Substrate effect on ultra-thin hydrogenated amorphous silicon solar cells

Solar Energy Materials and Solar Cells 171 (2017) 222–227

Contents lists available at ScienceDirect

Solar Energy Materials and Solar Cells journal homepage: www.elsevier.com/locate/solmat

Substrate effect on ultra-thin hydrogenated amorphous silicon solar cells a,b,c,d

a,b,c,d,⁎

a,b,c,d

a,b,c,d

MARK

a,b,c,d

Jia Fang , Baojie Yan , Tiantian Li , Changchun Wei , Dekun Zhang , Baozhang Lia,b,c,d, Qian Huanga,b,c,d, Xinliang Chena,b,c,d, Guofu Houa,b,c,d, Guangcai Wanga,b,c,d, ⁎ Ying Zhaoa,b,c,d, Xiaodan Zhanga,b,c,d, a

Institute of Photoelectronic Thin Film Devices and Technology, Nankai University, Tianjin 300071, PR China Key Laboratory of Photoelectronic Thin Film Devices and Technology of Tianjin, Tianjin 300071, PR China c Key Laboratory of Photoelectronic Thin Film Devices and Technology of Ministry of Education, Tianjin 300071, PR China d Collaborative Innovation Center of Chemical Science and Engineering (Tianjin), Tianjin 300072, PR China b

A R T I C L E I N F O

A B S T R A C T

Keywords: Ultra-thin solar cell Amorphous silicon Light trapping Scattering Transparent conductive oxide

We report a comprehensive study of ultra-thin a-Si:H solar cells on various transparent conductive oxide (TCO) coated glass substrates. We find that the highly textured boron doped zinc oxide (BZO) with multi-scale features made with metal organic chemical vapor deposition (MOCVD) outperforms the flat aluminum doped zinc oxide (AZO) deposited by sputtering and the highly textured AZO with only large features after the chemical etching with diluted HCl. We believe that the much larger features than the cell thickness provide only an enhancement of the effective area for increasing the light absorption, while the variable sizes of features similar to and smaller than cell thickness provide- a highly effective light trapping for thin film solar cells. In addition, the ultra-thin aSi:H solar cell on the BZO has a better p/i interface than the cell on the AZO substrates, which leads to a high cell performance. A high efficiency of 8.15% was obtained using an a-Si:H p-i-n structure on the BZO with the intrinsic layer of 70 nm.

1. Introduction With the increase of mass production and technology improvement, crystal silicon (c-Si) solar module efficiency has been increased [1–4] and their manufacturing cost has been steadily decreased in the last decade, which has accelerated the pace towards the grid parity on one hand, but pushed the other photovoltaic (PV) technologies out of the main stream, especially in the utility scale on the other hand. Hydrogenated amorphous silicon (a-Si:H) and microcrystalline silicon (μcSi:H) based film silicon PV technology is one of the technologies lost its market share significantly because of the low efficiency and metastability caused by the Staebler-Wronski effect [5]. However, thin film silicon solar cell has several unique properties over c-Si solar cells, which makes it suitable for some potential applications in the niche market. For example, a-Si:H based solar cells can be deposited on flexible substrates for flexible PV product [6,7]; on glass with back side transparent electrode for semitransparent solar panels [8–10]. Thin film silicon solar cells are normally deposited on transparent conductive oxide (TCO) coated glass substrates with a p-i-n structure, where the p, i, and n represent the p-layer, intrinsic layer, and n-layer, respectively, and with a back contact made of another TCO layer and a metal contact layer such as Ag and Al. The back contact could also use a highly ⁎

conductive TCO without the metal layer to make semitransparent solar cells for various applications such as building integrated PV (BIPV) with color windows for decoration and electricity generation at the same time and green house for agriculture with electricity generation as well. In order to have enough light transmission, reducing the a-Si:H absorber layer thickness is needed for the semi-transparent solar cells. In addition, people have tried various techniques to improve the cell efficiency and the stability further to keep thin film silicon solar PV industry alive [11,12]. One of the approaches is to design three dimensional (3D) solar cells on various micro- and nano-structures to make them optically thick to absorb more light and electrically thin to allow a high electric field to collect the photo-generated carriers [13–15]. For this reason, Ultra-thin a-Si:H solar cells (< 100 nm) have attracted a great deal of attentions. It has been well known that light trapping is one of the important techniques for improving thin film solar cell efficiency. The surface texture on the TCO substrate is the dominant factor for p-i-n structured thin film silicon solar cells. In the early days, Fluorine doped Tin dioxide (FTO) is widely used as the TCO for a-Si:H based p-i-n solar cells. With the invention of µc-Si:H as the absorber layer [16–18], Al doped Zinc Oxide (AZO) and Boron doped Zinc Oxide (BZO) have been widely used because they are stable under atomic hydrogen condition but FTO

Corresponding authors at: Institute of Photoelectronic Thin Film Devices and Technology, Nankai University, Tianjin 300071, PR China. E-mail addresses: [email protected] (B. Yan), [email protected] (X. Zhang).

http://dx.doi.org/10.1016/j.solmat.2017.06.065 Received 10 May 2017; Received in revised form 27 June 2017; Accepted 29 June 2017 0927-0248/ © 2017 Published by Elsevier B.V.

Solar Energy Materials and Solar Cells 171 (2017) 222–227

J. Fang et al.

reacts with the high density atomic hydrogen used in µc-Si:H solar cell deposition. AZO is normally deposited using a magnetron sputtering machine and it is relatively flat in the as-deposited state. A chemical etching process using diluted HCl was proposed by the Julisch group and proven to be an effectively method to increase the texture of AZO [19–21]. The chemically etched and highly textured AZO has been used in µc-Si:H deposition and results in high efficiency µc-Si:H singlejunction and a-Si:H/µc-Si:H double-junction solar cells [22,23]. While BZO has been deposited using Metal Organic Chemical Vapor Deposition (MOCVD), on which the textures are generated and controlled during the deposition process [24,25]. The highly textured BZO has also been used in µc-Si:H and results in high efficiency µc-Si:H single-junction and a-Si:H/µc-Si:H double-junction solar cells as well [26,27]. The effectiveness of light trapping depends not only on the textures of the TCO, but also on the solar cell thickness. The very large pyramidal feature has been proven to be a good structure for light trapping in c-Si solar cells [28], but may not be a good choice for thin film silicon solar cells. Because of the significantly reduction of the absorber layer thickness in ultra-thin a-Si:H solar cells, the normally optimized AZO and BZO for µc-Si:H solar cells might not be the good substrates for ultra-thin a-Si:H solar cells. The objective of this paper is to study the structural and optical properties of commonly used AZO and BZO and their correlations to ultra-thin a-Si:H solar cell performance and to search the best TCO substrate for ultra-thin a-Si:H solar cells.

Fig. 1. AFM surface morphology images of various TCO substrates and their RMS roughness values.

Number of Events (a.u.)

2. Experimental procedures AZO layers were deposited using a magnetron sputtering system (KJLC Lab-18, base pressure below 1 × 10−5 Pa) on cleaned glass substrate at 325 °C. In order to increase the surface textures, a chemical etching process was made in diluted HCl (5.0%) solution for 45 s. BZO layers were deposited using a MOCVD method on cleaned glass substrates at 150 °C. As a comparison, commercial FTO was used as the baseline. Ultra-thin a-Si:H solar cells with a 70-nm thick intrinsic layer was deposited on the selected substrates using a multi-chamber plasma enhanced chemical vapor deposition system. The substrate temperature was 210 °C for all of the silicon layers. The optimization of the cell performance was available elsewhere [29,30]. The surface structures of the TCO were measured using an atomic force microscopy (AFM) (SPA 400 AFM) with the surface roughness characterized by the root-mean-square (RMS) and the surface height distribution. The conformality of the ultra-thin a-Si:H solar cells on different TCO substrates were measured by using a cross-sectional transmission electron microscopy (X-TEM) (FEI Novanano lab 200). The optical properties of the TCO were measured using an optical spectrometer (Cary 5000) and analyzed by total transmittance and haze spectra. The solar cell performance was characterized by the current density versus voltage (J-V) characteristics measured under an AM1.5 solar simulator (WXS-156S-L2, AM1.45GMM by Wacom Co) with 100 mW/ cm2 of light intensity at 25 °C, and by the external quantum efficiency (EQE) measured with an EQE system (PV measurement QEX10). The EQE curves were normally measured at the short circuit condition (0 V bias), and here also under − 1.0 V and + 0.6 V for the study of carrier recombination losses.

MOCVD-ZnO:B AZO AZO+HCl FTO

1.01

2500

S Ratio 1.31

2000 1.15

1.07

1500 1000 500 0

0

100

200

300

400

500

600

700

800

Height (nm)

Transmittance, HAZE (%)

100

80

60 MOCVD-ZnO:B AZO AZO+HCl FTO

40

20

0 300

400

500

600

Wavelength (nm)

700

800

Fig. 2. Upper plot: the height distributions of the four TCO substrates, the lower plot: the transmittance and haze spectra of the four TCO substrates as shown in Fig. 1.

3. Results and discussion

structures after the chemical etching. The FTO has a medium texture with RMS = 41 nm. The upper panel in Fig. 2 plots the height distributions of the four TCO samples along with the S-ratio defined by the total surface area over the base flat area. With the increase of the surface texture, the distribution becomes broad and the peak shifts to a large height number. The etched AZO has the peak position at ∼ 500 nm and the distribution extends to 850 nm, which means the biggest distance from peak to valley could reach 850 nm. While the curve

Fig. 1 shows the surface morphologies of the four types of TCO films, where (a) is the MOCVD deposited BZO, (b) the sputtering deposited AZO before the chemical etching, and (c) the sputtering deposited AZO after the chemical etching, and (d) the reference FTO. The images show that the BZO has a rough surface of 83 nm RMS with various sized mountain-like structures. The AZO before the chemical etching is relatively flat with an RMS only 5.9 nm, while its surface becomes very rough (RMS = 135.4 nm) with large crater-like 223

Solar Energy Materials and Solar Cells 171 (2017) 222–227

J. Fang et al.

of the BZO shifts to the low height values with the peak position at 200 nm and the high value side tail extends to 500 nm. The distribution is very asymmetric with a long tail in the high value side and a sharp drop at the low value side, which is caused by various sizes of the micro-features. Additional structure information comes from the Sratio. The BZO has a larger S-ratio (1.31) than the etched AZO (1.07), which means that the BZO has steeper features than the etched AZO. From the microscopic structure analyses, we may conclude that the BZO has medium sized features with a wide range feature size distribution and steep feature structure; while the etched AZO has large size features with both large vertical and lateral dimensions. The lower panel in Fig. 2 shows the optical properties of the four TCO samples, characterized by the total transmittance and haze spectra. From the transmission point of view, the etched AZO is better than the BZO and the FTO in the wavelength > 400 nm, but not as good as the FTO in the wavelength < 390 nm. The chemical etching smoothed the transmission spectrum of AZO by removing the interference fringes. The short wavelength cut-on of the AZO is at ∼ 350 nm, while it is at ∼ 300 nm for FTO. The worst is the BZO with lowest total transmittance and longest short wavelength cut-on at ∼ 375 nm. The haze spectra follow to the surface texture as characterized by the RMS with the higher haze values corresponding to the higher surface roughness. Fig. 3 shows the schematic structure and the X-TEM images of four ultra-thin a-Si:H solar cells deposited on the different substrates aforementioned. One may note that the texture feature sizes for the solar cell on the BZO and on the FTO are in the similar order to the aSi:H thickness or at least not too much larger than the a-Si:H thickness, and the features are with irregular shapes; however, the feature sizes of the cell on the etched AZO is much larger than the a-Si:H thickness and reaches to 1–2 µm in the lateral direction, while the total a-Si:H thickness is ∼ 100 nm and consists of ∼ 10 nm of p layer, ∼ 70 nm of i layer and ∼ 20 of n layer. In addition, the a-Si:H layer is very conform on the AZO layer no matter with and without the chemical etching; but

2

Current Density (mA/cm )

4 2 0 BZO AZO AZO+HCl FTO

-2 -4 -6 -8 -10 -12 -14

0.0

0.2

0.4

0.6

Voltage (V)

0.8

1.0

80

EQE (%)

60

40 BZO(0V) AZO(0V) AZO+HCl(0V) FTO

20

0 300

400

500

600

700

800

Wavelength (nm) Fig. 4. (Upper panel) The J-V characteristics and (lower panel) EQE curves of the ultrathin a-Si:H solar cells on various TCO/glass substrates.

the a-Si:H layer on the BZO and FTO has some nonconformities. The differences in the feature sizes and the conformities could lead to a significant difference in terms of light trapping in the solar cells. Fig. 4 shows the performances of the four ultra-thin a-Si:H solar cells on the aforementioned TCOs, where the upper panel is the J-V characteristics under the AM1.5 solar simulator and the lower one the EQE curves, the performance parameters are summarized in Table 1. A few very interesting observations could be made as following. First, the solar cell on the FTO has a very poor solar cell performance with the lowest FF and Voc. As mentioned above, FTO could react with atomic hydrogen. Because we used a nc-SiOx:H p layer, which was deposited under a very high hydrogen dilution ratio with a high RF power for promoting the formation of nc-Si crystallites embedded in the a-SiOx:H tissues, the high hydrogen dilution and high RF power produced a high density of atomic hydrogen with high energy, which reacted with the FTO and reduced the properties of the FTO and degraded the a-Si:H solar cell performance. While the ultra-thin a-Si:H solar cells on the BZO and AZO substrates show reasonably good performances, confirming that the BZO and AZO substrates are stable under the high

Table 1 Performance parameters of ultra-thin a-Si:H solar cells on different substrates.

Fig. 3. (left) Schematics of the ultra-thin a-Si:H solar cells and (right) X-TEM images of ultra-thin a-Si:H solar cells on various TCO/glass substrates, where the i-layer of the solar cell is 70 nm.

224

TCO

Voc (mV)

Jsc (mA/cm2)

FF (%)

Eff. (%)

Rs (Ω cm2)

FTO Flat AZO Etched AZO BZO

869 930 956 915

12.11 8.56 11.09 12.82

44.96 63.74 66.59 69.50

4.73 5.07 7.06 8.15

10.05 10.00 8.80 6.00

Solar Energy Materials and Solar Cells 171 (2017) 222–227

J. Fang et al.

total transmission measurement was made because the back side medium was air with the diffractive index of 1, smaller than in the BZO. In this case, some of the lights were total reflected back to the film and trapped inside there. While when the silicon layers were deposited on the BZO, the light cannot be trapped in the BZO and all of them get into the silicon layer to produce photocurrent. Therefore, the light trapping effect of the BZO is higher than the etched AZO even the measured total transmittance and haze are lower. Comparing the EQE spectra of the four solar cells, one may note that the response difference between the BZO and other substrates occurs not only at the long wavelength region but also in the short wavelength region. The long wavelength difference reflects the effectiveness of light trapping from the textured substrates, while the short wavelength difference needs to be further investigated. As discussed above, the EQE difference in the wavelength shorter than 375 nm is mainly caused by the different bandgaps of the TCO materials, which limits the short wavelength cut-on in the EQE spectra. While one can see that the ultrathin a-Si:H solar cell on the BZO has much higher response in the wavelength range between 375 nm and 575 nm than the cells on the other substrates, which cannot be explained by the bandgap cut-on differences and by the enhanced light trapping. The EQE differences among the four solar cells could also be caused by other mechanisms. One is the total reflectance R(λ), which gives the total absorbance in the solar cell by A(λ) = 1 − R(λ). Fig. 5 plots the total absorption spectra of the four solar cells. First, the spectrum of the cell on the AZO without etching shows large interference fringes caused by the flat surface and is consistent with the EQE spectra. Second the long wavelength spectra show large differences, where the solar cell on the etched AZO spectrum is much lower than the cells on the BZO and the FTO, indicating an inefficient light trapping as being consistent with the EQE spectra. The absorbance of the BZO and FTO are similar, but their EQE spectra are difference as shown in Fig. 4, which means that the EQE difference between these two solar cells is not caused by the absorption, but other mechanisms such as recombination. The three solar cells on the etched AZO, BZO and FTO have very similar short wavelength absorbance, indicating the lights coupled into the solar cells are similar in the short wavelength region. Therefore, the EQE differences in the short wavelength are caused by other mechanisms such as recombination as well. To investigate the recombination losses, especially origins of the EQE differences in the short wavelength range, the biased EQE were measured under − 1.0 V and + 0.6 V. The biased EQE has been widely used to study the photocarrier losses in a-Si:H and nc-Si:H solar cells [33]. A reverse electric bias enhances the electric field and reduces the possibility of recombination by swiping out the photocarriers; while a forward electric bias has the opposite effect to reduce the electric field

density atomic hydrogen condition. Second, comparing the solar cells on the AZO substrates with and without chemical etching, it notes that the etched AZO substrate has improved all three factors of Jsc, Voc and FF of the ultra-thin solar cell; hence the efficiency is improved from 5.07% to 7.06%. The improvement of the Jsc is easy to be understood because the light trapping has been improved significantly by the increase of AZO texture. While the increase of Voc and FF is somewhat puzzling because normally the increase of texture would lead to additional micro-shunt paths and reduce Voc and FF [31,32]. However, as shown in Fig. 3, the feature sizes of the textures on the chemically etched AZO are much larger than the film thickness and with no sharp peaks and valleys, the micro-shunt issue is probably very minimum. But on the other hand, the chemical etching might improve the contact between the AZO and p-nc-SiOx:H by removing a high resistive layer at the as deposited AZO. The most important point is that the chemical etching AZO improves the ultra-thin a-Si:H solar cell efficiency significantly. Third, the ultra-thin a-Si:H solar cell on the BZO gives the highest Jsc, FF, and efficiency. The 8.15% efficiency is very respectable for the a-Si:H solar cell with only 70-nm thick intrinsic layer. However, the Voc is lower than the solar cells on the AZO substrates. Two factors could contribute to the lower Voc. First，the smaller features of the textures lead to a high density of peaks and valleys, which might result in some micro shunts and reduces the Voc. However, this assignment is inconsistent with the better FF of the cell on the BZO than on the AZO. Second，the work function difference between the BZO and p-ncSiOx:H layer could be larger than the value between AZO and the p-ncSiOx:H layer. The difference of work functions between the TCO and the p-nc-SiOx:H layer forms an opposite junction respecting to the p-i-n junction and reduces the Voc. The most interesting observation is the higher Jsc from the ultra-thin a-Si:H cell on the BZO than on the chemically etched AZO. As shown in Fig. 2, both of the total transmittance and the haze of the etched AZO are much higher than the BZO in the whole solar spectrum, from which one could expect a higher Jsc from the cell on the etched AZO than on the BZO substrates. However, the experiment showed an opposite result with a higher Jsc from the cell on the BZO than on the etched AZO substrates. We propose two explanations for the strange phenomenon. First, the feature sizes of the textures on the etched AZO are much larger than the light wavelengths in the response region of a-Si:H solar cells especially with the consideration of wavelength shrinking in the aSi:H layer. In this case, the ray optics works. Following the incident light, one could see that when the light reaches the textured surface, it is reflected (scattered) away from the direct transmission direction; hence a high haze was observed in the optical measurements. When the thin a-Si:H solar cell is deposited on the etched AZO substrate, as seen from Fig. 3, the top and bottom surfaces of the silicon layer are in parallel with large lateral feature sizes. Because of the feature sizes are much larger than the thickness of the silicon layer, the light scattering can be treated as following the ray optics and one could see that the light hits on the solar cell as if it reached on a tilted flat solar cell, and the reflected light is always coming out from a tilted flat solar cell with parallel silicon surfaces, which means the light scattering does not increase the light path significantly. The main effect of such texture on the light trapping is to increases the surface area, which is by 7% as the S-value suggested. While the feature sizes on the BZO are much smaller than those on the etched AZO, and some of them are comparable with the thickness of the silicon layers. In this case, even the light scattering is not as good as the etched AZO in the optical measurements, the light trapping in the solar cell is effective because of the scattered light does not reach a parallel top surface to the bottom reflecting surface. And some of lights reach the total reflection condition and are trapped in the solar cell. In addition, the S-value of 1.31 indicates a larger surface enhancement on the BZO than the etched AZO as well. The lower total transmittance may not mean that the BZO is more absorbing than the AZO. Instead it could result from the light trapping in the BZO when the

Absorptance (%)

80

60

BZO AZO AZO+HCl FTO

40

20 300

400

500

600

700

800

Wavelength (nm) Fig. 5. The total absorption spectra A(λ) = 1 − R(λ) of the four solar cells on different TCO substrates.

225

Solar Energy Materials and Solar Cells 171 (2017) 222–227

J. Fang et al.

EQE(-1V)/EQE(0V) (a.u.)

1.16

factor is the main reason for the difference in the recombination losses is not clear at this stage. As the Jsc of the ultra-thin solar cell with 70-nm thick intrinsic layer reaching 12.82 mA/cm2, one would ask how close the light trapping is to the classical limit of 4n2 [34,35], where n is the refractive index of silicon, and how much room exists for further improvement in such thin solar cells. First, the 4n2 enhancement was derived based on the classic ray optics, which is only valid for the cells thicker than the wavelength of the light in the medium. In case of the ultra-thin solar cells, the total silicon layers are much thinner than the light wavelength even with the consideration of wavelength shrinking, especially for the long wavelength region. An attempt to compare the EQE response of the ultra-thin solar cells with the classical limit might be a conceptual mistake. Second, the absorption in the doped layers (n and p) are in the same order of magnitude because the doped layers (p + n) are about 30 nm, while the i-layer is 20–70 nm; the light trapping also increases the absorbance in the doped layer, which makes the comparison between the EQE with the classical limit complicated. Based on these considerations, we do not try to add the comparison and leave this investigation to a future work.

BZO AZO AZO+HCl

1.12

1.08

1.04

1.00

400

500

600

700

Wavelength (nm)

EQE(+0.6V)/EQE(0V) (a.u.)

0.96 0.94 0.92

4. Summary

0.90

BZO AZO AZO+HCl

0.88 0.86 0.84

We have studied the substrate effect on ultra-thin p-i-n structured aSi:H solar cells with an intrinsic layer thickness of 70 nm. The commonly used chemically etched sputtering AZO and MOCVD deposited BZO. The HCl etched AZO has crater-like large features on the surface with the lateral dimensions of 1–2 µm, while the BZO has mountain-like structures with much smaller features with a wide size distribution. As a comparison, the commercially available FTO was used as a reference. Experimentally, we found that although the etched AZO has a better transmission and scattering effect as measured by the transmittance and haze spectrum than the BZO, the photocurrent density of the ultra-thin a-Si:H solar cell on the BZO is much higher than that on the etched AZO substrates. We believe that the much larger texture features on the etched AZO than the silicon layer thickness leads to the parallel TCO/Si interface to the Si/back reflector interfaces, where the light can escape the silicon layer without total reflection just as if the light gets into a tilted flat solar cell. Although the smaller features on the BZO does not produce as high scattering as the etched AZO as measured by the optical analyses, but the smaller features close the silicon layer thickness produces more effective light trapping than the large features on the etched AZO. In addition, the ultra-thin a-Si:H solar cells on AZO showed high EQE losses in the short wavelength, indicating a poor p/i interface, which was also consistent with the poor FF of the cell on the etched AZO. Because we used a nc-SiOx:H p-layer, the high hydrogen dilution and high RF power during the deposition of the p-SiOx:H resulted in a high density of atomic hydrogen and damaged the FTO. Therefore, the solar cell on the FTO had a poor performance with the lowest Voc and FF. Overall, we conclude that the MOCVD deposited BZO with moderate textures is the best TCO for the ultra-thin a-Si:H solar cells. With this optimized TCO substrate, we achieved an initial efficiency of 8.15% from an ultra-thin a-Si:H p-i-n solar cell with the i-layer thickness of 70 nm, which is one of the highest efficiencies for such thin solar cells. The results provide a guideline for selecting TCO substrates for ultrathin a-Si:H solar cells. And the knowledge gathered from this study is also applicable and useful for other thin film solar cells in terms of light management.

400

500

600

700

Wavelength (nm) Fig. 6. (upper panel) The EQE(− 1.0 V)/EQE(0 V) and (lower) the EQE(+ 0.6 V)/EQE (0 V) spectra of the ultra-thin a-Si:H solar cells on various TCO/glass substrates.

and enhances the recombination rate by retarding movement of the photocarriers. Therefore, the ratio of the biased EQE and zero biased EQE reveals some recombination information. For example, the ratio of EQE(− 1 V)/EQE(0 V) reflects the recombination difference between short circuit condition and − 1 V; and the EQE(+ 0.6 V)/EQE(0 V) reflects the recombination difference between short circuit condition and the condition close the maximum power. If the EQE loss occurs in the short wavelength, it suggests that recombination loss is in the p/i interface region, where the short wavelength light is absorbed; the middle-long wavelength loss reflects bulk recombination; and the very long wavelength EQE loss might give the recombination information near the i/n interface. Fig. 6 plots (upper panel) the EQE(− 1.0 V)/EQE (0 V) and (lower panel) the EQE(+ 0.6 V)/EQE(0 V) spectra of the ultra-thin a-Si:H solar cells on three TCO/glass substrates. One can see that the solar cells on the AZO (no matter etched or not) have much higher EQE losses in the short wavelength, slightly lower losses in the middle wavelength region, and much lower losses in the very long wavelength regions than the cell on the BZO. Correlating the EQE loss and the J-V characteristics, we conclude that the ultra-thin a-Si:H solar cells on the AZO have a poorer p/i interface, which not only reduces the short wavelength response but also reduces the FF. The poorer p/i interface could be due to additional defects in this region or B has gone into this region. The cell on the BZO shows a slightly more bulk recombination than the cell on the AZOs, which could be due to the bulk defect density is higher in the cell on the BZO, or the AZO cells have a higher electric field in the bulk than the BZO cell because the AZO cells might have a weaker electric field in the p/i interface region than the BZO cell in the same region as the total built-in potential is assumed the same. The lower electric field in the p/i interface region could result in an enhancement of electric field in the bulk of i-layer. However, which

Acknowledgements The authors gratefully acknowledge the supports from International Cooperation Project of the Ministry of Science and Technology (2014DFE60170), National Natural Science Foundation of China (61474065 and 61674084), Tianjin Research Key Program of Application Foundation and Advanced Technology (15JCZDJC31300), 226

Solar Energy Materials and Solar Cells 171 (2017) 222–227

J. Fang et al.

Appl. Phys. Lett. 99 (2011) 113512. [19] J. MuKller, O. Kluth, S. Wieder, H. Siekmann, G. Schope, W. Reetz, O. Vetterl, D. Lundszien, A. Lambertz, F. Finger, B. Rech, H. Wagner, Development of highly efficient thin film silicon solar cells on texture-etched zinc oxide-coated glass substrates, Sol. Energy Mater. Sol. Cells 66 (2001) 275–281. [20] J. Müller, G. Schöpea, O. Klutha, B. Recha, V. Sittingerb, B. Szyszkab, R. Geyerc, P. Lechnerc, H. Schadec, M. Rusked, G. Dittmare, H.-P. Bochemf, State-of-the-art mid-frequency sputtered ZnO films for thin film silicon solar cells and modules, Thin Solid Films 442 (2003) 158–162. [21] J. Hüpkes, H. Zhu, J.I. Owen, G. Jost, E. Bunte, Instabilities in reactive sputtering of ZnO:Al and reliable texture-etching solution for light trapping in silicon thin film solar cells, Thin Solid Films 520 (1913) (2012) 1913–1917. [22] Y. Mai, S. Klein, R. Carius, H. Stiebig, X. Geng, F. Finger, Open circuit voltage improvement of high-deposition-rate microcrystalline silicon solar cells by hot wire interface layers, Appl. Phys. Lett. 87 (2005) 073503. [23] A. Lambertz, V. Smirnov, T. Merdzhanova, K. Ding, S. Haas, G. Jost, R.E.I. Schropp, F. Finger, U. Rau, Microcrystalline silicon–oxygen alloys for application in silicon solar cells and modules, Sol. Energy Mater. Sol. Cells 119 (2013) 134–143. [24] J. Meier, S. Dubail, S. Golay, U. Kroll, S. Fay, E. Vallat-Sauvain, L. Feitknecht, J. Dubail1, A. Shah, Microcrystalline silicon and the impact on micromorph tandem solar cells, Sol. Energy Mater. Sol. Cells 74 (457) (2002) 457–467. [25] S. Nicolay, M. Despeisse, F.-J. Haug, C. Ballif, Control of LPCVD ZnO growth modes for improved light trapping in thin film silicon solar cells, Sol. Energy Mater. Sol. Cells 95 (2011) 1031–1034. [26] O. Caglar, P. Carroy, P.A. Losio, I. Sinicco, Nanocrystalline zinc oxide for surface morphology control in thin-film silicon solar cells, Sol. Energy Mater. Sol. Cells 144 (2016) 55–62. [27] K. Söderströmn, G. Bugnon, R. Biron, C. Pahud, F. Meillaud, F.-J. Haug, C. Ballif, Thin-film silicon triple-junction solar cell with 12.5% stable efficiency on innovative flat light-scattering substrate, J. Appl. Phys. 112 (2012) 114503. [28] J. Zhao, A. Wang, M.A. Green, 24.5% efficiency silicon PERT cells on MCZ substrates and 24.7% efficiency PERL cells on FZ substrates, Prog. Photovolt.: Res. Appl. 7 (1999) 471–474. [29] J. Fang, Z. Chen, N. Wang, L. Bai, G. Hou, X. Chen, C. Wei, G. Wang, J. Sun, Y. Zhao, X. Zhang, Improvement in performance of hydrogenated amorphous silicon solar cells with hydrogenated intrinsic amorphous silicon oxide p/i buffer layers, Sol. Energy Mater. Sol. Cells 128 (2014) 394–398. [30] J. Fang, Z. Chen, G. Hou, F. Wang, X. Chen, C. Wei, G. Wang, J. Sun, D. Zhang, Y. Zhao, X. Zhang, High-quality hydrogenated intrinsic amorphous silicon oxide layers treated by H2 plasma used as the p/i buffer layers in hydrogenated amorphous silicon solar cells, Sol. Energy Mater. Sol. Cells 136 (2015) 172–176. [31] L.-M. Huang, C.-W. Hu, H.-C. Liu, C.-Y. Hsu, C.-H. Chen, K.-C. Ho, Photovoltaic electrochromic device for solar cell module and self-powered smart glass applications, Sol. Energy Mater. Sol. Cells 99 (2012) 154–159. [32] H.B.T. Li, R.H. Franken, J.K. Rath, R.E.I. Schropp, Structural defects caused by a rough substrate and their influence on the performance of hydrogenated nanocrystalline silicon n–i–p solar cells, Sol. Energy Mater. Sol. Cells 93 (2009) 338–349. [33] G. Yue, B. Yan, L. Sivec, Y. Zhou, J. Yang, S. Guha, Effect of impurities on performance of hydrogenated nanocrystalline silicon solar cells, Sol. Energy Mater. Sol. Cells 104 (2012) 109–112. [34] E. Yablonovitch, Statistical ray optics, J. Opt. Soc. Am. 72 (1982) 899–907. [35] T. Teidje, E. Yablonovitch, G.D. Cody, B.G. Brooks, Limiting efficiency of silicon solar cells, IEEE Trans. Electron Devices 31 (1984) 711–716.

Key Project in the Science & Technology Pillar Program of Jiangsu Province (BE2014147-3), and the 111 Project (B16027). References [1] M.A. Green, The Passivated Emitter and Rear Cell (PERC): from conception to mass production, Sol. Energy Mater. Sol. Cells 143 (2015) 190–197. [2] K. Yoshikawa, H. Kawasaki, W. Yoshida, T. Irie, K. Konishi, K. Nakano, T. Uto, D. Adachi, M. Kanematsu, H. Uzu, K. Yamamoto, Silicon heterojunction solar cell with interdigitated back conacts for photoconversion over 26%, Nat. Energy 2 (2017) (17032-1-8). [3] F. Feldmann, M. Bivour, C. Reichel, M. Hermle, S.W. Glunz, Passivated rear contacts for high-efficiency n-type Si solar cells providing high interface passivation quality and excellent transport characteristics, Sol. Energy Mater. Sol. Cells 120 (2014) 270–274. [4] R. Müller, C. Reichel, J. Schrof, M. Padilla, M. Selinger, I. Geisemeyer, J. Benick, M. Hermle, Analysis of n-type IBC solar cells with diffused boron emitter locally blocked by implanted phosphorus, Sol. Energy Mater. Sol. Cells 142 (2015) 54–59. [5] D.L. Staebler, C.R. Wronski, Reversible conductivity changes in discharge-produced amorphous Si, Appl. Phys. Lett. 31 (1977) 292–294. [6] J. Yang, A. Banerjee, S. Guha, Amorphous silicon based photovoltaics-from earth to the “final frontier”, Sol. Energy Mater. Sol. Cells 78 (2003) 597–612. [7] S. Guha, J. Yang, B. Yan, High efficiency multi-junction thin film silicon cells incorporating nanocrystalline silicon, Sol. Energy Mater. Sol. Cells 119 (2013) 1–11. [8] J.-H. Yoon, J. Song, S.-J. Lee, Practical application of building integrated photovoltaic (BIPV) system using transparent amorphous silicon thin-film PV module, Sol. Energy 85 (2011) 723–733. [9] J.W. Lim, M. Shin, D.J. Lee, S.H. Lee, S.J. Yun, Highly transparent amorphous silicon solar cells fabricated using thin absorber and high-bandgap-energy n/i-interface layer, Sol. Energy Mater. Sol. Cells 128 (2014) 301–306. [10] J.W. Lim, G. Kim, M. Shin, S.J. Yun, Colored a-Si:H transparent solar cells employing ultrathin transparent multi-layered electrodes, Sol. Energy Mater. Sol. Cells 163 (2017) 164–169. [11] S. Kim, J.-W. Chung, H. Lee, J. Park, Y. Heo, H.-M. Lee, Remarkable progress in thin-film silicon solar cells using high-efficiency triple-junction technology, Sol. Energy Mater. Sol. Cells 119 (2013) (36-35). [12] J.S. Cashmore, M. Apolloni, A. Braga, et al., Improved conversion efficiencies of thin-film silicon tandem (MICROMORPH™) photovoltaic modules, Sol. Energy Mater. Sol. Cells 144 (2016) 84–95. [13] J. Zhu, C.-M. Hsu, Z. Yu, S. Fan, Y. Cui, Nanodome Solar Cells with Efficient Light Management and Self-Cleaning, Nano Lett. 10 (2010) 1979–1984. [14] Y. Kuang, K.H.M. van der Werf, Z.S. Houweling, R.E.I. Schropp, Nanorod solar cell with an ultrathin a-Si:H absorber layer, Appl. Phys. Lett. 98 (2011) 113111. [15] S.-Y. Wang, D.-A. Borca-Tasciuc, D.A. Kaminski, Optical properties of ultra-thin silicon films deposited on nanostructured anodic alumina surfaces, Appl. Phys. Lett. 104 (2014) 081119. [16] J. Meier, R. Fluckiger, H. Keppner, A. Shah, Complete microcrystalline p-i-n solar cell—crystalline or amorphous cell behavior? Appl. Phys. Lett. 65 (1994) 860–862. [17] A.V. Shah, J. Meier, E. Vallat-Sauvain, N. Wyrsch, U. Kroll, C. Droz, U. Graf, Improved conversion efficiencies of thin-film silicon tandem (MICROMORPH™) photovoltaic modules, Sol. Energy Mater. Sol. Cells S. 78 (2003) 469–491. [18] B. Yan, G. Yue, L. Sivec, J. Yang, S. Guha, C.-S. Jiang, Innovative dual function ncSiOx:H layer leading to a > 16% efficient multi-junction thin-film silicon solar cell,

227