Article pubs.acs.org/JPCC

Efficient Hydrogenated Amorphous Silicon Thin-Film Solar Cells Using Zinc Oxide Deposited by Atomic Layer Deposition as a Protective Interfacial Layer Young-Joo Lee,† Min-Seung Choi,† Dong-Ho Kim,† Chang-Su Kim,† Myung-Kwan Song,† Jae-Wook Kang,† Yongsoo Jeong,† Kee-Seok Nam,† Sung-Gyu Park,† Se-Hun Kwon,*,‡ Seung Yoon Ryu,*,§ and Jung-Dae Kwon*,† †

Advanced Functional Thin Films Department, Korea Institute of Materials Science (KIMS), Changwon, Gyeongnam, 641-831, Republic of Korea ‡ National Core Research Center for Hybrid Materials Solution, Pusan National University, Busan 609-735, Republic of Korea § Department of Information Display, Sunmoon University, Asan, Chungnam, 336-708, Republic of Korea S Supporting Information *

ABSTRACT: A single zinc oxide (ZnO) layer deposited by atomic layer deposition (ALD) was employed as a buffer layer on textured fluorine-doped tin oxide (FTO) glass in p-i-n-type hydrogenated amorphous silicon solar cells (a-Si:H SCs). ZnO was deposited between FTO glass and a p-type a-Si:H layer. The device with a 2 nm thick ZnO buffer layer deposition showed the highest cell efficiency as well as increased current density, showing considerable improvement in efficiency. This improvement was attributed to the protection against H2 plasma damage during plasma-enhanced chemical vapor deposition, which was confirmed by electrical impedance, conductivity, and optical transmittance measurements.

■

INTRODUCTION Solar cell technology has attracted much attention over recent years owing to its huge potential as a clean energy source for the future. Among the various categories of solar cells, p-i-ntype hydrogenated amorphous silicon (a-Si:H) thin-film solar cells (TFSCs) are one of the most feasible energy conversion devices due to the ease of large area deposition, low production cost, and large absorption coefficient, when compared to polycrystalline Si. Accordingly, a-Si:H TFSCs have been widely studied for enhancing photovoltaic performance. In conventional a-Si:H TFSCs, the superstrate configuration is in the following order: glass, transparent conductive oxide (TCO), p-i-n-type a-Si:H thin film alloy, and metal or TCO electrode. Fluorine-doped SnO2 is usually used as the TCO, owing to its advantageous excellent heat resistance, high optical transmittance, and high work function. Generally, the p-i-n-type a-Si:H thin film is deposited on the fluorine-doped tin oxide (FTO) by plasma-enhanced chemical vapor deposition (PECVD) methods using silane (SiH4) and H2 gases. During the growth of the Si thin films on the FTO film in the PECVD chamber, the FTO is exposed to the plasma, which contains hydrogen radicals and ions. The H+ ions in the hydrogen plasma atmosphere react with lattice oxygen atoms and reduce the surface, which results in the presence of metallic Sn.1 The chemical reduction leads to a less transparent film and a higher © 2012 American Chemical Society

sheet resistance because of the composition change in SnO2 and surface microstructure.2−5 In contrast, because the bonding energy between Zn and O is stronger than that between Sn and O, ZnO is resistant to H2 plasma and has been utilized to form the SnO2/ZnO bilayer to prevent hydrogen diffusion into the FTO films.6,7 Kubon et al. have already demonstrated the deposition of a 15 nm thick ZnO layer by sputtering to protect SnO2; however, no increase in cell performance was observed when only a ZnO layer was used because of the ZnO/p contact problem.8 In this paper, ZnO thin films grown by atomic layer deposition (ALD) were utilized as an interlayer to stabilize the FTO/p-Si:H interface in the a-Si TFSC. The ALD technique is able to grow thin, pinhole-free films with good step coverage because it is a self-limiting growth method that consists of saturated surface reactions. By keeping the precursors separate throughout the deposition process, the atomic layer thickness can be controlled by regulating the number of cycles. The ALD process has been extensively researched for studying the surface passivation of p-type silicon in crystalline silicon solar cells and the window layer of Cu(In,Ga)Se2 thin-film solar cells.9,10 Received: June 28, 2012 Revised: October 17, 2012 Published: October 22, 2012 23231

dx.doi.org/10.1021/jp306360h | J. Phys. Chem. C 2012, 116, 23231−23235

The Journal of Physical Chemistry C

Article

However, there are fews studies that have reported on the ALD process for the silicon thin-film solar cells. ZnO ALD films were grown as an interlayer on the FTO glass by carrying out 1, 2, or 5 nm of deposition. The dependence of photovoltaic performance on the thickness of the ZnO interlayer was investigated, with the thickness after 2 nm of deposition found to be optimal. Approximately, 9.3% enhancement of cell efficiency was achieved, in comparison to 6.697% for the reference device without the ZnO buffer layer and 7.319% for the FTO-based solar cell with a 2 nm ZnO layer. It was observed that the cell efficiency with the ZnO interlayer significantly increased up to 2 nm and then gradually decreased when the thickness was higher than 2 nm.

■

EXPERIMENTAL SECTION

The ZnO films were grown on surface textured FTO glass by using an ALD process [ALD system Lucida M100, NCD] at 100 °C. Diethyl zinc (DEZ, Zn(C2H5)2) and deionized (DI) water were used as the source of zinc and oxygen, respectively. One deposition cycle of ZnO consisted of a pulse and purge step of DEZ, DI water, and H2 plasma. The H2 plasma pulse was added after the pulse and purge step of DI water to improve the electrical resistivity. The H2 plasma time was fixed at 5 s with a plasma power of 100 W. The upper electrode was capacitively coupled with a radio frequency (RF) of 13.56 MHz. For the process of manufacturing the solar cells, the ZnO ALD coated onto FTO glass (5 cm × 5 cm) was used as the front electrode, and the p-i-n-type a-Si:H films were prepared by the three-chamber PECVD cluster system. After this, silver was thermally deposited in a high vacuum chamber (∼2 × 10−6 Torr) using a shadow mask to define a cell active area of 0.25 cm2. The FTO glass contained nine cells, and the cell performance was considered to be the average value determined after measuring the performances of nine cells. Current density−voltage (J−V) characteristics were measured using a Keithley 2400 source meter under 100 mW/cm2 (AM 1.5G) irradiation from a solar simulator (Pecell Technologies Inc., PEC-L11). In addition to the photovoltaic performance, internal and external quantum efficiency (IQE/EQE) of each photovoltaic device was obtained by using a 200 W Xe lamp and a grating monochromator, and the light intensity was measured by a calibrated Si solar cell (PV measurement). The work functions of devices were measured by the Kelvin probe method, using contact potential difference (CPD) of a previously ultraviolet photoemission spectroscopy (UPS) characterized Au surface as reference.

Figure 1. (a) Photocurrent vs voltage characteristics of solar cells with and without (reference) 9, 18, or 45 cycles of ZnO buffer layer deposition. The inset shows a vertical schematic illustration of a-Si:H TFSCs with ZnO ALD as a protection buffer layer. (b) The device performance of four kinds of devices in the fabricated p-i-n-type a-Si solar cells.

■

RESULTS AND DISCUSSION The J−V characteristics, including short-circuit current (Jsc), open-circuit voltage (Voc), and fill factor (FF), of four kinds of a-Si:H TFSC devices are shown in Figure 1. The cell efficiencies of the p-i-n-type a-Si:H SCs were improved with 9, 18, or 45 cycles of ZnO buffer layer and compared to a reference device. After 45 cycles of ZnO ALD, a 5 nm thick ZnO layer was observed by cross-sectional TEM analysis (not shown). Therefore, the thicknesses of ZnO thin films were estimated to be 1 and 2 nm when 9 and 18 cycles, respectively, of ZnO ALD were carried out. The Jsc and cell efficiency of the device with 18 cycles of ZnO buffer layer were higher than those of all the other devices; however, Voc and FF were almost the same, being approximately 0.804 V and 0.735, respectively. The Jsc of the device with a ZnO buffer layer from 18 cycles

(12.394 mA/cm2) was the highest among the devices and approximately 9.2% higher than that of the reference device (11.346 mA/cm2). Increasing the number of cycles to 45 gave a slightly lower Jsc of 12.275 mA/cm2, although this was still higher than the reference device. The work functions of devices were measured by the Kelvin probe method as shown in Figure 2(a). The Fermi levels of FTO and with 0, 1, 2, and 5 nm ZnO were around 5.0, 4.95, 4.83, and 4.64 eV, respectively. The holes generated at the intrinsic absorption layer were supposed to be easily extracted by the tunnel effect owing to a few nanometers thickness of ZnO layers. When a ZnO layer is inserted between the FTO/p-Si:H interface, the interface barrier (Φ2) formed between the ZnO and p-Si:H is higher than that (Φ1) of the cell without the ZnO layer due to the low 23232

dx.doi.org/10.1021/jp306360h | J. Phys. Chem. C 2012, 116, 23231−23235

The Journal of Physical Chemistry C

Article

higher than that of the FTO reference device; however, the IQE of the device with 5 nm of ZnO deposition demonstrated a poorer performance. This may be because the transmittance and impedance was heavily degraded, as shown in Figure 3(a)

Figure 2. (a) Band diagram variation due to the ZnO buffer layer at the FTO/p-Si:H interface. (b) Internal quantum efficiency spectra of solar cells with and without 1, 2, or 5 nm of ZnO buffer layer deposition. The inset shows the external quantum efficiency spectra of solar cells with and without 1, 2, or 5 nm of ZnO buffer layer deposition.

Figure 3. (a) Total transmittance of 1, 2, 5, or 20 nm of ZnO deposition on textured FTO glass. (b) Impedance spectra of solar cells with and without 1, 2, or 5 nm of ZnO buffer layer deposition.

work function of ZnO. The increase of the interface barrier might lead to the decrease of Voc according to the degraded built-in potential (Vbi).11 In addition, Voc also depends on photocurrent and dark saturation current density, according to the Shockley equation. The equation was the following equation. Voc =

⎞ nkT ⎛ Jph ln⎜⎜ + 1⎟⎟ q ⎝ Js ⎠

and (b). We strongly believe that the Jsc of the device is related to many things, i.e., not only to EQE and transmittance data, which are the optical properties, but also to series and shunt resistance in total impedance of device, which are the electrical properties.12 This relationship is very complicated and has not yet been exactly explained. The impedance of the device with 5 nm thick ZnO is much lower than that of the device with 1 nm thick ZnO. Therefore, in this work, we see that the electrical properties of a ZnO thin layer deposited by ALD, the interface between a textured FTO and p-type Si layer, are much more important than the optical properties, due to the thickness of a few nanometers. This ZnO thin layer has a thickness of a few nanometers. We do not believe that the transmittance data are not at all related to IQE or EQE spectra. Actually, IQE and EQE enhancement is not observed at just 500−700 nm but is observed at 300−700 nm, which is almost the entire wavelength range (except over 700 nm). That means that the device data, including IQE and EQE data, are mainly affected by the electrical properties that are impedance and avoiding plasma damage and not by the optical properties. In this work, the electrical interface between a textured FTO and p-type Si layer is much more critical than the optical properties. The transmittance of 1 and 2 nm thick ZnO films on FTO was almost the same as that for the reference; however, over 5 nm, the transmittance degraded. With increasing thickness, the transmittance of the ZnO film generally decreased due to the thickness effect.13 The light absorption by the 5 nm thick ZnO film might lead to the decrease in the quantum efficiency. Additionally, we measured the degree of elastic light scattering, i.e., haze ratio of FTO substrates with and without ZnO buffer layers (not shown). However, there is no discernible difference

(1)

where n is the ideality factor; q is the charge of an electron; k is the Boltzmann constant; Jph is photocurrent density; Js is the dark saturation current density; and T is the temperature.12 Considering the Shockley equation, because the dark current density was almost independent of the thickness of the ZnO interlayer, Voc would be increased. However, the Voc of devices with 1, 2, and 5 nm thick ZnO layers were almost unvaried at 0.803−0.808 V. The monotonic Voc value of 0.803−0.808 V would be affected by comprehensive factors such as work function and current density. However, the minimum and maximum Voc was 0.787 and 0.824 V, considering the measurement error. Therefore, it is also possible that the variation of Voc could not be detected using the measurement error. The internal quantum efficiency (IQE) and external quantum efficiency (EQE) values shown in Figure 2(b) can also be explained by Jsc. The ZnO layer influenced the IQE/ EQE values through its optical transmittance and electrical conductivity properties. The IQE of the device with a ZnO buffer layer from 2 nm was higher than that of any other devices closely following the trend of the Jsc and impedance data. Similar to the Jsc, the IQE of devices with 1 and 2 nm was 23233

dx.doi.org/10.1021/jp306360h | J. Phys. Chem. C 2012, 116, 23231−23235

The Journal of Physical Chemistry C

Article

depending on the ZnO film thickness of 0−5 nm. The haziness, defined as the ratio of diffuse (Tdiff) to total (Ttotal) transmittance, is the relative amount of light scattered by the surface texturing. It is thought that the decrease in Tdiff was as much as that in Ttotal because of the light absorption of the ZnO layer. Thus, the haze ratio did not change even when the absolute Tdiff and Ttotal were decreased by the coating of the ZnO layer. Impedance spectroscopy (IS) has been a powerful, nondestructive characterization tool for dye-sensitized solar cells and bulk heterojunction organic photovoltaic devices; this method can be used to calculate charge carrier lifetime, electronic densities of states, and charge carrier concentrations.14,15 In this study, we used the IS method to find the increase in the internal resistance due to the H2 plasma damage between FTO and p-type Si:H layers. The impedance of the devices with 1 and 2 nm was lower than that without the ZnO interlayer, indicating that the ZnO sufficiently protected the FTO from H2 plasma damage during the PECVD process. Previously, when a-Si:H thin film was grown on tin-doped indium oxide (ITO) or tin oxide (TO) by PECVD, the H2 plasma degraded the TCO surface due to the formation of metallic indium or tin.16 The surface of the degraded TCO altered the surface roughness and chemical state,and affected the properties of the interface between the TCO and a-Si:H films. However, interestingly, the impedance of the device with 5 nm of ZnO was increased compared to that with 2 nm, as shown in the inset of Figure 3 (b), although it was smaller than those of the reference device and that with 1 nm of ZnO. This impedance difference of 2 and 5 nm thick ZnO films clearly appeared, and the difference was reproduced in five measurements. This is because the ZnO interlayer that resulted from 5 nm acted in the same way as the bulk material, presenting a higher resistance than FTO. Another group simply utilized nitrogen-doped ZnO (ZnO:Al:N) as a buffer layer to adjust the potential barrier of the FTO at the FTO/a-Si interface.17 ZnO is generally wellknown for being an n-type semiconductor, and in that paper, there was little discussion involving H2 plasma damage. However, in this work, we point out that the ZnO interlayer is critical for protection against H2 plasma damage, also including the consideration of the band diagram between FTO and p-type Si by inserting the ZnO layers. In the ZnO ALD cycle, H2 plasma treatment was also carried out, similar to the PECVD process, after the reaction with DEZ and DI water to improve the electrical resistivity. The hydrogen ions could easily diffuse into the ZnO thin film, producing a shallow donor. The increase in free electron concentration in the donor level by hydrogen doping decreased the electrical resistivity.18 When a 40 nm thick ZnO film was deposited on bare glass, H2 plasma treatment caused a decrease in the ZnO film resistivity of 2 × 10−1 to 2 × 10−3 Ω cm. This decrease in resistivity indicated that the hydrogen ions successfully became free electron carriers. In both the ZnO ALD and a-Si:H PECVD processes, the FTO is exposed to the H2 plasma. To investigate the effects of only H2 plasma without the ZnO deposition, the optical characteristics of FTO in the ALD and PECVD chambers were compared under identical conditions of 100 °C (substrate temperature) and 100 W RF power, as shown in Figure 4. Total transmittance was measured with varying H2 plasma treatment time to investigate the reduction effect by the plasma on the FTO. There was little change in the total transmittance of FTO exposed to H2 plasma for 90 s in the ALD chamber. In the case of 900 s of treatment, the total

Figure 4. (a) Total transmittance of hydrogen plasma-treated FTO in the ALD and PECVD chambers depending on the plasma exposure time. (b) The conductivity of hydrogen plasma-treated FTO in the ALD and PECVD chambers depending on the plasma exposure time.

transmittance was slightly decreased in comparison with nontreated FTO. On the other hand, in the PECVD chamber, total transmittance was significantly decreased when H2 plasma time increased to more than 450 s. In addition, there was a change in FTO conductivity before and after the H2 plasma in the PECVD chamber, according to the results of Hall-effect measurements. The FTO conductivity decreased from 1400 to 1260 S/cm2 (10% decline) after treatment for 10 min. In contrast, the FTO conductivity in the ALD chamber was unaltered even after 10 min plasma treatment. This demonstrated that there was little effect on the FTO by H2 plasma in the ALD chamber when compared with the PECVD chamber. In spite of the same temperature and power, the FTO was more damaged by H2 plasma in the PECVD chamber, indicating that the hydrogen plasma reaction was more developed in the PECVD chamber than in the ALD chamber because of the different conditions, such as electrode dimensions, chamber volume, and process pressure. The power density of the ALD chamber was 0.12 W/cm2, which was lower than that of the PECVD chamber at 0.16 W/cm2, considering the electrode dimension. Moreover, the substrate temperature during the a-Si:H PECVD process was 250 °C, which is higher than ZnO ALD at 100 °C. It has been previously reported that the degree of FTO reduction is also affected by substrate temperature.16 Therefore, it seemed that the FTO was reduced more by the H2 plasma in the a-Si:H PECVD process than the ZnO ALD process, and the optical and electrical properties of FTO were not affected by H2 plasma during the ZnO interfacial layer deposition.

■

CONCLUSION

We investigated the role of ZnO as a protection buffer layer against H2 plasma damage on FTO during the PECVD process and also the dependence of the thickness of the ZnO on device performance. The device with 2 nm of ZnO deposition showed higher cell efficiency and Jsc than the reference device and the devices with 1 and 5 nm of deposition. This demonstrates that adequate optimization of the thickness of the ZnO buffer layer at the interface of the FTO/p-type Si can prevent H2 plasma damage to the FTO. This was confirmed by measuring optical transmittance, electrical impedance, and conductivity. 23234

dx.doi.org/10.1021/jp306360h | J. Phys. Chem. C 2012, 116, 23231−23235

The Journal of Physical Chemistry C

■

Article

ASSOCIATED CONTENT

S Supporting Information *

High-resolution TEM cross-sectional image and TEM EDX element mapping of 45 cycles ZnO ALD (Figure S1). This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] (S.H.K.); [email protected] (S.Y.R.); and [email protected] (J.D.K.). Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The present research was supported by the research fund (2012-PNK2860) of the Korea Institute of Materials Science, a subsidiary branch of the Korea Institute of Machinery and Materials.

■

REFERENCES

(1) Rajesh, D.; Tapati, J.; Swati, R. Sol. Mater. Sol. Cells 2005, 86, 207−216. (2) Bose, S.; Barua, A K J. Phys. D: Appl. Phys. 1999, 32, 213−218. (3) Lan, J.-H.; Kanicki, J. Thin Solid Films 1997, 304, 123−129. (4) Major, S.; Bhatnagar, M. C.; Kumar, S.; Chopra, K. L. J. Vac. Sci. Technol. A 1988, 6, 2415−2421. (5) Wallinga, J.; Arnoldbik, W. M.; Vredenberg, A. M.; Schorpp, R. E. I.; van der Weg, W. F. J. Phys. Chem. B 1998, 102, 6219−6224. (6) Ikeda, T.; Sato, K.; Hayashi, Y.; Wakayama, Y.; Adachi, K.; Nishimura, H. Sol. Mater. Sol. Cells 1994, 34, 379−384. (7) Natsuhara, H.; Matsumoto, K.; Yoshida, N.; Itoh, T.; Nonomura, S.; Fukawa, M.; Sato, K. Sol. Mater. Sol. Cells 2006, 90, 2867−2880. (8) Kubon, M.; Boehmer, E.; Siebke, F.; Rech, B.; Beneking, C.; Wagner, H. Sol. Mater. Sol. Cells 1996, 41/42, 485−492. (9) Schmidt, J.; Merkle, A.; Brendel, R.; Hoex, B.; van de Sanden, M. C. M.; Kessels, W. M. M. Prog. Photovoltaic: Res. Appl. 2008, 16, 461− 466. (10) Yousfi, E. B.; Weinberger, B.; Donsanti, F.; Cowache, P.; Lincot, D. Thin Solid Films 2001, 387, 29−32. (11) Fang, L.; Baik, S. J.; Lim, K. S.; Yoo, S. H.; Seo, M. S.; Kang, S. J.; Seo, J. W. Appl. Phys. Lett. 2010, 96, 193501−3. (12) Chao, H.; Chengmei, Z.; Hongbin, W.; Renqiang, Y.; Wei, Y.; Fei, H.; Guillermo, C. B.; Yong, C. J. Mater. Chem. 2010, 20, 2617− 2622. (13) Xiaotao, H.; Jin, M.; Deheng, Z.; Tianlin, Y.; Honglei, M.; Yingge, Y.; Chuanfu, C.; Jie, H. Appl. Surf. Sci. 2001, 183, 137−142. (14) Leever, B. J.; Bailey, C. A.; Marks, T. J.; Hersam, M. C.; Durstock, M. F. Adv. Energy Mater. 2012, 2, 120−128. (15) Kuwabara, T.; Kawahara, Y.; Yamaguchi, T.; Takahashi, K. ACS Appl. Mater. Interfaces 2009, 1, 2107−2110. (16) Kumar, S; Drevillon, B. J. Appl. Phys. 1989, 65, 3023−3034. (17) Baik, S. J.; Lim, K. S. J. Korean Phys. Soc. 2011, 59 (2), 443−447. (18) Ohashi, N.; Ishigaki, T.; Okada, N.; Taguchi, H.; Sakaguchi, I.; Hishita, S.; Sekiguchi, T.; Haneda, H. J. Appl. Phys. 2003, 93, 6386− 6392.

23235

dx.doi.org/10.1021/jp306360h | J. Phys. Chem. C 2012, 116, 23231−23235