Effect of additional HfO2 layer deposition on heterojunction c‐Si solar cells

The efficiency of a heterojunction with intrinsic thin‐layer (HIT) solar cell with an insulator/ITO structure was discussed in this paper. First, the efficiency was analyzed by OPAL 2 simulations. Second, an insulator/ITO structure was experimentally applied to a HIT solar cell. The OPAL 2 simulations using an insulator/ITO/Si structure and a TCO/ITO/Si structure showed that the generation current density in the Si substrate increased when the insulator or TCO layer had a proper thickness. Experimentally, HfO2, a type of insulator, was deposited on a HIT solar cell with a thickness varying from 3 to 15 nm. The average efficiency of the HIT solar cell improved from 18.21% to 20.75% after HfO2 deposition. The highest efficiency was achieved for the 3‐nm‐thick HfO2/HIT solar cell structure, which exhibited the best improvement in the current density of approximately 1.5 mA/cm2. For the external quantum efficiency of the HfO2/HIT solar cell, the total absorption was improved by HfO2 deposition. The results suggest that the HfO2 layer improves the solar cell efficiency of the HIT solar cell by increasing light absorption.


| INTRODUCTION
To increase the light absorption of solar cells, the reflectance of Si wafers should be reduced. The light absorption can be decreased by texturing and anti-reflection coating (ARC) techniques. 1 Anti-reflection coating technique uses the interference of lights to reduce the reflectance and is designed to maximize destructive interference at a specific wavelength, which is controlled by the ARC material and the coating thickness. Typically, ARC materials are insulators. Si 3 N 4 is a commercial material that is frequently applied as an ARC layer since it is easy to deposit at low temperatures (≤400) using plasma-enhanced chemical vapor deposition (PECVD) and is compatible with other conventional fabrication processes, such as metal screen printing. [2][3][4][5] A further decrease in the reflectance can be achieved by depositing a doublelayer ARC (DLARC). Various types of DLARCs, such as SiO 2 /SiN, SiO 2 /TiO 2 , ZnS/MgF 2 , and Al 2 O 3 /SiON, and the like for reducing the reflectance have been reported in many areas of research. [6][7][8][9][10][11][12][13] Even though some of the DLARC materials require further study, the DLARC technology can better enhance the generation current density by minimizing the reflectance than single-layer ARCs. Therefore, DLARCs are normally used to reduce the reflectance in conventional | 707 LEE Et aL. solar cell structures that do not use a transparent conductive oxide (TCO) layer. Passivated emitter solar cells (PESCs), passivated emitter and rear cells (PERCs), interdigitated back contact cells (IBCs), passivated emitter rear locally diffused cells (PERLs), and buried contact solar cells (BCSCs). [13][14][15][16][17][18][19][20] However, as shown in Figure 1, heterojunction with intrinsic thin-layer (HIT) solar cells do not have an insulation material for ARC layer since the TCO layer acts as an ARC layer. [21][22][23] To maximize the destructive interference at a specific wavelength, the materials and thickness of the TCO layer must be chosen carefully. Moreover, the TCO layer must have sufficiently high transmittance and electrical conductivity for application in a solar cell.
In this manuscript, the DLARC theory was applied to a HIT solar cell by depositing a thin insulator or an additional TCO layer on an initial TCO layer. We first used OPAL 2 24,25 to simulate the reflectance in various insulator/ITO/Si wafer structures and a TCO/ITO/Si wafer structure. Second, in this paper, we also reported the experimental results for a HfO 2 / HIT solar cell based on the simulations.

| Simulation by OPAL 2
We applied OPAL 2 24,25 to simulate the reflectance of insulator/TCO and TCO/TCO structures. Figure 1 shows a schematic of the structure employed in the simulation. The purpose of the simulation was to optimize the layer thickness and maximize the light absorption by the substrate. We set the substrate as 180-μm-thick Si (Crystalline, 300K, Gre08). ITO, a conventional TCO material, was selected as the TCO layer. A random upright pyramid structure with a 54.74° angle was applied to the surface morphology. AM 1.5G illumination (Gue95) with a 0° zenith angle was selected as the incident radiation. The light trapping model was given by the equation Z = 4 +{ln [n 2 + (1 − n 2 )e −4 W ]}∕ W. The simulation was adjusted to maximize the generation current density in the Si substrate. The total current density from the light source was fixed at 44 mA/cm 2 . The reflected current density, the current density absorbed by the film, and the current density absorbed by the Si (generation current density) were analyzed. Moreover, the reflectance as a function of the wavelength was also analyzed. The air/ITO/Si structure without an insulator was first simulated to optimize the ITO thickness. The doping concentration of the ITO was also varied in the simulation. Second, based on the optimized ITO thickness determined from the ITO simulation, the air/insulator or TCO/ITO/Si substrate structures ( Figure 1) were simulated. ZnS (evaporated, Siq88), HfO 2 (cubic hafnia, Woo90), Al 2 O 3 (atomic layer deposition (ALD) on glass, Kum09), SiO x (PECVD, McI14), SiO x N y (80% Sopra), MoO x (ALD, Mac15), and TiO 2 (ALD 75°C, Cui16) were applied in the air/insulator/ITO/Si substrate structures as insulators. ITO, ZnO (sputtered, EIA15), B-doped ZnO (Fan15), and ZnO:Al (EIA15) were applied to the air/TCO/ITO/Si substrate structure as TCO materials. In the case of the ITO/ ITO/Si substrate structure, the doping concentration of the top ITO layer was varied from 1.7 × 10 19 to 4.9 × 10 20 cm −3 , and the doping concentration of the bottom ITO was fixed at 6.1 × 10 20 cm −3 . The insulator and TCO thickness were optimized to maximize the generation current density in the Si substrate.

| HfO 2 deposition experiment
The HIT solar cell structure for the experiment is shown in Figure 2. A 180-μm-thick n-type Czochralski Si wafer with a resistivity of 3.8 ohm-cm was used for the experiment. Both surfaces of the Si wafer were randomly textured. Amorphous i/n and i/p Si layers were deposited on front and back surfaces, respectively, of the Si wafer. Afterward, an 80-nm-thick ITO layer was deposited on both sides. Front and back contacts were then formed by evaporation and plating after a photolithography process. Based on the simulation results, HfO 2 was chosen as one of the insulators and was deposited using ALD on the HIT solar cell after metallization. The thickness of the HfO 2 layer was controlled by varying the number of cycles. To analyze F I G U R E 1 OPAL 2 simulation structure with an insulator F I G U R E 2 Heterojunction with intrinsic thin-layer solar cell structure for the insulator experiment the solar cell characteristics, the solar cell parameters were measured before and after HfO 2 deposition using a solar simulator. The external quantum efficiency (EQE) was also measured to compare the efficacy of the ARC. Separately, the HfO 2 was deposited on a silicon substrate to analyze the HfO 2 characteristic. Figure 3 shows the results of the ITO simulation performed without an insulator to maximize the generation current density in the Si substrate. The absorbed current density changed as a function of the ITO doping concentration and ITO film thickness. The actual ITO doping concentration should be applied in an insulator/ITO/Si structure to simulate it more precisely. In general, doping concentrations above 1 × 10 20 cm −3 have been reported to give conductive and transparent ITO layers. 26-28 Therefore, ITO doping concentrations of 2.0, 4.0, and 6.1 × 10 20 were chosen in the insulator/ITO/Si structure. In the simulations of the insulator/ITO/Si structure, the optimized ITO thickness and the various ITO doping concentrations were used. Figure 4 shows the simulation results for the insulator/ ITO structures containing various insulators with an ITO doping concentration of 6.1 × 10 20 . As shown in Figure 4, the current density absorbed by Si (generation current density) gradually increased with increasing insulator thickness for all the insulators. As a representative example, in the results for HfO 2 , the reflected current density gradually decreased as the thickness of the HfO 2 layer increased, thereby increasing the generation current density. The cause of the reduction in the reflected current density was a decrease in the reflectance from approximately 200 to 300 nm and from 500 to 1400 nm, although the reflectance increased from approximately 300 to 500 nm ( Figure 5A). However, the generation current density decreased when the thickness of some of the insulators increased beyond the optimum value ( Figure 4B-F). For example, in the case of ZnS ( Figure 4F), the best generation current density of 41.18 mA/cm 2 was achieved when the thickness of the ZnS layer was 13 nm. However, as the thickness increased beyond 13 nm, the F I G U R E 5 Reflectance simulation results for (A) HfO 2 and (B) ZnS F I G U R E 6 Optimized simulation results before and after insulator deposition on the ITO/Si wafer as a function of the ITO doping concentration generation current density decreased, while the current density absorbed by the film and the reflected current density increased. This increase in the reflected current density resulted from the fact that the increase in reflectance from approximately 200 to 550 nm was larger than the decrease in the reflectance from approximately 550 to 1400 nm ( Figure 5B). Figure 6 shows the simulation results for the insulator/ ITO/Si substrate structures that maximized the generation current density. Regardless of the doping concentration of ITO, the deposition of an insulator with an appropriate thickness on the ITO layer improved the absorbed current density of the Si substrate because the total reflection was reduced by the ARC effect. Based on the above simulations, we proposed that the insulator/ITO structure could increase the current density of the HIT solar cell. Figure 7 shows the simulation results for the ITO/ITO/Si wafer structure. Even though the absorbed current density in the film increased from 1.55 to 1.7 mA/cm 2 , the generation current density increased with the deposition of an additional ITO layer since the decrease in the reflected current density was greater than the increase in the absorbed current density in the film ( Figure 7A,C). As the doping concentration of the additional ITO layer increased to 4.9 × 10 20 , the reflected current density increased gradually because although the reflectance decreased in the wavelength range of 300-500 nm, the reflectance increased to a larger degree in the other ranges at the 4.9 × 10 20 doping concentration ( Figure 7B).  Figure 8 shows the simulation results for the ZnO/ITO/ Si wafer structures with different ZnO materials. After the deposition of ZnO, B-doped ZnO, and ZnO:Al, the current density improved concomitantly with a decrease in the reflected current density. A similar trend was observed for the ITO/ITO/Si wafer structure. Even though the reflectance increased in the wavelength range of 300-500 nm, the reflectance decreased to a larger degree in the other ranges when ZnO was deposited on ITO.

| OPAL 2 simulation results with a TCO/ITO/Si wafer structure
As the results for the TCO or insulator/ITO/Si wafer simulations show, not only insulator deposition but also TCO deposition on ITO with proper thickness and doping concentration reduced the total reflection through the ARC effect.

| HfO 2 experimental results
As a result of the ellipsometry, the refractive index of the deposited HfO 2 is shown in Figure 9. The refractive index at 600 nm of deposited HfO 2 by ALD was 2.02. The HfO 2 was applied to the insulator/HIT solar cell structure. As we observed in the simulations with HfO 2 , the generation current density was experimentally shown to be improved by the deposition of HfO 2 ( Figure 11B). When the thickness of the HfO 2 layer was 3 nm, the highest generated current density of 38.75 mA/cm 2 was achieved, and this value was 1.5 mA/cm 2 higher than the original value ( Figure 13). However, when the thickness of the HfO 2 layer was 5 nm or more, the current density was only 0.25-0.5 mA/cm 2 higher than the original value. When comparing the simulation result ( Figure 5A) and the experiment result ( Figure 10), even though there were small differences at the level of the reflectance and the wavelength range, the reflectance decreases and increases showed the similar trend according to HfO 2 thicknesses. These changes in the reflectance were due to shifted maximum destructive interference wavelength and changed the reflectance of the wavelength region by HfO 2 . Table 1 shows average solar cell parameters before and after HfO 2 deposition. Figure 11 shows the difference of each parameter before and after HfO 2 . The reflectivity changed by HfO 2 deposition appeared to give increased current density (J sc ) in simulations and in experimental results because it had a lower reflectivity at higher intensities of sunlight. This effect caused changes in the EQE (Figure 12). Even though the EQE decreased in the wavelength region from 380 to 600 nm, the generation current density appeared to increase as a result of the increase in the EQE in the wavelength regions from 600 to 1100 nm and from 300 nm to 380 nm. Therefore, the current density of the solar cell was increased by the formation of an HfO 2 /ITO double layer rather than a single layer of ITO. Figure 11B shows that not only the current density but also the open-circuit voltage increased as HfO 2 was deposited. The reason for this increase was that the damage caused by sonication was repaired by HfO 2 deposition. The solar cell used in this paper required a photolithography process for the selective plating on ITO, which was necessary for the metallization process. In the photolithography process, a sonicator was used to remove the photoresist efficiently. The use of a sonicator decreased the lifetime and the open-circuit voltage of the solar cell according to the Suns-Voc plot (not shown in this paper). The initial open-circuit voltage of the HIT solar cell was restored when HfO 2 was deposited. As shown in Figure 11B, as the thickness of the HfO 2 layer increased, the open-circuit voltage gradually increased, eventually recovering and saturating at the original value. The initial opencircuit voltage was recovered because the sonicator damage was repaired by HfO 2 deposition. Therefore, the average solar cell efficiency was observed to increase from 18.21% to 20.75% after HfO 2 deposition, which was mainly due to increases in the open-circuit voltage and the current density ( Figure 11A).
Another cause of the increase in the efficiency of the solar cell was an increase in the fill factor. The series resistance before the HfO 2 deposition was 1.9-2.9 Ohm-cm 2 , and that after HfO 2 deposition was 1.2-1.3 Ohm-cm 2 . The high series resistance of the HIT solar cell before HfO 2 deposition was due to the lack of tips in the solar simulator measurements. However, since the change in the series resistances did not give rise to a significant change in the current density (approximately 0.01 mA/cm 2 ), the increase in the current density was attributed to HfO 2 deposition.  Therefore, based on the above results, the recovery of the open-circuit voltage and the fill factor did not have a significant influence on the increase in the current density. As HfO 2 , which is an insulator, was deposited, the total reflectance decreased and the current density increased. In conclusion, the solar cell's best efficiency was 21% by the improvements of current density, open-circuit voltage, and fill factor ( Figure 13). Moreover, based on the results of the simulations and HfO 2 experiments, we believe that not only HfO 2 but also other insulators could be the candidates for improving the efficiency of HIT solar cells. However, as shown in the simulations, the insulator must have an appropriate thickness.

| CONCLUSION
In this paper, OPAL 2 simulations were performed on insulator or TCO/ITO/Si wafer structures to increase the current density of HIT solar cells, and HfO 2 was experimentally deposited through ALD as an insulator. Based on the simulations, the generation current density increased due to ARC effect when not only an insulator but also when a TCO layer was deposited on the ITO/Si substrate. Based on the simulations, HfO 2 was chosen as the insulator, and the average efficiency of the solar cell increased from 18.21% to 20.75% after HfO 2 deposition. This increase was mainly caused by the increase in the current density and in the open-circuit voltage. The increase in the current density by 0.5-1.5 mA/cm 2 over that in the HIT solar cell without HfO 2 was due to a decrease in the reflectance caused by the deposition of HfO 2 on ITO. The increase in the open-circuit voltage caused by HfO 2 deposition was due to the recovery of the open-circuit voltage, which had initially decreased due to the damage caused by the sonication process.
Based on the simulations and experiments, we believe that HfO 2 deposition on a HIT cell can increase the current density by decreasing the reflectance. Additionally, if sonication is performed during the fabrication of a solar cell, the initial open-circuit voltage can be regained by HfO 2 deposition. We are performing further experiments with other insulator/HIT solar cell and TCO/HIT solar cell structures. Other TCO materials besides ITO can also be applied. We believe that these other TCO materials will improve the efficiency of the insulator/HIT solar cells that use other insulators, such as Al 2 O 3 , SiO x N y , SiO x , SiN x , ZnS, MoOx, and TiO 2 . Moreover, we believe that the efficiency of the TCO/HIT solar cell will also be improved. In particular, the ITO/HIT solar cell could effectively improve the efficiency since the additional ITO layer can be deposited after the original ITO deposition process by changing the deposition conditions. Furthermore, we also believe that the efficiency of the TCO/HIT solar cell can be improved based on the simulation results, and this improved efficiency is excellent for mass production because the simple addition of an insulator or TCO deposition process to the preor post-metallization process can improve the efficiency of the solar cell. Further experiments with new insulator/HIT solar cell and TCO/HIT solar cell structures will be conducted and reported.