Lei Zhao, Chunlan Zhou, Hailing Li, Hongwei Diao, and Wenjing Wang

Transcription

1 CHINESE JOURNAL OF PHYSICS VOL. 48, NO. 3 June 2010 Characterization on the Passivation Stability of HF Aqueous Solution Treated Silicon Surfaces for HIT Solar Cell Application by the Effective Minority Carrier Lifetime Measurement Lei Zhao, Chunlan Zhou, Hailing Li, Hongwei Diao, and Wenjing Wang Group of Solar Cell Technology, Key Laboratory of Solar Thermal Energy and Photovoltaic System of Chinese Academy of Sciences, Institute of Electrical Engineering, the Chinese Academy of Sciences, Beijing , P.R. China (Received May 27, 2009) The traditional passivation processes, standard RCA cleaning and 1% HF aqueous solution immersion, were introduced to passivate the surfaces of p-type Si(100) wafers for heterojunction with intrinsic thin-layer (HIT) solar cell application. The passivation stability was checked by monitoring the time decay of the effective minority carrier lifetime of the Si wafer in air via the microwave photoconductive decay (µpcd) method. The results show that the passivation effect is greatly dependent on the initial surface morphology of the Si wafer. During the subsequent exposure in air, the obtained effective minority carrier lifetime decays rapidly, corresponding to that the Si surface states increase greatly. Such decay occurs more severely on the textured surface than on the polished one, which gives a time limitation to the subsequent processes in the HIT solar cell fabrication. So, such passivation processes need further improvement. The results also prove that the effective minority carrier lifetime measurement can be adopted as an efficient and convenient method to check the passivation stability of the Si surface treated by wet-chemical methods. PACS numbers: jj, b I. INTRODUCTION Crystalline silicon solar cells still share over 90% of the rapidly growing photovoltaic market [1]. Heterojunction with intrinsic thin-layer (HIT) solar cells developed by Sanyo corporation have reached 6% of the whole world photovoltaic production [2]. HIT solar cells combine the good properties of crystalline or multicrystalline silicon with the advantages of thin-film silicon technology. In the HIT solar cell structure, an intrinsic a-si:h layer followed by a p-type or an n-type a-si:h layer is deposited on each surface of the crystalline Si wafer to form a p/n heterojunction or a back surface field (BSF), respectively. On the doped a-si:h layers, transparent conductive oxide (TCO) layers are formed, and finally, metal grid electrodes are prepared using a screen-printing method. By this technology, Sanyo has obtained a record efficiency up to 23% [3]. HIT processes are carried out under low temperatures (< 200 C), which can prevent the degradation of bulk quality that may occur in the high-temperature cycling processes. Compared with the conventional diffused cells, a much better temperature coefficient can be obtained with a higher open-circuit voltage [4, 5]. Hence, HIT solar cells have attracted more and more interest in the world [6, 7] c 2010 THE PHYSICAL SOCIETY OF THE REPUBLIC OF CHINA

2 VOL. 48 LEI ZHAO, CHUNLAN ZHOU, HAILING LI, et al. 393 FIG. 1: The surface morphology of (a) the polished wafer with the root-mean-square (rms) roughness of 0.4 nm obtained by AFM, and (b) the textured wafer with the pyramid size of 5-7 µm obtained by SEM. In the HIT structure, junctions are formed by growing thin films on the surfaces, which has the consequence that HIT solar cells are basically surface-(or interface-)dominated and the surface properties are absolutely critical for the performance of the resulting devices. So the Si wafer surface treatment prior to the thin-film deposition is thought to be one key fabrication process. Sanyo announced that a novel Si wafer cleaning process had been developed for obtaining the high HIT efficiency, but did not present any details [5]. The traditional method in most of the literature is to undertake the standard RCA cleaning processes followed by a short time HF dipping [6, 7]. Many characterization techniques, such as Raman spectroscopy, X-ray photoelectron spectroscopy (XPS), Fourier-transform infrared (FTIR), spectroscopic ellipsometry (SE), and so on, have been adopted to characterize the HF treated surfaces of the Si wafers [8 11]. Previous studies [12, 13] show that the Si surface is mainly H-terminated after HF aqueous solution treatment. Such a surface is relative stable in air. The regrowth of the oxide is very slow in the initial stage until a monolayer dioxide forms. Then the oxidation process accelerates to generate a thick dioxide layer. So one may wonder whether the passivation effect of HF treated surface can be kept for a long time. Obviously, long-time stable stable passivation is important for HIT solar cell fabrication. It can provide enough operation time for subsequent processes. In this study, one effective minority carrier lifetime measurement method, called microwave photoconductive decay (µpcd) was utilized to check the passivation effect of the HF treated Si surfaces and its stability.

3 394 CHARACTERIZATION ON THE PASSIVATION... VOL. 48 FIG. 2: The immersion time dependence of the effective minority carrier lifetime on the polished p-type Si(100) wafer treated by 1% HF aqueous solution, and the lifetime surface distribution at three particular time points. II. THE EXPERIMENT P-type Si(100) wafers with resistivity in the range of 1 10 Ω cm were adopted. The wafer surfaces were polished or textured to check the morphology effect. The textured wafers were obtained by alkaline etching. Fig. 1 gives the surface morphology of both the polished wafer and the textured wafer obtained by atomic force microscope (AFM) and scanning electron microscope (SEM) photographs. The root-mean-square (rms) roughness of the polished wafer surface is about 0.4 nm. On the textured wafer surface, the pyramid size is in the range of 5 7 µm. For all the samples, the standard RCA cleaning was carried out followed by 1% HF aqueous solution immersion. Such a 1% concentration is the most widely used one in HIT research. The µpcd measurement was conducted on a Semilab WT-2000 Wafer Scanner to characterize the immersion time dependence of the passivation effect. Then, the samples treated for the optimized immersion time were exposed in air to check the passivation stability. Lifetime measurement is a critical metrology used in the photovoltaic research. Com-

4 VOL. 48 LEI ZHAO, CHUNLAN ZHOU, HAILING LI, et al. 395 FIG. 3: The immersion time dependence of the effective minority carrier lifetime on the textured p-type Si(100) wafer treated by 1% HF aqueous solution, and the lifetime surface distribution at three particular time points. pared with XPS, FTIR, SE and so on, lifetime measurement can be carried out easily and can give more directly apparent results for the solar cell applications. The µpcd technique is based on the following mechanism [14, 15]: The laser pulse is absorbed by the Si wafer and thus free electron-hole pairs are generated. The free electrons and holes will recombine, their concentration and hence the conductivity of the Si wafer will decrease after the laser excitation. Since the reflected microwave power is proportional to the conductivity of the wafer, the decaying conductivity can be monitored by detecting the microwave reflection. It is measured as a function of time. The measured microwave reflectance decay is fitted with an exponential curve and the obtained time constant is recorded as the measured, effective lifetime. The measured lifetime characteristic of the whole recombination process 1 can be described as τ meas = 1 1 τ bulk + τ diff +τ surf, wherein τ meas is the measured effective lifetime, τ bulk is the bulk recombination lifetime, τ diff is the characteristic time necessary for carrier diffusion to the surface from the middle of the wafer, and τ surf is the surface recombination lifetime. For the same wafer, τ bulk and τ diff are both constant. So τ meas can be used to characterize τ surf, i.e., the surface passivation effect, under different passivation conditions.

5 396 CHARACTERIZATION ON THE PASSIVATION... VOL. 48 FIG. 4: The effective minority carrier lifetime decay as a function of the exposure time in air on the polished p-type Si(100) wafer treated by 1% HF aqueous solution for 2.5 min, and the lifetime surface distribution at three particular time points. III. RESULTS AND DISCUSSION Fig. 2 gives the immersion time dependence of the passivation effect on the polished wafer and the lifetime surface distribution at three time points. It can be seen that, firstly, τ meas increases gradually as the immersion time extends. At about 2 3 minutes, τ meas reaches the maximum value. A longer time has little influence on τ meas. Ideally, in the dilute HF aqueous solution, HF etches the native oxide on the Si surface, terminates the surface dangling bonds with H atoms, and hence induces the passivation effect. The maximal τ meas indicates the time point at which the native oxide is etched off completely. Some references have shown that overetching will induce many (111) microfacets on the surface [8] and strongly increase the surface microroughness, resulting in an increase of the surface state density [9]. This is not obvious here. Similar phenomena were also observed on the textured wafer, as shown in Fig. 3. But the optimized etching time needed for the textured one is over 5 minutes, which is consistent with the result obtained in reference [10], where the pyramid size is larger than 3 µm. The time difference indicates that the passivation effect

6 VOL. 48 LEI ZHAO, CHUNLAN ZHOU, HAILING LI, et al. 397 FIG. 5: The effective minority carrier lifetime decay as a function of the exposure time in air on the textured p-type Si(100) wafers treated by 1% HF aqueous solution for 5 min, and the lifetime surface distribution at three particular time points. is dependent on the surface morphology. Although it is difficult to passivate the textured surface, an excellent passivation effect can be obtained as long as the immersion time is long enough. The absolute value difference of the maximal lifetime between the two wafers is considered to be due to their nature nonuniformity. The stability of the surface passivation on HF wet chemically treated wafers is an important parameter in the HIT technological process. One polished wafer and one textured wafer treated by the HF aqueous solution with the optimized immersion time were monitored for the lifetime variation as a function of the exposure time in air, as shown in Fig. 4 and Fig. 5. The results are undesirable: the effective lifetime decreases very quickly once the wafer is exposed into the air, and reaches a very low value in just several dozens of minutes. Such a decay is more severe on the textured wafer than on the polished one. The passivation effect obtained by HF aqueous solution immersion is unstable in air. As mentioned above, the reoxidation of the H-terminated Si surface has an initial stage with a low growth rate at first. So the lifetime decrease must occur in the initial stage. This means that although the dioxide generates slowly, the density of surface states increases

7 398 CHARACTERIZATION ON THE PASSIVATION... VOL. 48 d drastically in this stage, because of τ surf = 2σν tn ss, where d is the thickness of the wafer, σ is the minority carrier capture cross section, υ t is the thermal velocity, and N ss is the density of surface states. Angermann et al. [9] reported that a continuous reduction in the number of SiH and SiH 2 bonds was observed by FTIR measurements during the formation of the first oxide monolayer the so-called initial stage of oxidation. These hydrogen vibrations completely disappear when the effective thickness of the native oxide film achieves one monolayer. It was also observed by the surface photovoltage measurement that the surface state distribution showed a strong increase in the defect density during the first 10 minutes of storage in clean-room air [16]. Hence, it can be concluded that the reduction of such H-terminated bonds increases the surface states on the silicon surface and thus weakens the passivation effect. The oxygen species in air (H 2 O and O 2 ) are expected to preferentially break the back bonds on the defects predominantly localized on the crystallographic steps and edges of the surface [11]. At defect sites, two-dimensional oxide islands start to form then extend to a complete monolayer. So the initial native oxide growth greatly depends on the surface morphology. The duration of the first monolayer growth time was found to be inversely proportional to the initial surface state density in the semi-logarithmic plot [17]. Besides, some positions on the surface may not be H-terminated, but rather F-terminated, depending on the environmental conditions, for example, the surface morphology and ph value of the solution [18]. Si-F bonds are not as stable as Si-H bonds. Such positions can be attacked easily by the oxygen species. All these are more likely to occur on the textured surface due to its larger roughness. That s why the passivation effect on the textured surface is more unstable than that on the polished one. However, textured surfaces are necessary for solar cells to reduce the light reflection. The instability of the passivation effect indicates that RCA cleaning followed by HF aqueous solution immersion or HF dipping may not be the perfect treatment for Si wafers in solar cell fabrication. Some improved technologies should be developed to lower the surface microroughness. And a new wet-chemical prescription is also needed to make a more nearly perfect H-terminated surface. IV. CONCLUSION In summary, the traditional passivation processes, standard RCA cleaning and 1% HF aqueous solution immersion, were introduced to passivate the surfaces of p-type Si(100) wafers for HIT solar cell application. The µpcd effective minority carrier lifetime measurement was utilized to characterize the passivation effect and its time decay in air. The results show that the passivation effect is greatly dependent on the initial surface morphology. And during the subsequent exposure in air, the obtained effective minority carrier lifetime of Si wafers decays rapidly, due to the fact that the Si surface states increase greatly. Such decay occurs more severely on the textured surface than on the polished one, which gives a time limitation to the subsequent processes in the HIT solar cell fabrication. So such passivation processes need further improvements. The results also prove the effective minority carrier lifetime measurement as an efficient and convenient method for checking the passivation

8 VOL. 48 LEI ZHAO, CHUNLAN ZHOU, HAILING LI, et al. 399 stability of the Si surfaces treated by wet-chemical methods. Acknowledgements This work was supported by the 863 high technology research program of China (Grant No. 2006AA05Z405). References Electronic address: [1] [2] P. D. Maycock, PV News 22, 4 (2003). [3] [4] C. Voz et al., Thin Solid Films , 415 (2006). [5] Y. Tsunomura et al., Sol. Energy Mater. Sol. Cells 93, 670 (2009). [6] Y. Xu et al., J. Non-Cryst. Solids 352, 1972 (2006). [7] K. v Maydell et al., J. Non-Cryst. Solids 352, 1958 (2006). [8] J. Wang et al., Mat. Sci. Eng. B 72, 193 (2000). [9] H. Angermann, W. Henrion, M. Rebiena, and A. Röseler, Solar Energy Mat. & Solar Cells 83, 331 (2004). [10] R. Barrio, C. Maffiotte, J. J. Gandía, and J. Cárabe, J. Non-Cryst. Solids 352, 945 (2006). [11] W. Henriona, M. Rebiena, H. Angermanna, and A. Röseler, Appl. Surf. Sci. 202, 199 (2002). [12] D. Graf, M. Grundner, and R. Schulz, J. Appl. Phys. 68, 5155 (1990). [13] S. I. Raider, R. Flitsh, and M. J. Palmer, J. Electrochem. Soc. 122, 413 (1975). [14] G. Beck and M. Kunst, Rev. Sci. Instrum. 57, 197 (1988). [15] M. Schofthaler and R. Bendei, J. Appl. Phys. 77, 3162 (1995). [16] H. Angermann, Anal. Bioanal. Chem. 374, 676 (2002). [17] G. F. Cerofolini, D. Mascolo, and M. O. Vlad, J. Appl. Phys. 100, (2006). [18] T. Yasaka., K. Kanda, K. Sawara, S. Miyazaki, and M. Hirose, Jpn. J. Appl. Phys. 30, 3567 (1991).