Structural and Optical Properties of Annealed Porous Silicon Bragg Reflector for Thin-Film Crystalline Silicon Solar Cells

Transcription

1 Available online at Energy Procedia 10 (2011 ) 8 13 European Materials Research Society Conference Symp. Advanced Inorganic Materials and Concepts for Photovoltaics Structural and Optical Properties of Annealed Porous Silicon Bragg Reflector for Thin-Film Crystalline Silicon Solar Cells Maïlys Grau a,b *, Nhat-Tai Nguyen a, Alain Straboni b, Mustapha Lemiti a a Université de Lyon; Institut des Nanotechnologies de Lyon INL-UMR5270, CNRS, INSA de Lyon, Villeurbanne, F-69621, France b S TILE, 3 rue Raoul Follereau, Poitiers Cedex, France Abstract Implementing an annealed porous silicon Bragg reflector between a crystalline silicon thin-film solar cell and its cheap silicon substrate improves greatly its performance. Such a multilayer features indeed a high reflectivity, allowing the optical confinement which is essential for these cells. Annealing the porous silicon above 900 C under H 2 modifies significantly the structure of the porous silicon. This phenomenon lowers its resistivity to avoid high cell series resistance and allows the subsequent epitaxial growth of the silicon cell. The different behaviors of porous silicon structure alteration are studied and rationalized. The pores shapes and sizes, the porosities and the thicknesses of layers are affected. These changes depend on the initial porosities and thicknesses and the annealing parameters. Pores can erupt at the surface when they are neighboring, thus causing a roughness of characteristic length according to the porosity of the layer. The pores sizes increase in all cases. Different multilayer reflectivities are presented, the highest ones being up to 98% over a wide wavelength range. It allows a better understanding of this annealing process and the optimization of the reflectivity of the Bragg reflector in order to apply it on solar cells Published by Elsevier Ltd. Open access under CC BY-NC-ND license. Selection and/or peer-review under responsibility of Organizers of European Materials Research Society (EMRS) Conference: Symposium on Advanced Inorganic Materials and Concepts for Photovoltaics. Solar cells; porous silicon * Corresponding author. Tel.: ; fax: address: mailys.grau@insa-lyon.fr Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Organizers of European Materials Research Society (EMRS) Conference: Symposium on Advanced Inorganic Materials and Concepts for Photovoltaics. Open access under CC BY-NC-ND license. doi: /j.egypro

2 Maïlys Grau et al. / Energy Procedia 10 ( 2011 ) Introduction Crystalline Silicon Thin-Film (CSiTF) solar cells are considered to be able to reduce the production costs of photovoltaic modules since they use less high purity silicon while performing similar efficiencies compared to conventional silicon solar cells. Indeed high efficiencies have already been found through simulation and achieved on high quality substrates [1,2]. To fabricate them, 10 to 30 μm-thick cells are grown epitaxially on an inactive low-cost substrate. This substrate allows mechanical holding for the fragile thin cell. Next the samples can be processed like conventional wafers. The substrate properties are crucial since it needs simultaneously to be cheap, have crystalline characteristics allowing epitaxy and feature thermal coefficients matching the cell, because of the high temperature steps of cell processing. Low-cost silicon wafers, such as these fabricated by S Tile [3] meet these demands, but for process demonstration, highly p-doped Czochralski wafers are used. An important condition for CSiTF cells to reach high efficiency is to perform high light confinement [4]. Indeed photons of long wavelength ( nm) need several tens of microns through silicon to be absorbed. Therefore in thin cells, they can be efficiently absorbed, only if they go several times through them, by being internally reflected. In this purpose, a Bragg reflector made out of porous silicon of two porosities alternated has been used in this work at the back surface of the solar cell, between the latter and its silicon substrate. The firstly obtained porous silicon structure will be reorganized in the experimental conditions of epitaxy which are a temperature above 900 C and a hydrogen atmosphere. Previous studies have reported this phenomenon for transfer solar cells from reusable substrates [5,6] and reflectors with low porosity [7,8]. Consequently the final porous structure will be different and so will be its properties. The work presented here focuses on the consequences of this restructuration on highly porous structures. The processing of restructuration is equivalent to the one described by the sintering theory, in the case of consolidation by coalescence [9]. The specific surface of as-etched porous silicon is very high, of several hundreds of m²/cm 3 and it results in a high surface energy. H 2 etches native silicon oxide so that the entire surface presents silicon dangling bonds. The ability of silicon atoms to move is thermally increased by the annealing, in consequence they migrate in order to lower this surface energy. The nanometer-scale pores merge into larger ones and their shapes are modified [10], leading to a more stable state. This behavior has been simulated with a 2D model by Müller and Brendel [11]. One of the outcomes of the restructuration is that the sample surface, which was containing a high density of pits from the porous silicon, performs a closure, therefore presenting a smooth surface enabling epitaxy, with the crystalline parameters of the substrate conserved. While previous studies showed the influence of restructuration on structures of alternated 22,5% and 50% of porosity [8], in this paper, we study the effect of the reorganization of a highly porous structure of 50% and up to 85% of porosity in alternance, on more doped substrates. The physical properties of the resulting structure will be different, which will have consequences on solar cell characteristics. 2. Experimental All samples described in this paper are 2 mω.cm p-doped Czochralski wafers. The porous silicon stack is made by anodic etching of the sample in a single tank anodization cell. The electrolyte used is a mixture of HF:ethanol at 36% of HF. Etch stops are applied to avoid local HF concentration reduction according to [12]. Successive high current (30 to 200 ma/cm²) and low current (3 ma/cm²) steps are applied, leading to alternated layers of two porosities until reaching 10 double layers. In this range, porosity increases with current density. Layers thicknesses are tuned by the porosification time so as to obtain the conditions of a Bragg reflector:

3 10 Maïlys Grau et al. / Energy Procedia 10 ( 2011 ) 8 13 n d = λ 0 /4 with n the refractive index, obtained through Maxwell-Garnett model of effective medium theory [13], d the layer thickness and λ 0 the wavelength we want the high reflection to be centered to. The restructuration and epitaxy take place in the atmospheric pressure CVD reactor. The sample is annealed in H 2 at 1100 C for 30 minutes with a temperature ramp up of 15 minutes. The beginning of the reorganization of porous silicon atoms can be visually monitored. If some epitaxy is needed after the reorganization of porous silicon, it is made directly after the restructuring step, by introducing the precursors without changing the temperature and pressure conditions. The samples are measured in direct reflectivity, using an integrating sphere to collect the reflected signal. Resistivity measurements are made across the wafer with a probing system developed at S Tile. The porous structure is considered as a resistive layer sandwiched between two bulk silicon parts, in analogy to contact resistivity measurements. The samples used for reflectivity measurements are without epitaxial layer, while these for resistivity present a 5μm-thick epitaxial layer. 3. Results 3.1. Microstructures obtained Given the high doping level of the substrates, all porosities used are higher than these described in most literature [7,8]. Porous multilayers are fabricated with layers of 55% of porosity on the one hand and 65 to 85% of porosity on the other hand (referred to in the following as low porosity layers and high porosity layers respectively). The morphology of a sample with 85% of initial porosity is shown in Fig. 1, before and after being reorganized. Fig. 1. SEM micrographs of the multilayer structure of porous silicon before (left) and after (right) restructuration, with high porosity layers at 85%. Layers seem thinner in the left because of SEM sample charging artifact considering the high resistivity of unannealed porous silicon. Fig. 2. Structures obtained by the secondary restructuration mode: cross-section (left, middle) and view from top (right). Images obtained by SEM imaging.

4 Maïlys Grau et al. / Energy Procedia 10 ( 2011 ) Before restructuring, the initial morphology is of pores channels is dendritic and perpendicular to the sample surface. The pores for the low porosity are too small to be seen at the SEM resolution, but the ones of the high porosity layers are larger, and visible in the picture. After the restructuration annealing, the cavities have broadened, as well as the silicon parts between them do, but not as much as described in literature with lower porosities (round shaped pores of typically ~150 nm diameter for 50% porosity layers and ~20nm for 22.5% porosity [8]). In our low porosity layers, pores are round shaped and up to 10nm large. In the high porosity layers however, pores are not round shaped and they keep their initial direction. This is related to the results of [10], saying that for high porosities, pores get elongated while for low porosities they take a more round shape. Nevertheless, given that pores are much smaller than the ones observed in literature on low porosities, it can be admitted that on these samples, the porosification mechanism leads to a structure intermediate between freshly anodized porous silicon and highly redistributed structure of [8]. Little change in microstructure is visible with SEM according to the high porosity used. It could be due to a porosity change during the restructuration, as have already been observed [11]. The restructuration of a porous multilayer can lead to two types of structures: the one mentioned in Fig. 1 with coarsened pores and a second one, which can be visualized in Fig. 2. In the latter, all the pores merge together across the layer, thus loosing the Bragg structure of alternated layers. Some pores seem to connect to the surface, as described in [14], leading to a high roughness at the surface of the porous silicon. This behavior is avoided by implementing a 400 nm-thick low porosity layer between the Bragg structure and the surface. It prevents pores to connect to the surface and enlarge Reflectivity of the structures The reflectivity of the samples restructured with no epitaxy, is shown in Fig. 3(a). The curves in dashed and solid grey of Fig. 3(a) represent the reflectivity of a sample with its higher porosity of 65%, respectively before and after restructuration. For this sample, the reflectivity is dramatically improved by the restructuration, both in peak height and width. It indicates that the difference of optical indexes, and therefore in porosities, is greatly increased. SEM showed no difference of thickness due to restructuration, therefore the overall porosity must have remained the same, by having simultaneously the high porosity layers increasing in porosity while the one of the low porosity layers decrease. The solid black curve, with high initial porosity of 85%, shows a reflectivity over 90% of reflectivity for a range as large as nm wavelengths. There is a drop in reflectivity, due to the thick top low porosity layer, creating a Fabry-Perot resonator. In the conditions of internal reflectivity in an epitaxial solar cell, this resonator would not occur given the lower refractive index change between the epilayer and the top porous layer. Such a high reflectivity would result in great increase in cell current. Reflectivity 1,0 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0, wavelength (nm) HP 85% (restructured) with HP 65% (before restructuration) with HP 65% (restructured) Layer resistance (Ohm.cm²) ,1 85% porosity 65% porosity Anodization current density (ma/cm²) Fig. 3. (a) Experimental reflectivity curves of the Bragg reflectors, with 10 double layers (HP: porosity of the high porosity layer); (b) Equivalent layer resistance of a structure made out of 10 double layers and one low porosity layer of 400nm, as a function of the current density used for the high porosity layers.

5 12 Maïlys Grau et al. / Energy Procedia 10 ( 2011 ) Resistivity Resistivity of the multilayer has been also measured. With samples of low initial porosity, resistivity of restructured porous silicon is reported to be identical to bulk silicon [15]. In our case however, a high layer resistance was found for multilayers with the highest porosities. Such high resistances are not desirable for solar cells fabrication since they lead to high series resistance, which reduces cell efficiency. With this configuration, local laser contacting prior to epitaxy would be necessary. Still, this resistance decreased strongly with the anodization current density, and therefore the porosity (Fig. 3(b)). In all cases, the resistance is lowered by the rearrangement, since resistivity of anodized porous silicon is said to be as high as the one of intrinsic silicon [16]. The remaining resistance is in agreement with the interpretation that the restructuration is less fulfilled than in the case of low porosity, then leading to higher resistance. The porosity value of the higher porosity layer leads therefore to two opposite effects on cells efficiency, since the reflectivity increases with porosity, but the conductivity reduces. 4. Conclusion Reorganization of porous silicon atoms has been performed on high porosities multilayers, intended to work as rear reflectors for thin solar cells; they perform high reflectivities. Their microstructures have been analyzed, and show a different behavior as described in literature for multilayers of lower porosities, by reorganizing to a lesser degree. The difference in the layers refractive indexes is increased by this process, which is in agreement with the interpretation that porosity changes during the restructuration. Different porosity values have been used for the layers of higher porosity. They show no difference by SEM imaging, but lower values lead to higher conductivity and lower reflectivity of the structure. To make efficient solar cells with such configuration, one can either choose to use low values of this porosity, or use higher values for better reflectivity, in conjunction with local laser contacting through the structure. References [1] van Nieuwenhuysen K, Payo MR, Kuzma-Filipek I, et al. Epitaxial Thin Film Silicon Solar Cells with Efficiencies up to 16.9 by Combining Advanced Light Trapping Methods and CVD Emitters. 24th Europ. Photovoltaic Solar Energy Conf.; [2] Faller FR, Henninger V, Hurrle A, Schillinger N. Optimization of the CVD process for low-cost crystalline silicon thinfilm solar cells. 2nd World Conf. On Photovoltaic Solar Energy Conversion; [3] Grau M, Blangis D, Lindekugel S, Janz S, Reber S, Straboni A. High Voc Crystalline Silicon Thin Film Solar Cells through Recrystallised Wafer Equivalent applied to Sintered Silicon. 24th Europ. Photovoltaic Solar Energy Conf.; [4] McCann MJ, Catchpole KR, Weber KJ, Blakers AW. A review of thin-film crystalline silicon for solar cell applications. Part 1: Native substrates. Sol Energ Mat Sol C. 2001;68(2): [5] Amtablian S. Du transfert de films minces de silicium monocristallin vers un procédé cellule à faible budget thermique. PhD Thesis, INSA de Lyon; [6] Solanki CS, Bilyalov RR, Poortmans J, Nijs J, Mertens R. Porous silicon layer transfer processes for solar cells. Sol Energ Mat Sol C. 2004;83(1): [7] Zettner J, Thoenissen M, Hierl T, Brendel R, Schulz M. Novel porous silicon backside light reflector for thin silicon solar cells. Prog. Photovolt: Res. Appl. 1998;6(6): [8] Duerinckx F, van Nieuwenhuysen K, Kim HJ, Kuzma-Filipek I, et al. Light Trapping For Epitaxial Thin Film Crystalline Silicon Solar Cells. 20th Europ. Photovoltaic Solar Energy Conf; 2005.

6 Maïlys Grau et al. / Energy Procedia 10 ( 2011 ) [9] Derbouz K. Etude du frittage du silicium pour application aux cellules solaires photovoltaïques. PhD Thesis, Université de Poitiers; [10] Ott N, Nerding M, Müller G, Brendel R, Strunk HP. Structural changes in porous silicon during annealing. phys stat sol (a). 2003;197(1): [11] Müller G, Brendel R. Simulated Annealing of Porous Silicon. phys stat sol (a). 2000;182(1): [12] Billat S, Thönissen M, Arens-Fischer R, Berger MG, Krüger M, Lüth H. Influence of etch stops on the microstructure of porous silicon layers. Thin Solid Films. 1997;297(1-2):22 5. [13] J. C. Maxwell-Garnett, Colours in Metal Glasses and in Metallic Films. Philos. Trans. R. Soc. London, Ser. A, 1904; 203:385. [14] Kraiem J. Epitaxie et transfert de films minces de silicium pour applications photovoltaïques. PhD Thesis, INSA de Lyon; [15] Duerinckx F, van Nieuwenhuysen K, Kim H, Kuzma-Filipek I, et al. Large-area epitaxial silicon solar cells based on industrial screen-printing processes. Prog. Photovolt: Res. Appl. 2005;13(8): [16] Canham L. Properties of Porous Silicon. Canham L, editor. INSPEC; 1997.