Open Access
Add to BibSonomyAbstract
Graphene oxide (GO) obtained by chemical exfoliation exhibits a quasi-2D structure. Its refractive index is very close to the theoretically predicted best refractive index for a single antireflection coating layer between air and Si. The robust honeycomb plane structure of GO makes it a promising mask candidate for surface texturizing. Here, we demonstrate different GO distributions on Si, and report the reflection properties before and after etching. For an etched Si substrate with suitable GO coating, the reflectance reached 2.1% at 667 nm. Preliminary 1.5-min-long etching of a p+nn+ solar cell with a GO mask boosted the efficiency from 7.09% to 7.55%.
© 2015 Optical Society of America
1. Introduction
Antireflection (AR) [1] is intensively investigated for applications in many fields such as detectors [2], lasers [3], and solar cells [4]. Two methods are commonly adopted for achieving AR. The first method uses single or multilayers of antireflection coatings (ARCs) [5–7]. The second method is based on incorporating the surface structure (texture) for reducing the reflection [8–10]. Researchers continually strive to develop low-cost and low-reflection structures. In the last decade, an interesting material, graphene oxide (GO), has been utilized in many applications owing to its simplicity and low-cost superiority [11, 12]. GO has a refractive index of ~1.8 in the visible range [13], which is very close to the theoretically predicted best refractive index [14] for a single AR layer between air and Si. In the past, graphene itself has been utilized as an antireflection coating on polished and textured Si, and it was found that the presence of (native) SiO2 is necessary for obtaining a significant reduction in reflection [15]. Amphiphilic GO can be more easily deposited on hydrophilic SiO2/Si substrates in solution compared with hydrophobic graphene, and the distribution of GO can be easily controlled by varying the extent of hydrophilic treatment [16]. In this work, we first investigated the AR property arising from the GO stacking. Next, we used GO as the etching mask for developing surface textures on Si for further reducing the reflection. GO obtained by chemical exfoliation can exhibit a quasi-2D structure [17]. Its robust honeycomb plane structure makes it a promising candidate as a mask for etching. In the past, nanoparticles such as polystyrene nanosphere monolayers have been adopted as the etching mask for demonstrating the possibility of nanorod arrays on Si [18]. This surface structure of Si results in the gradient refractive index or buffer effective refractive index between air and Si, thereby reducing the reflection. GO may provide even more advantages in terms of AR. First, the horizontal dimension scale of GO flakes could range from nm to µm, depending on the exfoliation and deposition processes. The size flexibility of GO flakes can be utilized for minimizing the reflection. Second, there is no need to remove GO because its refraction index is suitable for a single-layer ARC and it has low absorption characteristics. However, when the application requires removing GO, graphene-based materials can be easily removed by a suitable oxidation treatment [19, 20]. Hence, GO coating of Si substrates, serving as an AR layer, is an interesting topic. In the present study, we tuned the degree of hydrophilic treatment of Si wafers, obtaining Si surfaces with different GO distributions. The corresponding reflection properties (before and after etching) were determined. The beneficial AR effects were practically demonstrated on solar cells.
2. Fabrication process
First, a powder composed of graphite and KMnO4 was dissolved in a 1:3 ratio in H2SO4 (120 mL, 98%) for exfoliating graphite into GO sheets [21]. The mixture was continuously stirred for 2 h, and was placed in an ice bath for controlling the temperature, keeping the temperature below 35 °C. Then, the mixture was diluted by 250 mL of DI water and stirred for another 2 h. 700 mL of DI water and 20 mL of hydrogen peroxide (H2O2) were subsequently added and stirred for 12 h. The solution was then centrifuged at 10000 revolutions per minute (rpm) for 15 min. At such high rpm, both GO and unreacted solids precipitated, and the overlying solution was removed. The precipitation was then washed with HCl (660 mL, 3%) to remove the inorganic impurities. The procedures of centrifugation and HCl wash were repeated twice. Then, the procedures were changed to centrifugation and DI water washing for removing the acid. This was repeated until the pH value reached ~6. The solution was then centrifuged at 500 rpm for 15 min. At such low rpm, only the unreacted solids precipitated, and the hydrophilic GO flakes were well suspended in the upper part of the solution [22]. To attach the GO flakes to the Si substrates, the Si substrates were immersed in an SC1 solution at the temperature of 75 °C. The SC1 solution was composed of NH4OH, H2O2, and H2O with a 1:2:8 ratio, and the abundance of hydroxide (OH-) yielded hydrophilic Si surfaces. The hydrophilic surfaces attracted the hydrophilic GO flakes. Longer SC1 treatment was expected to yield a denser coverage of GO flakes on Si [16]. In this study, we prepared three silicon wafers with different treatment duration types: one without the SC1 treatment, another with 1-min-long SC1 immersion, and yet another one with 30-min-long SC1 immersion. The corresponding samples were labeled F0, F1, and F30, respectively. GO suspension was dropped on the top surfaces of the F0, F1, and F30 substrates, and they were dried at 70 °C.
Surface morphology was assessed by using atomic force microscopy (AFM) as shown in Fig. 1, and the root mean square (RMS) roughness values of 4 μm x 4 μm areas of F0, F1, and F30 samples were 3.9, 15.8, and 17.9 nm, respectively. Because F0 was not subjected to the SC1 treatment, GO flakes exhibited partial aggregation during deposition. A fraction of the F0 area was covered with small and well-distributed GO flakes (Fig. 1(a)) and another fraction of it was very rough. Hence, we only investigated the morphology and reflectance of the former (i.e., the following description of the F0 properties is for the area in which the GO flakes were small and well-distributed). The sample F1 was subjected to 1-min-long SC1 immersion, and it was possible to link larger GO flakes (Fig. 1(b)). Because the immersion time of F30 was much longer than that of F1, there were both large and small GO flakes distributed on F30 (Fig. 1(c)).
Fig. 1 AFM images of the (a) F0, (b) F1, and (c) F30 samples. The left part shows the surface morphology and the right part shows the height profile (The profile is cut along the dashed lines shown in the morphology figures on the left side of the figure).
3. Reflectance reduction contributed only by GO deposition
First, we investigated the extent to which GO deposition affected the reflection of light incident on Si. The refractive index of air is 1, and the refractive index of Si is ~3.4 [23]. If an ARC layer with a refractive index between 1 and 3.4 is deposited on Si, the reflectance of the resulting structure should be reduced relative to that of Si.
For a single-layer thin-film, the antireflection condition for the air/substrate interface is the existence of an optical quarter-wave layer with a refractive index equal to the square root of the substrate's refractive index. We have simulated the relation between the reflectance and the refractive index for antireflection layers with different thickness (Fig. 2). For two representative GO thicknesses in Fig. 2, the refractive index of 1.8 indeed resulted in a low reflectance. Since GO has the refractive index of ~1.8 in the visible range, as described in [13], it may itself contribute initially to the antireflection. The reflectance spectra of F0, F1, and F30 were measured as shown in Fig. 3. A control Si sample, without any GO deposition, was also measured for comparison. The reflectance spectra were measured by using a reflective Fourier transform infrared (FTIR) spectrometer. The incident IR-beam was 15° relative to the Si surface normal, and the spot size was ~1 mm in diameter. In general, normal incidence is adopted in reflectance studies. However, the smallest angle provided by our reflective FTIR spectrometer (V70, Bruker) was 15°. Larger incidence angles yield larger deviation from normal incidence. The oscillation at ~1 μm in the reflectance spectra in Fig. 3 is owing to the Si bandgap energy of 1.1 eV.
Fig. 2 Dependence of reflectance on the refractive index and wavelength, for different Si antireflection layer thicknesses: (a) 50 nm and (b) 100 nm. The incident light was assumed to be at 15° relative to the Si surface normal.
Fig. 3 The reflectance of GO-covered and control samples (without etching). The deposition of GO reduced the sample reflectance.
The sample F0 was uniformly covered by GO flakes, and its reflectance was smaller, compared with the control sample (Fig. 3). On F0, the distance between adjacent GO flakes was smaller than the incident light wavelength; thus, the GO flakes formed a moth-eye structure [24, 25]. This means that the incident light experience an effective refractive index (nGO,eff) between nGO (the refractive index of GO) and nair, before entering Si. Such an intermediate refractive index could underlie the reflectance reduction.
The GO flakes on F1 are much larger than the flakes on F0, but the inter-flake distance increases as well. The larger inter-flake separation means that the horizontal distribution of GO flakes on F1 should be taken into account. For the area with GO deposition, the reflection is reduced owing to the intermediate GO between the air and Si. However, for the area without GO deposition, the reflection cannot be reduced. Owing to the encumbrance of the area without GO, the reflectance of F1 is between that of F0 and that of the control.
On the other hand, F30 is covered by GO flakes of different sizes. From the profile in Fig. 1(c), it is seen that F30 is densely covered with small flakes, just like F0. In addition, some large flakes are present as well, similar in size to those observed on F1. Hence, the reflection characteristics of F30 combine antireflection benefits of the F0 and F1 samples. This results in F30 having the smallest reflectance among the three samples, as shown in Fig. 3.
4. Further reduction of reflection using GO as the etching mask
To further reduce the surface reflectance, the surface should be texturized by using GO as the etching mask. Owing to the robust honeycomb plane structure of GO, it can be used as a mask on Si. F0, F1, F30, and control samples were etched by using an aqueous KOH (1 M) solution at 70 °C for 5 min [26]. The surface morphologies and profiles of the etched F0, F1, F30, and control samples are shown in Fig. 4. The RMS roughness values of the 4 μm x 4 μm areas of the etched F0, F1, F30 and control samples were 27.2, 47.6, 49.8, and 3.78 nm, respectively. The height of the islands on the etched F0 sample was obviously larger than that before etching. These islands were sharp and dense. This means that the dense and small GO flakes in the F0 sample played a role in forming these dense islands after etching. More than 100 nm in depth of Si could be etched away around these sharp islands. As described in [27], the graded refractive index of a tapering structure can effectively reduce the structure reflectance. Our F0 structure could also benefit from the graded refractive index because the surface morphology was basically similar to that of [27], as shown in Fig. 4(a). The reflectance of the etched F0 (Fig. 5) was significantly reduced, compared with that of F0 before etching (Fig. 3).
Fig. 4 AFM images of the (a) etched F0, (b) etched F1, (c) etched F30, and (d) etched control samples. The islands on the etched F0 sample are sharper than those on the etched F30 sample.
Fig. 5 The reflectance of the etched GO and control samples. The samples were etched by applying a KOH solution for 5 min.
The large inter-flake separation on F1 also resulted in the large separation of Si islands on the etched F1 sample. Similarly, the reflectance of the etched F1 was between those of the etched F0 and etched control samples.
On the other hand, crowded but wide Si islands were formed on the etched F30. The islands on the etched F30 were not as sharp as those on the etched F0. The reflectance of the etched F30 was smaller than that of the etched F0 at short wavelengths, but became larger than that of the etched F0 at long wavelengths (Fig. 5). Light at short wavelengths is more sensitive to the graded refractive index than light at long wavelengths. Hence, the gentle Si islands of the etched F30 increased the short-wavelength light sensitivity to the gradual changes in the refractive index, resulting in a relatively low reflectance. The reflectance was as low as 2.1% at the wavelength of 667 nm. With increasing the wavelength, the effect of gradual changes on the refractive index slowly diminished.
In principle, the Si/air volume ratio could be used for estimating the effective refractive index. However, the excellent antireflection property of F30 is owing to the vertically gradual change of the effective refractive index of each thin part of the gentle Si islands. Therefore, characterization of the overall refractive index cannot successfully address the effective antireflection ability.
The ratio of reflectance reduction (antireflection property) was larger in the visible range than those in the NIR and mid IR ranges, because shorter-wavelength light was more sensitive to the vertical changes in the refractive index. For example, at 667 nm (visible), the reflectance of F30 was 0.253, 29.9% lower than the control sample reflectance of 0.361. At 2500 nm (IR), the reflectance of F30 was 0.371, 13.3% lower than the control sample reflectance of 0.428. After etching, at 667 nm, the reflectance of the etched F30 sample was 0.021, 92.9% lower than the value of 0.297 for the etched control. At 2500 nm, the reflectance of the etched F30 sample was 0.145, 64.5% lower than the value of 0.408 for the etched control.
The antireflection properties contributed by GO and the graded refractive index were demonstrated for Si substrates. For the studied visible and IR bands, especially for interesting applications to solar cells and detectors, Si is not transparent. Of course, GO can be deposited on transparent substrates such as glass for studying the transmission characteristics. However, the refractive index profile between air and glass is different from that between air and Si. Therefore, the optical impedances of the air/GO/glass and air/GO/Si interfaces are different. Thus, we have not evaluated the transmission characteristics of GO on a transparent substrate.
In general, isopropyl alcohol (IPA) could be added to the KOH etchant for improving the uniformity of pyramid textures, although this method is expensive [28]; thus, we did not add IPA to our KOH etchant. The scanning electron microscope (SEM) photographs in Fig. 6 show that the etched F30 sample (Fig. 6(c)) is characterized by far fewer pyramid textures (bright spots), compared with the etched F0 sample (Fig. 6(a)). However, the etched F30 sample still yields a very low reflectance, which indicates that the graded refractive index contributed by the sub-micrometer sized gentle islands shown in Fig. 4 is the main reason for the reflectance reduction.
Fig. 6 SEM images of the (a) etched F0, (b) etched F1, and (c) etched F30 samples. The dense small GO flakes in addition to some large GO flakes on F30 yield fewer pyramid textures (bright spots) on the etched F30 sample.
5. Application to solar cells
The significantly improved reflectance reduction in Fig. 5 implies that the AR technique described herein can be utilized for near-infrared detectors and solar cells. In what follows, we demonstrate the technique for solar cells. First, 525-μm-thick n-Si wafers with resistivity in the 1–10 Ω-cm range were prepared as the substrates. The polished front sides of the wafers were implanted with BF2+ with the energy of 40 keV and the dose of 5 × 1015 /cm2. The un-polished back sides of the wafers were implanted with P+ with the energy of 40 keV and the dose of 5 × 1015 /cm2. Then, the wafers were annealed at 1050 °C for 30 min for activating the dopant, and the p+nn+ structures were obtained. The p+ sides of the p+nn+ structures were then covered with GO, by using the process described in Section 2 for forming the F0, F1, and F30 cells. Subsequently, the samples were etched in a KOH solution for 1.5 min (a p+nn+ structure without GO deposition was also etched for fabricating the etched control sample). The etching duration (1.5 min) was less than that for the samples in Fig. 4 (5 min) because 5-min-long etching would cause the cells to be broken (the partial p + layer would be totally removed). Actually, in the standard solar-cell process flow, doping is usually performed after the etching process. However, the inherent limitations of the available facilities made it necessary to perform etching after doping. The front and back sides were then screen-printed with Al and Ag pastes, respectively. After co-firing in rapid thermal annealing (RTA), p+nn+ solar cells were ready for measurement of their electrical properties. Figure 7(a) shows the current density-voltage (J-V) characteristics of the etched F0, F1, F30, and control samples. The solar cell performance parameters are listed in Table 1. The etched GO samples (etched F0, F1, and F30) all exhibited better efficiencies than the etched control sample. The enhancement was most remarkable for JSC, and it must be attributed to the antireflection treatment. The JSC trend, F30 > F0 > F1 > control, matches the inverse trend of the reflectance, control > F1 > F0 > F30. The extent of JSC enhancement is closer to the degree of reflectance reduction of the un-etched case shown in Fig. 3 than to the etched case shown in Fig. 5, because the 1.5-min-long etching of p+nn+ cells could not induce undulations as large as those formed during 5-min-long etching of Si substrates. For example, the morphology of the 1.5-min-long etched F30 (p+nn+ Si substrate) is shown in Fig. 7(b), and its roughness of 9.1 nm is smaller than the roughness of 49.8 nm corresponding to the 5-min-long etched F30 sample in Fig. 4(c). Performance enhancement of samples with GO acting as the etching masks is likely to further increase if future technological advances will allow doping to be performed after etching.
Fig. 7 (a) Current density-voltage characteristics of 1.5-min-long etched F0, F1, F30, and control p + nn + cells under AM 1.5 G illumination. (b) An AFM image of the 1.5-min-long etched F30 p + nn + substrate. Shorter etching duration results in smaller undulations, compared with those in Fig. 4(c).
Table 1. Solar Cell Performance Parameters.
6. Conclusion
Antireflection of Si substrates can be achieved by using GO. The GO deposition in itself has already yielded the reflectance reduction because GO possesses an ideal refractive index as a single-layer ARC between air and Si. GO can act as a texturizing mask for further reducing the reflectance. With suitable GO distribution, the IPA-free 5-min-long KOH-etched Si substrate exhibited a relatively low reflectance of 2.1% at the light wavelength of 667 nm. The antireflection effects on practical p+nn+ Si solar cells have also been demonstrated. Due to the limited facilities, only a shorter-duration of 1.5-min-long etching could be performed after the pn junction formation. Notwithstanding, the cell efficiency increased from 7.09% to 7.55% with the contribution of GO. More significant improvement is likely to be obtained with longer-duration etching. Our results demonstrate that the GO-based AR technology bears a great promise for use in infrared detectors and solar cells.
Acknowledgments
This work was supported by the Ministry of Science and Technology of the R.O.C. under contract Nos. NSC 101-2221-E-259 −023 -MY3 and MOST 104-2221-E-259-030-MY3.
References and links
1. H. K. Raut, V. A. Ganesh, A. S. Nair, and S. Ramakrishna, “Anti-reflective coatings: A critical, in-depth review,” Energy Environ. Sci. 4(10), 3779–3804 (2011). [CrossRef]
2. X. H. Xie, Z. Z. Zhang, B. H. Li, S. P. Wang, M. M. Jiang, C. X. Shan, D. X. Zhao, H. Y. Chen, and D. Z. Shen, “Mott-type MgxZn1-xO-based visible-blind ultraviolet photodetectors with active anti- reflection layer,” Appl. Phys. Lett. 102(23), 231122 (2013). [CrossRef]
3. M. Nakahama, H. Sano, S. Inoue, T. Sakaguchi, A. Matsutani, and F. Koyama, “Tuning characteristics of monolithic MEMS VCSELs with oxide anti-reflection layer,” IEEE Photonics Technol. Lett. 25(18), 1747–1750 (2013). [CrossRef]
4. W. Huang, Y. Xue, X. Wang, and X. Ao, “Black silicon film with modulated macropores for thin-silicon photovoltaics,” Opt. Mater. Express 5(6), 1482–1487 (2015). [CrossRef]
5. H. P. Zhou, D. Y. Wei, L. X. Xu, Y. N. Guo, S. Q. Xiao, S. Y. Huang, and S. Xu, “Low temperature SiNx:H films deposited by inductively coupled plasma for solar cell applications,” Appl. Surf. Sci. 264, 21–26 (2013). [CrossRef]
6. S. Duttagupta, F.-J. Ma, B. Hoex, and A. G. Aberle, “Excellent surface passivation of heavily doped p+ silicon by low-temperature plasma-deposited SiOx/SiNy dielectric stacks with optimised antireflective performance for solar cell application,” Sol. Energy Mater. Sol. Cells 120, 204–208 (2014). [CrossRef]
7. L. J.-H. Lin and Y.-P. Chiou, “Optical design of GaN/In(x)Ga(1-x)N/cSi tandem solar cells with triangular diffraction grating,” Opt. Express 23(11), A614–A624 (2015). [CrossRef] [PubMed]
8. Y. Liu, A. Das, Z. Lin, I. B. Cooper, A. Rohatgi, and C. P. Wong, “Hierarchical robust textured structures for large scale self-cleaning black silicon solar cells,” Nano Energy 3, 127–133 (2014). [CrossRef]
9. K.-S. Han, J.-H. Shin, W.-Y. Yoon, and H. Lee, “Enhanced performance of solar cells with anti-reflection layer fabricated by nano-imprint lithography,” Sol. Energy Mater. Sol. Cells 95(1), 288–291 (2011). [CrossRef]
10. U. Schulz, F. Rickelt, H. Ludwig, P. Munzert, and N. Kaiser, “Gradient index antireflection coatings on glass containing plasma-etched organic layers,” Opt. Mater. Express 5(6), 1259–1265 (2015). [CrossRef]
11. K. Jiao, X. Wang, Y. Wang, and Y. Chen, “Graphene oxide as an effective interfacial layer for enhanced graphene/silicon solar cell performance,” J. Mater. Chem. C Mater. Opt. Electron. Devices 2(37), 7715–7721 (2014). [CrossRef]
12. Z. Yu, H. Di, Y. Ma, Y. He, L. Liang, L. L. X. Ran, Y. Pan, and Z. Luo, “Preparation of graphene oxide modified by titanium dioxide to enhance the anti-corrosion performance of epoxy coatings,” Surf. Coat. Tech. 276, 471–478 (2015). [CrossRef]
13. I. Jung, M. Vaupel, M. Pelton, R. Piner, D. A. Dikin, S. Stankovich, J. An, and R. S. Ruoff, “Characterization of thermally reduced graphene oxide by imaging ellipsometry,” J. Phys. Chem. C 112(23), 8499–8506 (2008). [CrossRef]
14. D. K. Cheng, Field and Wave Electromagnetics, 2nd ed. (Addison-Wesley, 1989), p. 457.
15. R. Kumar, A. K. Sharma, M. Bhatnagar, B. R. Mehta, and S. Rath, “Antireflection properties of graphene layers on planar and textured silicon surfaces,” Nanotechnology 24(16), 165402 (2013). [CrossRef] [PubMed]
16. C.-H. Lin, W.-T. Yeh, and M.-H. Chen, “Metal-insulator-semiconductor photodetectors with different coverage ratios of graphene oxide,” IEEE J. Sel. Top. Quantum Electron. 20(1), 3800105 (2014).
17. S. Stankovich, D. A. Dikin, G. H. B. Dommett, K. M. Kohlhaas, E. J. Zimney, E. A. Stach, R. D. Piner, S. T. Nguyen, and R. S. Ruoff, “Graphene-based composite materials,” Nature 442(7100), 282–286 (2006). [CrossRef] [PubMed]
18. M.-A. Tsai, P.-C. Tseng, H.-C. Chen, H.-C. Kuo, and P. Yu, “Enhanced conversion efficiency of a crystalline silicon solar cell with frustum nanorod arrays,” Opt. Express 19(S1Suppl 1), A28–A34 (2011). [CrossRef] [PubMed]
19. L. R. Radovic, “Active sites in graphene and the mechanism of CO2 formation in carbon oxidation,” J. Am. Chem. Soc. 131(47), 17166–17175 (2009). [CrossRef] [PubMed]
20. Y. Cui, Q. Fu, H. Zhang, D. Tan, and X. Bao, “Dynamic characterization of graphene growth and etching by oxygen on Ru(0001) by photoemission electron microscopy,” J. Phys. Chem. C 113(47), 20365–20370 (2009). [CrossRef]
21. C.-Y. Su, Y. Xu, W. Zhang, J. Zhao, X. Tang, C.-H. Tsai, and L.-J. Li, “Electrical and spectroscopic characterizations of ultra-large reduced graphene oxide monolayers,” Chem. Mater. 21(23), 5674–5680 (2009). [CrossRef]
22. C.-H. Lin, W.-T. Yeh, C.-H. Chan, and C.-C. Lin, “Influence of graphene oxide on metal-insulator-semiconductor tunneling diodes,” Nanoscale Res. Lett. 7(1), 343 (2012). [CrossRef] [PubMed]
23. G. T. Reed, Silicon Photonics: The State of the Art (Wiley, Atrium, 2008), p. 182.
24. K. Forberich, G. Dennler, M. C. Scharber, K. Hingerl, T. Fromherz, and C. J. Brabec, “Performance improvement of organic solar cells with moth eye anti-reflection coating,” Thin Solid Films 516(20), 7167–7170 (2008). [CrossRef]
25. K. Askar, B. M. Phillips, Y. Fang, B. Choi, N. Gozubenli, P. Jiang, and B. Jiang, “Self-assembled self-cleaning broadband anti-reflection coatings,” Colloid Surf. A-Physicochem, Eng. Asp. 439, 84–100 (2013). [CrossRef]
26. G. K. Mayer, H. L. Offereins, H. Sandmaier, and K. Kühl, “Fabrication of non‐underetched convex corners in anisotropic etching of (100)‐silicon in aqueous KOH with respect to novel micromechanic elements,” J. Electrochem. Soc. 137(12), 3947–3951 (1990). [CrossRef]
27. J.-Y. Jung, Z. Guo, S.-W. Jee, H.-D. Um, K.-T. Park, and J.-H. Lee, “A strong antireflective solar cell prepared by tapering silicon nanowires,” Opt. Express 18(S3), A286–A292 (2010). [CrossRef] [PubMed]
28. A. K. Chu, J. S. Wang, Z. Y. Tsai, and C. K. Lee, “A simple and cost-effective approach for fabricating pyramids on crystalline silicon wafers,” Sol. Energy Mater. Sol. Cells 93(8), 1276–1280 (2009). [CrossRef]
