Materials Letters 231 (2018) 146–149 Contents lists available at ScienceDirect Materials Letters journal homepage: www.elsevier.com/locate/mlblue Direct soldering of screen-printed Al-paste layer on back-side of silicon solar cell using SnAg solder Weibing Guo a,b, Xinran Ma a, Mingze Gao a, Jiuchun Yan a,⇑ a b State Key Laboratory of Advanced Welding and Joining, Harbin Institute of Technology, Harbin 150001, China School of Material Science and Engineering, Hebei University of Technology, Tianjin 300130, China a r t i c l e i n f o Article history: Received 15 August 2017 Received in revised form 9 July 2018 Accepted 28 July 2018 Available online 30 July 2018 Keywords: Solar energy materials Sn-Ag solders Ultrasonic-assisted soldering Microstructure Electrical properties a b s t r a c t Direct joining of Al back surface field (Al BSF) in a polycrystalline silicon solar cell using a green Sn-3.5Ag solder by the assistance of ultrasound was investigated. SEM, peel force, electrical resistance and open-circuit voltage (Voc) tests were used to study the effect of ultrasonic action time on the performance of the solar cell. The results show that with the increasing of ultrasonic action time, more Al particles in the paste residual layer dissolved into the solder layer. The dissolved Al existed in the bond metal as forms of a-Al and Ag-Al compound phases. The solder bonded directly with the Al-Si eutectic layer under ultrasonic action time of 6 s. The resistance of the joints was 1.27 mX and the peel force could reach as high as 1.44 N/mm. The Voc of solar cell was 524 mV, which was higher than the 467 mV of solar cell soldered with Ag electrode. Ó 2018 Elsevier B.V. All rights reserved. 1. Introduction Crystalline silicon solar cell is trending towards high efficiency, low cost  and environmental friendliness. In the packaging process, the rear contact of cell requires to be bonded with the top contact of another cell to conduct current in a solar power module . There will be three layers after sintering aluminum powder suspension on the rear side of Si: the Al-doped p+-layer, the eutectic layer and the layer of paste residuals. The sintering process realizes the doping on the rear side of solar cells and produces Al back surface field [3,4]. However, the Al BSF is very difficult to solder. In order to make soldering easier, silver busbars are printed on the back surface, and Cu interconnector ribbons coated with SnPb solder are used to join the silver busbars . The metallization processes are complicated and the noble metals of silver and copper are used. Moreover, the Sn-Pb solders are not environment-friendly. In addition, the silver busbars prevent the formation of BSF, which would raise the recombination rate  and lead to a lower open-circuit voltage . If the solder can form direct bond with the Al BSF underneath the metallization, the metallization processes can be simplified, costly Ag can be avoided and the performance and reliability of solar power modules can be improved. However, the oxide film on the surface of Al is very stable, resulting in poor solderability. In recent years, Al and its ⇑ Corresponding author. E-mail address: email@example.com (J. Yan). https://doi.org/10.1016/j.matlet.2018.07.127 0167-577X/Ó 2018 Elsevier B.V. All rights reserved. alloys were successfully soldered with Sn-based alloys. Pure Sn was used to solder Al 2024 directly assisted by ultrasound [8,9]. Sn-3.5Ag solder was also used to solder Al 1070 . Recently, green solders such as Sn-Zn, Sn-Ag and Sn-Ag-Cu have been developed to join the Ag electrodes of solar cells [11,12] and Al ribbon was developed for lower cost than the tin-coated Cu ribbon . Here, we soldered the Al BSF and Al ribbon directly using a Sn-3.5Ag solder with the assistance of ultrasonic waves. The effect of ultrasonic action time on the microstructure, mechanical properties, electric conductivity and photoelectric properties of joints was investigated in details. The work is helpful to realize the packaging of solar cells with low cost, high performance and less pollution. 2. Experimental procedure The commercial polycrystalline solar cells were provided by Shanghai Suiying Photovoltaic Co. Ltd. The solar cell was 200 lm in thickness and the nominal Voc was 0.5 V. The rear side was printed with Al pastes. Commercial Sn-3.5Ag (wt%) alloys were used as filler metals, whose melting point is 221 °C. Pure Al 1060 alloy was used as ribbons, which were 100 lm in thickness and 2 mm in width. In the process of soldering, 30 mg solders were placed on the surface of Al BSF. Then the ultrasonic waves were applied into the melting solders with a ultrasonic iron. The ultrasonic generator in the iron was equipped with a power supply, a vibrator, a W. Guo et al. / Materials Letters 231 (2018) 146–149 transducer, a booster and a sonotrode. The frequency of the ultrasonic vibration was 60 kHz with an amplitude of 3 lm and the power of induction heating was 60 W. After the application of ultrasound for 2 s, 4 s and 6 s during soldering, the solder layer had dimensions of 6 4 mm2. Then the interconnector ribbon 147 was joined to the solder layer on Al BSF surface with the ultrasonic iron. The microstructures of the interfaces were observed by scanning electron microscopy (FEI-Quanta 200F) equipped with an energy dispersive X-ray spectroscopy (EDS). X-ray diffraction (XRD) tests were carried out by Bruker D8 Advance multi crystal diffractometer. A four-point resistance measurement was used to evaluate the electrical conductivity of the joint, as shown in Fig. 1. The values of resistance were provided by KEITHLEY-2420 multifunctional digital meter. Conventional 90° peel tests were performed to evaluate the mechanical properties of the joints using an adapted method of DIN EN 50 461 . The tests were conducted using an microforce tester (Instron-5948). The Voc of solar cell was determined with Oriel Sol1A sunlight simulator (p = 23.5 mW/cm2) and Versa STAT 3 electrochemical workstation. 3. Results and discussion Fig. 1. (a) and (b) Schematic of 4-point resistance measurement. Fig. 2a–c show the microstructure of the interfaces and the element distribution along the corresponding lines under the ultrasonic action for 2 s, 4 s, and 6 s. The thickness of the original Al paste residual layer was about 35 lm and this layer was composed of loosely sintered Al particles. When the ultrasonic action time was 2 s, the solder could wet and bond with the Al layer. The solder layer was mainly composed of b-Sn (Fig. 2a). The EDS results show the phase at point G had compositions of 64% Ag, 32% Al and 4%Sn (at.%), which can be identified as Ag2Al phase. The Al paste residual layer was still complete, as the element distribution along line AB indicates that the existence of the Al particle layer, with the thickness of about 35 lm. As the ultrasonic action time increased to 4 s, more Al particles dissolved into the solder layer and the solder had Fig. 2. Microstructure of the interfaces and the element distribution along the corresponding lines under the ultrasonic action for (a) 2 s, (b) 4 s and (c) 6 s; EDS Color maps for (b) 4 s of (d) Al, (e) Si, (f) Ag, and (g) Sn; (h) XRD patterns of joints for different ultrasonic action time. 148 W. Guo et al. / Materials Letters 231 (2018) 146–149 penetrated into the Al paste residual layer, as shown in Fig. 2b. According to the element distribution along line CD, the Al-Si eutectic layer and Al particle layer can be identified and the penetration depth of the solder was about 20 lm. Fig. 2d–g present the distribution of Sn, Ag, Si, Al around the joint in Fig. 2b. It could be learnt from Fig. 2d that Al diffusion layer in the Si substrate existed and it formed Al BSF. Sn penetrated through the Al particles (Fig. 2f) and Ag concentrated on the interface of Al particle layer and solder alloy (Fig. 2g). Thus, Ag-Al phases existed near this interface, which could also be seen in Fig. 2b by distinct contrast. After further increasing the ultrasonic action time to 6 s, the large amount of Al particles could no longer be observed. Only few incomplete Al particles existed near the interface, as shown in Fig. 2c. Almost the entire Al paste residual layer was dissolved into the solder layer. The phase at point I had compositions of 60% Ag, 37%Al, and 3%Sn (at.%), which can be identified as Ag3Al2 phase. The chemical compositions of phase at point H were: 92% Al, 2% Ag, and 6%Sn (at.%), indicating that the phase was a-Al phase. The EDS scanning result along line EF in Fig. 2c shows the dissolution and diffusion layer of Al and Si at the interface, indicating that the Al-Si eutectic layer and p+-layer were not broken during the soldering process. XRD tests were carried out to provide evidence of Ag-Al phases in the solder layer, as shown in Fig. 2h. It could be seen that Sn phase (JCPDS 04-0673, I41/amd space group, a = b = 0.5831 nm, c = 0.3182 nm) existed, and as ultrasonic time was prolonged and more Al (JCPDS 65-2869, Fm-3 m space group, a = b = c = 0.40497 nm) particles dissolved into the solder alloys, Ag2Al (JCPDS 14-0647, P63/mmc space group, a = b = 0.2885 nm and c = 0.4624 nm) reduced gradually and disappeared. Small amount of Ag3Al2  was found when the ultrasonic action time was 6 s. The formation of Ag-Al compound is complicated for different atomic ratios of Ag and Al. As soon as Al dissolved into the solder, Al would react with Ag and Ag-Al compound formed. The compound was Ag-rich firstly [15,16], including Ag2Al and Ag3Al (JCPDS 28-0034, P4132 space group, a = b = c = 0.693400 nm). As Ag-Al continued to dissolve into the solder, Ag-rich compounds transformed into Al-rich Ag3Al2 gradually. The ultrasonic action was a key factor to realize the soldering of screen-printed Al-paste layer. It contributed to the wetting, oxide film breaking and dissolution. The oxide film on the surface of Al particles was wetted by the liquid solder with the assistance of ultrasound. The physical mechanism is such as: the continuous propagation of ultrasonic wave in liquid alloys induced cavitation bubbles. When the bubbles collapsed in a very short time, a micro-jet with very high shock pressure could be generated near the interface to break the oxide film of the aluminum alloys. The solder filled the gaps between the Al particles for the ultrasonicinduced capillary effect . For a small Al particle, the liquid solder could surround it very quickly and the notches appeared on the oxide film of Al particles. More notches on oxide film accelerated the dissolution of Al particles. As shown in Fig. 2a–c, the effect of ultrasonic waves on the joints was not uniform. The inhomogeneity of the microstructure could be attributed to two main factors. Firstly, the diffusion of these spheres started as soon as ultrasonic cavitation caused notches on the oxide film . However, the cavitation occurred Fig. 3. (a) Experimental setup for peel force test; (b) Peel force of the joints by different ultrasonic action time; Fracture surfaces of joints under ultrasonic action for (c) 2 s and (d) 6 s. W. Guo et al. / Materials Letters 231 (2018) 146–149 149 interface directly. For the sound interfacial bonding and good conductibility of the SnAg alloy, the resistance dropped to 1.27 mX. The Voc of solar cells soldered with Al BSF by ultrasonic action time of 2 s, 4 s and 6 s were 483 mV, 535 mV and 524 mV, while Voc of solar cells soldered with Ag electrode was only 467 mV. The Ag busbars on the rear side prevented the formation of BSF, which would raise the recombination rate, leading to a lower opencircuit voltage. Therefore, direct soldering Al BSF can lower costs and raise Voc of the solar cells. 4. Conclusions Fig. 4. Resistance of the joints and VOC of solar cells under different ultrasonic action time. randomly and the diffusion process might vary among different spheres in different sizes. Secondly, the ultrasonic field in the liquid is not uniform ; the microstructure of joints may show inhomogeneity with different distances away from the ultrasonic iron. Peel force tests were carried out with a constant speed of 1 mm/min, as shown in Fig. 3a. Fig. 3b shows the relationship of the ultrasonic action time and the peel force of the joints. By short ultrasonic action time of 2 s, the peel force of the joint was 0.97 N/mm. The solder bonded with the Al paste residual layer indeed. The bonding between Al particles was extremely weak, resulting in low peel force of the joint. The joints fractured inside the Al paste residual layer, and the fracture surface shows the spherical Al paste residual particles (Fig. 3c). When the ultrasonic action time was 4 s, more Al particles were dissolved into the solder layer. The Sn-Ag solder had penetrated into the Al paste residual layer for the ultrasonic-induced capillary effect. The peel force of the joint was about 1.17 N/mm, which was slightly higher than that of the joint under ultrasonic action for 2 s. After further increasing the ultrasonic action time to 6 s, the solder bonded with the Al-Si eutectic layer directly. The peel force of the joints increased to about 1.44 N/mm, which reflects the metallurgical bonding strength of SnAgAl/AlSi alloy interface. The fracture occurred at the interface of the solder layer and the solar cell. The chemical compositions of phase at point J were: 87% Al and 13%Si (at.%), indicating that the fracture surface shows the Al-Si eutectic layer (Fig. 3d). Fig. 4 demonstrates the trends of joint resistance and VOC under different ultrasonic action time. The resistance of joint was 1.79 mX under ultrasonic action for 2 s. The resistance was relatively large for most of the Al paste particles still existed beneath the solder layer. The oxide film on the surface of Al particles is rather thick and stable. The conductivity of the oxide film is poor and the bonding between the oxide films is very weak, resulting in high resistance. When the ultrasonic action time was 4 s, the electrons needed to go through less Al particles and the resistance of joint decreased to 1.66 mX. Further increasing the ultrasonic action time to 6 s could lead to the direct bond between the solder and the Al-Si eutectic layer. The electrons could pass through the The Al back surface field was soldered using Sn-3.5Ag solder by the assistance of ultrasound. The ultrasonic action time had great influence on the interfacial microstructure and properties of the joints. When the ultrasonic action time was 2 s, the solder bonded with the Al paste residual layer for the little dissolution. The joints had poor mechanical properties and electrical conductivity. The Al paste residual layer could be dissolved completely under ultrasonic action for 6 s and the solder bonded with the Al-Si eutectic layer directly. The resistance of the joints was 1.27 mX and the peel force could reach 1.44 N/mm. The Voc of solar cell was 524 mV, which was higher than the 467 mV of solar cell soldered with Ag electrode, for the existence of Al BSF. Acknowledgment This research was sponsored by the National Natural Science Foundation of China (Grant No. 51435004). References  Y. Wu, S. Yuan, X. Yu, X. Qiu, H. Zhu, J. Qian, D. Yang, Sol. Energy Mater. Sol. Cells 154 (2016) 94–98.  M.T. Zarmai, N.N. Ekere, C.F. Oduoza, E.H. Amalu, Appl. Energy 154 (2015) 173– 182.  Y. Chen, H. Shen, P.P. Altermatt, Sol. Energy Mater. Sol. Cells 120 (2014) 356– 362.  J. Krause, R. Woehl, M. Rauer, C. Schmiga, J. Wilde, D. Biro, Sol. Energy Mater. Sol. Cells 95 (2011) 2151–2160.  H.H. Hsieh, F.M. Lin, F.Y. Yeh, M.H. Lin, Sol. Energy Mater. Sol. Cells 93 (2009) 864–868.  R. Hoenig, A. Kalio, J. Sigwarth, F. Clement, M. Glatthhaar, J. Wilde, D. Biro, Sol. Energy Mater. Sol. Cells 106 (2012) 7–10.  P. Schmitt, D. Eberlein, C. Ebert, M. Tranitz, U. Eitner, H. Wirth, Energy Procedia. 38 (2013) 380–386.  H.R. Faridi, J.H. Devletian, H. Le, Weld. J. 79 (2000) 41–45.  Y.X. Li, X.S. Leng, S. Cheng, J.C. Yan, Mater. Des. 40 (2012) 427–432.  Y. Yao, J. Zhou, F. Xue, X. Chen, J. Alloy. Compd. 682 (2016) 627–633.  K.J. Chen, F.Y. Hung, T.S. Lui, L.H. Chen, Y.W. Chen, Sol. Energy Mater. Sol. Cells 143 (2015) 561–566.  L. Zhang, J.G. Han, Y.H. Guo, L. Sun, Sol. Energy Mater. Sol. Cells 130 (2014) 397–400.  H. Schulte-Huxel, S. Blankemeyer, R. Bock, A. Merkle, S. Kajari-Schröder, R. Brendel, Sol. Energy Mater. Sol. Cells 106 (2012) 22–26.  D.I.F. Normung, Solar cells-Datasheet information and product data for crystalline silicon solar cells DIN EN 50461, 2007, 11.  L. Yang, T.L. Guo, J. Funct. Mater. 42 (2011) 2307–2310 (in Chinese).  M.R. Choi, H.G. Kim, T.W. Lee, Y.J. Jeon, Y.K. Ahn, K.W. Koo, Y.C. Jang, S.Y. Park, J. H. Yee, N.K. Cho, I.T. Kang, S. Kim, S.Z. Han, S.H. Lim, Microelectron. Reliab. 55 (2015) 2306–2315.  Z.W. Xu, L. Ma, J.G. Yang, J. Zhang, J.C. Yan, Weld. J. 95 (2016) 264–272.  X.G. Chen, J.C. Yan, F. Gao, J.H. Wei, Z.W. Xu, G.H. Fan, Ultrason. Sonochem. 20 (2013) 144–154.  L. Bai, W. Lin, P. Wu, J. Deng, C. Li, D. Xu, D. Wang, L. Chen, Ultrason. Sonochem. 32 (2016) 213–217.