Study on the front contact mechanism of screen-printed multi-crystalline silicon solar cells

Solar Energy Materials & Solar Cells 141 (2015) 80–86

Contents lists available at ScienceDirect

Solar Energy Materials & Solar Cells journal homepage: www.elsevier.com/locate/solmat

Study on the front contact mechanism of screen-printed multi-crystalline silicon solar cells Shiliang Wu a, Li Li a, Wei Wang a, Dong Yu a, Wenchao Liu b, Xiaoshan Wu a, Fengming Zhang a,n a

National Laboratory of Solid State Microstructures, Center of Photovoltaic Engineering and School of Physics, Nanjing University, Nanjing 210093, China National Laboratory of Solid State Microstructures, Center of Photovoltaic Engineering and School of Modern Engineering and Applied Sciences, Nanjing University, Nanjing 210093, China

b

art ic l e i nf o

a b s t r a c t

Article history: Received 24 November 2014 Received in revised form 5 May 2015 Accepted 6 May 2015

The two key issues of front contact of screen-printed multi-crystalline silicon have been studied. Firstly, the effects of doping profiles in the emitter on the contact-resistance have been systematically investigated. Constant source diffusion is used to explain the doped phosphorus profile in the textured emitter by simulation. Secondly, the penetration mechanism of silver into SiNx layer was studied by using SEM and EDX to reveal the microstructure of the Ag/Si contact. A model has been proposed to explain the formation of the ohmic contact. It is found that the effectiveness of the contact is determined by the direct contact area of the silver crystallites to both the emitter and the finger. For a commercial multicrystalline silicon solar cell of 156
156 mm2, the optimal contact-resistance of about 1.2 mΩ is suggested, corresponding to the doping sheet resistance of around 80–90 Ω/□. Multi-crystalline Si solar cells with excellent results, up to 18.51% in conversion efficiency, have been achieved. Moreover, to further understand the model, effects of the firing temperature and the thickness of SiNx layer on the contactresistance have been investigated in the solar cells. It is proved that the formation of the silver crystallites is related to the existence of SiNx layer. & 2015 Elsevier B.V. All rights reserved.

Keywords: Contact-resistance Silver crystallites Phosphorus diffusion Multi-crystalline Si solar cells

1. Introduction In the production process of crystalline silicon solar cells, a thin layer of SiNx is deposited on the textured emitter in order to passivate the dangling bonds and reduce the light reflection. After the deposition step, screen-printing of silver paste, followed by a firing process, is widely used in the front electrode fabrication. In the firing process, silver particles in the paste grow together due to the driving force of the minimization of surface energy. At the same time, the SiNx layer is etched through and an electrical contact is formed. With the rapid development of processing technology in Si solar cells, emitters of higher and higher sheet resistance have been adopted for increasing the conversion efficiency of solar cells by reducing the minority recombination in the emitter areas. Previously, emitters of low sheet resistance (40–50 Ω/□) were widely used to achieve good contacts for industrial mass production of silicon solar cells. However, low sheet resistance leads to the electrical losses in the cell performance because heavy doping concentration could result in high recombination velocity [1]. The motivations to get high electrical performance promoted the development of the silver paste. n

Corresponding author. E-mail address: [email protected] (F. Zhang).

http://dx.doi.org/10.1016/j.solmat.2015.05.012 0927-0248/& 2015 Elsevier B.V. All rights reserved.

Nowadays, a new type of commercial silver paste is available for the production of the solar cells with high sheet resistance emitters (80–90 Ω/□). The performance of solar cells is significantly affected by the series resistance, one of the key parameters of solar cells. However, the series resistance strongly depends on the contact-resistance between the silver and emitter, and the effectiveness of the silver penetration. The contacts of the solar cells based on the fundamental of metal–semiconductor contacts were first investigated in some early studies [2–5]. In the metal–semiconductor contacts, the conduction mechanism depends on the doping concentration (ND) of the semiconductor substrate. For ND 41
1019 cm  3, the field emission (FE) is dominant. For ND o1
1017 cm  3, the thermionic emission (TE) is dominant. For 1
1017 cm  3 oND o1
1019 cm  3, it is a combination of TE and FE, defined as the thermionic field emission (TFE). In the recent research works [6–9], silver crystallites were observed in the interface layer between the silicon emitter and the silver fingers. It is believed that silver crystallites played an important role in the contact formation and current transport. Most of these results are based on mono-crystalline silicon, while the silver crystallites, with shape of inverted pyramid and penetrating into the silicon, are believed to be metal–semiconductor ohmic contact to the emitter [10].

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

Although the contacts between the silver finger and the silicon emitter have been investigated widely, it is still difficult to give a completely clear picture of the formation of the Ag/Si contact due to: (a) the unclear current transport mechanism [11], (b) the formation mechanism of the silver crystallites [12], especially during the complexities of the latest commercial silver paste, (c) the textured surface topography and (d) the uncertainty of crystal orientation [13] in multi-crystalline silicon are taken into account. Based on the experimental and theoretical studies, here presents a detailed investigation on the contacts in the multi-crystalline silicon solar cells.

2. Experimental

3. Results and discussion

Before studying the contact-resistance between the fingers and the emitter, the relationship between sheet resistance and phosphorus diffusion profile in the textured silicon wafers was first examined by theoretical simulation. According to the diffusion theory [14], the phosphorus diffusion in the industrial productions can be approximated to a model of constant source diffusion. The diffusion concentration can be written as follows:

x = Ns 2 Dt

∫z

z=

x 2 Dt

∞

2

e−z dz

(2)

with Ns being the surface concentration, D the diffusion coefficient, x the depth and t the time. The diffusion coefficient changing with diffusion temperature can be written as follows:

⎛ Ea ⎞ D = D0 exp ⎜ − ⎟ ⎝ kT ⎠

(3)

where D0 is a pre-exponential constant, Ea is the activation energy, T is the temperature of diffusion, and k is the Boltzmann coefficient. According to the semiconductor theory [15], the sheet resistance (dR□(x)) of a thin layer (dx) at the depth x can be written as follows:

dR, (x) =

1 1 1 ρ = = = dx nqμ n dx N (x) eμ n dx σdx

(1)

(4)

where μn is the mobility of the electron and can be written as an empirical equation [16].

μn = 65 +

1265

(1 +

N (x ) 8 . 5 × 1016

0.72

)

(5)

So, the total sheet resistance (R□) in the phosphorus diffusion emitter can be written as follows:

1 = R□

∫ dR□1 (x) = ∫ N (x) eμn dx

(6) 2

In the simulations, D0 ¼3.85 cm /s, Ea ¼ 3.15 eV, and T¼ 1093 K. According to Eqs. (1) and (6), the profiles of the phosphorus diffusion in the emitter can be simulated numerically by spreadsheet software. Fig. 1 shows the correspondence of sheet-resistance to the profiles of the phosphorus diffusion in the emitter. The simulated sheet resistances with diffusion times of 28 min, 20 min and 12 min are 56.7 Ω/□, 67.1 Ω/□ and 86.6 Ω/□, respectively, consistent to the experimental values of 56.4 Ω/□, 66.1 Ω/□ and 87.7 Ω/□, respectively. In the former studies [17,18], it is known that silver paste can etch silicon with the result of silver crystallites penetrated into the 1E21

12min 20min 28min

1E20

1E19

1E18

56.7 Ω /

1E17

3.1. Phosphorus diffusion in the textured wafers

N (x, t ) = Ns erfc

where

N(x)

156
156 mm2 p-type industrial multi-crystalline silicon wafers, with bulk resistivity of 1–3 Ω cm, were used in the fabrication of the samples in a typical process of five steps, using the industrial production facilities. Step 1, the wafers were first textured or saw damage-etched by acid etching using a mixture HNO3/HF system. Step 2, the n-type emitter was formed by the POCl3 diffusion. Step 3, the rear junction and phosphosilicate glass were removed by a standard wet-chemical step. Step 4, the SiNx antireflection and passivation layer was deposited via plasma-enhanced chemical vapor deposition (PECVD). Step 5, the front and back metallization process was carried out by screen-printing commercial silver and aluminum pastes, respectively. After the printing, the front and back contacts were simultaneously formed during a fast firing process in an IR heated belt furnace. In our studies, the commercial silver paste (Giga 590A), suitable for high sheet resistance emitters (80–90 Ω/□), was used for the front contacts. Two series of the samples were prepared in the experiment: (a) after conventional texturing, wafers diffused with emitters of different sheet resistances (56.4 Ω/□, 66.1 Ω/□ and 87.7 Ω/□ with the diffusion times 28 min, 20 min and 12 min, respectively) and coated with (without) a thin layer of 80 nm SiNx were printed with four short silver fingers at the central region. The contact area is 3
0.06 mm2 for each short finger and the spacing is 1.73 mm. (b) Solar cells with different SiNx layer thicknesses (0 nm, 80 nm, 160 nm, 240 nm, 320 nm, 400 nm, 480 nm and 560 nm) were prepared with different firing temperatures (745 °C, 775 °C, 805 °C, 835 °C and 865 °C), respectively. The sheet resistance of the wafers and the electrical performance of the solar cells were tested using 4-point probe method after step 3 in the processing, while the electrical performance of the solar cells was characterized with I–V tester after the step 5. The contact-resistance was measured using a 2-point and 4-point probe method. Field-emission scanning electron microscopy (SEM) was used to study the microstructure of the Ag/Si contact and energy dispersive X-ray spectrometry (EDX) was used to identify the Ag content of the micro-crystallites of the samples.

81

67.1 Ω / 86.6 Ω /

1E16

1E15

0

50

100

150

200

250

300

Depth (nm) Fig. 1. Correspondence of sheet-resistance to the profiles of the phosphorus diffusion in the emitter simulated with different diffusion times of 28 min, 20 min and 12 min at temperature of 1093 K.

82

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

Table 1 Doping concentration (ND) for the solar cells with different sheet resistances. R□ (Ω/□)

56.4

66.1

87.7

ND (cm  3)

2.105
1020

1.964
1020

1.694
1020

8

1.2

Isc = 8.99 A FF = 79.01 % Eta = 18.51 % Rs = 0.00164 Ω Rsh = 68.73 Ω

without SiN x

2

1.0

Rc (mΩ)

6

4

with 80nm SiNx

I-V Curve

Uoc = 0.634 V Isc (A)

1.4

0.8

0 0.0

0.6

IRev1 = 0.084 A IRev2 = 0.207 A 0.1

0.2

0.3

0.4

0.5

0.6

0.7

Voc (V)

0.4

Fig. 3. I–V curve and electrical parameters of a typical multi-crystalline Si solar cell prepared at optimized condition of 87.7 Ω/□ in sheet resistance and 80 nm SiNx layer.

0.2 0.0

10

55

60

65

70

R (Ω /

75

80

85

90

)

Fig. 2. Dependence of contact-resistance (Rc) between the fingers and emitter on doping profile (sheet resistance) in the solar cells with and without SiNx layer, with inset as the diagrams of the measurement of the contact-resistance.

surface of the emitter at a range of around 50 nm. In our simulations, it is reasonable to take 50 nm as the average depth of the etching. So the average doping concentration (ND) in the area of crystallites–emitter contacts can be obtained from the profiles of the simulations, corresponding to the sheet resistance of 56.4 Ω/□, 66.1 Ω/□ and 87.7 Ω/□ as the results shown in Table 1. 3.2. Contact-resistance between the finger and the emitter The effects of different sheet resistances on the contact-resistance between the fingers and emitters of the multi-crystalline silicon wafers were examined, using the 2-point and 4-point probe methods, respectively, as shown in the inset in Fig. 2. The difference between the values obtained using 2-point and 4-point probe methods could be taken as the practical contact-resistance between the fingers and emitters. In the measurement, the contact-resistance between the probe and silver finger is metal–metal contact, which can be ignored. Correspondingly, the total front metallization contact-resistance for a piece of 156
156 mm2 full size multi-crystalline silicon solar cell can be calculated by considering the total contact area between the fingers (including the bus-bar) and emitters which is calculated as 14.97 cm2. Shown in Fig. 2 is the dependence contactresistance between the finger and the textured emitter on the sheet resistance for the solar cells with and without a SiNx layer. As shown in Fig. 2, the contact-resistance increases with the rise in the sheet resistance and the samples without SiNx layer show lower contactresistance compared with the samples with SiNx layer. From Fig. 2, it can be seen that in the production an optimal contact-resistance of about 1.2 mΩ, corresponding to sheet resistance of about 85 Ω/□, is suggested to match the silver paste (Giga 590A) and high sheet resistance (490 Ω/□) will reduce the conversion efficiency of the solar cells due to high contact-resistance. Typically, multi-crystalline Si solar cells with doping at 87.7 Ω/□ in sheet resistance and 80 nm SiNx layer were prepared. Conversion efficiency up to 18.51% has been achieved, as shown in Fig. 3.

To better understand the effect of sheet resistance on the ohmic contact, it is necessary to have a clear image of the microstructure of the Ag/Si interface. Fig. 4 shows the typical top-view and cross-section images at the finger areas with SEM. Fig. 4(a) and (b) are the topview image nearby the finger and cross-section image of the Ag/Si interface without SiNx layer, respectively. While, Fig. 4(c) and (d) are the corresponding images for the samples with SiNx layer. It can be seen that for the sample with SiNx layer there are some crystallites formed in the areas nearby the fingers (Fig. 4(c)), with variation in size from tens nanometers to nearly five hundred nanometers. However, no crystallite with the same size has been observed in Fig. 4(a). Similarly, no crystallite has been observed at the interface of the sample without SiNx, while crystallites are found at the interface of the sample with SiNx, as shown in Fig. 4(c) and (d). From the EDX analysis, the crystallites in Fig. 4(c) and (d) were found to be silver content. According to the former studies [12,19], the silver crystallites, formed due to the chemical reaction between the SiNx and silver paste, were supposed to have relation with the current transport. In light of the results above and the SiNx firing-through process in the literature [19], a simple model of the contact formation can be proposed in the multi-crystalline silicon solar cells. During the firing process, the silver crystallites and a glass layer are formed during the redox reaction in the SiNx firing-through process. With the progressing of the reaction, the silver crystallites grow up and move closer to the emitter. When the firing process ends, the silver crystallites penetrate in the glass layer. Those crystallites, with enough size to pass through the glass layer formed during the redox reaction, could play a role of bridge for the current transport from the emitter to the finger. The quality of the contact relies on the amount and the area of the direct crystallite–silicon contact and the crystallite–finger contact. According to the model above, it is easy to explain the difference between the wafers with and without a SiNx layer. The fingers in the wafers without SiNx layer have a better interconnection because there is no interruption caused. However, the contactresistance of the samples without SiNx does not show a significantly lower value. It may be explained by the wetting effect of the liquid glass phase and the etching effect of the silver paste to the silicon. Both effects could limit the direct contact between the finger and the emitter. Therefore, the wafer without SiNx layer shows a little lower contact-resistance. The sheet resistance dependence of the contact-resistance is caused by the different concentration of phosphorus diffusion.

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

83

Fig. 4. Typical images at the finger areas. (a) and (b) are the top-view image nearby the finger and cross-section image of the Ag/Si interface for the samples without SiNx layer, respectively. (c) and (d) are the corresponding images for the samples with SiNx layer.

To give a clear picture of the effect of diffusion concentration on the contact-resistance, the contact-resistance is studied by theoretical calculation. According to the theory of metal–semiconductor contact, the specific resistance is defined as follows [4]:

⎛ ∂J ⎞− 1 ρc = ⎜ ⎟ ⎝ ∂V ⎠V =0

⎞ ⎟⎟ ⎠

(8)

1010 cm−3/2 eV −1 are chosen as the reference. The contact-resistance can be written as follows:

ρc ASt

1.1

Rc

1.0 0.9 0.8

where ρc0 is a function of ND, T and ΦB. In the studies, ρc0 E2
10  7 Ω cm2 [10], ΦB ¼0.78 eV and 4π εSi m0 * /h ≈ 7×

Rc =

Experiment Calculation

1.2

(7)

For silicon solar cells with high doping level ND 41
1020 cm  3, field emission (FE) is the dominant conduction mechanism. The specific contact-resistance depends on the barrier height (ΦB)of the metal–semiconductor contact and the doping concentration (ND) at the emitter surface. It can be obtained by [20,21]

⎛ 4π εSi m* ϕ B ρc = ρc 0 exp ⎜⎜ h ND ⎝

1.3

1.7

1.8

1.9 20

ND/10

2.0

2.1

-3

(cm )

Fig. 5. Contact-resistance (Rc) of experiment and simulation in a multi-crystalline silicon solar cell.

results of simulation agree with those of experiment very well when the effective area ratio is 0.071%.

(9)

where St is the total contact area between the silver finger and the silicon emitter. A is ratio of the effective area. The contact-resistance is a function of doping concentration and effective area ratio. The effective area ratio depends on the crystallite–emitter contact and the crystallite–finger contact. In the model of contact formation, only a part of the silver crystallites directly contact to both the emitter and the finger. In the simulation, the ratio of 0.071% permits an analogous contact-resistance as the situation of the solar cells with SiNx layer in Fig. 2. Fig. 5 shows the simulated and experimental results in the multi-crystalline silicon solar cells. Convincingly, both the results of simulation and experiment have the similar trend with the rise of the doping concentration and the

3.3. Contact-resistance of the solar cells with different prepare conditions To further improve the understanding of the model of contact formation, two series of solar cells with different firing temperatures and thicknesses of SiNx layer were prepared and investigated. Fig. 6 shows the SiNx thickness dependence of the series resistance (Rs) of the solar cells. It can be seen that with the rise of the thickness of SiNx layer, the series resistance increases monotonically. When the thickness exceeds 400 nm, the series resistance increases rapidly and there is nearly no electrical output for the cells with 560 nm SiNx layer. Fig. 7 shows the cross-sectional

84

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

4

Rs

Rs (

)

3

2

1

0 100

200

300

400

500

600

Thickness (nm) Fig. 6. Dependence of the series resistance (Rs) on the SiNx thickness in multicrystalline silicon solar cells.

micrographs of the samples with a SiNx layer of 240 nm (a), 400 nm (b) and 560 nm (c). It can be seen that the silver crystallites penetrate into SiNx layer with the average size of about 200 nm, the same size as the silver crystallites given in Fig. 4. Considering all the conditions except the thickness of the cells are identical, the increase of the series resistance can be recognized as the result of the increase of the contact-resistance due to the thickness increase. According to the model we build in the paper, it is easy to understand the phenomenon. In the firing process, the size of the silver crystallites mainly depends on the reaction time. It will take more firing time for the silver crystallites to move to the emitter because of the long distance, which was also proved by the second firing of the solar cell with 320 nm SiNx layer. The series resistance drops to 0.01 Ω from 0.06 Ω. Therefore, only the crystallites with the size larger than the thickness of SiNx layer could directly contact to both the finger and emitter. In other words, the direct contact area decreases with the rise of the thickness, resulting in high contact-resistance. Meanwhile, it also can be seen that the 560 nm SiNx layer is too thick to fire through in the conventional firing process. It is proved that the formation of the silver crystallites mainly depends on the SiNx layer in the firing-trough process and it does not seem to have relationship with the silicon or the phosphorus diffusion. Fig. 8 shows the firing temperature dependence of the series resistance (Rs) for the samples with 80 nm, 160 nm and 240 nm SiNx layer. It can be seen that the series resistance decreases monotonically with the rise of the firing temperature and the decrease is also attributed to the difference of the contact-resistance. Compared with the cells with 80 nm and 160 nm SiNx, it appears that the cells with 240 nm SiNx have more sensitive temperature effects. The microstructure of the solar cells fired at different temperatures was investigated with SEM. The complexity of the contact microstructure, owing to the textured surface and the small space, means that it is still difficult to give a macroscopical difference of firing temperature in the cross-sectional micrographs under the silver fingers. In the studies, the microstructure nearest to the silver finger in the top-view is very close to the microstructure of the crosssection under the finger. Hence, the microstructure nearby the silver finger can give a clear difference for different firing temperatures, as shown in Fig. 9. Fig. 9(a)–(c) are the top-view images for the solar cells with 160 nm SiNx at peak firing temperature of 745 °C (a), 805 °C (b) and 865 °C (c), respectively. It can be found that the average size of the silver crystallites near to the finger becomes larger with the rise of peak firing temperature. On the basis of the model of the contact formation, the direct interconnection is

0.030

80 nm SiN x

0.025

160 nm SiNx 240 nm SiNx

Rs (Ω )

0.020 0.015 0.010 0.005 0.000

740

760

780

800

T( Fig. 7. Cross-sectional SEM micrographs of the solar cells with different thicknesses of SiNx layer, (a) 240 nm, (b) 400 nm and (c) 560 nm.

820

840

860

880

)

Fig. 8. Firing temperature dependence of series resistance (Rs) on the thickness of SiNx layer (80 nm, 160 nm and 240 nm) in multi-crystalline silicon solar cells.

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

85

diffusion, from which the doping concentration is obtained. In the studies, a simple model is proposed to explain the contact formation. It is found that the values of the contact-resistance depend on the direct contact area of crystallite–silicon contact and crystallite– finger contact. The wafers without SiNx layer have better contact compared with the ones with SiNx layer, due to the highly direct contact without the interruption. For commercial silicon solar cell production, the optimal sheet resistance of 80–90 Ω/□, corresponding to a contact-resistance of about 1.2 mΩ, is suggested to match the latest commercial silver paste. Solar cells with conversion efficiency up to 18.51% have been obtained. In the end, the solar cells with different firing temperatures and thicknesses of SiNx layer were investigated for the further understanding of the model. It is revealed that the direct contact area, related to the size and amount of the crystallites, determines the contact-resistance (or series resistance) in the silicon solar cells. High temperature or long reaction time is helpful to get low contact-resistance. Meanwhile, from the studies of the solar cells with thick SiNx layer, it is proved that the formation of the silver crystallites is related to the existence of SiNx layer.

Acknowledgments We acknowledge support from the NKPBRC (2010CB923404), National Natural Science Foundation of China (Nos. 11274153, 1120412, and 51202108).

References

Fig. 9. Surface SEM micrographs of the figure areas for solar cells with 160 nm SiNx layer and different firing temperatures, (a) 745 °C, (b) 805 °C and (c) 865 °C.

achieved through the silver crystallites formed in the SiNx firingthrough process. High temperature accelerates the redox reactions between the silver paste and SiNx layer. Therefore, higher temperature means larger size of the silver crystallites and lower contact-resistance. As a consequence, series resistance is improved with the rise of temperature. Moreover, the average size of the silver crystallites formed in the conventional firing process is about 200 nm. So it is easy to pass through the 80 nm or 160 nm SiNx layer (or glass layer) in the firing process and get a good contact even when the peak firing temperature is lower than the optimal firing temperature in the conventional process. But it is difficult to fire through the SiNx layer thicker than 200 nm in the conventional firing process and it needs high temperature or long reaction time to get well Ag/Si contacts. That can explain why the solar cells with 240 nm SiNx are more sensitive to firing temperature.

4. Conclusion The latest commercial silver paste is used in the fabrications of the 156
156 mm2 p-type industrial multi-crystalline silicon solar cells. The effect of sheet resistance to the contact-resistance has been investigated experimentally and theoretically. The profiles of phosphorus diffusion in the textured emitters with different sheet resistances are simulated using the model of constant source

[1] A. Luque, S. Hegedus, Handbook of Photovoltaic Science and Engineering, John Wiley & Sons, England, 2003. [2] R.B. Campbell, A. Rohatgi, Investigation of contact metallization systems for solar cells, J. Electrochem. Soc. 127 (1980) 2702–2704. [3] D.L. Meier, R.B. Campbell, J.R. Davis, P. Rai-Choudhury, L.J. Sienkiewicz, Solar cell contacts, in: Proceedings of the 16th Photovoltaic Specialists Conference, 1982, pp. 904–910. [4] D.K. Schroder, D.L. Meier, Solar cell contact resistance – a review, IEEE Trans. Electron Devices 31 (1984) 637–647. [5] A.Y.C. Yu, Electron tunneling and contact resistance of metal-silicon contact barriers, Solid-State Electron. 13 (1970) 239–247. [6] Z.G. Li, L. Liang, A.S. Lonkin, B.M. Fish, M.E. Lewittes, L.K. Cheng, K.R. Mikeska, Microstructural comparison of silicon solar cells’ front-side Ag contact and the evolution of current conduction mechanisms, J. Appl. Phys. 110 (2011) 074304. [7] M.M. Hilali, S. Sridharan, C. Khadilkar, A. Shaikh, A. Rohatgi, S. Kim, Effect of glass frit chemistry on the physical and electrical properties of thick-film Ag contacts for silicon solar cells, J. Electron. Mater. 36 (2006) 2041–2047. [8] S. Kontermann, M. Hörteis, M. Kasemann, A. Grohe, R. Preu, E. Pink, T. Trupke, Physical understanding of the behavior of silver thick-film contacts on n-type silicon under annealing conditions, Sol. Energy Mater. Sol. Cells 93 (2009) 1630–1635. [9] M.M. Hilali, K. Nakayashiki, C. Khadilkar, R.C. Reedy, A. Rohatgi, A. Shaikh, S. Kim, S. Sridharan, Effect of Ag particle size in thick film Ag paste on the electrical and physical properties of screen printed contacts and silicon solar cells, J. Electrochem. Soc. 151 (2006) A5–A11. [10] C. Ballif, D.M. Huljić, G. Willeke, A. Hessler-Wyser, Silver thick-film contacts on highly doped n-type silicon emitters: structural and electronic properties of the interface, Appl. Phys. Lett. 82 (2003) 1878. [11] G. Schubert, F. Huster, P. Fath, Physical understanding of printed thick-film front contacts of crystalline Si solar cells – review of existing models and recent developments, Sol. Energy Mater. Sol. Cells 90 (2006) 3399–3406. [12] K.K. Hong, S.B. Cho, J.S. You, J.W. Jeong, S.M. Bea, J.Y. Huh, Mechanism for the formation of Ag crystallites in the Ag thick-film contacts of crystalline Si solar cells, Sol. Energy Mater. Sol. Cells 93 (2009) 898–904. [13] K.T. Butler, P.E. Vullum, A.M. Muggerud, E. Cabrera, J.H. Harding, Structural and electronic properties of silver/silicon interfaces and implications for solar cell performance, Phys. Rev. B 83 (2011) 235307. [14] L.R. Yan, W. Zhou, D.G. Liu, IC Fabrication, Tsinghua University Press, Beijing, 2010. [15] S.M. Sze, Kwok K. Ng, Physics of Semiconductor Devices, third ed., John Wiley & Sons, England, 2006. [16] M.A. Green, Solar Cells: Operating Principles, Technology, and System Applications, Prentice-Hall, New Jersey, 1982.

86

S. Wu et al. / Solar Energy Materials & Solar Cells 141 (2015) 80–86

[17] E. Cabrera, S. Olibet, J. Glatz-Reichenbach, R. Kopecek, D. Reinke, G. Schubert, Current transport in thick film Ag metallization: direct contacts at silicon pyramid tips? J. Appl. Phys. 110 (2011) 114511. [18] M.I. Jeong, S.E. Park, D.H. Kim, J.S. Lee, Y.C. Park, K.S. Ahn, C.J. Choi, Transmission electron microscope study of screen-printed Ag contacts on crystalline Si solar cells, J. Electrochem. Soc. 157 (2010) H934–H936. [19] S.L. Wu, W. Wang, L. Li, D. Yu, L. Huang, W.C. Liu, X.S. Wu, F.M. Zhang, Investigation of the mechanism of the Ag/SiNx firing-through process of screen-printed silicon solar cells, RSC Adv. 4 (2014) 24384.

[20] H. Yang, X.D. Lei, H. Wang, M.Q. Wang, Investigation of contact resistance between silicon and silver paste made by screen-printing, Clean Technol. Environ. Policy 15 (2013) 1049–1053. [21] P.N. Vinod, Specific contact resistance measurements of the screen-printed Ag thick film contacts in the silicon solar cells by three-point probe methodology and TLM method, J. Mater. Sci.: Mater. Electron. 22 (2011) 1248–1257.