Investigation of the mechanism of the Ag/SiN_x firing-through process of screen-printed silicon solar cells

Abstract

The mechanism of a Ag/SiN_x firing-through process in the manufacture of multi-crystalline silicon solar cells has been studied. The effect of firing temperature on the electrical performance of screen printed multi-crystalline silicon solar cells, with a conversion efficiency of up to 17.2%, has been investigated. It is revealed that with an increase in firing temperature both the series resistance and shunting resistance of the solar cells decrease monotonically, while the reverse leakage current rises gradually. SEM and EDX are used to study cross-sections of the Ag/Si interface under the fingers. It is revealed that hexagonal like silver crystallites are formed due to the chemical reaction between SiN_x and Ag₂O in the Ag paste during the firing process, through which the direct interconnection between the emitter and silver particles contained in the paste is achieved. The physical process of the firing-through is discussed. Moreover, the diffusion coefficients at different temperatures are obtained by fitting the diffusion profiles.

---

Introduction

In the photovoltaic industry, screen-printing has been the most commonly used method to deposit metallic paste onto silicon with SiN_x for front metallization. For large scale manufacturing of solar cells, a thin layer of SiN_x is needed to passivate the dangling bonds and reduce the light reflection at the front surfaces of the silicon wafers to increase the light–electricity conversion efficiency. Silver fingers are then printed on the SiN_x layer for soldering in the inter-connection between cells for making modules. Therefore, due to the electrical insulation of SiN_x, a so-called firing process is essential to drive the Ag through the SiN_x layer to generate the carriers in the cells which are transmitted out. The firing process is a rapid annealing process which occurs at a high temperature.¹

For screen-printed solar cells, it is well known that the peak temperature in the firing process is extremely important for the performance of the solar cells. In particular, the fill factor (FF) of the solar cell is strongly affected by both the series resistance and the shunting resistance. As an important part of the series resistance, the contact resistance depends on the behavior of the Ag/Si contact interface after the firing process.² In addition, the junction leakage and shunting characteristics are also heavily affected by the firing process. It has been concluded that in the manufacturing of solar cells, there exists an optimal firing temperature to reasonably ensure good front contact and low reverse leakage current for the junction.

The current commercial silver paste consists of silver particles, glass frit, solvents, binders and relevant oxides including Ag₂O etc. With firing, a good ohmic contact between the silver fingers and the emitter of the cells is formed. It is generally believed that with the increase in reaction temperature, the SiN_x layer is gradually etched and a glass layer (mainly SiO₂) is formed between the emitter and the silver finger.³ The glass layer at the interface is assumed to be insulating and is therefore responsible for high contact resistance. To date, two main hypotheses have been proposed to explain the mechanism of current transport from the n-type emitter of the silicon solar cells to the silver fingers: (1) the current is supposed to be transported via (local) direct interconnections between the silver fingers and the silicon,^4,5 (2) the current is supposed to be transported via a multi-step tunnelling process from the emitter to the silver fingers through the glass layer.^6,7

In fact, quite a few reports have been trying to reveal the physical picture of the Ag/Si interface. Butler and his colleagues have investigated the contact potential between the Ag and Si interface both experimentally and theoretically, and it was found that the Schottky barrier height (SBH) depends upon the orientation of the Si surface.⁸ In detailed work, including observations of the microscopic structure of the formed front contacts on flat and textured mono-crystalline Si solar cell surfaces with SEM, and also measurements of the contact resistivity after sequential etch-back of the metallization from different silver thick film pastes, Cabrera et al. presented convincing experimental evidence that the major current flow into the silver finger is through these direct contacts.^9,10 Moreover, some other works have studied nanoscale silver crystals at the interface of silver thick film contacts on n-type silicon.^11–13

Although the works mentioned above and those presented in some other reports provide some information,^14–16 a complete and clear picture of the contact formation between silver and the Si emitter is still not available. Furthermore, little work has been carried out quantitatively on the process of current transport from the emitter of the solar cells to the silver fingers, and on the precise correlation between silver and the quality of the ohmic contact, in particular, a report on practical textured multi-crystalline solar cells has not been found. Based on cross-sectional observations using SEM, we present here the first and preliminary investigation on the mechanism of the Ag/SiN_x firing-through process, results showing the influence of firing temperature on the electrical properties, and the microstructure of the Ag/Si contact interface, giving a complete physical picture of the firing process.

Experimental

Solar cell samples were fabricated on a typical industrial production line, using 156 × 156 mm² p-type polycrystalline silicon wafers. The wafers were first textured or saw damage-etched by acid etching using a mixed HNO₃–HF system. The emitter was formed by standard POCl₃ diffusion with 45 Ω □⁻¹ sheet resistance, and then selective-emitter technology was used in order to obtain good cell performance. The SiN_x anti-reflection and passivation layer was deposited via plasma enhanced chemical vapour deposition (PECVD) with an average thickness of 82 nm and a refractive index of 2.08. The front and back metallization process was carried out by screen-printing commercial silver and aluminium pastes, respectively. Finally, the front and back contacts were simultaneously formed during a fast firing process in an IR heated belt furnace.

There were five groups of sample cells with firing temperatures of 745 °C, 775 °C, 805 °C, 835 °C, and 865 °C, while the other preparation conditions were identical. The electrical performance of the sample cells was characterized with an I–V tester. Field-emission scanning electron microscopy (SEM) and energy dispersive X-ray spectrometry (EDX) were used to study the cross-section of the Ag/Si interface.

Results and discussion

The electrical performance of the sample cells is shown in Table 1. It can be noted from the table that the open circuit voltage (V_oc) and the short circuit current (I_sc) are almost the same between samples, which is due to the consistently similar quality of the start wafer materials and identical manufacturing conditions, except for the firing temperature. Fig. 1(a)–(d) show the firing temperature dependence of the individual electrical parameters of efficiency and fill factor, series resistance, shunting resistance and reverse leakage current, respectively.

Fig. 1 Temperature dependence of the electrical properties of the sample cells. (a) Fill factor (FF) and conversion efficiency (E_ff), (b) series resistance (R_s), (c) shunting resistance (R_sh) and (d) reverse leakage current (I_rev1 and I_rev2 correspond to the bias voltage ¬10 V and ¬12 V).

Table 1 The electrical performance of the solar cells at different firing temperatures

T (°C)	Isc (A)	Voc (V)	FF (%)	Eff (%)	Rs (Ω)	Rsh (Ω)	Irev1 (A)	Irev2 (A)
745	8.426215	0.624533	78.17645	16.90490	0.002056	210.8812	0.102252	0.418127
775	8.434872	0.624247	78.32449	16.94662	0.001990	149.0577	0.156951	0.511428
805	8.47421	0.626127	78.89935	17.20229	0.001923	138.2943	0.162102	0.521614
835	8.466094	0.621599	78.67301	17.01256	0.001840	106.0098	0.170251	0.536030
865	8.280422	0.623448	78.61310	16.67700	0.001720	62.11593	0.236446	0.963737

---

It can be seen from Fig. 1(a) that with the rise in temperature the values for both the conversion efficiency and fill factor increase first and then decrease after a peak is reached. It is demonstrated that both the series resistance and shunting resistance decrease monotonically with the increase in firing temperature, as shown in Fig. 1(b) and (c). As we know, the lower the series resistance is, the better the cell performs. Therefore, from the ohmic contact point, a high firing temperature is desired. However, on the other hand, a firing temperature which is too high is harmful to the performance of the cells because over diffusion of silver into the surface region of the cells can cause a higher probability of shunting at the junction. Therefore, an optimal firing temperature exists to compromise for the series resistance and shunting resistance, which is indicated in Fig. 1(a). Fig. 1(d) shows the dependence of the reverse currents at bias voltages of −10 V and −12 V, on the firing temperature. Clearly, the reverse leakage currents increase with the rise in firing temperature. A higher reverse current is caused by a higher firing temperature, which is consistent with the observations of the firing temperature dependence of the shunting resistance, as discussed above.

In order to better understand the effect of firing temperature on series resistance, shunting resistance and reverse leakage current, it is helpful to study the interface region formed by the diffusion of silver at different firing temperatures. The areas underneath the fingers of the sample cells obtained at different firing temperatures were analyzed with EDX. From the EDX data, the profiles of silver diffusion into the surface areas were plotted, as shown in Fig. 2. From Fig. 2, it can be seen that the diffusion of silver mainly concentrates in the surface area of the emitter. A higher temperature corresponds to higher silver content and this is responsible for a higher reverse leakage current and a lower shunting resistance.

Fig. 2 Semi-logarithmic plot of the silver diffusion profiles in the surface of silicon.

To fit the profile of silver, it is supposed that the surface concentration of silver is unchanged in the firing process. According to the diffusion theory,¹⁷ the diffusion concentration can be written as

where N_s is the surface concentration, D is the diffusion coefficient, x is the depth and t is the time. From the profiles of different firing temperatures in Fig. 2, the diffusion coefficient changes with firing temperature can be obtained and are shown in Fig. 3.

Fig. 3 Temperature dependence of the diffusion coefficients from experiments at different firing temperatures, with comparison to the theoretical values for mono-crystalline silicon.

The diffusion coefficient can be written as

where D₀ is a pre-exponential constant, E_a is the activation energy, and k is the Boltzmann coefficient. For single crystalline silicon, according to the literature,¹⁸ D₀ = 1 × 10⁻³ cm² s⁻¹ and E_a = 1.6 eV. Therefore, the temperature dependence of the diffusion coefficient can be plotted from eqn (2) and the plotted line is also shown in Fig. 3. Convincingly, the diffusion coefficient obtained from multi-crystalline silicon is nearly consistent with that from mono-crystalline silicon.

The cross sections of the silver diffused areas under the fingers of the sample cells, obtained at different firing temperatures, were further studied with SEM. Fig. 4(a) and (b) show the SEM cross-sectional images of the samples at 745 °C and 805 °C, respectively. The big voids observed in the figures are due to the missing silver particles, caused during sample preparation for the cross sectional studies. Although there is a non-flat interface due to the feature of the textured surface on the solar cells, it is clear that there exists a thin layer between the silver and the silicon. It was revealed with EDX that there is a large amount of nitrogen (about 11.75 atm% at the cross mark point) in the thin layer of the sample in Fig. 4(a), the EDX result is shown in Fig. 5. However, no nitrogen was detected in the thin layer of the sample in Fig. 4(b), in accordance with previous studies.¹⁹ It can be seen that for both cases, in addition to the larger silver particles, there are a lot of small hexagonal like crystallites embedded into the thin layers. Moreover, it can be seen that the hexagonal like crystallites in Fig. 4(b) are larger than those in Fig. 4(a).

Fig. 4 Cross-sectional images of the Ag/Si contact interface. (a) Firing at 745 °C, (b) firing at 805 °C (the optimal firing condition) and (c) simple current transport model.

Fig. 5 The analysis result of EDX at the cross mark point.

The interpretation of the interaction between the silver and the SiN_x thin layer before diffusing into the silicon is interesting and meaningful. For the larger silver particles located at the void positions, they are obvious from the original paste. However, the small hexagonal like crystallites could only come from the reaction during the firing. From the EDX analysis of the sample in Fig. 4(b), the hexagonal like crystallites were found to be silver in content, as shown in Fig. 6. According to the literature,²⁰ the most likely redox reaction seems to be:

2Ag₂O_{(in glass)} + SiN_x(s) → 4Ag_(s) + SiO_{2(in glass)} + x/2N₂ (3)

Fig. 6 Elemental mapping of the microstructure by EDX. (a) SEM image of the interface, (b) mapping of silicon and (c) mapping of silver.

Therefore, the hexagonal like silver crystallites are formed due to the reaction of Ag₂O and SiN_x. It is known that firing at a higher temperature could increase the aggressiveness of the etching reaction between the paste and the SiN_x layer. This could explain the relatively larger silver crystallites in Fig. 4(b) than those in Fig. 4(a), due to the higher firing temperature of the samples in Fig. 4(b). It is also noted that for both cases all hexagonal like silver crystallites are located at the interface between the larger silver particles and the emitter. With the increase in reaction temperature, the SiN_x layer is gradually etched from the top surface and the original silver particles from the paste are immersed in molten glass. At the same time the hexagonal like silver crystallites are nucleated and grow up at the reaction front between the Ag₂O and SiN_x. As long as the reaction is progressing, the reaction front moves closer to the emitter, as do the hexagonal like silver crystallites while they grow. After the completion of the reaction, the SiN_x could be completely consumed and all the hexagonal like silver crystallites reach their destination at the emitter surface. However, because the glass composites are always in a molten state at the reaction temperature, all the hexagonal like silver crystallites are embedded as shown in the figures. Finally, good ohmic contact could be achieved through the electrical connection between the large silver particles and the emitter via the small hexagonal like silver crystallites, as shown schematically in Fig. 4(c). At the same time, it is reasonable that the silver crystallites grow and the contact area becomes larger with the increase in firing temperature. Consequently, with increasing firing temperatures, the series resistance reduces because better direct interconnection between the emitter and the silver finger is realized for a larger contact area. However, higher temperature means more silver diffusion into the emitter, giving rise to a higher chance for the junction area to be shunted. Therefore, the reverse leakage current would increase and the shunting resistance would decrease due to greater silver diffusion.

Conclusion

The effect of firing temperature on the electrical performance of multi-crystalline silicon solar cells has been investigated. It was found that with the temperature increase both the efficiency and fill factor increase and then decrease gradually after a peak value is reached. A higher firing temperature could induce lower series resistance due to a larger contact area between the silver particles and the emitter. A clear picture of the firing process, i.e. that the hexagonal like silver crystallites are nucleated and grow up at the reaction front between the Ag₂O and SiN_x with movement closer to the emitter, has been revealed. From studies of the microstructure, it is confirmed that the current transport mechanism is through interconnections between the emitter and silver fingers via hexagonal like silver crystallites.

Acknowledgements

We acknowledge support from NKPBRC (2010CB923404) and NNSFC (nos 11274153, 11204124, 51202108).

References

1. E. Cabrera, Doctoral thesis, Constanz University, 2013.
2. K. Firor, S. J. Hogan, J. M. Barrett, and R. T. Coyle, Proceedings of the 16th IEEE Photovoltaic Specialists Conference, 1982, pp. 824–827.
3. G. Schubert, F. Huster and P. Fath, Sol. Energy Mater. Sol. Cells, 2006, 90, 3399–3406.
4. G. C. Cheek, R. P. Mertens, R. Van Overstraeten and L. Frisson, IEEE Trans. Electron Devices, 1984, 31(5), 602–609.
5. R. J. S. Young and A. F. Carroll, Proceedings of the 16th EC PVSEC, Glasgow, 2000, pp. 1731–1734.
6. M. Prudenziati, L. Moror, B. Morten and F. Sirotti, Act. Passive Electron. Compon., 1989, 13, 133–150.
7. T. Nakajima, A. Kawakami and A. Tada, Int. J. Hybrid Microelectron., 1983, 6(1), 580–586.
8. K. T. Butler, P. E. Vullum, A. M. Muggerud, E. Cabrera and J. H. Harding, Phys. Rev. B: Condens. Matter Mater. Phys., 2011, 83, 235307.
9. E. Cabrera, S. Olibet, J. Glatz-Reichenbach, R. Kopecek, D. Reinke and G. Schubert, J. Appl. Phys., 2011, 110, 114511.
10. E. Cabrera, S. Olibet, J. Glatz-Reichenbach, R. Kopecek, D. Reinke and G. Schubert, Energy Proc., 2011, 8, 540–545.
11. H. Yang, X. D. Lei, H. Wang and M. Q. Wang, Clean Technol. Environ. Policy, 2013, 15, 1049–1053.
12. S. Kontermann, R. Preu and G. Willeke, Appl. Phys. Lett., 2011, 99, 111905.
13. S. Kontermann, A. Ruf, R. Preu and G. Willeke, Appl. Phys. Lett., 2012, 101, 121907.
14. P. Narayanan Vinod, RSC Adv., 2013, 3, 14106–14113.
15. S. Kontermann, G. Willeke and J. Bauer, Appl. Phys. Lett., 2010, 97, 191910.
16. P. Narayanan Vinod, RSC Adv., 2013, 3, 3618–3622.
17. L. R. Yan, W. Zhou and D. G. Liu, IC Fabrication, Tsinghua University Press, 2010, pp. 68–69.
18. S. W. Jones, Diffusion in silicon, IC Knowledge LLC, April 25, 2008.
19. C. H. Lin, S. Y. Tsai, S. P. Hsu and M. H. Hsieh, Sol. Energy Mater. Sol. Cells, 2008, 92, 1011–1015.
20. 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.

---