The liquid phase crystallization (LPC) of silicon is an emerging technology for fabricating 10 − 20 µm thin multi-crystalline silicon layers on glass. LPC silicon solar cells exhibit similar electronic performance to multi-crystalline wafer-based devices. Due to the reduced absorber thickness, however, effective measures for light trapping have to be taken. We present tailor-made micro-structures for light trapping at the LPC silicon back-side, whereby a nano-imprinted resist layer serves as a three-dimensional etching mask in subsequent reactive ion etching. Contrary to state-of-the-art random pyramid textures produced by wet-chemical etching, this method allows to produce tailor-made textures independent of grain orientation. Differently shaped micro-textures were replicated in LPC silicon. Absorptance and external quantum efficiency of periodic honeycomb patterns and random pyramids were found to be equivalent. Thus, the method enables the potential to further optimize light trapping in LPC silicon solar cells.
Published by The Optical Society under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.
Liquid phase crystallized (LPC) silicon thin-film solar cells on glass are a promising candidate for high-efficiency thin-film devices, providing power conversion efficiencies up to 14.2% so far . While the open-circuit voltage and the fill factor are already close to values reached on multi-crystalline wafer cells , the short-circuit current density of 31.3 mA cm−2 is about 30% lower than for record devices on high performance multi-crystalline wafers. Improving the short-circuit current density in thin-film solar cells requires light management techniques, in particular nano- or micro-texturing. Up to now, tailor-made, periodic light management textures were implemented in LPC silicon solar cells at the sun-facing glass-silicon interface [3–5], improving light in-coupling into the device. At the rear-side, both planar and glass-textured devices relied on random pyramid textures produced by wet-chemical etching in potassium hydroxide (KOH) for light trapping [1, 6, 7]. The pyramid texture resulting from wet-chemical etching, however, depends on the silicon crystal grain orientation [1, 6], leading to varying pyramid angles and, potentially, optical properties . Furthermore, wet-chemical etching with KOH does not allow individual design of tailored light trapping structures (optimized for the respective thin-film solar cell geometry) with periodic or deterministic aperiodic geometry. In micro-crystalline silicon thin-film solar cells, tailored light trapping textures efficiently increasing light absorption in the thin absorber layer could be identified [9–13]. One investigated texture was the honeycomb texture introduced by Sai et al. , which allowed to reach short-circuit current densities above 30 mA cm−2 in 4 µm thick micro-crystalline silicon thin-film solar cells .
In the present publication, a method to replicate micro-textures in LPC silicon thin-film solar cells on glass is presented. The method relies on a combination of nano-imprint lithography (NIL) and reactive ion etching (RIE), whereby a resist layer structured by NIL serves as a three-dimensional etching mask during RIE. It is demonstrated that this NIL+RIE method allows to produce tailor-made micro-textures in LPC silicon independent of grain orientation. Honeycomb textures with different periods and height-to-period ratios are replicated and analyzed with respect to their absorption properties. An optimum honeycomb texture for light trapping in LPC silicon is identified and compared to random pyramids produced by KOH etching.
Samples were produced on Corning Eagle XG glasses. After a glass cleaning in a commercial lab dish-washer, an interlayer stack of 250 nm silicon oxide (SiOx) / 70 nm silicon nitride (SiNx) was deposited via plasma-enhanced chemical vapor deposition (PECVD). Subsequently, the SiNx layer was exposed to an oxygen-containing plasma, leading to an oxidation of the first ∼ 10 nm of the SiNx layer, providing an optimized interface passivation . A 16 µm thick nano-crystalline silicon layer was deposited by electron-beam evaporation at a heater temperature of 600 °C. Using PECVD, a highly-doped thin amorphous silicon layer and silicon oxide capping layer are deposited. During the ensuing liquid phase crystallization, the dopant atoms from the a-Si layer distribute over the whole silicon absorber, thereby ensuring a homogenous doping concentration. Liquid phase crystallization is performed with a 5 cm wide line shaped laser source with a wavelength of 808 nm under vacuum. Subsequently, the silicon oxide capping layer is removed by wet-chemical etching in hydrofluoric acid. A hydrogen plasma treatment at 600 °C is carried out in order to reduce defects in the LPC silicon absorber . Reference samples were wet-chemically etched in a KOH-containing solution, resulting in a random pyramid texture (Fig. 1, KOH), representing the state-of-the-art texturing method [1, 6].
Honeycomb structures were produced on silicon wafers for use as master structures in nano-imprint lithography . For this, a thermally grown oxide layer is wet-chemically etched through a hexagonal hole pattern scribed by photolithography. Master structures with periods 1.5 µm, 3.0 µm and 6.0 µm and height-to-period ratios between 0.1 and 0.3 were produced. For details of the honeycomb structure production process, the reader is referred to Ref. .
For pattern replication of honeycomb structures in LPC silicon, as sketched in Fig. 1, a method employing a combination of NIL and reactive ion etching (RIE) was developed. In a first step, the desired texture was replicated in a commercially available organic resist (UVcur06 by microresist technology GmbH) by UV-NIL. The textured resist serves as a three-dimensional etching mask in the subsequent reactive ion etching (RIE) process.
For texture replication with high structural fidelity, the etch rates of the resist and LPC silicon need to be the equivalent. Additionally, the RIE process needs to be highly anisotropic to avoid texture distortions. The RIE process was optimized for an Oxford Plasmalab 80 Plus etcher. A gas mixture of 35.7 sccm CHF3 / 14.3 sccm SF6, chamber pressure of p = 20 mTorr, and plasma powers of PRF = 100 W and PICP = 50 W were identified as optimal conditions for texture replication from UVcur06 to LPC silicon. Possible resist residues are removed by an oxygen plasma treatment.
Textured LPC silicon surfaces were characterized using a KEYENCE laser scanning confocal microscope and Park Systems XE-70 atomic force microscope (AFM). Absorption measurements were performed using a Lambda 1050 photospectrometer with the samples placed in an integrating sphere. Heterojunction solar cells with a lithography-free processing were produced according to the cell design described in Haschke et al. , therein denoted as test structure. While these cells typically exhibit a low fill factor, they allow to determine the solar cell characteristics, in particular the external quantum efficiency (EQE). EQE was measured using a home-made setup with a spot size of 3 × 2 mm2.
3. Structural analysis
The surface of LPC silicon layers with textured rear-sides were analyzed with respect to the dependency of the texturing method on the silicon crystal grain orientation and are illustrated in Fig. 2. Additionally to the random pyramid texture by wet-chemical etching in KOH and honeycomb textures, the study is extended to a random micro-texture based on reactive ion etching in a SF6/O2 plasma proposed by Moreno et al. [19, 20] and a periodic array of inverted pyramids with a period of 1.25 µm and a pyramid base of 1.0 µm produced by NIL+RIE. For the sample textured by wet-chemical etching in KOH, the surface measured by laser scanning confocal microscopy reveals that pyramids on different grains are tilted with a distinct tilting angle. This is attributed to the wet-chemical texturing mechanism stemming from different etch rates on different grain orientations, giving rise to pyramids with (111) facets . The SF6/O2 plasma texturing leads to small random pyramid-like shapes and larger bowl-shaped features, whereby the texturing is independent of grain orientation. The same holds for LPC silicon layers textured by NIL+RIE (second row), where the same texture is found on both grains. The successful replication of both, an inverted pyramid array and honeycomb textures, demonstrates that the NIL+RIE method is suitable for smooth as well as sharp-edged textures.
The surface of different honeycomb textures was investigated using laser scanning confocal microscopy in Fig. 3(a). Using NIL+RIE, a wide variety of honeycomb textures with periods between P = 1.5 µm and P = 6.0 µm and heights from h = 0.13 µm to h = 1.4 µm could successfully be reproduced in LPC silicon. Apart from single defects that are attributed to the mastering process, the textures could be replicated with high structural fidelity and homogeneity. As discussed earlier, the NIL+RIE textures are independent of grain orientation, in contrast to the random pyramid texture produced by wet-chemical etching in KOH.
Further investigation of the honeycomb textures etched into LPC silicon surface is performed using atomic force microscopy. Figure 3(b) exhibits surface profiles of two distinct honeycomb textures, one with a period of 1.5 µm and a height of 125 nm (blue) and one with a period of 6.0 µm and height of 1.4 µm (black). The profiles of the respective sample types (the height-to-period ratio differs for the samples shown in Fig. 3(a) and 3(b)) are sketched as blue lines in the surface images in Fig. 3(a). For better comparison of the structures, the axes of the smaller texture (right axes) were stretched by a factor of 4, the ratio of the samples’ periods. Both textures exhibit the characteristic surface profile of the honeycomb pattern. For the sample with a period of 1.5 µm, the height in the LPC silicon is reduced by 25 nm to 100 nm due to height reduction during nano-imprinting or reactive ion etching. For the larger texture, in contrast, no height reduction could be observed. In parts, the structure is even higher than on the master by about 100 nm. These deviations are attributed to inhomogeneities in manufacturing and characterization of both masters and NIL+RIE samples.
Nonetheless, the surface analysis proves the applicability of the NIL+RIE method to master structures with a wide range of textures, periods, and heights. In contrary to the KOH pyramid texturing employed in state-of-the-art LPC silicon solar cells, the quality of the micro-textures produced by NIL+RIE does not depend on silicon crystal grain orientation.
4. Optical properties of LPC silicon with honeycomb textures
The absorption characteristics of LPC silicon absorbers with honeycomb textures produced by NIL+RIE (see inset in Fig. 4 for the measured layer stack) were studied with respect to their optimum geometrical parameters. Figure 4 illustrates the mean absorptance from 800 nm to 1100 nm of 16 µm thick silicon absorbers with honeycomb textures for varied height-to-period ratios and periods of 1.5 µm (green), 3.0 µm (blue), and 6.0 µm (red). For 1.5 µm and 3.0 µm, the shallow textures with a height-to-period ratio of 0.1 or less do not increase absorption. All other structured silicon layers absorb more light than the reference with a planar back-side (black line). For all periods, a higher height-to-period ratio leads to increased absorption. This is attributed to an increase in scattering and scattering into higher angles. Comparing the periods for a constant height-to-period ratio of 0.2, it is found that absorptance in the long wavelength range is higher for larger periods. While absolute differences are small, the absorber with a period of 6.0 µm and a height-to-period ratio of 0.2 provides the best light trapping ability and will therefore be further evaluated in the following.
Figure 5 represents the absorption of 16 µm thick LPC silicon absorbers with the produced light-trapping textures at the back-side. For clarity, only the optimized honeycomb texture (period 6.0 µm and height-to-period ratio 0.2) is depicted next to the inverted pyramid array produced by NIL+RIE (yellow), the random micro-texture (SF6/O2 texture, magenta) introduced by Moreno et al.  and the random pyramid texture produced by wet-chemical etching in KOH (KOH texture, black).
While the random micro-texture produced using a SF6/O2 plasma provided anti-reflective properties for silicon wafer solar cells , they do not exhibit a strong light-trapping effect in LPC silicon (Fig. 5, magenta curve). The inverted pyramid array does enhance absorption in the long wavelength range, but absorbs less light than the honeycomb structure. Compared to the state-of-the-art KOH texture employed in current LPC silicon record devices , the honeycomb texture exhibits a slightly lower absorption in the wavelength regime from 700 nm to 900 nm. Optical properties of the LPC silicon absorbers discussed here are found to be comparable to those of interdigitated back contacted (IBC) heterojunction solar cell presented in Ref. . Hence, it is assumed that characteristics of the back-side textures presented in this study can be transferred to IBC solar cells.
As the pyramid tilting angle of the KOH texture was found to depend on the grain orientation, its light trapping characteristics are expected to vary over the sample, depending on the tilting angle of the pyramids. Absorption measurements were performed on four different spots on the samples with KOH pyramid texture and the optimized honeycomb texture, whereby a spot diameter of 0.4 cm was used to measure on individual grains. Mean absorptance and standard deviation over these four measurements are depicted in Fig. 6.
Contrary to the expected variation in absorption for the grain-orientation dependent KOH texture, no significant changes were measured on the four grains. Designated experimental designs or simulations of pyramids with varying angles in the future will shed light on the dependency of optical properties on the tilting angles of the pyramids on differently oriented silicon crystal grains. For both textures, absorption varies by about ±3% and within the error range, absorption values are the same. The difference in absorptance in the wavelength range from 700 nm to 900 nm found for the individual measurement presented in Fig. 5 is less pronounced for the averaged absorptance. Therefore, absorptance of the LPC silicon layers with honeycomb textured back-side is comparable to KOH pyramids if statistical variations and measurement inaccuracies are considered.
As the position of the sun varies during the day and year, the angular dependency of the absorption should be taken into account to further assess the suitability of honeycomb and KOH textures for application as back-side structures in LPC silicon thin-film solar cells. Absorption was measured depending on the incident angle of the light. Figure 7 presents the mean value of absorptance from 700 nm to 1100 nm for a 16 µm thick LPC silicon absorber with a planar (grey), KOH textured (black) and optimized honeycomb textured (red) back-side.
Mean absorption is higher by about 20 % (absolute) for both textures compared to a planar back-side over the whole range of incident angles. Comparing the change of absorption with the angle of incidence, only small differences are found, indicating that the angular dependency is dominated by the front side of the devices, which is identical for all samples. As discussed for the optical properties at normal incidence, the KOH texture and honeycomb texture exhibit equivalent absorption characteristics also for oblique angles of incidence.
5. Solar cell results
Solar cells were produced on LPC silicon absorbers with the optically most promising textures, namely the KOH texture and a honeycomb texture with a period of 6.0 µm and height of 1.2 µm produced by NIL+RIE. To investigate the influence of the NIL+RIE method on the LPC silicon absorber, in particular the ion bombardment during RIE, measurements of EQE (solid) were performed. EQE results for a LPC silicon thin-film solar cell with a KOH texture (black) and honeycomb texture (red) at the back-side are illustrated in Fig. 8. For reference, the absorptance (dashed) is depicted.
EQE of both textured cells is analogous. Small deviations in the wavelength ranges from 350 nm to 450 nm and 700 nm to 1000 nm lie within the variation on one substrate and are therefore considered negligible. As both EQE and absorptance of the textured solar cells are equivalent, this also holds for the material quality of the LPC silicon absorber layers. Thus, the RIE process does not cause additional defects at the LPC silicon surface and is therefore suitable as a production method for tailor-made micro-structures for light trapping at the back-side of LPC silicon thin-film solar cells. Short-circuit current densities calculated from EQE amount to 28.7 mA cm−2 and 28.8 mA cm−2 for the solar cells with a honeycomb texture produced by NIL+RIE and KOH pyramid texture, respectively. For this study, no additional light management measures, e.g. an anti-reflective foil at the air-glass interface, were employed, making these values comparable to state-of-the-art record LPC devices .
A method for the replication of tailor-made micro-textures in LPC silicon thin-film solar cells on glass based on a combination of nano-imprint lithography (NIL) and reactive ion etching (RIE) was presented. A nano-imprinted resist layer served as a three-dimensional etching mask in the subsequent reactive ion etching, allowing to replicate the imprinted pattern in the underlying crystalline silicon absorber material. In particular, the method was applied to honeycomb structures that were developed as light trapping textures in micro-crystalline silicon thin-film solar cells and an inverted pyramid array with a period of 1.25 µm. It was demonstrated that both the smooth honeycomb texture and the sharp-edged pyramid structure could be replicated in LPC silicon with high structural fidelity. For the honeycomb texture, patterns with periods between 1.5 µm and 6.0 µm and height-to-period ratios from 0.1 to 0.3 were replicated. It was shown that the NIL+RIE method is independent of grain orientation, in contrast to the state-of-the-art KOH texturing employed in current LPC silicon record devices. Hence, the NIL+RIE method allowed to produce tailor-made micro-textures for a wide range of texture geometry, periods, and heights, while being independent of the orientation of the silicon crystal grains.
A detailed analysis of the absorption characteristics of honeycomb textured LPC silicon absorbers was carried out. Optimum parameters for light trapping, with a period of 6.0 µm and a height-to-period ratio of 0.2, could thereby be identified. The absorption of this optimized honeycomb texture was compared to the inverted pyramid array, a random micro-texture produced by RIE etching in a SF6/O2 plasma, and a random micro-pyramid texture produced by wet-chemical etching. It was found that the honeycomb and random pyramid texture are the most promising textures for light trapping.
Statistical analysis and angle-dependent absorptance measurements revealed that the optical properties of both, the LPC silicon absorber with KOH pyramids and with an optimized honeycomb texture, are equivalent. Additionally, solar cell results in form of external quantum efficiency were presented. Material quality for the honeycomb texture produced by NIL+RIE was found to be equal to that of the state-of-the-art KOH texture, as demonstrated by short-circuit current densities of 28.7 mA cm−2 and 28.8 mA cm−2, respectively.
Combining NIL and RIE therefore provides a suitable method to produce tailor-made light trapping textures in multi-crystalline silicon independent of the crystal grain orientation. The NIL+RIE method enables to systematically tailor the texture dimensions and thus yields the potential to improve light trapping and, thereby, short-circuit current density in liquid phase crystallized silicon thin-film solar cells.
Bundesministerium für Bildung und Forschung (501100002347) (No. 03X5520).
The authors thank M. Krüger, M. Muske, I. Rudolph, and C. Klimm for their support with experimental work and SEM imaging. GP Solar is acknowledged for providing the Alkatex IPA-free texturing agent. Part of the photolithography process was conducted at the AIST Nano-Processing Facility, supported by the ”Nanotechnology Network Japan” of the Ministry of Education, Culture, Sports, Science and Technology, Japan (MEXT). K. Masuda is ackknowledged for his assistance in photolithography.
References and links
1. C. T. Trinh, N. Preissler, P. Sonntag, M. Muske, K. Jäger, M. Trahms, R. Schlatmann, B. Rech, and D. Amkreutz, “Potential of interdigitated back-contact silicon heterojunction solar cells for liquid phase crystallized silicon on glass with efficiency above 14%,” Sol. Energy Mat. Sol. Cells 174, 187–195 (2018). [CrossRef]
2. J. Benick, A. Richter, R. Müller, H. Hauser, F. Feldmann, P. Krenckel, S. Riepe, F. Schindler, M. C. Schubert, M. Hermle, A. W. Bett, and S. W. Glunz, “High-Efficiency n-Type HP mc Silicon Solar Cells,” IEEE J. Photovoltaics 7, 1171–1175 (2017). [CrossRef]
3. V. Preidel, D. Amkreutz, J. Haschke, M. Wollgarten, B. Rech, and C. Becker, “Balance of optical, structural, and electrical properties of textured liquid phase crystallized Si solar cells,” J. Appl. Phys. 117, 225306 (2015). [CrossRef]
5. D. Eisenhauer, G. Köppel, K. Jäger, D. Chen, O. Shargaieva, P. Sonntag, D. Amkreutz, B. Rech, and C. Becker, “Smooth anti-reflective three-dimensional textures for liquid phase crystallized silicon thin-film solar cells on glass,” Sci. Rep. 7, 2658 (2017). [CrossRef] [PubMed]
6. P. Sonntag, N. Preissler, M. Bokalič, M. Trahms, J. Haschke, R. Schlatmann, M. Topič, B. Rech, and D. Amkreutz, “Silicon Solar Cells on Glass with Power Conversion Efficiency above 13% at Thickness below 15 Micrometer,” Sci. Rep. 7, 873 (2017). [CrossRef]
7. G. Köppel, D. Eisenhauer, B. Rech, and C. Becker, “Combining tailor-made textures for light incoupling and light trapping in liquid phase crystallized silicon thin-film solar cells,” Opt. Exp. 25, A467–A472 (2017). [CrossRef]
8. P. Sonntag, J. Haschke, S. Kühnapfel, O. Gabriel, D. Amkreutz, and B. Rech, “Properties of liquid phase crystallized interdigitated back-contact solar cells on glass,” Energy Procedia 77, 487–492 (2015). [CrossRef]
9. M. Python, O. Madani, D. Dominé, F. Meillaud, E. Vallat-Sauvain, and C. Ballif, “Influence of the substrate geometrical parameters on microcrystalline silicon growth for thin-film solar cells,” Sol. Energy Mat. Sol. Cells 93, 1714–1720 (2009). [CrossRef]
11. Z. Yu, A. Raman, and S. Fan, “Fundamental limit of light trapping in grating structures,” Opt. Exp. 18, A366–A380 (2010). [CrossRef]
12. C. Battaglia, C.-M. Hsu, K. Söderström, J. Escarré, F.-J. Haug, M. Charriè, M. Boccard, M. Despeisse, D. T. L. Alexander, M. Cantoni, Y. Cui, and C. Ballif, “Light Trapping in Solar Cells: Can Periodic Beat Random?” ACS Nano 6, 2790–2797 (2012). [CrossRef] [PubMed]
13. H. Sai, K. Saito, N. Hozuki, and M. Kondo, “Relationship between the cell thickness and the optimum period of textured back reflectors in thin-film microcrystalline silicon solar cells,” Appl. Phys. Lett. 102, 053509 (2013). [CrossRef]
14. H. Sai, T. Matsui, K. Saito, M. Kondo, and I. Yoshida, “Photocurrent enhancement in thin-film silicon solar cells by combination of anti-reflective sub-wavelength structures and light-trapping textures,” Prog. Photovolt: Res. Appl. 23, 1572–1580 (2015). [CrossRef]
15. O. Gabriel, T. Frijnts, N. Preissler, D. Amkreutz, S. Calnan, S. Ring, B. Stannowski, B. Rech, and R. Schlatmann, “Crystalline silicon on glass - interface passivation and absorber material quality,” Prog. Photovolt: Res. Appl. 24, 1499–1512 (2015). [CrossRef]
16. L. J. Guo, “Nano-imprint lithography: Methods and Material Requirements,” Adv. Mater. 19, 495–513 (2007). [CrossRef]
17. J. Haschke, D. Amkreutz, L. Korte, F. Ruske, and B. Rech, “Towards wafer quality crystalline silicon thin-film solar cells on glass,” Sol. Energy Mat. Sol. Cells 128, 190–197 (2014). [CrossRef]
18. Y. Wang, R. Luo, J. Ma, and S.-Q. Man, “Fabrication of the pyramidal microstructure on silicon substrate using KOH solution,” 5th International Conference on Advanced Engineering Materials and Technology 5, 302–307 (2015).
19. M. Moreno, D. Daineka, and P. Roca i Cabarrocas, “Plasma texturing for silicon solar cells: From pyramids to inverted pyramids-like structures,” Sol. Energy Mat. Sol. Cells 94, 733–737 (2010). [CrossRef]
20. M. Moreno, D. Muias, J. Martínez, C. Reyes-Betanzo, A. Torres, R. Ambrosio, P Rosales, P. Roca i Cabarrocas, and M. Escobar, “A comparative study of wet and dry texturing processes of c-Si wafers for the fabrication of solar cells,” Solar Energy 101, 182–191 (2014). [CrossRef]