- Email: [email protected]
Influences of oxygen contamination on evaporated poly-Si thin-film solar cells by solid-phase epitaxy
Thin Solid Films 518 (2010) 4351–4355
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Influences of oxygen contamination on evaporated poly-Si thin-film solar cells by solid-phase epitaxy Song He a,⁎, Johnson Wong a, Daniel Inn b, Bram Hoex c, Armin G. Aberle c, Alistair B. Sproul a a b c
ARC Photovoltaics Centre of Excellence, The University of New South Wales (UNSW), Sydney, NSW 2052, Australia IBM-Thomas J. Watson Research Center, Yorktown Heights, NY, USA Solar Energy Research Institute of Singapore and National University of Singapore, Singapore 117576, Singapore
a r t i c l e
i n f o
Article history: Received 11 January 2009 Received in revised form 7 January 2010 Accepted 12 January 2010 Available online 18 January 2010 Keywords: Evaporated poly-Si thin-film solar cell Solid-phase epitaxy Oxygen contamination Deposition rate Base pressure
a b s t r a c t The influences of the oxygen contaminations on the crystal quality and performances of the evaporated polycrystalline silicon (poly-Si) thin-film solar cells prepared by solid-phase epitaxy were investigated by applying different deposition rates and base pressures. The experimental results show that although the evaporated poly-Si thin-film solar cell obtained at high base pressures (9.33 × 10− 5 Pa) and high deposition rate (300 nm/min) has small amount of SiO2 precipitations, it still shows the similar good material quality and performances as the cell prepared at low base pressure (1.33 × 10− 6 Pa) and high deposition rate (300 nm/min) with oxygen interstitials. On the other hand, the poly-Si thin-film solar cell deposited at low base pressure (1.33 × 10− 6 Pa) and low deposition rate (50 nm/min) has large amount of SiO2 precipitations and resulting worse material quality and hence cell performances. Therefore, the high deposition rate is desirable to maximize the solar cell performance, as well as the throughput. It is a more influential factor than the base pressure. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Polycrystalline silicon (Poly-Si) thin-film solar cells are becoming increasingly important in recent years due to their potential of significantly reduced manufacturing cost when compared to Si wafer based solar cells [1,2]. At the University of New South Wales, poly-Si thin-film solar cells are being developed utilizing solid-phase epitaxy (SPE) of a-Si films by both e-beam evaporation [3] and plasma-enhanced chemical vapor deposition (PECVD) [4]. Comparing with the PECVD method, the key advantage of e-beam evaporation is the exceptionally high Si deposition rate of up to 1 μm/min, which is obviously beneficial from a cost point of view due to the capability of high throughput. Additional advantages of the method include the absence of toxic gases, optimal Si source material usage, and compatibility with a continuous in-line deposition mode. Thus, if performed in a non-ultra-high vacuum (nonUHV) environment (base pressure> 1.33 × 10− 6 Pa, pressure during Si evaporation > 1.33 × 10− 5 Pa), e-beam evaporation could potentially be a low-cost Si deposition method for thin-film solar cells. The base pressure and Si evaporation rate are the two key parameters during deposition, which can affect impurities (e.g. oxygen) and defect densities in films and hence the crystal quality and resulting minority carrier lifetime of the poly-Si thin-film solar cells. It has been established that Czochralski Si wafer solar cells still exhibit a high minority carrier
⁎ Corresponding author. E-mail address: [email protected] (S. He). 0040-6090/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2010.01.017
lifetime with oxygen concentration levels of up to several 1018 cm− 3 as a result of strong Si–O bonds that saturate the dangling bonds [5]. In poly-Si thin-film solar cells, high short-circuit current densities can be achieved with a small minority carrier diffusion length in the range of the film thickness of ∼2 µm in conjunction with good light trapping [6]. Moreover, the poly-Si thin-film solar cells are more than two orders of magnitude thinner than Si wafer solar cells. Therefore, the tolerable oxygen contamination levels could be much higher than in Si wafer solar cells. In this work, different base pressures and deposition rates will be tested and the impact of the oxygen contaminations on the crystal quality and performance of the resulting SPE poly-Si thin-film solar cells will be investigated using various in-situ and ex-situ characterization techniques. 2. Experimental details Initial experiments were conducted with SPE poly-Si films on double-side polished intrinsic (100) float zone Si wafers for the purpose of Fourier Transformation Infrared Spectroscopy (FTIR) measurement. Prior to the Si evaporation process, the Si wafers were chemically cleaned with piranha solution (a 1:1 mixture of H2SO4 and H2O2), followed by a dip into 5% HF to remove the surface oxide. The a-Si films of 1.5 µm thickness were then deposited at either low rate (50 nm/min) or high rate (300 nm/min) by e-beam evaporation, using a substrate temperature in the range of 200–220 °C. The base pressures of the evaporator was ∼1.33 × 10− 6 Pa for the low base pressure setting, and deliberately increase up to ∼9.33× 10− 5 Pa with continuous air flow via
4352
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
a capillary valve for the high base pressure setting. Three sets of Si thin films were prepared with the following conditions: (1) high deposition rate and low base pressure; (2) high deposition rate and high base pressure; (3) low deposition rate and low base pressure. After the depositions, the a-Si thin film samples were transferred in air to a tube furnace for crystallization at 575 °C for 24 h with continuous nitrogen flow. In order to observe the impacts of base pressure and deposition rate on cell performances, poly-Si thin-film solar cells were fabricated with the lightly-doped absorber layers being deposited at different based pressures and deposition rates by aluminium-induced crystallization epitaxy (ALICE). The concept of ALICE comes from the SPE poly-Si thin film on seed layer by aluminium-induced crystallization (AIC). Both heavily-doped emitter and back surface field (BSF) layers were deposited at a low rate due to the heating limitation of doping sources. The settings of the base pressures and deposition rates corresponded to the above experiments with poly-Si thin films on Si wafers. The structure of the ALICE cells is glass (superstrate configuration)/SiN(70 nm)/p+ (75 nm, ∼1 × 1019 cm− 3, Al, AIC seed layer)/p+ (50 nm, ∼1 × 1019 cm− 3, B, emitter)/n (1500 nm, ∼8 × 1016 cm− 3, P, absorber)/n+ (100 nm, ∼1 × 1020 cm− 3, P, BSF). Al, B, P stand for aluminium, boron and phosphorous doping, respectively. After crystallization in tube furnace, the ALICE cells received a rapid thermal annealing (RTA) process at 900 °C for 4 min to activate doping and reduce the defect density. Then, high-temperature hydrogenation treatments were applied to the ALICE cells to passivate the grain boundaries in the poly-Si thin films. The hydrogenation processes were performed in a cold-wall vacuum system featuring an inductively coupled remote plasma source with glass temperature of 610 °C for 20 min and plasma power of 3300 W, hydrogen flow of 200 sccm and argon flow of 60 sccm. The hydrogenated ALICE cells were bifacially metallized by contacting the emitter layers and BSF layers with 900 nm Al separately. This metallization scheme has less than 5% Al shading as the air-side contact in order to be compatible with back surface reflectors. The total size of the test structure is 1.25 × 0.8 cm2 [7]. A final phosphoric etching was applied to the metallized ALICE cells to eliminate the local shunts by removing the Al residues left in the voids of poly-Si films. The partial pressures of background gases during the deposition process were monitored in-situ using residual gas analyzer (RGA) from Stanford Instruments. The bonded oxygen contents in the polySi films were measured by FTIR, using a Nicolet 5700 system. The UV reflectance measurements were applied with a Varian Cary 5G double-beam spectrophotometer in the wavelength range from 250 nm to 400 nm to determine the crystal quality of poly-Si films. Moreover, cross-sectional Transmission Electron Microscopy (XTEM) was used to reveal the structural quality of poly-Si films by Philips CM12 and CM200. Open-circuit voltage (Voc) of ALICE cells was measured using a quasisteady-state Voc (Sun-Voc) method. Sun-Voc curves were fitted with a two-diode model, (ideality factors n = 1, n = 2) which includes a shunt resistance, illuminated current–voltage (I–V) curves. External quantum efficiency (EQE) measurements of the ALICE solar cells were performed on an Oriel system, with a wavelength step size of 5 nm. The EQE spectra were converted to internal quantum efficiency (IQE) spectra by calculating with the total hemispherical reflectance. PC1D simulations were applied to calculate the minority carrier diffusion lengths. 3. Experimental results and discussions 3.1. Vapor compositions in the non-UHV Si evaporator Using a RGA, the partial pressures of selected molecules (N2, O2, H2, and H2O) were monitored during a standard a-Si evaporation process onto a SiN-coated glass substrate. The base pressure of the Si evaporator just prior to the deposition was about 3.99 × 10− 6 Pa. The pump system on the Si evaporator chamber consists of a mechanical
Fig. 1. Partial pressures of selected background gases (H2, N2, O2, and H2O) in the vacuum chamber during the a-Si precursor thin-film deposition process on SiN-coated glass.
roughing pump and a liquid helium-cooled cryogenic pump. The Si deposition occurs in a stationary mode (i.e., non-moving substrate). The results obtained with the RGA are shown in Fig. 1. It can be seen that the base pressure of the system, prior to the loading of the sample, is dominated by the presence of N2 gas. Upon insertion of the SiN-coated glass substrate into the evaporation chamber via a loadlock and switching on of the heaters, the sample mainly releases H2 and H2O. Switching on the e-beam eliminates O2 and H2O from the gas phase, but leaves the N2 concentration unaffected. This shows that O2 and H2O are very efficiently gettered by the Si vapor and are thus incorporated into the deposited Si film. The partial pressure of H2 increases drastically during the Si melting process as H2 releases from the Si pellets in the crucible. H2 is considered to be benign impurity in Si, hence this pressure increase during Si evaporation is not expected to have a detrimental effect on the Si thin film material quality. The RGA clearly is a very useful tool for the optimization and the control of the Si evaporation process. 3.2. Oxygen contents in Si films FTIR measurements are non-destructive and can yield information on impurities in poly-Si thin films. It can distinguish the difference between interstitial (Oi) and precipitated SiO2 cluster, which cannot be performed in SIMS. Furthermore, the oxygen detection limit for FTIR is about 1016 atoms/cm3, which is one order of magnitude better than SIMS detection limit (∼1017 atom/cm− 3). Oxygen can dramatically affect the charge carrier diffusion length by either acting as recombination center (e.g., a B–O complex) [8] or by disrupting the Si crystal by inclusion of SiO2-like clusters [9,10]. FTIR was used to identify the oxygen contents in Si films on intrinsic Si wafers prepared at different deposition rates and base pressures. Prior to the measurements, the Si films were dipped into a 5% HF solution to remove the surface SiO2 layer. From Fig. 2, it can be seen that there is acceptable interstitial oxygen [11] in poly-Si thin films prepared at high deposition rate and low base pressure, while detrimental SiO2 precipitates [12] in films prepared at either high rate/high pressure or low rate/low pressure. It is noted that the high base pressure in this experiment is 17 times higher than low base pressure, while low deposition rate is 6 times lower than high deposition rate. This indicates that most of oxygen was not gettered by Si vapor during deposition in the high base pressure environment. This phenomenon is confirmed by RGA showing that the oxygen partial pressures are consistent during the deposition process. It is most likely that the deposition rate of Si is so high at 300 nm/min that oxygen has much less time/chance to be gettered by the Si vapor although there is plenty of oxygen in the chamber. Additional experiments show that the oxygen content levels are proportional to the base pressure if the deposition rate is
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
Fig. 2. FTIR absorbance spectrum of c-Si films on intrinsic Si wafer substrates prepared at different deposition rates and base pressures. Absorption peaks related to SiO2 and Oi can clearly be distinguished. The samples were cleaned in diluted HF prior to the measurement to remove the surface oxide.
decreased simultaneously. On the other hand, the Si vapor has more time to getter the oxygen in the chamber at low deposition rate, which causes the oxygen content level to be inversely proportional to the deposition rate. 3.3. Crystal quality of c-Si films on Si wafers The crystal quality of c-Si films obtained from evaporated a-Si films is possibly strongly affected by oxygen contaminations in the a-Si. Therefore, UV reflectance measurement is used to observe the possible crystal quality different of the Si films on Si wafer presented in Section 3.2. The probed sample region of UV light has a typical area of 106 µm2 and hence the UV reflectance measurement determines the spatially averaged crystal quality of the Si films, particularly at the wavelengths corresponding to the characteristic UV reflection e1 and e2 peaks of c-Si. Structural disorder in the surface region of the Si film can cause a broadening and height reduction of the e1 and e2 peaks. The crystal quality figure of merit (Q) is calculated according to [13]: Q =
1 Re1 Re2 + 2 Re1 · cSi Re2 · cSi
ð1Þ
Re1 and Re2 are the values of the UV reflectance values of the investigated c-Si film at the e1 and e2 peaks, respectively. Re1.cSi and Re2. cSi are the values of the UV reflectance of a high-quality single crystalline Si wafer at e1 and e2 peaks, respectively. Fig. 3 shows the measured UV reflectance of the c-Si films obtained from SPE of a-Si evaporated onto intrinsic Si wafers at different deposition rates and base pressures. The cSi thin films obtained from high rate and low base pressure has high Q value of 98.1%, whilst the c-Si thin films obtained at high rate and high base pressure has a slightly smaller Q value of 96.9% (The weaker e2 peak response might be related to the surface oxide layer, hence the calculated Q value is slightly lower bounded.), while the poly-Si film obtained from low rate and low base pressure has a much worse Q value of only 92.0%. The UV reflectance measurements show that the high rate deposited films at either high or low base pressures have similar good crystal quality, while film deposited at low rate and low base pressure has much worse crystal quality. 3.4. XTEM analysis of c-Si films XTEM analysis was taken to assess the grain structure of Si films on Si wafers discussed in Sections 3.2 and 3.3. It is noted that the XTEM sample preparation method for the sample in Fig. 4(b) is different
4353
Fig. 3. UV reflectance measurements of c-Si films on intrinsic Si wafer substrates obtained at different deposition rates and base pressures, with Si wafer as reference.
from the method used for samples in Fig. 4(a) and (c). The XTEM sample in Fig. 4(b) is prepared using focused ion beam (FIB), while traditional polishing method is used for the preparation of samples in Fig. 4(a) and (c). Fig. 4(a) shows XTEM image of the sample prepared at low rate and low base pressure. Lots of the fine-grained poly-crystal with amplitude contrast from the small crystallites show up as dark regions in the bright field image. The epitaxial quality of Si film is generally rather poor, with a high density of structural defects. It is believed that the large amount of SiO2 clusters in the low rate deposited films retard the crystallization process of a-Si films, leading to a disorder of the Si atom structure. Fig. 4(b) shows the XTEM image of the sample prepared at high low base pressure and high deposition rate. The Pt coating on the surface of Si film is for the Si film protection during FIB milling process. The dark lines in Si wafer are bend contours which arise from a locally bent wafer foil because the sample is thin and tend to flex slightly. In the Si epitaxial layer, some fringes could be end contours and others could be grain boundaries or other types of defect. Overall, the Si film looks predominantly single crystal despite of the detrimental SiO2 precipitations in the film, indicating that a small amount of SiO2 precipitation does not significantly affect the crystal quality of Si film. Fig. 4(c) shows the XTEM image of the sample prepared at high deposition rate and low base pressure. The local distortion of the crystal around the defect brings some crystal planes into Bragg condition and hence efficient diffraction of these electrons obtained. Consequently, the dark contrast is seen in the image. The Si film is also single crystal dominated, similar with the film in Fig. 4(b). The oxygen interstitials do not affect the lattice structure of Si films. There is a very good agreement between the UV reflectance measurement and XTEM imaging. 3.5. Electrical properties of the evaporated ALICE solar cells prepared at different deposition rates and base pressures Sun-Voc measurements show that the hydrogenated ALICE cells at different deposition conditions have the effective ideality factors neff of 1.8–2.0 (Fig. 5), meaning that half of the minority carrier recombination occurs in the bulk regions of the absorber layers and the surfaces, the other half happens in the depletion regions of the PN junctions [14]. Hydrogen is still quite effective to passivate the massive amount of dangling bonds and other types of defects at the grain boundaries even for ALICE prepared at low deposition rate. Fig. 6 and Table 1 show the I–V performance of ALICE cells prepared at different deposition conditions. The poor fill factors (FF) of ∼ 56% are partially due to the high series resistances presented by the very thin Al electrode (∼ 100 nm) as the Pseudo FF (FF without
4354
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
Fig. 5. Sun-Voc curves with two-diode model fittings. There are very good fits of Voc, V1, and V2 curves, indicating that the two-diode model is effective.
Fig. 6. I–V curves of the ALICE cells prepared at different deposition rates and base pressures.
Table 1 Electrical properties of ALICE solar cells prepared at different deposition rates and base pressures. Types of ALICE cells
Voc (mV)
Jsc (mA/cm2)
neff
FF (%)
ŋ (%)
High R low P High R high P Low R low P
437 436 394
9.7 9.6 8.5
2.0 1.9 1.8
55.9 55.8 55.9
2.26 2.25 1.99
pressure have the worse performance. Considering with the oxygen content levels determined with the Si thin films on Si wafers, it looks that ALICE solar cells can tolerate oxygen level up to 3 × 1018 cm− 3 without deteriorating the performance. It must be noted that the oxygen content distributions in ALICE solar cells on glass is more complicated than those in Si films on Si wafers.
3.6. Spectral response measurement of ALICE cells Fig. 4. XTEM images of c-Si thin films on intrinsic Si wafer substrates prepared at (a) low deposition rate and low base pressure; (b) high deposition rate and high base pressure; and (c) high deposition rate and low base pressure.
calculating the series resistance) of ALICE cells are ∼ 66%. The cell performances (Voc, Jsc, ŋ) appear to be corresponded to the material qualities of poly-Si thin films. It shows that high rate deposited ALICE cells at either high or low base pressures have similar and better performances, while ALICE cells prepared at low rate and low base
The spectral response measurements of the ALICE cells are presented in Fig. 7. The three types of ALICE cells investigated in this work have strong absorption in the visible wavelengths, but weak absorption in infrared region due to the small thickness of poly-Si films and lack of back surface reflectors. They all have similar peak positions at ∼ 490 nm, indicating that the emitter thicknesses for these ALICE cells are similar. The high rate ALICE cells prepared at either high or low pressures are very similar and have the peak IQE of 48.8%, while ALICE cells prepared at low rate and low pressure have a lower
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
4355
Fig. 7. Spectral response measurements of ALICE cells prepared at different deposition rates and base pressures.
peak IQE of 45.7% as a result of more minority recombination in the structural defects and oxygen recombination centers.
Fig. 8. Example of the experimental EQE and IQE of ALICE cell with the measured front surface reflectance, together with PC1D simulation curves.
3.7. Diffusion lengths of ALICE cells The minority carrier diffusion lengths in absorber layers of the ALICE cells were calculated via the one-dimensional PC1D simulations [15] of spectral response spectra (Fig. 8). For these fits, a three-region device (vertical stack) is used where each region represents emitter (including AIC), absorber and BSF layers, respectively. The main fit parameters in PC1D simulations are the defect (i.e. Shockley–Read–Hall recombination) related bulk lifetimes in the three regions. The material parameters of mono-crystalline Si were assumed. The mobility of free carriers in poly-Si materials is significantly lower than in high-purity monocrystalline Si and is dependent on grain size and grain boundary properties. Due to this assumption in the simulations, the diffusion lengths obtained are lower bounds to the actual values. The PC1D simulation results show that high rate deposited ALICE cells prepared at either high or low base pressure have the minority carrier diffusion length of 1.01 µm, while ALICE cells prepared at low rate and low base pressure has diffusion lengths of 0.93 µm, respectively. From the above results, it appears that the high deposition rate is much more critical than the low base pressure in determining the material quality and the resulting performance of the SPE evaporated poly-Si thin-film solar cells. It is found that the crystal quality of evaporated poly-Si thin film goes to saturation when the deposition rate is higher than 300 nm/min. It is easy to realize the high rate deposition for the lightly-doped absorber layer, however, the low deposition rate has to be applied to the heavily-doped emitter layers and BSF layers due to the heating limitation of the dopant sources. Efforts need to be taken to increase the deposition rate for the emitter and BSF layers for better material quality, which could potentially improve the poly-Si thin-film solar cells. Furthermore, the upper limit for the base pressure which has no or little impact on the crystal quality and cell performance of evaporated poly-Si thin-film solar cells needs to be determined in the future. 4. Conclusions Evaporated poly-Si thin-film solar cells obtained at high base pressures and high deposition rate have small amount of SiO2 precipitations, however still show the similar good material quality and performances as the cells prepared at low base pressure and high
deposition rate with oxygen interstitials. On the other hand, the poly-Si thin-film solar cells deposited at low deposition rate have large amount of SiO2 precipitations and resulting worse material quality and cell performances. Therefore, during a-Si thin-film deposition at non-UHV environment, the low base pressure is not very critical as long as the high evaporation rate can be obtained. This can significantly reduce the capital cost on the vacuum equipment which can directly reduce the manufacturing cost of the poly-Si thin-film solar cells.
Acknowledgement Song He acknowledges a Ph.D. scholarship from the Faculty of Engineering at The University of New South Wales (UNSW). Song He also would like to give thanks to Xiaojing Hao, Tom Puzzer, Charlie Kong and Yidan Huang for their assistance on the measurements of FTIR and TEM imaging.
References [1] P.A. Basore, Proc. 29th IEEE Photovoltaic Specialists Conference, New Orleans, U.S.A, 2002, p. 49. [2] M.J. Keevers, T.L. Young, U. Schubert, M.A. Green, Proc. 22nd European Photovoltaic Solar Energy Conference, Milan, Italy, 2007, p. p.1783. [3] S. He, B. Hoex, D. Inns, I.C. Brazil, P.I. Widenborg, A.G. Aberle, Sol. Energy. Mater. Sol. Cells 93 (2009) 1116. [4] P.I. Widenborg, A. Straub, A.G. Aberle, J. Cryst. Growth 276 (2005) 19. [5] R. Brendel, Thin-Film Crystalline Silicon Solar Cells, Wiley-VCH, Berlin, 2003. [6] P.A. Basore, Proceeding 3rd World Conference on Photovoltaic Solar Energy Conversation, Osaka, Japan, 2003, p. 935. [7] O. Kunz, J. Wong, J. Janssens, J. Bauer, O. Breitenstein, A.G. Aberle, Prog. Photovolt. Res. Appl. 17 (2009) 35. [8] K. Bothe, R. Sinton, J. Schmidt, Prog. Photovolt. Res. Appl. 13 (2005) 287. [9] A. Borghesi, B. Pivac, A. Sassella, A. Stella, J. Appl. Phys. 77 (1995) 4169. [10] P. Gaworzewski, E. Hild, F.G. Kirscht, L. Vecsernyes, Phys. Stat. Sol. (a) 85 (1984) 133. [11] C.T. Kirk, Phys. Rev. B 38/2 (1988) 1255. [12] W.S. Lau, Infrared Characterization for Microelectronics, World Scientific Publishing Co., Singapore, 1999. [13] A. Straub, P.I. Widenborg, A. Sproul, Y. Huang, N.P. Harder, A.G. Aberle, J. Cryst. Growth 265 (2004) 168. [14] R.A. Sinton, A. Cuevas, 16th European Photovoltaic Solar Energy Conference, Glasgow, UK, 2000, p. 124. [15] P.A. Basore, IEEE Trans. Electron Devices 37 (1990) 337.
Contents lists available at ScienceDirect
Thin Solid Films j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / t s f
Influences of oxygen contamination on evaporated poly-Si thin-film solar cells by solid-phase epitaxy Song He a,⁎, Johnson Wong a, Daniel Inn b, Bram Hoex c, Armin G. Aberle c, Alistair B. Sproul a a b c
ARC Photovoltaics Centre of Excellence, The University of New South Wales (UNSW), Sydney, NSW 2052, Australia IBM-Thomas J. Watson Research Center, Yorktown Heights, NY, USA Solar Energy Research Institute of Singapore and National University of Singapore, Singapore 117576, Singapore
a r t i c l e
i n f o
Article history: Received 11 January 2009 Received in revised form 7 January 2010 Accepted 12 January 2010 Available online 18 January 2010 Keywords: Evaporated poly-Si thin-film solar cell Solid-phase epitaxy Oxygen contamination Deposition rate Base pressure
a b s t r a c t The influences of the oxygen contaminations on the crystal quality and performances of the evaporated polycrystalline silicon (poly-Si) thin-film solar cells prepared by solid-phase epitaxy were investigated by applying different deposition rates and base pressures. The experimental results show that although the evaporated poly-Si thin-film solar cell obtained at high base pressures (9.33 × 10− 5 Pa) and high deposition rate (300 nm/min) has small amount of SiO2 precipitations, it still shows the similar good material quality and performances as the cell prepared at low base pressure (1.33 × 10− 6 Pa) and high deposition rate (300 nm/min) with oxygen interstitials. On the other hand, the poly-Si thin-film solar cell deposited at low base pressure (1.33 × 10− 6 Pa) and low deposition rate (50 nm/min) has large amount of SiO2 precipitations and resulting worse material quality and hence cell performances. Therefore, the high deposition rate is desirable to maximize the solar cell performance, as well as the throughput. It is a more influential factor than the base pressure. © 2010 Elsevier B.V. All rights reserved.
1. Introduction Polycrystalline silicon (Poly-Si) thin-film solar cells are becoming increasingly important in recent years due to their potential of significantly reduced manufacturing cost when compared to Si wafer based solar cells [1,2]. At the University of New South Wales, poly-Si thin-film solar cells are being developed utilizing solid-phase epitaxy (SPE) of a-Si films by both e-beam evaporation [3] and plasma-enhanced chemical vapor deposition (PECVD) [4]. Comparing with the PECVD method, the key advantage of e-beam evaporation is the exceptionally high Si deposition rate of up to 1 μm/min, which is obviously beneficial from a cost point of view due to the capability of high throughput. Additional advantages of the method include the absence of toxic gases, optimal Si source material usage, and compatibility with a continuous in-line deposition mode. Thus, if performed in a non-ultra-high vacuum (nonUHV) environment (base pressure> 1.33 × 10− 6 Pa, pressure during Si evaporation > 1.33 × 10− 5 Pa), e-beam evaporation could potentially be a low-cost Si deposition method for thin-film solar cells. The base pressure and Si evaporation rate are the two key parameters during deposition, which can affect impurities (e.g. oxygen) and defect densities in films and hence the crystal quality and resulting minority carrier lifetime of the poly-Si thin-film solar cells. It has been established that Czochralski Si wafer solar cells still exhibit a high minority carrier
⁎ Corresponding author. E-mail address: [email protected] (S. He). 0040-6090/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.tsf.2010.01.017
lifetime with oxygen concentration levels of up to several 1018 cm− 3 as a result of strong Si–O bonds that saturate the dangling bonds [5]. In poly-Si thin-film solar cells, high short-circuit current densities can be achieved with a small minority carrier diffusion length in the range of the film thickness of ∼2 µm in conjunction with good light trapping [6]. Moreover, the poly-Si thin-film solar cells are more than two orders of magnitude thinner than Si wafer solar cells. Therefore, the tolerable oxygen contamination levels could be much higher than in Si wafer solar cells. In this work, different base pressures and deposition rates will be tested and the impact of the oxygen contaminations on the crystal quality and performance of the resulting SPE poly-Si thin-film solar cells will be investigated using various in-situ and ex-situ characterization techniques. 2. Experimental details Initial experiments were conducted with SPE poly-Si films on double-side polished intrinsic (100) float zone Si wafers for the purpose of Fourier Transformation Infrared Spectroscopy (FTIR) measurement. Prior to the Si evaporation process, the Si wafers were chemically cleaned with piranha solution (a 1:1 mixture of H2SO4 and H2O2), followed by a dip into 5% HF to remove the surface oxide. The a-Si films of 1.5 µm thickness were then deposited at either low rate (50 nm/min) or high rate (300 nm/min) by e-beam evaporation, using a substrate temperature in the range of 200–220 °C. The base pressures of the evaporator was ∼1.33 × 10− 6 Pa for the low base pressure setting, and deliberately increase up to ∼9.33× 10− 5 Pa with continuous air flow via
4352
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
a capillary valve for the high base pressure setting. Three sets of Si thin films were prepared with the following conditions: (1) high deposition rate and low base pressure; (2) high deposition rate and high base pressure; (3) low deposition rate and low base pressure. After the depositions, the a-Si thin film samples were transferred in air to a tube furnace for crystallization at 575 °C for 24 h with continuous nitrogen flow. In order to observe the impacts of base pressure and deposition rate on cell performances, poly-Si thin-film solar cells were fabricated with the lightly-doped absorber layers being deposited at different based pressures and deposition rates by aluminium-induced crystallization epitaxy (ALICE). The concept of ALICE comes from the SPE poly-Si thin film on seed layer by aluminium-induced crystallization (AIC). Both heavily-doped emitter and back surface field (BSF) layers were deposited at a low rate due to the heating limitation of doping sources. The settings of the base pressures and deposition rates corresponded to the above experiments with poly-Si thin films on Si wafers. The structure of the ALICE cells is glass (superstrate configuration)/SiN(70 nm)/p+ (75 nm, ∼1 × 1019 cm− 3, Al, AIC seed layer)/p+ (50 nm, ∼1 × 1019 cm− 3, B, emitter)/n (1500 nm, ∼8 × 1016 cm− 3, P, absorber)/n+ (100 nm, ∼1 × 1020 cm− 3, P, BSF). Al, B, P stand for aluminium, boron and phosphorous doping, respectively. After crystallization in tube furnace, the ALICE cells received a rapid thermal annealing (RTA) process at 900 °C for 4 min to activate doping and reduce the defect density. Then, high-temperature hydrogenation treatments were applied to the ALICE cells to passivate the grain boundaries in the poly-Si thin films. The hydrogenation processes were performed in a cold-wall vacuum system featuring an inductively coupled remote plasma source with glass temperature of 610 °C for 20 min and plasma power of 3300 W, hydrogen flow of 200 sccm and argon flow of 60 sccm. The hydrogenated ALICE cells were bifacially metallized by contacting the emitter layers and BSF layers with 900 nm Al separately. This metallization scheme has less than 5% Al shading as the air-side contact in order to be compatible with back surface reflectors. The total size of the test structure is 1.25 × 0.8 cm2 [7]. A final phosphoric etching was applied to the metallized ALICE cells to eliminate the local shunts by removing the Al residues left in the voids of poly-Si films. The partial pressures of background gases during the deposition process were monitored in-situ using residual gas analyzer (RGA) from Stanford Instruments. The bonded oxygen contents in the polySi films were measured by FTIR, using a Nicolet 5700 system. The UV reflectance measurements were applied with a Varian Cary 5G double-beam spectrophotometer in the wavelength range from 250 nm to 400 nm to determine the crystal quality of poly-Si films. Moreover, cross-sectional Transmission Electron Microscopy (XTEM) was used to reveal the structural quality of poly-Si films by Philips CM12 and CM200. Open-circuit voltage (Voc) of ALICE cells was measured using a quasisteady-state Voc (Sun-Voc) method. Sun-Voc curves were fitted with a two-diode model, (ideality factors n = 1, n = 2) which includes a shunt resistance, illuminated current–voltage (I–V) curves. External quantum efficiency (EQE) measurements of the ALICE solar cells were performed on an Oriel system, with a wavelength step size of 5 nm. The EQE spectra were converted to internal quantum efficiency (IQE) spectra by calculating with the total hemispherical reflectance. PC1D simulations were applied to calculate the minority carrier diffusion lengths. 3. Experimental results and discussions 3.1. Vapor compositions in the non-UHV Si evaporator Using a RGA, the partial pressures of selected molecules (N2, O2, H2, and H2O) were monitored during a standard a-Si evaporation process onto a SiN-coated glass substrate. The base pressure of the Si evaporator just prior to the deposition was about 3.99 × 10− 6 Pa. The pump system on the Si evaporator chamber consists of a mechanical
Fig. 1. Partial pressures of selected background gases (H2, N2, O2, and H2O) in the vacuum chamber during the a-Si precursor thin-film deposition process on SiN-coated glass.
roughing pump and a liquid helium-cooled cryogenic pump. The Si deposition occurs in a stationary mode (i.e., non-moving substrate). The results obtained with the RGA are shown in Fig. 1. It can be seen that the base pressure of the system, prior to the loading of the sample, is dominated by the presence of N2 gas. Upon insertion of the SiN-coated glass substrate into the evaporation chamber via a loadlock and switching on of the heaters, the sample mainly releases H2 and H2O. Switching on the e-beam eliminates O2 and H2O from the gas phase, but leaves the N2 concentration unaffected. This shows that O2 and H2O are very efficiently gettered by the Si vapor and are thus incorporated into the deposited Si film. The partial pressure of H2 increases drastically during the Si melting process as H2 releases from the Si pellets in the crucible. H2 is considered to be benign impurity in Si, hence this pressure increase during Si evaporation is not expected to have a detrimental effect on the Si thin film material quality. The RGA clearly is a very useful tool for the optimization and the control of the Si evaporation process. 3.2. Oxygen contents in Si films FTIR measurements are non-destructive and can yield information on impurities in poly-Si thin films. It can distinguish the difference between interstitial (Oi) and precipitated SiO2 cluster, which cannot be performed in SIMS. Furthermore, the oxygen detection limit for FTIR is about 1016 atoms/cm3, which is one order of magnitude better than SIMS detection limit (∼1017 atom/cm− 3). Oxygen can dramatically affect the charge carrier diffusion length by either acting as recombination center (e.g., a B–O complex) [8] or by disrupting the Si crystal by inclusion of SiO2-like clusters [9,10]. FTIR was used to identify the oxygen contents in Si films on intrinsic Si wafers prepared at different deposition rates and base pressures. Prior to the measurements, the Si films were dipped into a 5% HF solution to remove the surface SiO2 layer. From Fig. 2, it can be seen that there is acceptable interstitial oxygen [11] in poly-Si thin films prepared at high deposition rate and low base pressure, while detrimental SiO2 precipitates [12] in films prepared at either high rate/high pressure or low rate/low pressure. It is noted that the high base pressure in this experiment is 17 times higher than low base pressure, while low deposition rate is 6 times lower than high deposition rate. This indicates that most of oxygen was not gettered by Si vapor during deposition in the high base pressure environment. This phenomenon is confirmed by RGA showing that the oxygen partial pressures are consistent during the deposition process. It is most likely that the deposition rate of Si is so high at 300 nm/min that oxygen has much less time/chance to be gettered by the Si vapor although there is plenty of oxygen in the chamber. Additional experiments show that the oxygen content levels are proportional to the base pressure if the deposition rate is
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
Fig. 2. FTIR absorbance spectrum of c-Si films on intrinsic Si wafer substrates prepared at different deposition rates and base pressures. Absorption peaks related to SiO2 and Oi can clearly be distinguished. The samples were cleaned in diluted HF prior to the measurement to remove the surface oxide.
decreased simultaneously. On the other hand, the Si vapor has more time to getter the oxygen in the chamber at low deposition rate, which causes the oxygen content level to be inversely proportional to the deposition rate. 3.3. Crystal quality of c-Si films on Si wafers The crystal quality of c-Si films obtained from evaporated a-Si films is possibly strongly affected by oxygen contaminations in the a-Si. Therefore, UV reflectance measurement is used to observe the possible crystal quality different of the Si films on Si wafer presented in Section 3.2. The probed sample region of UV light has a typical area of 106 µm2 and hence the UV reflectance measurement determines the spatially averaged crystal quality of the Si films, particularly at the wavelengths corresponding to the characteristic UV reflection e1 and e2 peaks of c-Si. Structural disorder in the surface region of the Si film can cause a broadening and height reduction of the e1 and e2 peaks. The crystal quality figure of merit (Q) is calculated according to [13]: Q =
1 Re1 Re2 + 2 Re1 · cSi Re2 · cSi
ð1Þ
Re1 and Re2 are the values of the UV reflectance values of the investigated c-Si film at the e1 and e2 peaks, respectively. Re1.cSi and Re2. cSi are the values of the UV reflectance of a high-quality single crystalline Si wafer at e1 and e2 peaks, respectively. Fig. 3 shows the measured UV reflectance of the c-Si films obtained from SPE of a-Si evaporated onto intrinsic Si wafers at different deposition rates and base pressures. The cSi thin films obtained from high rate and low base pressure has high Q value of 98.1%, whilst the c-Si thin films obtained at high rate and high base pressure has a slightly smaller Q value of 96.9% (The weaker e2 peak response might be related to the surface oxide layer, hence the calculated Q value is slightly lower bounded.), while the poly-Si film obtained from low rate and low base pressure has a much worse Q value of only 92.0%. The UV reflectance measurements show that the high rate deposited films at either high or low base pressures have similar good crystal quality, while film deposited at low rate and low base pressure has much worse crystal quality. 3.4. XTEM analysis of c-Si films XTEM analysis was taken to assess the grain structure of Si films on Si wafers discussed in Sections 3.2 and 3.3. It is noted that the XTEM sample preparation method for the sample in Fig. 4(b) is different
4353
Fig. 3. UV reflectance measurements of c-Si films on intrinsic Si wafer substrates obtained at different deposition rates and base pressures, with Si wafer as reference.
from the method used for samples in Fig. 4(a) and (c). The XTEM sample in Fig. 4(b) is prepared using focused ion beam (FIB), while traditional polishing method is used for the preparation of samples in Fig. 4(a) and (c). Fig. 4(a) shows XTEM image of the sample prepared at low rate and low base pressure. Lots of the fine-grained poly-crystal with amplitude contrast from the small crystallites show up as dark regions in the bright field image. The epitaxial quality of Si film is generally rather poor, with a high density of structural defects. It is believed that the large amount of SiO2 clusters in the low rate deposited films retard the crystallization process of a-Si films, leading to a disorder of the Si atom structure. Fig. 4(b) shows the XTEM image of the sample prepared at high low base pressure and high deposition rate. The Pt coating on the surface of Si film is for the Si film protection during FIB milling process. The dark lines in Si wafer are bend contours which arise from a locally bent wafer foil because the sample is thin and tend to flex slightly. In the Si epitaxial layer, some fringes could be end contours and others could be grain boundaries or other types of defect. Overall, the Si film looks predominantly single crystal despite of the detrimental SiO2 precipitations in the film, indicating that a small amount of SiO2 precipitation does not significantly affect the crystal quality of Si film. Fig. 4(c) shows the XTEM image of the sample prepared at high deposition rate and low base pressure. The local distortion of the crystal around the defect brings some crystal planes into Bragg condition and hence efficient diffraction of these electrons obtained. Consequently, the dark contrast is seen in the image. The Si film is also single crystal dominated, similar with the film in Fig. 4(b). The oxygen interstitials do not affect the lattice structure of Si films. There is a very good agreement between the UV reflectance measurement and XTEM imaging. 3.5. Electrical properties of the evaporated ALICE solar cells prepared at different deposition rates and base pressures Sun-Voc measurements show that the hydrogenated ALICE cells at different deposition conditions have the effective ideality factors neff of 1.8–2.0 (Fig. 5), meaning that half of the minority carrier recombination occurs in the bulk regions of the absorber layers and the surfaces, the other half happens in the depletion regions of the PN junctions [14]. Hydrogen is still quite effective to passivate the massive amount of dangling bonds and other types of defects at the grain boundaries even for ALICE prepared at low deposition rate. Fig. 6 and Table 1 show the I–V performance of ALICE cells prepared at different deposition conditions. The poor fill factors (FF) of ∼ 56% are partially due to the high series resistances presented by the very thin Al electrode (∼ 100 nm) as the Pseudo FF (FF without
4354
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
Fig. 5. Sun-Voc curves with two-diode model fittings. There are very good fits of Voc, V1, and V2 curves, indicating that the two-diode model is effective.
Fig. 6. I–V curves of the ALICE cells prepared at different deposition rates and base pressures.
Table 1 Electrical properties of ALICE solar cells prepared at different deposition rates and base pressures. Types of ALICE cells
Voc (mV)
Jsc (mA/cm2)
neff
FF (%)
ŋ (%)
High R low P High R high P Low R low P
437 436 394
9.7 9.6 8.5
2.0 1.9 1.8
55.9 55.8 55.9
2.26 2.25 1.99
pressure have the worse performance. Considering with the oxygen content levels determined with the Si thin films on Si wafers, it looks that ALICE solar cells can tolerate oxygen level up to 3 × 1018 cm− 3 without deteriorating the performance. It must be noted that the oxygen content distributions in ALICE solar cells on glass is more complicated than those in Si films on Si wafers.
3.6. Spectral response measurement of ALICE cells Fig. 4. XTEM images of c-Si thin films on intrinsic Si wafer substrates prepared at (a) low deposition rate and low base pressure; (b) high deposition rate and high base pressure; and (c) high deposition rate and low base pressure.
calculating the series resistance) of ALICE cells are ∼ 66%. The cell performances (Voc, Jsc, ŋ) appear to be corresponded to the material qualities of poly-Si thin films. It shows that high rate deposited ALICE cells at either high or low base pressures have similar and better performances, while ALICE cells prepared at low rate and low base
The spectral response measurements of the ALICE cells are presented in Fig. 7. The three types of ALICE cells investigated in this work have strong absorption in the visible wavelengths, but weak absorption in infrared region due to the small thickness of poly-Si films and lack of back surface reflectors. They all have similar peak positions at ∼ 490 nm, indicating that the emitter thicknesses for these ALICE cells are similar. The high rate ALICE cells prepared at either high or low pressures are very similar and have the peak IQE of 48.8%, while ALICE cells prepared at low rate and low pressure have a lower
S. He et al. / Thin Solid Films 518 (2010) 4351–4355
4355
Fig. 7. Spectral response measurements of ALICE cells prepared at different deposition rates and base pressures.
peak IQE of 45.7% as a result of more minority recombination in the structural defects and oxygen recombination centers.
Fig. 8. Example of the experimental EQE and IQE of ALICE cell with the measured front surface reflectance, together with PC1D simulation curves.
3.7. Diffusion lengths of ALICE cells The minority carrier diffusion lengths in absorber layers of the ALICE cells were calculated via the one-dimensional PC1D simulations [15] of spectral response spectra (Fig. 8). For these fits, a three-region device (vertical stack) is used where each region represents emitter (including AIC), absorber and BSF layers, respectively. The main fit parameters in PC1D simulations are the defect (i.e. Shockley–Read–Hall recombination) related bulk lifetimes in the three regions. The material parameters of mono-crystalline Si were assumed. The mobility of free carriers in poly-Si materials is significantly lower than in high-purity monocrystalline Si and is dependent on grain size and grain boundary properties. Due to this assumption in the simulations, the diffusion lengths obtained are lower bounds to the actual values. The PC1D simulation results show that high rate deposited ALICE cells prepared at either high or low base pressure have the minority carrier diffusion length of 1.01 µm, while ALICE cells prepared at low rate and low base pressure has diffusion lengths of 0.93 µm, respectively. From the above results, it appears that the high deposition rate is much more critical than the low base pressure in determining the material quality and the resulting performance of the SPE evaporated poly-Si thin-film solar cells. It is found that the crystal quality of evaporated poly-Si thin film goes to saturation when the deposition rate is higher than 300 nm/min. It is easy to realize the high rate deposition for the lightly-doped absorber layer, however, the low deposition rate has to be applied to the heavily-doped emitter layers and BSF layers due to the heating limitation of the dopant sources. Efforts need to be taken to increase the deposition rate for the emitter and BSF layers for better material quality, which could potentially improve the poly-Si thin-film solar cells. Furthermore, the upper limit for the base pressure which has no or little impact on the crystal quality and cell performance of evaporated poly-Si thin-film solar cells needs to be determined in the future. 4. Conclusions Evaporated poly-Si thin-film solar cells obtained at high base pressures and high deposition rate have small amount of SiO2 precipitations, however still show the similar good material quality and performances as the cells prepared at low base pressure and high
deposition rate with oxygen interstitials. On the other hand, the poly-Si thin-film solar cells deposited at low deposition rate have large amount of SiO2 precipitations and resulting worse material quality and cell performances. Therefore, during a-Si thin-film deposition at non-UHV environment, the low base pressure is not very critical as long as the high evaporation rate can be obtained. This can significantly reduce the capital cost on the vacuum equipment which can directly reduce the manufacturing cost of the poly-Si thin-film solar cells.
Acknowledgement Song He acknowledges a Ph.D. scholarship from the Faculty of Engineering at The University of New South Wales (UNSW). Song He also would like to give thanks to Xiaojing Hao, Tom Puzzer, Charlie Kong and Yidan Huang for their assistance on the measurements of FTIR and TEM imaging.
References [1] P.A. Basore, Proc. 29th IEEE Photovoltaic Specialists Conference, New Orleans, U.S.A, 2002, p. 49. [2] M.J. Keevers, T.L. Young, U. Schubert, M.A. Green, Proc. 22nd European Photovoltaic Solar Energy Conference, Milan, Italy, 2007, p. p.1783. [3] S. He, B. Hoex, D. Inns, I.C. Brazil, P.I. Widenborg, A.G. Aberle, Sol. Energy. Mater. Sol. Cells 93 (2009) 1116. [4] P.I. Widenborg, A. Straub, A.G. Aberle, J. Cryst. Growth 276 (2005) 19. [5] R. Brendel, Thin-Film Crystalline Silicon Solar Cells, Wiley-VCH, Berlin, 2003. [6] P.A. Basore, Proceeding 3rd World Conference on Photovoltaic Solar Energy Conversation, Osaka, Japan, 2003, p. 935. [7] O. Kunz, J. Wong, J. Janssens, J. Bauer, O. Breitenstein, A.G. Aberle, Prog. Photovolt. Res. Appl. 17 (2009) 35. [8] K. Bothe, R. Sinton, J. Schmidt, Prog. Photovolt. Res. Appl. 13 (2005) 287. [9] A. Borghesi, B. Pivac, A. Sassella, A. Stella, J. Appl. Phys. 77 (1995) 4169. [10] P. Gaworzewski, E. Hild, F.G. Kirscht, L. Vecsernyes, Phys. Stat. Sol. (a) 85 (1984) 133. [11] C.T. Kirk, Phys. Rev. B 38/2 (1988) 1255. [12] W.S. Lau, Infrared Characterization for Microelectronics, World Scientific Publishing Co., Singapore, 1999. [13] A. Straub, P.I. Widenborg, A. Sproul, Y. Huang, N.P. Harder, A.G. Aberle, J. Cryst. Growth 265 (2004) 168. [14] R.A. Sinton, A. Cuevas, 16th European Photovoltaic Solar Energy Conference, Glasgow, UK, 2000, p. 124. [15] P.A. Basore, IEEE Trans. Electron Devices 37 (1990) 337.