Acta Polytechnica doi:10.14311/AP.2014.54.0341 Acta Polytechnica 54(5):341–347, 2014 © Czech Technical University in Prague, 2014 available online at http://ojs.cvut.cz/ojs/index.php/ap ANNEALING OF POLYCRYSTALLINE THIN FILM SILICON SOLAR CELLS IN WATER VAPOUR AT SUB-ATMOSPHERIC PRESSURES Peter Piknaa, b, ∗, Vlastimil Píčb, Vítězslav Bendaa, Antonín Fejfarb a Department of Electrotechnology, Czech Technical University, Technicka 2, 166 27 Prague, Czech Republic b Institute of Physics, Academy of Sciences of CR, Cukrovarnicka 10, 162 53 Prague, Czech Republic ∗ corresponding author: pikna@fzu.cz Abstract. Thin film polycrystalline silicon (poly-Si) solar cells were annealed in water vapour at pressures below atmospheric pressure. The PN junction of the sample was contacted by measuring probes directly in the pressure chamber filled with steam during passivation. The Suns-VOC method and a Lock-in detector were used to monitor the effect of water vapour on VOC of the solar cell during the whole passivation process (in-situ). The tested temperature of the sample (55–110 °C) was constant during the procedure. The open-circuit voltage of a solar cell at these temperatures is lower than at room temperature. Nevertheless, the voltage response of the solar cell to the light flash used during the Suns-VOC measurements was well observable. The temperature dependences for multicrystalline wafer-based and polycrystalline thin film solar cells were measured and compared. While no significant improvement to the parameters of the thin film poly-Si solar cells from annealing in water vapour at below-atmospheric pressures has been observed in the past, in-situ observation proved that there was the required sensitivity to changing VOC at elevated temperatures during the process. Keywords: passivation, water vapour, thin film solar cell, polycrystalline silicon (poly-Si), multicrys- talline silicon (m-Si), Suns-VOC . 1. Introduction The efficiency of a solar cell is directly influenced by the semiconductor material that is used. However, high quality materials tend to be rather expensive, while cheaper materials are of lower quality. One possible way to deal with this issue is to reduce the number of imperfections and defects in the structure of a solar cell. Sufficient photovoltaic efficiency of thin film solar cells in connection with a sufficiently low production price is the key condition for compet- itiveness on the market dominated by wafer-based solar cells. Plasma hydrogenation is usually used for passivation, i.e., to reduce the electrical activity of re- combination centres, e.g., grain boundaries, impurities and crystallographic defects in silicon [1–5]. However, this relatively expensive passivation procedure could be replaced by a cheaper alternative — passivation in water vapour [6–8]. 2. Experimental The use of passivation by water vapour was inves- tigated using thin film silicon solar cells fabricated by CSG Solar AG, Germany. The structure of the solar cell is shown in Fig. 1 [9]. These solar cells were placed in a stainless steel pressure chamber filled with deionized water. The chamber was heated to prepare water vapour. Fig. 2 shows the passivation apparatus with the solar cell contacted directly in the pressure chamber. The steam pressure in the chamber was regulated by the temperature of the chamber heater Figure 1. Structure of a thin film polycrystalline sili- con solar cell fabricated by CSG Solar AG without met- allization to allow passivation in water vapour [9]. according to the water phase diagram [10–12]. The solar cell was heated separately by the pyrolytic Boron Nitride sample heater [13] supplied by wiring passing into the chamber through the feedthroughs. The tem- perature of the sample was measured using a coaxial thermocouple [13]. The temperature of the cell was at all times kept around 10 °C higher than the temperature of the steam, in order to prevent it condensing on the sam- ple and to ensure passivation strictly in vapour. The steam pressures ranged from 0.01 to 0.1 MPa. The per- formance of water vapour passivation was measured by the Suns-VOC method and a Lock-in detector during the process every 15 minutes for 2 hours. Open-circuit 341 http://dx.doi.org/10.14311/AP.2014.54.0341 http://ojs.cvut.cz/ojs/index.php/ap P. Pikna, V. Píč, V. Benda, A. Fejfar Acta Polytechnica Figure 2. Apparatus comprising the stainless steel pressure chamber, the heater, the xenon flash lamp, the thermocouple, the pyrolytic Boron Nitride sample heater, and stainless steel probes to contact the investi- gated thin film poly-Si solar cell used for investigating water vapour passivation. voltage VOC is commonly used as a silicon quality fac- tor. The higher the VOC, the lower the concentration of defects [3]. This key parameter was measured at light intensity 1000 W/m2 = 1 SUN for the Suns-VOC method (a xenon flash used as a light source) and at the light intensity of a white LED determined only ap- proximately to be around 1000 W/m2 for the Lock-in detector. 3. Suns-VOC method The principle of the Suns-VOC method is based on a light flash about 10 times longer than the lifetime of the charge carriers in the investigated semiconduc- tor [14]. Free charge carriers are generated in the semiconductor by absorption of light, and it takes some time — the lifetime of the free charge carrier - to recombine with the oppositely charged carrier. The decrease in light intensity during the light flash is a slow process from the point of view of the generated charge carriers. A quasi-steady state of the charge car- rier concentration can therefore be assumed in every moment of the measurement during the light decay. Two parameters — the light intensity, and the cor- responding VOC — are measured in time, as shown in Fig. 3. They are also presented in Fig. 4. for bet- ter understanding, where the data is needed for the Suns-VOC measurement in comparison with a common current-voltage measurement. In principle, Suns-VOC measurements can be under- stood as a sequence of ordinary current-voltage mea- surements at different light intensities (from 100 W/m2 to 1000 W/m2 in Fig. 4), where only VOC is noted at the corresponding light intensities. It is very impor- tant to know that the light intensity and the generated solar cell short-circuit current ISC are linearly propor- tional. If some value of the current generated by the solar cell at the corresponding light intensity is known, e.g., ISC at 1000 W/m2 measured at a com- Figure 3. Suns-VOC measurement: the decrease in light intensity and the corresponding open-circuit volt- age VOC of the solar cell in time [15]. Figure 4. Explanation of the Suns-VOC method: two parameters — the light intensity and the correspond- ing open-circuit voltage marked by the red ellipses — are measured during the Suns-Voc measurement, and one parameter — the short-circuit current ISC at light intensity of 1000 W/m2 (blue circle) — is needed from some other measurement, e.g., a common current- voltage measurement, to recalculate the detected light intensity of the flash into ISC of the solar cell [16]. mon current–voltage (I–V ) measurement, any light intensity illuminating the solar cell can be recalculated to the current generated by the solar cell, as is done to compare a Suns-VOC measurement with a common I–V measurement, see Fig. 5. This picture consists of two graphs. The graph on the left-hand side presents a measurement of the light intensity illuminating a solar cell and the corresponding VOC of the solar cell during a light flash around 14 ms in duration. The graph on the right-hand side of the picture shows the current-voltage curve calculated from the data measured by the Suns-VOC method, and also explains the differences between this Suns-VOC current-voltage characteristic and the I–V curve of the same solar cell measured in the common way. According to the I–V curves presented here, the Suns-VOC measurement gives better results (efficiency and fill factor). The 342 vol. 54 no. 5/2014 Annealing of Polycrystalline Thin Film Silicon Solar Cells Figure 5. Creation of a Suns-VOC current-voltage characteristic from the measured light intensity and the corresponding VOC. Comparison of a Suns-VOC measurement with a current-voltage characteristic of the same solar cell measured in the common way [15]. Figure 6. Temperature dependence of open-circuit voltage VOC for a tested non-passivated thin film poly- crystalline silicon solar cell. reason is that the Suns-VOC method omits the influ- ence of the series resistance of the solar cell, because no current is measured, just voltage. An advantage of this method is that no solar cell metallization is needed, just a pn junction. A comparison of the Suns- VOC measurement performed before the metallization step (deposition of metal electrodes) and a common I–V curve measured after metallization provides in- formation about the quality of this manufacturing procedure. 4. Lock-in detector A MODEL SR830 DSP Lock-in detector was used to measure the open-circuit voltage generated by the tested solar cell during passivation. The open-circuit voltage VOC of a solar cell decreases as its temperature increases. This is presented in Fig. 6 for a tested non- passivated thin film polycrystalline silicon solar cell. This data was measured by the Suns-VOC method, VOC was measured at light intensity 1000 W/m2 of a xenon flash lamp. According to the measured data, VOC becomes comparable with/lower than the parasitic thermo- electrical voltage Vth at higher temperatures (VOC ≈ 5 mV, Vth ≈ 23 mV at 150 °C). To measure VOC at elevated temperatures, the generated photovoltage signal VOC has to be amplified or measured selectively by a Lock-in detector. The basic idea of this mea- surement is that the Lock-in detector supplies a light source (white LED in our case) with a specific fre- quency (70 Hz in our case). Simultaneously, the same Lock-in detector measures the VOC of the investigated solar cell. The measured VOC has an alternating (AC) character with the same frequency as the light source (70 Hz). The photogenerated AC voltage is separated from the parasitic thermo-electrical voltage of direct (DC) character, and also from any other parasitic signals by the Lock-in detector measuring only sig- nals with the same frequency as it generates (70 Hz of the white LED). Selective measurement of VOC for a tested solar cell can be done this way, and even very small signals that normally vanish in noise can be detected with high accuracy. 5. Results and discussions The development of the open-circuit voltage of a solar cell passivated at different temperatures (55–110 °C) and steam pressures (0.01–0.1 MPa) is shown in Fig. 7. These investigations were realized by the Suns-VOC method. According to photovoltaic theory, average VOC decreases with increasing temperature. No clear trend of VOC in time at particular temperatures was observed, and previous expectations of increasing VOC during annealing due to passivation of solar cell defects were not fulfilled. It is assumed that the changes in VOC in time are probably caused by instability of the contacting probes. Some details about the process should be noted in order to provide a picture of the complexity of the experiments that were performed, and to discuss possi- ble reasons for the negative results that were obtained. VOC was measured at sample passivation tempera- tures higher than room temperature (55–110 °C). The 343 P. Pikna, V. Píč, V. Benda, A. Fejfar Acta Polytechnica Figure 7. Open-circuit voltage VOC of the thin film poly-Si solar cell CSG measured in time at light intensity 1000 W/m2 for different temperatures by the Suns-VOC method; legend for the measurements: VOC at room temperature before passivation (e.g., 171 mV), water vapour pressure (e.g., 0.01 MPa), sample temperature (e.g., 55 °C) during passivation, and VOC at room temperature after passivation (e.g., 172 mV). Voltage VOC at room temperature is not depicted in the graph. It was measured at light intensity 1 SUN by the Suns-Voc method, as during annealing. intensity of the measured signal was therefore lim- ited, see Fig. 6. The thermal conductivity of the stainless steel chamber is relatively poor. Due to the non-homogeneous temperature distribution in partic- ular parts of the chamber, the temperature gradient changed in time. The influence of the thermo-electrical voltage should therefore also be taken into account. The thermo-electric voltage can be explained in the following way. When two different metals are connected in two points, and the temperature of these connections is different, the electrical voltage can be measured as the difference in the electrical potential at these two places [17]. A MODEL SR830 DSP Lock-in detector with 70 Hz flashing frequency of the white LED was used to sepa- rate the “AC VOC” alternating part of the signal and the “DC VOC” direct part (thermo-electrical voltage and illumination background). The “AC VOC” part of the signal (bold curves) measured by the Lock-in detec- tor, and the “DC VOC” part of the signal, measured by Suns-VOC purged of parasitic thermo-electrical voltage (fine curves), are shown in Fig. 8. The open-circuit voltage of a solar cell can be mea- sured in direct (DC) mode under constant illumina- tion, e.g., by the Suns-VOC method, or under alter- nating (AC) light, e.g., by a Lock-in detector. The measurement of VOC under a slow decaying light flash used for the Suns-VOC method can be understood as a sequence of many separate VOC measurements un- der smaller and smaller constant light intensities (DC mode). If the decay of the light intensity during the light flash is appropriately slow, the semiconductor material has enough time to approach the equilibrium state — a balance between generated and recombined free charge carriers, electrons and holes, and this mea- surement is called a quasi-steady-state measurement. Suns-VOC measurements are quasi-steady state mea- surements, and the appropriate duration of the light flash is around 10 ms for polycrystalline silicon. Another way to measure the VOC of a solar cell is under a sequence of relatively fast flashes (duration in the range of µs). The solar cell generates VOC with the same frequency as the frequency of the fast light flashes. The solar cell generates AC voltage. To com- pare the VOC of a solar cell measured as a quasi-steady state measurement (Suns-VOC method) and as an AC measurement under a sequence of fast light flashes, it is necessary to transfer the AC voltage to DC, or vice versa. In principle, this is similar to comparing the voltage measured in an electricity network (230 VAC) and the DC voltage of some accumulator. In order to compare AC and DC voltage, it is necessary to calculate the effective AC voltage, see [18]. A comparison between the effective alternating part of VOC measured by a Lock-in detector and a direct signal measured by the Suns-VOC method purged of parasitic DC signals (thermo-electrical voltage) was performed on the basis of the following recalculation, using VOC measured at room temperature. The open- circuit voltage of the sample measured under a slow light flash at room temperature (RT) was 201 mV, and it corresponded to the alternating voltage measured by a 60.7 mV Lock-in detector, also at RT. All data measured by the Lock-in detector during passivation was therefore multiplied by factor F to compare them with the reference “DC VOC” of the non-passivated 344 vol. 54 no. 5/2014 Annealing of Polycrystalline Thin Film Silicon Solar Cells Figure 8. Annealing the thin film solar cell (sample CSG 30a) in water vapour at vapour pressures 0.06 and 0.08 MPa. The temperature of the sample was 95 °C (for 0.06 MPa) and 105 °C (for 0.08 MPa). The bold curves represent data measured by the Suns-VOC method purged of the parasitic thermo-electrical voltage Vth. The fine curves are data measured by the Lock-in detector, subsequently recalculated into DC voltage according to formula (1). Legend: VOC at room temperature before passivation (e.g., 201 mV), water vapour pressure (e.g., 0.06 MPa) and sample temperature (e.g., 95 °C) during passivation and VOC at room temperature after passivation (e.g., 193 mV). Voltage VOC at room temperature is not depicted in the graph. It was measured at light intensity 1 SUN by the Suns-Voc method, as during annealing. solar cell at RT equal to 201 mV, see below: F = DC VOC AC VOC = 201 mV 60.7 mV = 3.31 (1) This recalculation allowed a comparison between “DC VOC” data measured by Suns-VOC and “AC VOC” data measured by the Lock-in detector. Factor F = 3.31 allowed to omit some inconsistences of the measurements performed by the Lock-in detector and the Suns-VOC method. The Suns-VOC measurement used a xenon flash lamp, but the Lock-in detector used a white LED; the intensity and the spectra of the light sources were different - the Lock-in detector expected a sinus-shaped detected signal, but the white LED flashed with a rectangle shape function All these discrepancies are included in factor F. Data measured by the Suns-VOC method and data from the Lock-in detector recalculated on the basis of the formula presented above are shown below. Fine curves represent data measured by the Lock-in de- tector, and the results were recalculated according to formula (1). Bold curves are data measured by the Suns-VOC method. Thermo-electrical voltage Vth at an appropriate temperature was subtracted from them. The Vth data was measured in the dark, di- rectly on the contacts used for VOC measurements. The red curves are correlated, with the exception of some points; there is an even better correlation between the blue curves. The discrepancy between the “AC VOC” and “DC VOC” data is probably due to changes in the thermo-electrical voltage in time, or mechanical insta- bility of the contacts. Another explanation could be an impact of the illumination background. These im- perfections should be removed for future experiments, and then a more exact determination of VOC would be approached. Better correlation of “AC VOC” and “DC VOC” could then be expected. Temperature dependences of the open-circuit volt- age for a polycrystalline thin film solar cell and a multicrystalline wafer-based solar cell are shown in Fig. 9. Thermo-electrical voltage Vth at different temperatures was measured for both solar cell types (green curves). Curves with triangles correspond to the multi-Si wafer-based solar cell, and curves with points represent the poly-Si thin film data. The blue curves represent data measured by the Lock-in detector recalculated on the basis of formula (1) and the red data are from the Suns-VOC method including Vth. The results from measurements of VOC and Vth for multi-Si and poly-Si solar cells are summarized in Table 1. The VOC data at room temperatures, the VOC decrease at increasing temperature in unit mV/°C and the relative expression in unit %/°C and also the thermo-electrical voltage at 150 °C for both solar cell types provide a comparison of the different behaviour of thin film and wafer-based solar cells. The VOC decrease at increasing temperature is stronger for poly- Si than for multi-Si for both types of measurements (Suns-VOC, and by the Lock-in detector). However, stronger VOC temperature dependence was expected 345 P. Pikna, V. Píč, V. Benda, A. Fejfar Acta Polytechnica Figure 9. Temperature dependences of VOC for multicrystalline wafer-based (triangles) and polycrystalline thin film (points) solar cells. Red curves represent data measured by the Suns-VOC method, including the thermo-electrical voltage; blue curves represent data measured by the Lock-in detector recalculated on the basis of formula (1). The green curves are measurements of thermo-electrical voltage. The “DC VOC” Suns-VOC measurements were realized at light intensity 1000 W/m2 (xenon flash lamp). The light intensity during measurements performed by the Lock-in detector was around 1 SUN (white LED). VOC (mV) dVOC/dT dVOC/VOC dT (%/°C) Vth (mV) Solar cell (23 °C) (mV/°C) (23 °C) (150 °C) Multi-Si AC VOC 531 2.8 0.5 0.7 Multi-Si DC VOC 531 2.3 0.4 0.7 Poly-Si AC VOC 218 3.2 1.5 23.1 Poly-Si DC VOC 218 2.6 1.2 23.1 Table 1. Overview of VOC at room temperature, dVOC/dT , dVOC/VOC dT and thermo-electrical voltage at 150 °C for multicrystalline and polycrystalline solar cells. from the wafer-based solar cell type. The measured thermo-electrical voltage was significantly different for poly-Si and multi-Si. While Vth of the wafer-based solar cell was less than 1 mV at 150 °C, the thin film cell approached more than 23 mV of Vth. This big difference could be explained by an influence from the borofloat glass substrate (thermal insulator) used for poly-Si. The temperature gradients were probably bigger in this case. 6. Conclusions Polycrystalline silicon thin film solar cells were an- nealed in water vapour to improve their quality by passivation, in other words by de-activating the re- combination centres. The slight improvement in VOC detected in some cases is more likely to be a mea- surement inaccuracy than a real improvement. The tested pressure range was 0.01–0.1 MPa, and the sam- ple temperature ranged from 55 to 110 °C. In spite of the elevated temperatures, the open-circuit voltage VOC was well detectable by the Suns-VOC method and when using the Lock-in detector. Since no sig- nificant improvement was observed in the range of sub-atmospheric pressures, our next experiments will be focused on the more promising supra-atmospheric pressures of water vapour. A comparison of the tem- perature dependences for thin film poly-Si solar cells and wafer-based multi-Si solar cells has been given. All investigations of water vapour passivation were re- alized using an in-situ setup: the changing VOC of the solar cell was measured directly during annealing in steam. An advantage of this approach is elimination of a time as a passivation parameter. Therefore, only a steam pressure and a temperature of the process have to be optimised. Acknowledgements This research work was supported by SGS grant num- ber SGS13/072/OHK3/1T/13, and by projects AV CR M100101216, MPO Moremit FR-TI2/736 and MSMT LNSM. 346 vol. 54 no. 5/2014 Annealing of Polycrystalline Thin Film Silicon Solar Cells References [1] B. Gorka, Hydrogen Passivation of Polycrystalline Si Thin Film Solar Cells, PhD thesis, Technische Universität Berlin, 2010. [2] L. Carnel, I. Gordon, K. Van Nieuwenhuysen, D. Van Gestel, G. Beaucarne, and J. Poortmans, Defect passivation in chemical vapour deposited fine-grained polycrystalline silicon by plasma hydrogenation, Thin Solid Films, 487, pp. 147–151, 2005, doi:10.1016/j.tsf.2005.01.081 [3] L. Carnel, I. Gordon, D. Van Gestel, K. Van Nieuwenhuysen, G. Agostinelli, G. Beaucarne, and J. Poortmans, Thin-film polycrystalline silicon solar cells on ceramic substrates with a Voc above 500 mV, Thin Solid Films, 511–512, pp. 21–25, 2006, doi:10.1016/j.tsf.2005.12.069 [4] B. Gorka, B. Rau, P. Dogan, C. Becker, F. Ruske, S. Gall, and B. Rech, Influence of Hydrogen Plasma on the Defect Passivation of Polycrystalline Si Thin Film Solar Cells, Plasma Processes and Polymers, 6, pp. S36–S40, 2009, doi:10.1002/ppap.200930202 [5] C. H. Seager, D. S. Ginley: Passivation of grain boundaries in polycrystalline silicon: Applied Physics Letters 34(5), 1979, doi:10.1063/1.90779 [6] T. Sameshima, H. Hayasaka, M. Maki, A. Masuda, T. Matsui, and M. Kondo: Defect Reduction in Polycrystalline Silicon Thin Films by Heat Treatment with High-Pressure H2O Vapor, Jpn. J. Appl. Phys., 46, pp. 1286–1289, 2007, doi:10.1143/JJAP.46.1286 [7] S. Honda, T. Mates, B. Rezek, A. Fejfar, J. Kocka: Microscopic study of the H2O vapor treatment of the silicon grain boundaries, Journal of Non-Crystalline Solids 354, pp. 2310–2313, 2008, doi:10.1016/j.jnoncrysol.2007.09.107 [8] S. Honda, A. Fejfar, J. Kocka, T. Yamazaki, A. Ogane, Y. Uraoka, T. Fuyuki: Annealing in water vapor as a new method for improvement of silicon thin film properties, Journal of Non-Crystalline Solids 352, pp. 955–958, 2006, doi:10.1016/j.jnoncrysol.2006.01.061 [9] M. J. Keevers, T. L. Young, U. Schubert, M. A. Green: 10% Efficient CSG Minimodules, 22nd European Photovoltaic Solar Energy Conference, Milan, Italy, pp. 1783-1790, 2007. [10] http://www1.lsbu.ac.uk/water/phase.html [2014-07-20]. [11] http://ergodic.ugr.es/termo/lecciones/water1. html [2014-07-20]. [12] http://en.wikipedia.org/wiki/Phase_diagram# mediaviewer/File:Phase_diagram_of_water.svg [2014-07-20]. [13] http://www.tectra.de/heater.htm [2014-07-20]. [14] R. Sinton: User Manual to Photoconductance Lifetime Tester and Optional Suns-Voc Stage, http://www.sintonconsulting.com [2014-07-20]. [15] R. Sinton, A. Cuevas: A Quasi-Steady-State Open- Circuit Voltage Method for Solar Cell Characterization, 16th European Photovoltaic Solar Energy Conference, Glasgow, United Kingdom, 2000. [16] H. Schmidt, B. Burger, U. Bussemas, S. Elies: How fast does an MPP tracker really need to be? 24th European Photovoltaic Solar Energy Conference, Hamburg, Germany, 2009. [17] R. M. Park, R. M. Carroll, P. Bliss, G. W. Burns, R. R. Desmaris, F. B. Hall, M. B. Herzkovitz, D. MacKenzie, E. F. McGuire, R. P. Reed, L. L. Sparks, T. P. Wang, Manual on the Use of Thermocouples in Temperature Measurement, Fourth Edition, ISBN: 0-8031-1466-4, 1993. [18] N. N. Bhargava, D. C. Kulshreshtha, S. C. Gupta, Basic Electronics and Linear Circuits, Tata McGraw- Hill Education, ISBN: 0-07-451965-4, 1984. 347 http://dx.doi.org/10.1016/j.tsf.2005.01.081 http://dx.doi.org/10.1016/j.tsf.2005.12.069 http://dx.doi.org/10.1002/ppap.200930202 http://dx.doi.org/10.1063/1.90779 http://dx.doi.org/10.1143/JJAP.46.1286 http://dx.doi.org/10.1016/j.jnoncrysol.2007.09.107 http://dx.doi.org/10.1016/j.jnoncrysol.2006.01.061 http://www1.lsbu.ac.uk/water/phase.html http://ergodic.ugr.es/termo/lecciones/water1.html http://ergodic.ugr.es/termo/lecciones/water1.html http://en.wikipedia.org/wiki/Phase_diagram#mediaviewer/File:Phase_diagram_of_water.svg http://en.wikipedia.org/wiki/Phase_diagram#mediaviewer/File:Phase_diagram_of_water.svg http://www.tectra.de/heater.htm http://www.sintonconsulting.com Acta Polytechnica 54(5):341–347, 2014 1 Introduction 2 Experimental 3 Suns-VOC method 4 Lock-in detector 5 Results and discussions 6 Conclusions Acknowledgements References