Effect of glass frit in Ag paste on the electrical properties of front-side Ag contacts for crystalline-silicon solar cells

Abstract

The effect of glass frit, used in front-side silver pastes, on the electrical properties of front-side silver contacts in silicon solar cells was studied. Glass frits with the same composition but three different degrees of crystallization were prepared by introducing some nucleating agents to provide nucleation sites for the formation of the crystal structure of the glass-frit. The phase structure of the glass frits was characterized using X-ray diffraction. The Ag–Si contacts were investigated using scanning electron microscopy, and the results indicated that glass frit with a moderate crystallization degree facilitated to acquire optimal size crystalline Ag, which was distributed uniformly in the glass layer and silicon substrate. Therefore, an excellent metallized contact was achieved. In addition, the particle size and proportion of glass frit in the paste also had an impact on the electrical properties of the solar cells. The results demonstrated that a silver paste consisting of 3 wt% crystalline glass frit with a particle size of 3.12 μm not only ensured sufficient adhesion of the silver electrode, but also improved the Ag–Si contact to enhance the conversion efficiency of the solar cells.

---

Introduction

With the increasing concerns regarding environmental pollution and global climate change, more attention is being paid to the development and utilization of new energy resources. Crystalline-silicon solar cells are a main source of alternative energy because of their high efficiency and low cost.^1–4 Screen-printed thick-film silver metallization is widely used for the front-side contacts of silicon solar cells because it is more cost- and time-effective than other metallization techniques such as photolithography or light-induced electroplating.^5,6 In general, a silver paste primarily consists of three constituents:^7–9 (a) a powder of silver particles, which provides the conductive phase for collecting electrons owing to silver’s superior conductivity among the noble metals; (b) glass frit, as a binder phase to adhere the silver phase to the silicon substrate and ensure the formation of an Ag–Si contact; (c) an organic medium, which disperses the silver powder and glass frit to obtain a good aspect ratio of the grids and is required to attain the desired rheological properties of the paste.

Several studies^10–12 have reported the role of glass frit in Ag–Si contact formation. During firing, the fluidized glass frit is known to etch through the SiN_x antireflection coating (ARC) and react with the silicon emitter, which enables silver crystallites to nucleate at the glass/silicon interface to form an electrical contact with the emitter. The distribution and size of the silver crystallites formed at the silicon emitter interface are believed to be important factors for achieving good-quality ohmic contact with the emitter.^13,14 Ballif et al.¹⁵ reported that the most appropriate size of silver crystallites is around 200–500 nm. Small silver crystallites cannot adequately penetrate the silicon substrate to form a good ohmic contact with the silicon substrate, and silver crystallites that overgrow into the emitter can cause junction shunting. In addition, Li et al.¹⁶ confirmed that a glass layer formed between the silicon substrate and bulk silver in their study. Because the glass layer thickness is critical to the tunneling mechanism of current flow,¹⁷ Li et al.¹⁸ proposed a plausible tunneling conduction model in which photoelectrons are extracted from the silicon emitter through a thin interfacial glass layer that is richly decorated with nanoscale silver colloids.¹⁹ Therefore, an ideal Ag–Si eutectic structure at the interface between the silver and silicon layers for crystalline-silicon solar cells requires silver crystallites of an appropriate size and a glass layer of an appropriate thickness.

In this work, we focused on the preparation of a functional glass frit that could crystallize in the sintering process. This kind of glass frit can freeze the glass at an early stage to prevent silver crystallites from fully precipitating from the glass to form thinner glass regions. We also used scanning electron microscopy (SEM) and X-ray diffraction (XRD) to study the nucleation mechanism of the crystalline glass frit to determine the route to form an ideal Ag–Si eutectic structure. In addition, we studied other factors through which the glass frit can affect the electrical performance of a silver paste, such as the particle size of the glass frit and the content of glass frit in the paste. Finally, this paper will present a few ways to optimize the silver-paste material for forming high-quality front contacts in crystalline-silicon solar cells.

Experimental

Synthesis of glass-frit powders

The glass frit powders were prepared by traditional melting. A uniform stoichiometric oxide mixture consisting of PbO, TeO₂, SiO₂, CaO, and Al₂O₃ was heated in an aluminum crucible placed in a 1200 °C muffle furnace for 1.5 h. The melt was then removed rapidly from the furnace and poured into deionized water to cool to room temperature. After grinding by ball milling for different cycles (measured in hours), a series of glass frit powders with different sizes was prepared.

Synthesis of organic medium

The organic medium was prepared using mixed solvents, thixotropic agents, a surfactant and a coupling agent. The solvents (methyl carbitol, N-butyl butyrate, terpineol and triethanolamine), thickener (EC), thixotropic agents, surfactant (sorbitan trioleate), and coupling agent (silane coupling agent KH-570) were added into a three-necked flask successively. The materials completely dissolved around 90–110 °C in a water bath with stirring for 3 h and the solution was then cooled to room temperature.

Synthesis of crystalline silver

In a typical procedure, silver nitrate and ascorbic acid were separately dissolved in deionized water. Gelatin was subsequently dissolved in both of these solutions according to the stoichiometric ratio. The silver nitrate and gelatin solution was added to the ascorbic acid and gelatin solution slowly, and stirred for 30 min. The pH of the reaction system was controlled using aqueous ammonia during the reaction process. Finally, the silver particles were filtered off and washed with alcohol to remove any impurities and then dried at 70 °C for 8 h.

Synthesis of silver paste

The silver paste was prepared by mixing the as-prepared microscale Ag crystalline particles, the glass frit powder with a specific size, and the organic medium, using a three-roll mixer. This process was intended to produce silver pastes by using different glass frits with different crystallization degrees, particle sizes and proportions, while the Ag particles and the organic medium remained unchanged. During the mixing procedure, the organic medium was responsible for wetting and dispersing the silver particles and glass frits homogeneously, and then silver pastes with optimal viscosity and thixotropy were acquired. Crystalline Si solar cells were then fabricated using the different silver pastes mentioned above on single-crystalline Cz-Si wafers (125 mm × 125 mm) with an emitter whose resistivity was approximately 60 Ω sq⁻¹, using screen-printing and high-temperature firing steps. The organic medium evaporated during the drying and firing step.

Measurements

The crystal structures of the silver particles and the glass-frit powders were characterized with an X-ray diffractometer (D/Max-3C, Rigaku, Japan) using Cu K_α radiation. The morphology of the silver particles, the glass-frit powders and the silver-electrode samples were examined using a scanning electron microscope (JSM-6390, JEOL, Japan). The distribution of the glass frit’s particle size was measured using a laser particle analyzer (Mastersizer 3000, Malvern). The welding tension between the silver electrode and the Si substrate was measured using FDV-50 force apparatus (Wagner Instruments, USA).

The surface morphology of the cells was obtained by chemical etching, which consisted of two steps: (a) etching by aqua regia (32% HCl, 65% HNO₃, v/v = 3 : 1; t = 1 h) to remove silver from the finger bulk, removing the glass layer at the emitter surface; (b) subsequent etching of the glass layer by a 3% HF solution for 10 min at room temperature, removing Ag crystallites grown in the emitter. The conversion efficiency of the single-crystalline Si solar cells were measured using battery testing equipment (ITA, Baccini Applied Materials, Italy/USA).

Results and discussion

Crystal structure and morphology of silver particles

Fig. 1 displays the XRD pattern of the as-prepared silver particles. All the peaks from the XRD pattern were in accordance with JCPDS card no. 04-0783 and no impurities were observed, indicating that the sample was silver crystals with high purity. Furthermore, the strong intensity of the peaks demonstrated the good crystallinity of the silver particles. Fig. 2 presents the SEM images of the silver particles’s surface morphology. It can be seen from Fig. 2 that the silver particles dispersed uniformly with an average size of around 1.5 μm. According to reports, silver crystallites with diameters of 1–2 μm are appropriate for forming metallized contacts.^19–21

Fig. 1 XRD pattern of the silver particles obtained under optimal reaction conditions.

Fig. 2 SEM image of the silver particles prepared under optimal reaction conditions (a) low resolution and (b) high resolution.

Effect of crystal structure of glass frits on Ag crystallites grown on Si substrate

Analysis of glass structure and effect on silver crystallites in glass layer. Table 1 shows the different constituents of the glass-frit powders, named as GS-1, GS-2 and GS-3. The glass-frit powders were prepared by adding a specific reagent that provided nucleation sites to facilitate the formation of the crystal structure, as shown in Fig. 3. The phases of the glass-frit samples were identified using XRD. As shown in Fig. 3, only a broad peak was detected at around 30° in the pattern of GS-1, indicating the material’s amorphous structure. With the nucleation content increasing from 7% mol to 11% mol, GS-2 and GS-3 patterns had a little background noise due to the presence of a partial glassy phase. Two different crystalline phases appeared in the glass structure that were confirmed to be two kinds of TeO₂ (JCPDS no. 41-0945 and JCPDS no. 09-0433). The peak intensity of GS-3 was stronger than that of GS-2, which was ascribed to the excellent crystallization property resulting from the introduction of more nucleation. According to the reactions proposed by Hong et al.,¹⁹ during the sintering process, the glass frit softened and melted first and then began to dissolve the silver particles. Upon further heating, the lead oxide of the glass frit started to etch the silicon nitride layer via a redox reaction as below:

2PbO + SiN_x → 2Pb + SiO₂ + N₂

PbO + Si → Pb + SiO₂

Table 1 Experimental constituents of glass frit powders labelled as GS-1, GS-2, GS-3

Glass frits	Constituent (mol%)
	PbO	TeO2	SiO2	CaO	Al2O3	Others
GS1	43	37	12	2	4	2
GS2	43	37	9	2	2	7
GS3	43	37	6	2	1	11

---

Fig. 3 XRD patterns of amorphous glass GS-1 and crystalline glass GS-2 and GS-3.

At the same time, the silver particles also sintered or coalesced according to the interdiffusion of atoms between contacted metal particles. Finally, during the cooling process, silver and lead separated according to the phase diagram and silver crystallized in the glass layer and on the silicon surface. As for the typical amorphous structure of glass frit, the melting and cooling process was slow and provided sufficient time for sintering and coalescing of the silver particles, which resulted in large silver crystallites forming during the cooling process and this increased the probability of junction shunting for shallow emitters. During the firing step, the silicon nitride layer was etched by PbO in the glass frit and some silver crystallites embedded in the silicon substrate and formed metallized contacts. Generally speaking, the depth of the p–n junction is 100 nm,¹⁵ so when large crystalline silver penetrated the silicon substrate deeper than 100 nm, the p–n junction was broken down and resulted in the degradation of the electrical properties. In contrast, the crystalline glass accelerated the melting process and promoted its flow, which tended to decrease the coalescence of the silver particles but allowed them to distribute uniformly in the glass layer.

Furthermore, the rapid crystallization of the crystalline glass frit could also freeze silver particles and prevent them form aggregating. As a result, smaller silver crystallites diffused into the glass layer and the silicon emitter uniformly, which reduced the probability of junction shunting for shallow emitters. In addition, more silver precipitates in the glass could favor metal-assisted tunneling through the glass layer. Therefore, the crystalline glass frit should have optimal ingredients for forming good metallized contacts.

Investigation of crystalline glass frits’ effect on silver–silicon contact

In order to investigate the influence of the glass structure on the silver crystallite distribution in the glass layer and on the silicon substrate, the different kinds of glass frits (GS-1, GS-2, and GS-3) were mixed with the same silver particles and organic medium to prepare front-side silver pastes, denoted as GS-1a, GS-2b, and GS-3c for printing onto the silicon substrate to form the positive emitter. Fig. 4a–c show the cross-sectional microstructures of solar cells fabricated with these silver pastes. In Fig. 4a₁–c₁, the bulk silver and glass layer were removed by a chemical etching technique using aqua regia to expose the silver crystallites on the surface of the silicon substrate. As shown in Fig. 4a₂–c₂ after removing the silver crystallites by a second chemical etching with 3% HF, some pits appeared on the silicon substrate as a result of Ag-crystallite penetration.

Fig. 4 SEM images of fired solar cells: (a–c) SEM cross section images of the silver electrode, the insets display the distribution of silver crystallites on the silicon substrates and glass layers; (a₁–c₁) SEM top-view images of the c-Si solar cell after the first chemical etching using aqua regia, demonstrating the silver crystallites attached to the surface of the silicon substrate; and (a₂–c₂) SEM top-view images after the second chemical etching using 3% HF to expose the surface of the emitter, some pits appeared on the silicon substrate.

Based on the above understanding, a comparison of the cross-sectional microstructures of the solar cells can be made using the SEM images displayed in Fig. 4. It can be seen clearly from Fig. 4(a) and (a₁), that much larger silver crystallites were distributed in the glass layer and embedded in the silicon substrate, which would break-down the p–n junction and lead to electric leakage of the solar cell.^10,15 The result was proven by the big pits in the silicon substrate shown in Fig. 4(a₂). As for the GS-2, a large number of small silver crystallites were distributed uniformly in the glass layer, as shown in Fig. 4(b) and (b₁), which offered conducting channels and mild etching to the silicon substrate (Fig. 4(b₂)), and met the requirements for good ohmic contact.¹⁵ The result was confirmed by the shallower pits on the silicon emitter surface in Fig. 4(a₂) than Fig. 4(a₁). However, the sharp diffraction peaks indicated that the crystallinity of GS-3 increased substantially and this further shortened the melting and cooling time. Therefore, much smaller crystalline silver particles were obtained compared with GS-2 (Fig. 4(c) and (c₁)). These smaller silver crystallites precipitated but did not penetrate the glass layer (Fig. 4(c₂)), so an effective metallized contact could not be acquired. A simple schematic illustration of the formation of metallized contacts is presented in Fig. 5. Put simply, the GS-2 was the optimal crystal glass with appropriate proportions, which not only provided conducting channels but also mild etching of the silicon substrate to form good ohmic contacts.

Fig. 5 Schematic diagram of the formation mechanism of conducted tunneling for an electrode emitter by different glass frits.

Effect of particle size of glass frits on electrical properties of solar cells

GS-2 was chosen for studying the effect of particle size on solar-cell electrical properties. The cooled melt was ground by a planetary ball mill for different periods of time to obtain glass-frit powders with different particle sizes. The size distribution of the ground glass frit powders was plotted as a function of ball milling time (h). It was noticed that the size decreased gradually and then increased after 12 h of ball milling due to agglomeration. Based on the curve (Fig. 6) of ground glass-frit powders, we chose the powders that were ground for 6 h (D₉₀ = 7.57 μm), 8 h (D₉₀ = 4.94 μm), 10 h (D₉₀ = 3.12 μm), and 12 h (D₉₀ = 1.58 μm), hereafter referred to as GF-1, GF-2, GF-3, and GF-4, respectively. Front-side silver pastes were prepared using the above glass frits and then screen printed onto silicon wafers.

Fig. 6 The change curve of the glass frit particle size decreased with increasing grinding time.

In the high-energy ball-milling process, collision and squeezing between the glass frit and the zirconia grinding media changed the morphology of the glass-frit powder. The morphologies of the glass frits after different ball milling times are shown in Fig. 7 and the insets in the top right corners are the size distribution of the glass frits measured using a laser particle analyzer. Fig. 7(a) displays the morphology and particle size of GF-1 that experienced 6 h of ball milling. It could be seen clearly that large clumps of the glass-frit powder appeared and the average diameter of the glass frit was 7.57 μm (D₉₀) from the inset. It could also be seen from the inset that there was a big difference between D₁₀ (0.36 μm) and D₉₀ (7.57 μm), indicating that the glass frit particles were distributed nonuniformly. After being ground for 8 h the particle size of the glass frit decreased dramatically to 4.94 μm (D₉₀) and some fine particles appeared, with big clumps and fine particles mixing together, which resulted in nonuniform distribution, as shown in Fig. 7(b). As for GF-3, after being ground for 10 h, the large clumps disappeared completely in Fig. 7(c) and the particle size was distributed uniformly with an average diameter of 3.12 μm (D₉₀). When the glass frit was ground for 12 h, as shown in Fig. 7(d), the particle size of the glass frit decreased significantly and became an ultrafine powder with an average diameter of 1.58 μm (D₉₀). As the specific surface area increased, the particles tended to aggregate seriously during the firing step.

Fig. 7 SEM images and particle size distribution of different glass frits: (a) GF-1; (b) GF-2; (c) GF-3; and (d) GF-4. The insets display the distribution of the glass frit particle size after different ball milling times.

Fig. 8 shows plane-view images of silver fingers on the fired solar cells. Some holes could be observed clearly in Fig. 8a, b, and d, exposing the bare silicon surface. Fig. 8a and b indicates that the particle size of the glass frit was too large and could not match with the silver particles (1–2 μm), therefore, compact contact between the glass frit and the silver particles could not be acquired and this resulted in big holes in the silver electrode. Because the particle size of GS-2 was smaller than GS-1, the smaller holes presented in GS-2. However, when the size of glass frit particles became smaller in GS-4, the particles tended to agglomerate seriously and this decreased the specific surface area during the sintering process. The agglomeration of small glass frit particles also resulted in big holes in the silver electrode, as demonstrated in Fig. 7(d). As for the GS-3, the size of the particles of glass frit were moderate and matched well with the silver particles, meanwhile, the particles could not cause agglomeration during the firing step. Therefore, GS-3 with an average diameter of 3.12 μm (D₉₀) was suitable for the prepared silver paste and formed a smooth and regular silver electrode without holes. In order to further testify the excellence of GS-3, a comparison of the performance parameters of the solar cells was investigated, as shown in Fig. 9, including the conversion efficiency (E_ff), fill factor (FF), open-circuit voltage (V_oc), and series resistance (R_s). Results showed that the silver electrode prepared with GS-3 led to high electrical resistivity and a low conversion efficiency. Therefore, the optimal size of crystalline glass was 3.12 μm with uniform distribution.

Fig. 8 SEM images of the front-side silver fingers screen printed by pastes containing glass frits: (a) GF-1; (b) GF-2; (c) GF-3; and (d) GF-4.

Fig. 9 A comparison of the performance parameters of solar cells: (a) conversion efficiency (E_ff); (b) fill factor (FF); (c) open-circuit voltage (V_oc); and (d) series resistance (R_s).

Effect of glass-frit-powder content in paste on electrical properties of solar cells

The as-prepared crystalline glass frit (GS-3) was mixed with a certain amount of the organic medium and silver particles to prepare a series of front-side silver pastes with a different content of glass frit, as shown in Table 2. The glass frit content was decreased from 4.0 wt% to 2.0 wt%, and was denoted as GJ1 to GJ5. The adhesion of the silver electrode prepared with different silver pastes containing a different content of glass frit (GS-3) was tested with FDV-50 force apparatus. The results are demonstrated in Fig. 10, the left bar graphs are the tensile force between the silver electrodes and the silicon substrates and the right are the silver electrodes that experienced drawing by a tension test. As Fig. 10 presents, when the glass-frit content in the paste was 2 wt% (GJ5), the adhesion was too low (1.26 N mm⁻²) and the solder ribbon fell off easily without being broken. As the content of GS-3 increased to 3 wt% in GJ3, the viscosity of the silver paste was moderate and bound the silver particles effectively to the silicon substrate, providing appropriate adhesion (3.39 N mm⁻²) to ensure that the solder ribbon was firmly affixed to the wafer throughout the long working life of the encapsulated solar-cell panel. When the proportion of glass frit was increased to 4 wt% in GJ1, the tension break force reached 7.76 N mm⁻², which was too strong to result in the solar cell’s fragmentation. Therefore, with the GS-3 content increasing in the silver paste, the tension between the silver electrode and silicon substrate subsequently increased. Although high adhesion is beneficial for prolonging the working life of solar cells, the series resistance was also increased, lowering the conversion efficiency of the cells, as shown in Fig. 11. Hence, the 3.0 wt% content of GS-3 in the silver paste could not only provide moderate tension but also led to low series resistance, which was the optimum composition.

Table 2 The different content of components in the paste and corresponding conversion efficiency

Paste	Composition
	Silver particles (wt%)	Glass frit (wt%)	Organic medium (wt%)	Conversion efficiency (Eff%)
GJ1	90	4.0	6.0	12.51
GJ2	90	3.5	6.5	16.81
GJ3	90	3.0	7	18.48
GJ4	90	2.5	7.2	17.92
GJ5	90	2.0	7.4	17.60

---

Fig. 10 The change of the Ag fingers’ adhesion with increasing glass frit content. The left bar graphs show the tensile force between the silver electrodes and the silicon substrates; the right bar graphs show the silver electrodes that experienced drawing by a tension test.

Fig. 11 Curves of the electrical properties of solar cells screen printed with pastes GJ1, GJ2, GJ3, GJ4, and GJ5.

A few reasonable assumptions can be proposed to explain the effect of glass frit on the adhesion of cells. The glass frit in the silver paste is melted first during firing, promoting the silver particles to melt rapidly. Hence, a paste with low glass-frit content cannot provide enough contact points to completely melt the particles. The main reason for the decreased adhesion was that the solder ribbon could not be tightly welded to the fingers because the silver particles of some areas melted unevenly. With the glass-frit content increasing, the adhesion of fingers also gradually enhanced, demonstrating that more contact points for melting among silver particles offered them a chance to integrate more easily. However, a thicker glass layer deposited on the silicon substrate due to the high content of glass frit in the paste, and the photo-generated carriers cannot be transferred through the glass layer and create a conducting channel from silicon substrate to silver electrode. Consequently, the adhesion is high enough with higher glass-frit content, while the electrical properties of the cells degrade, as shown in Fig. 11.

It could be deduced that a low glass-frit content of 2.0–2.5 wt% in the silver paste provided an insufficient liquid phase to wet the silver particles. The dense fingers could not be formed because of the low driving force. Hence, the nonuniform silver crystallites appearing on the silicon substrate would not draw and collect electrons smoothly and resulted in high series resistance. When the glass-frit content was increased to 3.5–4.0%, dense fingers were formed during sintering because the silver particles melted completely. Fig. 11 shows that these parameters were not ideal, however, it was well known that the amount of silver crystallites formed on the Si substrate was determined by the glass frit in the paste. A large amount of silver crystallites were beneficial in drawing and collecting electrons. However, a thicker glass layer precipitated between the silicon substrate and the bulk silver with increasing glass frit content, which blocked the electrons from being drawn owing to the absence of assisted multi-step tunneling in the glass layer. Meanwhile, the thick glass layer reduced the probability of contact between the silver crystallites and the bulk silver. Our results showed that around 3 wt% of glass frit in the silver paste yielded excellent electrical properties of the solar cells.

Conclusions

As an important component in silver paste, glass frit powder played a significant role in the formation of Ag–Si contacts in crystalline-silicon solar cells. Glass frits with different degrees of crystallization were prepared by introducing a nucleating agent to provide nucleation sites for the formation of the crystal structure. Through SEM observations of the cross-sectional and surface microstructures of solar cells, it was speculated that a glass frit with moderate crystallization could control the growth and distribution of silver crystallites in the glass layer and silicon substrate, which facilitated conduction tunneling and provided optimal etching conditions for the silicon substrate. Meanwhile, it could be seen from the optimization of the parameters that 3 wt% of glass frit with a particle size of 3.12 μm in the silver paste ensured the best electrical performance of the solar cells. Therefore, the preparation of crystalline glass frit with different diameters and proportions provided a new way for improving the conversion efficiency of solar cells.

Acknowledgements

This work was supported by National Hi-Tech Research and Development Program (863) Key Project of China (No. 2012AA050301-SQ2011GX01D01292), China International Science and Technology Cooperation Special Program (No. 2010DFB60400), Major Science and Technology Innovation Subject Fund of Shaanxi Province (No. 2010ZKC03-14) and Xi’an Industrial Technology Innovation Project-technology transfer promoting program (No. CX1242, CXY1123-5, CX12182-3, CX12182-2).

Notes and references

1. J. Zhou, W. Gan, Y. Li, B. Huang and C. Yang, J. Mater. Sci.: Mater. Electron., 2015, 26, 234–241.
2. J. Wang, L. Zhou, H. Zhu, R. Yang, Y. Zhou, L. Liu and J. Chen, Photonics Res., 2015, 3, 58–62.
3. G. Zheng, Y. Tai, H. Wang and J. Bai, J. Mater. Sci.: Mater. Electron., 2014, 25, 3779–3786.
4. J. Wu, P. Cao, T. Pan, Y. Yang, C. Qiu, C. Tremblay and Y. Su, Photonics Res., 2015, 3, 9–14.
5. W. Li, T. Wu, R. Jiao, B. P. Zhang, S. Li, Y. Zhou and L. Li, Colloids Surf., A, 2015, 466, 132–137.
6. B. Blank, C. Ulbrich, T. Merdzhanova, C. Zahren and B. E. Pieters, Sol. Energy Mater. Sol. Cells, 2015, 143, 1–8.
7. J. T. Tsai and S. T. Lin, J. Alloys Compd., 2013, 548, 105–109.
8. J. Y. Huh, K. K. Hong, S. B. Cho, S. K. Park, B. C. Lee and K. Okamoto, Mater. Chem. Phys., 2011, 131, 113–119.
9. A. S. Ionkin, B. M. Fish, Z. R. Li, M. Lewittes, P. D. Soper, J. G. Pepin and A. F. Carroll, ACS Appl. Mater. Interfaces, 2011, 3, 606–611.
10. M. M. Hilali, S. Sridharan, C. Khadilkar, A. Shaikh, A. Rohatig and S. Kim, J. Electron. Mater., 2006, 35, 2041–2047.
11. E. Cabrera, S. Olibet, J. Glatz-Reichenbach, R. Kopecek, D. Reinke and G. Schubert, J. Appl. Phys., 2011, 110, 114511.
12. C. H. Lin, S. Y. Tsai, S. P. Hsu and M. H. Hsieh, Sol. Energy Mater. Sol. Cells, 2008, 92, 1011–1015.
13. E. Cabrera, S. Olibet, J. Glatz-Reichenbach, R. Kopecek, D. Reinke and G. Schubert, Energy Procedia, 2011, 8, 540–545.
14. D.-Y. Shin, J.-Y. Seo, H. Tak and D. Byu, Sol. Energy Mater. Sol. Cells, 2015, 136, 148–156.
15. C. Ballif, D. M. Huljić, G. Willeke and A. Hessler-Wyser, Appl. Phys. Lett., 2003, 82, 1878–1880.
16. Z. G. Li, L. Liang and L. K. Cheng, J. Appl. Phys., 2009, 105, 066102.
17. A. S. Ionkin, B. M. Fish, Z. Rick Li and L. Liang, Sol. Energy Mater. Sol. Cells, 2014, 124, 39–47.
18. Z. G. Li, L. Liang, A. S. Ionkin, B. M. Fish, M. E. Lewittes, L. K. Cheng and K. R. Mikeska, J. Appl. Phys., 2011, 110, 074304.
19. K. K. Hong, S. B. Cho, J. S. You, J. W. Jeong, S. M. Bea and J. Y. Huh, Sol. Energy Mater. Sol. Cells, 2009, 93, 898–904.
20. G. Guo, W. Gan and F. Xiang, J. Mater. Sci.: Mater. Electron., 2011, 22, 527–530.
21. J.-T. Tsai and S.-T. Lin, J. Alloys Compd., 2013, 548, 105–109.

---