Impact of adding TiO2 on the front and rear surfaces of the PERC cell
In order to gauge the true merits of using TiO2 as a capping layer in PERC solar cells, the first set of simulations entail adding a TiO2 layer on the front and rear side of the solar cell separately, while leaving the original stack (80 nm SiNx (n = 2.1) on the front and 5 nm Al2O3/95 nm SiNx (n = 2.1) on the rear side) intact. Figure 2 (a) and (b) demonstrate the effect of adding ALD, APCVD and mesoporous TiO2 layers for a range of thicknesses, on the front and rear sides of the cell respectively, in terms of the photocurrent absorbed in the silicon substrate (Jsub). Jsub can also be defined as the maximum short circuit current that the cell can deliver if there are no shading and electrical losses (such as loss of current due to recombination).
As observed from Figs. 3 (a) and (b), mesoporous TiO2 improves Jsub of the original PERC cell both on the front and rear side. A maximum Jsub of 43.19 mA/cm2 is obtained for 80 nm mesoporous TiO2 deposited on the front side, which is 0.65% higher than the Jsub of the original PERC cell (42.91 mA/cm2). A 1.65% increase in Jsub, for 90 nm mesoporous TiO2 is observed on the rear side as well. It is important to point out that the Jsub values for the front side are generally higher because it is a textured surface whereas the rear side of the cell is planar. Despite being the same material, the APCVD and ALD based TiO2 capping layers do not perform well, as the mesoporous ones, and the Jsub values reduce even further when the APCVD or ALD based layers are added to the original stack. This is because, the optical properties of these films are heavily dependent on their deposition methods, due to the phase of the resulting film. As observed in the refractive index plots in Fig. 1, mesoporous TiO2 has the lowest refractive index of 1.6 at the design wavelength (λ0 = 600 nm), whereas it is much higher (2.3) for ALD TiO2.
For solar cells with a double antireflective coating system (shown in Fig. 4), the refractive index value for the top layer (ntop) that minimizes the reflection at the design wavelength, is given by [22]:
\({n}_{\text{t}\text{o}\text{p}}^{3}={n}_{Si}{n}_{0}^{2}\) (Eq. 1)
Using nSi (λ0) = 3.94 [23] and n0 = 1, the optimal ntop value is calculated to be 1.579. Among the three TiO2 films investigated in this work, mesoporous TiO2 exhibits a n value very close to the optimal refractive index, hence it performs better than APCVD and ALD based films. However, it is also important to explore why the APCVD and ALD TiO2 based films do not lead to any improvement when compared to the original PERC cell with (n = 2.1) SiNx. To understand this, Fig. 5 compares the reflection, absorption, and transmission spectra for (a) uncapped PERC solar cell and after capping the front surface with 80 nm of (b) mesoporous (c) APCVD and, (d) ALD based TiO2 layers.
As evident in Fig. 5(b), the absorption in the 300–500 nm range is lower for the mesoporous SiNx/TiO2 stack when compared to the ‘SiNx only’ case in Fig. 5(a). Even though the reflection in Fig. 5(b) for the mesoporous SiNx/TiO2 stack is higher in the 300–350 nm range, it still leads to a higher Jsub because the spectral intensity in this wavelength is quite low. Therefore, no significant photocurrent is lost due to the higher reflection in this wavelength range. Moving on to Figs. 5(c) and (d), the stacks based on ALD and APCVD TiO2 films lead to a peak reflection of ~ 8–10% which spreads across the visible and early IR range (500–800 nm), causing significant loss in Jsub, as this is a region of high solar spectral intensity.
The trends in Jsub observed for different types of TiO2 capping layers can also be seen in the efficiencies of the PERC solar cells with these layers. Figure 6 (a) and (b) shows the front and rear side illuminated JV curves, respectively, for the original PERC solar cell (with 80 nm of (n = 2.1) SiNx on the front and 5 nm/95 nm of Al2O3/SiNx (n = 2.1) on the rear side) and the same cell when it is capped with 80 nm of mesoporous, ALD or APCVD TiO2 layers. The corresponding conversion efficiencies and solar cell parameters for each case is summarized in Table 1.
Table 1
Front and rear side solar cell parameters for the original PERC solar cell and the same cell when it is capped with 80 nm of mesoporous, ALD or APCVD TiO2 layers.

Stack definition

Jsc (mA/cm2)

Voc (V)

FF (%)

Efficiency (%)

Front (textured)

SiNx (original PERC)

41.12

0.675

80.00

22.19

SiNx/mesoporous TiO2

41.37

0.675

79.99

22.33

SiNx/APCVD TiO2

40.63

0.674

80.03

21.92

SiNx/ALD TiO2

38.78

0.673

80.13

20.91

Rear (planar)

Al2O3/SiNx (original PERC)

28.67

0.665

80.48

15.34

Al2O3/SiNx/mesoporous TiO2

29.06

0.666

80.47

15.57

Al2O3/SiNx/APCVD

27.00

0.664

80.55

14.44

Al2O3/SiNx/ALD TiO2

23.51

0.660

80.56

12.49

The table suggests that a 0.14/0.23% increase in the front/rear side efficiency is possible when 80 nm mesoporous TiO2 layer is added on the SiNx passivated PERC cell. Moreover, as illustrated in Figs. 3 (a) and (b), the optical properties, and hence the Jsub is not very sensitive to the changes in thickness of the mesoporous TiO2 layer which is a desirable property since this TiO2 layer is prepared using chemical methods where precise control of thickness is often difficult.
Front side optimization of SiN and TiO mesoporous together
Considering the advantages of using mesoporous TiO2 as a capping layer for the PERC solar cells, and the mature surface passivation provided by the underlying SiNx layer, the next step in harnessing the full potential of this stack is to optimize the thicknesses of both these layers and the refractive index of the SiNx layer, simultaneously. This is because, the optimal design point (i.e. refractive index and thickness) of SiNx when it is used as a standalone ARC will be different when it is used in combination with the TiO2 layer. In order to find this optimal design point, the SiNx refractive index value is fixed, and then the thickness of the SiNx and TiO2 layers is varied to maximize Jsub. This process is then repeated for SiNx layers with different refractive indices (in the range of 1.91–2.37). As an example, Fig. 7 demonstrates the impact of varying the thicknesses of mesoporous TiO2 and SiNx (n = 1.91) on the front Jsub value.
The Jsub value does not vary significantly for SiNx films > 50 nm, irrespective of the TiO2 film thickness, as evident in the figure, and a maximum value of 43.13 mA/cm2 is achieved when the SiNx/TiO2 film thickness is 55/60 nm, respectively. This Jsub value is 0.22 mA/cm2 higher than that of the original SiNx (n = 2.1) PERC cell. To improve the performance of the SiNx/TiO2 stack further, contour plots similar to Fig. 5 are generated for SiNx layers with different refractive indices, to identify the optimal thicknesses of the SiNx and TiO2 layers in each case. These results are summarized in Fig. 8.
As illustrated in the figure, the highest photocurrent possible for the SiNx/TiO2 is 43.39 mA/cm2, where the thickness of the stack is 58/76 nm, respectively, and the refractive index of the SiNx layer is 1.99. It is worth noting that ideally, the refractive index (at λ0) of bottom layer which minimizes reflection in a DLARC system is given by Eq. 2 [22], and found to be 2.49.
\({n}_{\text{b}\text{o}\text{t}}^{3}={n}_{0}{n}_{Si}^{2}\) (Eq. 2)
The optimal n value of 1.99 for SiNx in the SiNx/TiO2 stack found through optical simulations deviates significantly from the value found using Eq. 2. This is because, the goal of Eq. 2 is to minimize the reflection at λ0, however, it does not account for the photons that fail to transmit because they are absorbed by the films. Hence this deviation could be attributed to the relatively high absorption of the n = 2.37 film (Note: The SiNx layer with n = 2.37 is chosen as it is closest to the optimal n of 2.49, available in the OPAL2 database). Figures 9(a) and (b) show the reflection and absorption, respectively, for the front surface SiNx/meso TiO2 stack with SiNx n = 1.99 and n = 2.37.
As observed in Fig. 9(a), the reflection around λ0 is expectedly lesser for SiNx with the higher refractive index value. Moreover, the total reflection loss for the stack with SiNx n = 1.99 and 2.37 is 1.2 and 0.4%, respectively, corroborating well with the results found using Eq. 2. Whereas, the absorption for the stack with SiNx n = 2.37 is higher, as shown in Fig. 9(b) and remains nonzero, which falls into the visible region where the solar intensity is high. The total losses due to absorption in the n = 1.99 and n = 2.37 based SiNx stack are 0.2 and 3.2%, making the cumulative losses 1.4% and 3.6%, respectively. This makes SiNx, with n = 1.99, more suitable as a bottom layer in the SiNx/TiO2 stack for PERC solar cells with textured surfaces. It is also noteworthy that the reflection for both stacks in Fig. 9(a) remains < 10% throughout the wavelength range indicating that the loss in Jsub for solar cells with textured surfaces is primarily dominated by absorption in the films.
Though, one might argue that the Jsub value obtained via optimizing the SiNx/TiO2 stack, can also be achieved if the refractive index and thickness of the standalone SiNx layer is optimized, without having to deposit the additional mesoporous TiO2 film. Hence, to investigate the full potential of the SiNx layer, the refractive index and thickness of the layer in each case is optimized and the results are displayed in Fig. 10.
As shown in the figure, the highest photocurrent possible for a single SiNx layer (n = 1.92) is 43.2 mA/cm2, which is still lower than the photocurrent possible for most of the SiNx/TiO2 stack in Fig. 6. Therefore, it can be concluded that capping the SiNx based PERC cells with mesoporous TiO2 can enhance the performance of the solar cells even further. In terms of the solar cell performance, Table 2, and Fig. 11 demonstrate the effect of SiNx and TiO2 layers optimization on the cell parameters and front side illuminated JV curves, respectively.
Table 2
Front side solar cell parameters of original SiNx (n = 2.1), optimized SiNx (n = 1.92) and optimized SiNx/meso TiO2 based PERC solar cell.
Stack definition  Jsc (mA/cm2)  Voc (V)  FF (%)  Efficiency (%) 
SiNx (original PERC: n = 2.1, t = 80 nm)  41.12  0.675  80.00  22.19 
SiNx (optimized: n = 1.92, t = 79 nm)  41.38  0.675  79.99  22.33 
(n = 1.99) SiNx/mesoporous TiO2 (t = 58/76 nm)  41.57  0.675  79.98  22.43 
Optimizing SiN and TiO mesoporous together for rear side
The process for optimizing the SiNx and TiO2 layers on the rear side is very similar to that for the front side presented in the previous section. Though it is necessary to repeat the optimization process as the rear side of the PERC cell used in this work is planar whereas the front side is textured. Figure 12 shows the rear side Jsub values obtained for the Al2O3/SiNx/mesoporous TiO2 stack with different SiNx refractive indexes. The Al2O3 layer thickness is kept intact (at 5 nm) to ensure that the rear surface recombination and fixed charge values used in the TCAD simulations are still valid.
Unlike the front surface, the refractive index value for SiNx that gave the best results for the rear side is 2.37, which corroborates well with the value obtained using Eq. 2. For the sake of comparison with the front textured surface, Figs. 13(a) and (b) show the reflection and absorption, respectively for the stacks based on SiNx with n = 1.99 and n = 2.37 on the rear side.
While the absorption for the stack with SiNx n = 2.37 is still higher (absorption loss = 2.8%), when compared to n = 1.99 (0.2%), it is the loss due to reflection which dominates for the planar surface, as the light rays do not get a chance to enter the substrate the second time after it is reflected once, especially in the visible region. The total loss due to reflection for the stack with SiNx n = 1.99 and n = 2.37 is 8.2 and 3.8%, respectively, which results in the highest Jsub value of 41.08 mA/cm2 for the Al2O3/n = 2.37 SiNx/TiO2 stack on the rear surface. In terms of the solar cell parameters, the rear side efficiency for the PERC solar cell improves from 15.34 (for the original Al2O3/(n = 2.1) SiNx stack) to 16.59% when the refractive index value for SiNx and thicknesses of both SiNx and TiO2 are optimized together, as shown in Table 3, and Fig. 14.
Table 3
Rear side solar cell parameters of original Al2O3/SiNx (n = 2.1), optimized Al2O3/SiNx (n = 1.92) and optimized Al2O3/ SiNx/meso TiO2 based PERC solar cell.
Stack definition  Jsc (mA/cm2)  Voc (V)  FF (%)  Efficiency (%) 
Al2O3/SiNx n = 2.1 (original PERC: t = 5/95 nm)  28.67  0.665  80.48  15.34 
Al2O3/SiNx (optimized: n = 1.92, t = 77 nm)  29.94  0.666  80.42  16.03 
Al2O3/n = 2.37 SiNx/TiO2 (t = 5/47/85 nm)  30.59  0.667  80.39  16.59 