1. Introduction

Due to the excellent control of the rear-side recombination losses and enhanced longer wavelength photon absorption [1–4], the passivated emitter and rear cell (PERC) silicon (Si) solar cell is predicted to become the dominant cell technology for mass production in the coming years [5,6]. Moreover, the preference of directly employing high throughput industrial laser machines, to ablate precise laser contact openings in the dielectric layers for metallization contact, over techniques such as lithography or mechanical scribing has further accelerated this trend.

To maximize the power conversion gains from a PERC solar cell, it is vital to consider three main factors, namely (1) the recombination losses at the rear-side Si/ dielectric interface, (2) the internal reflectivity of infrared light at the rear-side dielectric/ metal interface and (3) the metal recombination losses at the laser contact openings, as illustrated in Fig. 1.

 

Fig. 1 Cross-sectional schematic of mc-Si PERC solar cell with (1) rear-side Si/dielectric interface, (2) internal reflectivity of rear-side dielectric/metal interface and (3) rear-side Si/metal interface.

Download Full Size | PDF

With reference to the aforementioned point (2), PERC solar cells with various rear-side dielectric passivation schemes were studied to improve the external quantum efficiency (EQE) at wavelengths between ~900 nm to 1180 nm [7,8]. It was also reported that a 160 nm thick capping silicon nitride (SiN_x) film over a 15 nm alumina (AlO_x) on a flat yet highly scattering rear surface of the Si solar cell can improve Lambertian reflection, giving a current gain of up to 1.0 mA/cm² [9]. In addition, it has been reported that the PERC cell efficiency decreases linearly when the laser irradiation energy increased [10]. While it has been demonstrated that femtosecond (fs) laser ablation in the gentle regime is able to obtain high-efficiency PERC Si solar cells [11], it is also important to consider the high cost to deploy fs laser ablation for industrial mass production. Thus, this paper explores an optical constraint approach to design a multi-layer rear-side dielectric stacks to synergistically improve the near infra-red (NIR) absorption of the multicrystalline silicon (mc-Si) absorber and also to mitigate laser-induced damage caused by the nanosecond laser ablation at the laser contact openings. Based on the aforementioned optical constraints, the OPAL 2 simulation program [12] was used to compute the optical reflectance of the multi-layer dielectric stack to obtain a higher reflectance at the wavelength of the laser ablation source and lower reflectance in the NIR region. A total of four different groups of rear-side multi-layer dielectric stacks were computed using OPAL 2, as follows:

2. Experimental details

2.1 Solar cell fabrication

The substrates used were 156.7 mm by 156.7 mm, 185 µm thick, solar-grade p-type mc-Si wafers with nano-spherical surface texture using the SERIS developed nanotexture solution [13]. The nanotextured mc-Si wafers sequentially underwent a standard RCA (Radio Corporation of America) clean 1 and 2, and a 2% hydrofluoric (HF) acid dip. The single-sided n-type emitter was formed using POCl₃ diffusion on the nanotextured surface with a silicon nitride (SiN_x) masked rear surface that was previously planarized using alkaline potassium hydroxide (KOH). This 160 nm SiN_x masking layer was deposited using an inline plasma-enhanced chemical vapor deposition (PECVD) tool (MAiA) from Meyer Burger, Germany. Both the phosphosilicate glass and rear-side SiN_x mask were simultaneously removed in heated 10% HF solution. The final emitter sheet resistance after this 10% HF step was 105 ± 5 Ω/□. The rear surface was passivated with one of the aforementioned PECVD stacks, followed by a 64 nm SiN_x / 87 nm SiO_z PECVD stack deposited on the front-side nanotextured surface. The rear-side dielectric stack was subsequently ablated using a nanosecond green laser (532 nm) from Innolas Solutions, Germany, using a fixed average laser power of 4 W. The rear laser pattern was chosen to be a narrow continuous line pattern with line width of 33 µm and a pitch of 0.9 mm. The laser ablated wafers were then screen printed with a full-area aluminium (Al) paste on the rear side, while the front-side metallization grid consists of 106 fingers (45 µm wide) and 5 floating busbars (800 µm wide) using two different silver pastes. The samples were sequentially fired using an industrial firing furnace at a peak temperature of 740°C.

2.2 Characterization

The reflectance spectrum of the different rear-side dielectric stacks was measured using a UV–Vis–NIR spectrometer to determine the weighted average reflectance (WAR) for the wavelength range 900 to 1200 nm. The electrical performance of the cells was characterized to assess the impact of the various rear-side dielectric stacks. Specifically, the 1-Sun and dark current-voltage (J–V) characteristics were measured at SERIS using a calibrated LED based solar simulator from WaveLabs. The external quantum efficiency (EQE) of the solar cells was also measured at short-circuit condition, using a spectrometer from Bentham. In addition, the effective electron diffusion length in the base region of these cells was calculated based on a SERIS-developed short-circuit current loss analysis method [14,15]. Photoluminescence (PL) imaging at open-circuit condition was also performed on these solar cells. Lastly, scanning electron microscopy (SEM) imaging was performed on the cells after metallization and firing to assess the characteristics of the local back surface field (LBSF) at the rear-side Si/metal interface within the laser contact opening.

3. Results and discussion

UV-Vis-NIR reflectance measurements were performed before laser ablation on the four groups with different rear-side dielectric stacks and their corresponding weighted average reflectance (WAR) calculated, as illustrated in Fig. 2. WAR is defined as the solar weighted reflectance in accordance to the incident photon flux of the AM 1.5G solar spectrum. As mentioned earlier, the four groups of rear-side dielectric stacks were designed in such a manner that they have different reflectance values at the laser ablation wavelength of 532 nm, with the constraint that their WAR in the 900 - 1200 nm wavelength range is similar. It should be highlighted that adding an additional silicon oxide (SiO_z) layer reduces the reflectance at 532 nm, which consequently increases the laser power absorption by the triple-layer dielectric stack, as compared to a double-layer dielectric stack. Based on their reflectance value at 532 nm, a simple calculation of the absorbed laser power revealed that G4 has a ~29% reduction of absorbed laser power, as compared to G2. In addition, the calculated WAR values of G1 are approximately 16.20%, whereas the WAR value of G2 – G4 ranges between 15.37 and 15.70%.

 

Fig. 2 UV-Vis-NIR reflectance curves of the four different rear-side dielectric stacks and their corresponding WAR values in the 900 - 1200 nm wavelength range. The green dashed line at 532 nm indicates the target wavelength of laser ablation source.

Download Full Size | PDF

The electrical parameters of the mc-Si PERC solar cells from the four groups were measured at 1-Sun (AM 1.5G) illumination, as shown in Fig. 3. It is worth highlighting that G3 and G4 yielded a higher open-circuit voltage (V_oc) of ~3 mV and higher short-circuit current density (J_sc) of ~0.3 mA/cm², when compared to G1 and G2. This gain in cell performance corroborates well with the aforementioned example whereby the reduction in absorbed laser power by G4, as compared to G2, could possibly moderate the laser-induced damage at the laser contact openings during the nanosecond laser ablation. The WAR value for the 900 - 1200 nm wavelength range of the rear-side dielectric stack does not have significant impact on the electrical parameter of the cells. For example, although both G2 and G3 have similar WAR values of 15.47% and 15.70% respectively, but G3 has ~0.25% (absolute) higher cell efficiency than G2. Furthermore, both G3 and G4 yielded similar cell efficiencies even though their WAR values are similar. Hence, it is evident that the approach to evaluate the reflectance of the rear-side dielectric stack at the wavelength of the laser ablation source while keeping the NIR reflectance of the four groups similar has merit. In addition, a statistical analysis of variance (ANOVA) was performed, with α = 0.05, showing that variations in thickness (e.g. 95 nm to 105 nm for G3 and G4, respectively) or refractive index (e.g. n = 1.85 to n = 2.05 for G1 and G2, respectively) of the rear-side capping SiN_x do not increase the J_sc significantly.

 

Fig. 3 The measured electrical 1-Sun parameters of the mc-Si PERC solar cells with different rear-side dielectric stacks.

Download Full Size | PDF

Figure 4 shows the measured external quantum efficiencies of the mc-Si PERC cells with different rear-side dielectric stacks. These results corroborate the J_sc gain shown in Fig. 3. Moreover, the effective electron diffusion length (L_eff) in the base region of these cells from the four groups was calculated, based on short-circuit current loss analysis method developed at SERIS [15]. The results are as follows:

 

Fig. 4 The measured external quantum efficiencies of the mc-Si PERC solar cells with different rear-side dielectric stacks at short-circuit condition.

Download Full Size | PDF

Figure 5 depicts photoluminescence (PL) images of the cells from the different groups. The PL intensity is influenced by the effective minority carrier lifetime which is affected by both the bulk and surface quality of the mc-Si wafer, and in turn alludes to the voltage potential of the solar cell. All else constant, the higher PL counts for G3 and G4 compared to G1 and G2 indicate a reduction in laser-induced damage. The results correlate well to the measured open-circuit voltages of the mc-Si PERC cells (see Fig. 3). In addition, the grain distribution characteristic of the sister cells is also revealed in Fig. 5.

 

Fig. 5 The photoluminescence maps of the mc-Si PERC solar cells with different rear-side dielectric stacks at open-circuit condition.

Download Full Size | PDF

Figure 6 shows SEM images of mc-Si PERC solar cells from the four different groups. It is clearly evident from these images that the LBSF thickness increases with increasing reflectance at the laser wavelength of 532 nm, which explains the higher V_oc of their respective groups, as shown in Fig. 3. It has been reported that the relatively steady energy injection from a nanosecond laser pulse can transmit energy through the passivation dielectrics with large optical bandgap. The transmitted energy is mostly absorbed by the Si underlying the dielectrics and can lead to the melting of the Si [16]. Another study reported that it is important to manage the laser-induced damage in the Si so as to minimize the thickness of the amorphous Si layer formed, which could in turn affect the Al/Si alloying process during the metallization firing step, and hence negatively affect the PERC cell efficiency [10]. Therefore, the higher reflectance of the rear-side dielectric stack at the laser wavelength could reduce the transmitted laser energy to the underlying Si and increase the thickness of the LBSF layer, which in turn would give higher PERC solar cell efficiency.

 

Fig. 6 SEM images of the mc-Si PERC solar cells with different rear-side dielectric stacks, showing different LBSF thicknesses at the laser contact openings.

Download Full Size | PDF

4. Conclusions

In summary, we have demonstrated that by tuning the rear-side dielectric stack to achieve a higher reflectance at the wavelength of the nanosecond laser source used for laser ablation can reduce the laser-induced damage, as indicated by higher V_oc and higher PL counts for mc-Si PERC solar cells, which is beneficial to attain higher cell efficiency. Moreover, a current loss analysis and SEM images of these mc-Si PERC cells with higher reflectance at the laser ablation wavelength reveal a thicker LBSF layer and a longer L_eff in the base region. Our results thus demonstrate the feasibility to increase the working window for nanosecond laser ablation of the rear-side dielectric passivation of mc-Si PERC solar cells. Optimization of the rear-side dielectric schemes to increase the Lambertian reflection and to reduce the laser-induced damage is expected to further improve the efficiency of industrial mc-Si PERC solar cells.