveal that there are significant changes in the hydrogen concentration near the a-Si:H/c-Si heterointerfaces. The near-interfacial hydrogen concentration values range from 3.50 × 1021 cm−3 to 4.55 × 1021 cm−3, as given in Table III. In the IB-exposed samples, the near-interfacial H concentration is decreased by 23% compared to the as-deposited sample and increases again slightly after the deposition of the 80 nm LP-ITO layer. These changes in the hydrogen distribution in Fig. 5 and Table III are entirely consistent with the trend of the τeff changes during ITO sputtering (Fig. 3) and the above described mechanism of de-hydrogenation and re-hydrogenation of interfacial DBs.

FIG. 5.

Hydrogen depth profile of n-a-Si:H/i-a-Si:H/c-Si heterojunctions (a) before sputtering, (b) after 10- and (c) 80-nm-thick deposition in sample A, and (d) after 10- and (e) 80-nm-thick deposition in sample B. For the samples of (b) and (c), ITO was removed by dipping the samples in a 5% HF solution.

FIG. 5.

Hydrogen depth profile of n-a-Si:H/i-a-Si:H/c-Si heterojunctions (a) before sputtering, (b) after 10- and (c) 80-nm-thick deposition in sample A, and (d) after 10- and (e) 80-nm-thick deposition in sample B. For the samples of (b) and (c), ITO was removed by dipping the samples in a 5% HF solution.

Close modal
TABLE III.

Hydrogen concentration at the a-Si:H/c-Si heterointerface with LP-ITO thickness.

dLP-ITO (nm)010 (w/o glass)80 (w/o glass)10 (w/glass)80 (w/glass)
CH at the a-si:H/c-Si      
heterointerface (×1021 cm−3     
IH in the a-Si:H layer (×10197.6 8.18 7.87 7.59 7.2 
dLP-ITO (nm)010 (w/o glass)80 (w/o glass)10 (w/glass)80 (w/glass)
CH at the a-si:H/c-Si      
heterointerface (×1021 cm−3     
IH in the a-Si:H layer (×10197.6 8.18 7.87 7.59 7.2 

We understand these trends on the basis of the following scenario: Strong IB damage during initial ITO deposition (10–20 nm) creates de-passivated Si-DBs in the interfacial region that represents stable trap sites for H, which are not re-populated in this early stage, however, because the defect gradient drives diffusing H away from the interface. As the ITO thickness increases, the sputtered ITO ions can no longer penetrate down to the interface, but their mobilizing effect on H in the a-Si:H layer by cleaving Si–H bonds persists. This still generates diffusible H atoms in the a-Si:H layer that may become trapped at the de-hydrogenated Si-DBs in the interfacial region in later deposition stages. The resulting back-diffusion of H to the interface explains the re-passivation of the DBs in the recovery stage. Quantitatively, this slight return of interfacial H is a small effect compared to the much larger H loss during the initial ITO deposition step, very similar to the decrease/recovery trend of τeff in Fig. 3. Comparison with the glass-covered samples shows that the influence of VUV radiation on the H-distribution and τeff is much smaller than that of IB but qualitatively similar. From these insights into the damaging/recovery mechanisms, we hypothesize that the residual sputtering damage can be reduced by injecting more energy into the a-Si:H/c-Si heterostructures in the recovery stage.

To verify the aforementioned hypothesis, we performed two-step deposition by changing the RF power during sputtering. Figure 6 shows the relative τeff values of the ITO deposited samples before [Figs. 6(a) and 6(b)] and after annealing [Fig. 6(c)] normalized by τeff before ITO sputtering (τ0). The samples used for the lifetime measurements are illustrated as samples C and D in Fig. 1. Figures 6(a) and 6(b) show the τeff/τ0 values of the samples without and with glass covers, respectively. The τeff of all the samples using the ITO double layers is observed to increase after HP-ITO sputtering. Furthermore, τeff after HP-ITO sputtering is higher than that of the sample after 80-nm-thick LP-ITO deposition [the bar graph with black diagonal lines shown on the right of Fig. 6(a)]. In Fig. 6(b), a similar behavior was confirmed before and after HP-ITO sputtering. However, the τeff of the sample after 10-nm-thick LP-ITO sputtering was decreased after 70-nm-thick HP-ITO sputtering. We consider that this difference is due to the absorption of plasma light in the ITO layer during HP-ITO sputtering. According to the results shown in Fig. 4 and Table II, the transmittance of 10-nm-thick ITO on glass at 220 nm and 330 nm is almost the same as that for bare glass. Therefore, the passivation degradation can be explained by plasma VUV light irradiation with higher photon flux density during the damage stage. In the recovery stage, the plasma light reaching the a-Si:H/c-Si heterointerface was reduced by absorption in the ITO layer on glass covers thicker than 20 nm, whereupon re-termination of DBs can take place.

FIG. 6.

Ratios of effective carrier lifetime for samples C and D at a minority carrier density of 1.0 × 1015 cm−3 as a function of the thickness of LP-ITO and HP-ITO. (a) and (b) show τeff of samples after sputtering without and with glass covers, respectively, divided by τeff of samples before sputtering. (c) τeff of annealed samples normalized using τeff before sputtering.

FIG. 6.

Ratios of effective carrier lifetime for samples C and D at a minority carrier density of 1.0 × 1015 cm−3 as a function of the thickness of LP-ITO and HP-ITO. (a) and (b) show τeff of samples after sputtering without and with glass covers, respectively, divided by τeff of samples before sputtering. (c) τeff of annealed samples normalized using τeff before sputtering.

Close modal

Figure 6(c) shows the results of all the samples that were annealed at 200 °C for 15 min. For most samples with the ITO double layer, while τeff at the MCD of 1.0 × 1016 cm−3 was almost comparable, that at the MCD of 1.0 × 1015 cm−3 was higher than the samples with the ITO single layer after annealing regardless of whether glass covers were used or not. In particular, the τeff deposited 40-nm-thick LP-ITO and HP-ITO were the highest in all samples. We attribute these results to the change in the initial stage before annealing at the a-Si:H/c-Si heterointerface in the recovery stage. The increased energy of accelerated ions and plasma light irradiation during HP-ITO sputtering enhances the slight diffusion of hydrogen atoms toward the a-Si:H/c-Si heterointerface and, thus, aids the re-termination of the DBs created in the damage stage. The resulting increase in H atoms near the a-Si:H/c-Si heterointerface then further enhances the recovery in the annealing process. These results show that the improvement of τeff in the recovery stage is likely to be enhanced by injecting higher energy generated by two-step sputtering with changing RF power, thereby offering a low-damage process and high cell performance.

Figure 7 shows (a) the electron density Ne, (b) the Hall mobility μ, and (c) the resistivity of ITO single and double layers on glass substrates before and after annealing. For the as-deposited ITO double layers, lower μ and higher Ne were observed than LP-ITO single layers both before and after annealing. Then, a decrease in Ne and an increase in μ are observed by annealing. These are probably attributable to the effect of oxygen vacancies by HP-ITO deposition and annealing in air. It is known that oxygen vacancies act as donors in ITO and, thus, increase Ne and decrease μ.38,39 Furthermore, the oxygen vacancies in ITO films are annihilated during annealing in air, and then, Ne decreases and μ increases.40,41 Hence, in HP-ITO deposition, accelerated negatively charged particles (e.g., electrons, oxygen ions, and secondary electrons) with higher energy could generate many oxygen vacancies by collisions at the growing ITO surface. Meanwhile, the increase in μ after annealing is considered to be due to the decreased oxygen vacancies and the crystallization of ITO. In terms of the resistivity, ITO double layers with 40-nm-thick LP-ITO and HP-ITO have the best conductivity after annealing.

FIG. 7.

(a) Electron density, (b) Hall mobility, and (c) resistivity of ITO single and double layers on glass substrates as a function of the thickness of the ITO layer. The unfilled and filled circles represent the results of the samples using ITO single and double layers, respectively.

FIG. 7.

(a) Electron density, (b) Hall mobility, and (c) resistivity of ITO single and double layers on glass substrates as a function of the thickness of the ITO layer. The unfilled and filled circles represent the results of the samples using ITO single and double layers, respectively.

Close modal

Figure 8 shows the transmittance and reflectance spectra of the ITO single and double layers (a) before and (b) after annealing. The reflectance spectra are observed to decrease from 350 nm to 700 nm by changing the single LP-ITO layer into HP-ITO/LP-ITO double layers; this is caused by the differences in the refractive index n and extinction coefficient k for LP-ITO and HP-ITO. Figure 9 shows the n and k of single LP-ITO and HP-ITO layers on glass substrates (a) before and (b) after annealing. The values of n and k at 400 nm and their mean square errors (MSEs) are summarized in Table IV. Because all the MSE values are small enough, the experimental data are in good agreement with the model fit curves, confirming the validity of the optical constants and the model. The value of n was higher for LP-ITO than for HP-ITO (Fig. 9), both before and after annealing. Because of the difference in n between HP-ITO and LP-ITO at the interface, the increase in transmittance and the decrease in reflectance observed for the ITO double layers arose from a better anti-reflection effect. Moreover, the single HP-ITO layer after annealing exhibited high transparency in the near-infrared region, possibly because of decreased optical loss due to free-carrier absorption. However, in terms of the TCO layer for SHJ solar cells, the transparency for the wavelength absorbed by c-Si has much importance than the near-infrared region. From the above, ITO double layers are superior to the ITO single layer also in terms of optical properties.

FIG. 8.

Optical properties of ITO single and double layers on glass substrates (a) before and (b) after annealing. Solid and dotted lines represent the transmittance and reflectance, respectively, of a sample. The layer thickness of each film is 80 ± 5 nm.

FIG. 8.

Optical properties of ITO single and double layers on glass substrates (a) before and (b) after annealing. Solid and dotted lines represent the transmittance and reflectance, respectively, of a sample. The layer thickness of each film is 80 ± 5 nm.

Close modal
FIG. 9.

Refractive index n and extinction coefficient k for LP-ITO and HP-ITO (a) before and (b) after annealing.

FIG. 9.

Refractive index n and extinction coefficient k for LP-ITO and HP-ITO (a) before and (b) after annealing.

Close modal
TABLE IV.

Refractive index n and extinction coefficient k of LP-ITO and HP-ITO at a wavelength of 400 nm and MSEs before and after annealing.

LP-ITOHP-ITO
 n at 400 nm 2.148 2.145 
Before annealing k at 400 nm 0.028 0.040 
 MSE 2.703 6.215 
 n at 400 nm 2.207 2.187 
After annealing k at 400 nm 0.028 0.013 
 MSE 6.175 5.478 
LP-ITOHP-ITO
 n at 400 nm 2.148 2.145 
Before annealing k at 400 nm 0.028 0.040 
 MSE 2.703 6.215 
 n at 400 nm 2.207 2.187 
After annealing k at 400 nm 0.028 0.013 
 MSE 6.175 5.478 

The results show that the better electrical properties were obtained by changing ITO single into double layers. Furthermore, in terms of the transparency in the wavelengths absorbed by c-Si, better optical properties were also obtained with ITO double layers compared with LP-ITO and HP-ITO single layers due to the difference in refractive index n in ITO double layers. Regarding the passivation effect, ITO double layers are superior to an ITO single layer because of the lower sputtering damage with the ITO double layers. Furthermore, in this work, we did not introduce O2 gas during ITO deposition at all. Hence, optimization of the process gas flow may improve some parameters of ITO double layers (e.g., transparency and the Hall mobility) to realize the further great cell performance. From the above, ITO double layers deposited with changing RF power can offer improved performance of SHJ solar cells.

We investigated the effect of ITO sputtering and two-step deposition on the passivation performance of the a-Si:H/c-Si heterojunctions. The value of τeff decreased with increasing thickness up to dLP-ITO = 10 nm and then increased as the ITO sputtering process proceeded, suggesting that damage and recovery stages were present in the ITO sputtering process. In the damage stage, IB dominantly created defects in a-Si:H layers and a-Si:H/c-Si heterointerfaces. In the recovery stage, the energy of IB and VUV presumably contributed to the increase in τeff because of re-hydrogenation of DBs using the energy generated from braking high-energy ions on the deposited ITO layer and absorption of VUV in ITO.

The ITO double layers gave relatively high τeff in comparison with an ITO single layer, presumably because of the injection of a larger amount of energy to enhance re-hydrogenation of DBs during HP-ITO deposition. In terms of the electrical properties, the resistivity of ITO double layers after annealing was approximately 4.0 × 10−4 Ω cm. The transparency of the ITO double layers in the wavelengths absorbed by c-Si was superior to that of the ITO single layer. From the above, ITO double layers with changing RF power can further improve the SHJ solar cell performance. Furthermore, the two-step deposition with modulating RF power may be useful for the materials containing weakly bonded hydrogen.

See the supplementary material for the simulating condition and results, the schematic procedures of the double layer, and sample preparation for NRA measurements.

The data that support the findings of this study are available from the corresponding author upon reasonable request.

This work was supported by Industrial Technology Development Organization (NEDO) and “Hydrogenomics” (Grant Nos. JP18H05514 and JP18H05518) from the Japan Society for the Promotion of Science (JSPS) in Japan. We would like to thank H. Miura and A. Shimizu for technical support.

1.
S.
De Wolf
,
A.
Descoeudres
,
Z. C.
Holman
, and
C.
Ballif
,
Green
18
,
7
(
2012
).
2.
M. A.
Green
,
Y.
Hishikawa
,
W.
Warta
,
E. D.
Dunlop
,
D. H.
Levi
,
J.
Hohl-Ebinger
, and
A. W. H.
Ho-Baillie
,
Prog. Photovoltaics
25
,
668
(
2017
).
3.
L.
Korte
,
E.
Conrad
,
H.
Angermann
,
R.
Stangl
, and
M.
Schmidt
,
Sol. Energy Mater. Sol. Cells
93
,
905
(
2009
).
4.
E.
Kobayashi
,
S.
De Wolf
,
J.
Levrat
,
A.
Descoeudres
,
M.
Despeisse
,
F.-J.
Haug
, and
C.
Ballif
,
Sol. Energy Mater. Sol. Cells
173
,
43
(
2017
).
5.
M.
Tanaka
,
M.
Taguchi
,
T.
Matsuyama
,
T.
Sawada
,
S.
Tsuda
,
S.
Nakano
,
H.
Hanafusa
, and
Y.
Kuwano
,
Jpn. J. Appl. Phys., Part 1
31
,
3518
(
1992
).
6.
S.
De Wolf
and
M.
Kondo
,
J. Appl. Phys.
105
,
103707
(
2009
).
7.
A.
Descoeudres
,
L.
Barraud
,
S.
De Wolf
,
B.
Strahm
,
D.
Lachenal
,
C.
Guérin
,
Z. C.
Holman
,
F.
Zicarelli
,
B.
Demaurex
,
J.
Seif
,
J.
Holovsky
, and
C.
Ballif
,
Appl. Phys. Lett.
99
,
123506
(
2011
).
8.
T. F.
Schulze
,
H. N.
Beushausen
,
C.
Leendertz
,
A.
Dobrich
,
B.
Rech
, and
L.
Korte
,
Appl. Phys. Lett.
96
,
252102
(
2010
).
9.
S.
De Wolf
,
S.
Olibet
, and
C.
Ballif
, “
Stretched-exponential a-Si:H/c-Si interface recombination decay
,”
Appl. Phys. Lett.
93
,
032101
(
2008
).
10.
S.
De Wolf
,
C.
Ballif
, and
M.
Kondo
, “
Kinetics of a-Si:H bulk defect and a-Si:H/c-Si interface-state reduction
,”
Phys. Rev. B
85
,
113302
(
2012
).
11.
K.
Gotoh
,
M.
Wilde
,
S.
Kato
,
S.
Ogura
,
Y.
Kurokawa
,
K.
Fukutani
, and
N.
Usami
,
AIP Adv.
9
,
075115
(
2019
).
12.
B.
Demaurex
,
S.
De Wolf
,
A.
Descoeudres
,
Z. C.
Holman
, and
C.
Ballif
,
Appl. Phys. Lett.
101
,
171604
(
2012
).
13.
S.
De Wolf
,
C.
Demaurex
,
A.
Descoeudres
, and
C.
Ballif
,
Phys. Rev. B
83
,
233301
(
2011
).
14.
T.
Konishi
and
K.
Ohdaira
, “
Indium tin oxide sputtering damage to catalytic chemical vapor deposited amorphous silicon passivation films and its recovery
,”
Thin Solid Films
635
,
73
(
2017
).
15.
D. E.
Carlson
,
Appl. Phys. A: Solids Surf.
41
,
305
(
1986
).
16.
D.
Zhang
,
A.
Tavakoliyaraki
,
Y.
Wu
,
R. A. C. M. M.
van Swaaij
, and
M.
Zeman
,
Energy Procedia
8
,
207
(
2011
).
17.
T.
Kamioka
,
Y.
Hayashi
,
Y.
Isogai
,
K.
Nakamura
, and
Y.
Ohshita
,
AIP Adv.
7
,
095212
(
2017
).
18.
S.
De Wolf
and
M.
Kondo
,
Appl. Phys. Lett.
91
,
112109
(
2007
).
19.
H.
Fujiwara
and
M.
Kondo
,
Appl. Phys. Lett.
90
,
013503
(
2007
).
20.
G. S.
Belo
,
B. J. P.
da Silva
,
E. A.
de Vasconcelos
,
W. M.
de Azevedo
, and
E. F.
da Silva
,
Appl. Surf. Sci.
255
,
755
(
2008
).
21.
F.
Wang
,
R.
Du
,
Q.
Ren
,
C.
Wei
,
Y.
Zhao
, and
X.
Zhang
,
Appl. Surf. Sci.
426
,
856
(
2017
).
22.
Y.
Zhou
,
H.
Cheun
,
W. J.
Potscavage
, Jr.
,
C.
Fuentes-Hernandez
,
S.-J.
Kim
, and
B.
Kippelen
,
J. Mater. Chem.
20
,
6189
(
2010
).
23.
M.
Huang
,
Z.
Hameiri
,
S.
Venkataraj
,
A. G.
Aberle
, and
T.
Mueller
,
Energy Procedia
33
,
91
(
2013
).
25.
R.
Sinton
and
A.
Cuevas
,
Appl. Phys. Lett.
69
,
2510
(
1996
).
26.
T.
Koida
,
M.
Kondo
,
K.
Tsutsumi
,
A.
Sakaguchi
,
M.
Suzuki
, and
H.
Fujiwara
,
J. Appl. Phys.
107
,
033514
(
2010
).
27.
K.
Fukutani
,
Curr. Opin. Solid State Mater. Sci.
6
,
153
(
2002
).
28.
M.
Wilde
and
K.
Fukutani
,
Surf. Sci. Rep.
69
,
196
(
2014
).
29.
A. H. T.
Le
,
V. A.
Dao
,
D. P.
Pham
,
S.
Kim
,
S.
Dutta
,
C. P. T.
Nguyen
,
Y.
Lee
,
Y.
Kim
, and
J.
Yi
,
Sol. Energy Mater. Sol. Cells
192
,
36
(
2019
).
30.
C.
Leendertz
,
R.
Stangl
,
T. F.
Schulze
,
M.
Schmidt
, and
L.
Korte
,
Phys. Status Solidi C
7
,
1005
(
2010
).
31.
M. J.
Kerr
and
A.
Cuevas
,
J. Appl. Phys.
91
,
2473
(
2002
).
32.
E.
Kobayashi
,
S.
De Wolf
,
J.
Levrat
,
G.
Christmann
,
A.
Descoeudres
,
S.
Nicolay
,
M.
Despeisse
,
Y.
Watabe
, and
C.
Ballif
,
Appl. Phys. Lett.
109
,
153503
(
2016
).
33.
Y.
Kuang
,
B.
Macco
,
B.
Karasulu
,
C. K.
Ande
,
P. C. P.
Bronsveld
,
M. A.
Verheijen
,
Y.
Wu
,
W. M. M.
Kessels
, and
R. E. I.
Schropp
,
Sol. Energy Mater. Sol. Cells
163
,
43
(
2017
).
34.
R.
Varache
,
C.
Leendertz
,
M. E.
Gueunier-Farret
,
J.
Haschke
,
D.
Muñoz
, and
L.
Korte
,
Sol. Energy Mater. Sol. Cells
141
,
14
(
2015
).
35.
P.
Mahtani
,
R.
Varache
,
B.
Jovet
,
C.
Longeaud
,
J.
Kleider
, and
N. P.
Kherani
,
J. Appl. Phys.
114
,
124503
(
2013
).
36.
M.
Stutzmann
,
W. B.
Jackson
, and
C. C.
Tsai
,
Phys. Rev. B
32
,
23
(
1985
).
37.
B. W.
Clare
,
J. C. L.
Cornish
,
G. T.
Hefter
,
P. J.
Jennings
,
C. P.
Lund
,
D. J.
Santjojo
, and
M. O. G.
Talukder
,
Thin Solid Films
288
,
76
(
1996
).
38.
W.-F.
Wu
and
B.-S.
Chiou
,
Semicond. Sci. Technol.
11
,
196
(
1996
).
39.
R. B. H.
Tahar
,
T.
Ban
,
Y.
Ohya
, and
Y.
Takahashi
,
J. Appl. Phys.
83
,
2631
(
1998
).
40.
S.
Li
,
Z.
Shi
,
Z.
Tang
, and
X.
Li
,
Vacuum
145
,
262
(
2017
).
41.
S.
Ray
,
R.
Banerjee
,
N.
Basu
,
A. K.
Batabyal
, and
A. K.
Barua
,
J. Appl. Phys.
54
,
3497
(
1983
).

Supplementary Material