Impact of B₂H₆ plasma treatment on contact resistivity in silicon heterojunction solar cells

The importance of photovoltaics is increasing since it is considered the primary power source for a decarbonized society. Among many varieties of solar cells, silicon heterojunction (SHJ) solar cells using hydrogenated amorphous Si (a-Si:H) exhibit an extremely high conversion efficiency exceeding 25%, ^1–4) leading to high power generation by photovoltaics. The stack of a doped a-Si:H layer and an intrinsic a-Si:H (i-a-Si:H) layer is employed in conventional SHJ solar cells due to its high passivation performance against the c-Si surface. The excellent passivation performance results in a high open-circuit voltage (V_OC) value, which is an attractive feature of SHJ solar cells. ^5–7) Another feature of SHJ solar cells is the low-temperature coefficient of V_OC compared to homojunction c-Si solar cells, which is advantageous for generating a higher power under practical operating conditions. ^8,9) Therefore, SHJ can be expected to accelerate the utilization of photovoltaics and thus the realization of a decarbonized society.

The p-type a-Si:H (p-a-Si:H) is regarded as an issue that limits the performance of SHJ solar cells. Boron doped a-Si:H is commonly used as the p-a-Si:H for SHJ solar cells, and its typical conductivity is in the order of about 10⁻⁴ S cm⁻¹. This requires transparent conductive oxide (TCO) films to collect the carriers into metal electrodes. Gogolin et al. reported that the largest resistive loss is caused by the carrier transport through the a-Si:H and across the TCO/a-Si:H interface. ¹⁰⁾ In particular, larger contact resistance is originated not from TCO/n-a-Si:H/i-a-Si:H contact but from TCO/p-a-Si:H/i-a-Si:H contact. ^10–12) Hence, the electrical contact properties at the TCO/p-a-Si:H interfaces are the bottleneck to achieving low resistive loss. Bivour et al. carried out a numerical analysis of the TCO/p-a-Si:H contact properties and revealed that better contact properties at the TCO/p-a-Si:H are accomplished by heavily doped p-a-Si:H. ¹³⁾ In the numerical device simulation, defect formation depending on the doping variation of the p-a-Si:H layer is not taken into account. ¹³⁾ However, in reality, degradation of passivation performance is often caused by depositing boron-doped a-Si:H. ¹⁴⁾ De Wolf argued that the degraded passivation performance is attributed to Fermi energy-dependent Si–H bond rupture. ^15,16) Based on this notion, B doping in a-Si:H tends to facilitate the destruction of Si–H bonds, leading to the degradation of passivation levels in SHJ solar cells. This suggests that lightly doped p-a-Si:H is beneficial to passivation performance. Thus, simultaneous attainment of good contact properties at the TCO/p-a-Si:H and a high level of passivation is challenging and crucial for high-performance SHJ solar cells.

In this study, we proposed a B₂H₆ plasma treatment on the surface of the p-a-Si:H layer to improve contact properties at the TCO/p-a-Si:H heterointerface. The B₂H₆ plasma treatment can create highly concentrated doping atoms in an extremely thin layer. Indeed, a heavily B-doped layer is employed in a-Si:H thin film solar cells and successfully enhances the conversion efficiency. ^17,18) In SHJ solar cells, the B₂H₆ plasma treatment can suppress the doping level in the p-layer, while increasing the doping level at the p-a-Si:H surface. Hence, the contact properties of the TCO/p-a-Si:H heterointerface are expected to be improved without significant degradation of its passivation performance.

All depositions were performed on c-Si(100) substrates by a plasma-enhanced CVD (PECVD) system with a frequency of 27.12 MHz (ULVAC Inc., CME-200J). SiH₄, H₂, PH₃, and B₂H₆ were used as source gases. The i-a-Si:H, n-type a-Si:H (n-a-Si:H), and p-a-Si:H layers were deposited on c-Si substrates by the PECVD system. The deposition parameters are given in Table I. In this study, an i-a-Si:H bilayer was employed to enhance the passivation performance. ^19,20)

2.1. Doping concentration and passivation performance measurements

Double-sided mirror-polished and floating zone grown n-type c-Si(100) substrates were cleaned by Semicoclean-23 (Furuuchi Chemicals) for 6 min in an ultra-sonic bath. The wafer thickness and resistivity were 280 μm and 1–5 Ω·cm, respectively. The native surface oxide layer was then removed by immersing the substrates into 5.0% HF for 1 min, followed by dipping into ozonized and deionized water (DI-O₃) for 10 min to clean the surface and form a protective oxide layer. The O₃ concentration in the DI-O₃ was about 25 ppm. The protective oxide layer was removed by immersing the substrates in 2.5% HF for 1 min, and then oxidizing pre-treatment (OPT) was conducted by immersing the substrates in 1.5% H₂O₂ solutions for 35 s to prevent the desorption of hydrogen around the a-Si:H/c-Si interfaces. ^21,22) Subsequently, the substrates were quickly loaded into a chamber for the PECVD system. The i-a-Si:H and p-a-Si:H layers were deposited on only the front side of c-Si substrates for doping concentration measurements and on both sides for passivation performance measurements. After depositing the p-a-Si:H layer, B₂H₆ plasma treatment was carried out by stopping SiH₄ flow with RF power on for 15 s. Figures 1(a) and 1(b) show the schematic processes without and with the B₂H₆ plasma treatment, respectively. Post-deposition annealing (PDA) was performed in the air using a hot plate at 200 °C for 30 min.

Zoom In Zoom Out Reset image size

Boron concentrations (C_B) and Si secondary ion intensity (I_Si) were examined by a secondary ion mass spectroscopy (SIMS) system (ATOMIKA 4500) using O₂ ⁺ ions of 0.5 keV as primary ion species. A boron ion implanted c-Si reference was used to evaluate C_B in the heterostructures. Quasi-steady-state photoconductance measurements were performed in 1/64 generalized and transient modes (Sinton Instruments, WCT-120TS) to characterize the effective lifetime (τ_eff). ^23,24)

2.2. Contact resistivity measurements

2.3. Solar cell fabrication and characterization

2.4. Numerical simulation of solar cells

Figure 2 shows the depth profiles of B concentration (C_B) and Si secondary ions intensity (I_Si) in the i-a-Si:H/p-a-Si:H/n-c-Si heterostructures fabricated (a) without and (b) with B₂H₆ plasma treatment. For the SIMS measurements, 30 nm thick, undoped a-Si:H layers were deposited on the p-a-Si:H surfaces to mitigate the modulation of the concentration at the surface. A unimodal C_B peak was observed for the conventional heterostructure, whereas a bimodal C_B peak was observed for the heterostructure employing the B₂H₆ plasma treatment. The C_B peaks of ∼9 × 10²⁰ cm⁻³ at about 36 nm are observed for the two structures, indicating that the peaks arise from the p-a-Si:H layer. The high C_B peak of approximately 1 × 10²¹ cm⁻³ at 30 nm stemmed from the B₂H₆ plasma treatment. Interestingly, a bimodal C_B peak was observed by using the B₂H₆ plasma treatment. We speculate that this dip in the C_B peak was originated from reduction of input plasma power. A swift stop of SiH₄ flow possibly causes a reduction in pressure, which leads to an unstable plasma state. The B was incorporated in an undoped a-Si:H protective layer at the order of 10¹⁸ cm⁻³ possibly due to residual B₂H₆ in the deposition chamber. This doping level is significantly lower than those of the p-a-Si:H layer, and thus the doping profiles were hardly affected by B in the protective layers. Therefore, a heavily doped p-a-Si:H layer was formed on the p-a-Si:H surface by the B₂H₆ plasma treatment.

Zoom In Zoom Out Reset image size

Figure 3 shows the I–V characteristics of the Ag/IWO/p-a-Si:H/i-a-Si:H/p-c-Si heterostructures (a) without and (b) with the B₂H₆ plasma treatment after post-annealing at 200 °C for 30 min. The inset shows the schematic sample structure for TLM measurements. Figure 3(c) shows the total resistance (R_total) derived from the I–V characteristics as a function of electrode spacing (d). In the conventional TLM model, R_total and ρ_c are expressed as follows,

$$\begin{eqnarray}&&{R}_{\mathrm{total}}=\frac{{R}_{\mathrm{sh}}}{W}d+2{R}_{{\rm{c}}}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{\rho }_{{\rm{c}}}={R}_{{\rm{c}}}{L}_{{\rm{T}}}W\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&{L}_{{\rm{T}}}=\frac{{R}_{{\rm{c}}}W}{{R}_{\mathrm{sh}}},\end{eqnarray} \tag{ 3 }$$

where R_sh is the sheet resistance of the Si substrate, W is the electrode width, R_c is the contact resistance, and L_T is the transfer length. The ρ_c values of the samples without and with the B₂H₆ plasma treatment were 2.793 and 0.859 Ω‧cm², respectively. The ρ_c was reduced to about one-third by the heavily B-doped layer.

Zoom In Zoom Out Reset image size

Hall measurements were carried out for the IWO layer before and after post-annealing at 200 °C for 30 min. The electron density, mobility, and resistivity of the IWO layer are summarized in Table III. The electron density and mobility are slightly improved after the post-annealing, which is ascribed to the crystallization of the IWO layer and an increase in activated dopants. Figure 4 shows (a) the schematic structures of the Ag/IWO heterostructure for TLM measurements. I–V characteristics of the heterostructures (b) before and (c) after annealing at 200 °C for 30 min (d). Total resistance (R_total) of the heterostructures as a function of electrode distance (d) before and after annealing at 200 °C for 30 min. From Fig. 4(d), the slopes of the linear regression became smaller after annealing, meaning the R_sh of the IWO layer became smaller. This reduced R_sh is explained by the reduced resistivity observed by Hall measurements without contradiction. Note that the decreasing rate of R_sh values measured by TLM before and after annealing was smaller than that obtained from Hall measurements. This difference is possibly arisen from currents from the side of the pads in addition to current between pad electrodes, which contribute to reduce the resistance. The decrease in resistivity due to annealing enhances currents from the side of the pads. Furthermore, the resistance may have fluctuated due to a deviation of the electrode size from the design value for the Hall and TLM measurements. These effects result in the large decreasing rate of R_sh in the TLM measurement than the Hall measurement. The ρ_c values were slightly improved from 0.024 to 0.016 Ω‧cm² after the post-annealing. The ρ_c value of the Ag/IWO structure was significantly smaller than that of the IWO/p-a-Si:H/i-a-Si:H heterostructure, indicating that the contact resistance of the heterostructures was dominated by the IWO/p-a-Si:H/i-a-Si:H heterointerfaces.

Zoom In Zoom Out Reset image size

Figure 5 shows (a) J–V curves and (b) EQE of the SHJ solar cells with and without the B₂H₆ plasma treatment. As can be readily seen, the J–V characteristics are improved by using the B₂H₆ plasma treatment. The short-circuit current density (J_SC), V_OC, fill factor (FF), conversion efficiency (η), series resistance (R_s), shunt resistance (R_p), and ideal factor (n) are summarized in Table IV. The R_s, R_p, and n were determined by comparing the I–V curve of a solar cell at one Sun with the one at 0.1 Sun. ³¹⁾ The B₂H₆ plasma treatment significantly improved FF due to lower R_s and higher R_p. Furthermore, there were no remarkable differences in EQE spectra between the SHJ solar cells with and without B₂H₆ plasma treatment. Figure 6 shows τ_eff of symmetrically passivated c-Si using the B₂H₆ plasma treatment as a function of minority carrier density (MCD). The τ_eff at MCD of 1 × 10¹⁵ cm⁻³ was improved from 1.4 to 2.1 ms by employing the B₂H₆ plasma treatment. The implied V_OC (i-V_OC) was 0.694 V and 0.695 V for the sample without and with the B₂H₆ plasma treatment, respectively. Almost identical i-V_OC was obtained irrespective of the B₂H₆ plasma treatment, suggesting that the difference in the passivation performances between both is less distinct at high injection level. On the other hand, the passivation performance at low injection level was enhanced by using the B₂H₆ plasma treatment. The improvement in solar cell performance shown in Fig. 5(a), i.e. improved FF, can be explained by the improved passivation performance at low injection level. Hence, the B₂H₆ plasma treatment can reduce resistive loss without sacrificing passivation performance. The reduced resistive loss is attributed to the lowered resistance of the p-a-Si:H layer and contact resistance at the IWO/p-a-Si:H heterointerfaces, as shown in Fig. 3. In addition, the reduction of leak current and recombination at the heterointerfaces seem to be responsible for the improved FF, since decreased n was observed for the sample with the heavily B-doped layer.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 7 shows the numerically simulated band diagrams of the n-c-Si/p-a-Si:H/IWO heterointerfaces (a) without and (b) with the B₂H₆ plasma treatment by the AFORS-HET simulator. Their band diagrams were obtained at each maximum power point (MPP) under 1-Sun illumination. It is seen that upward band bending is enhanced by employing the B₂H₆ plasma treatment in Fig. 7(b) compared with Fig. 7(a). This is probably caused by the increase in Fermi energy difference between p-a-Si:H and n-c-Si and the suppression of the depleted region in p-a-Si:H at the TCO/p-a-Si:H heterointerface owing to the higher hole concentration in the heavily B-doped layer. The upward band bending is responsible for the mitigated resistive loss and subsequently increased FF observed in Fig. 5. Note that the band diagrams are idealized. The impact of the TCO work function was not considered since ideal TCO/p-a-Si:H contact was assumed. Furthermore, the doping levels of the heavily B-doped layer and the p-layer are not accurate. Since the activated doping level was unclear, the doping levels were referred to as the peak B concentration shown in Fig. 2 and they were modified to simulate conversion efficiency in Table IV. The work function of TCO and the concentration of activated dopants affect the real band diagrams. The simulated J–V curves of both two solar cells are shown in Fig. 8 and the solar cell parameters are summarized in Table V. The V_OC and FF, and consequently η were improved by employing the B₂H₆ plasma treatment. The improved V_OC in the numerical simulation was probably attributed to the assumption of no defects at the heterointerfaces.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The efficient hole collection is caused by facilitated tunneling from c-Si to p-a-Si:H and the recombination junction of the TCO/p-a-Si:H interface. It is considered that the extraction of holes in c-Si is ruled by tunneling from c-Si to p-a-Si:H via i-a-Si:H, work function matching between TCO and p-a-Si:H, and trap-assisted tunneling (TAT) at the TCO/p-a-Si:H. ^12,32) Since the VB offset of 400–600 meV is created at the a-Si:H/c-Si heterointerface, ^33,34) tunneling of holes through the i-a-Si:H layer is required. The upward band bending observed in Fig. 7(b) suppresses band flattening at the MPP, which facilitates the tunneling from c-Si to p-a-Si:H. Furthermore, electrons can be repelled out of the n-c-Si/p-a-Si:H heterointerface as a by-product of the enhanced upward band bending, and hence carrier recombination at the a-Si:H/c-Si interface is reduced. It is known as well that the work function mismatch between TCO and p-a-Si:H results in a depleted a-Si:H due to the comparatively low carrier density of p-a-Si:H than that of TCO. Heavily doped p-a-Si:H, as observed in Fig. 2(b) can render the depletion width in the p-a-Si:H layer narrow. Furthermore, recombination between holes in p-a-Si:H and electrons in TCO is necessary for solar cell operation via TAT, since the majority carrier of TCO and p-a-Si:H are electrons and holes, respectively. It is reported that energy states in doped a-Si:H are additionally introduced near the band edges by heavy doping, ^35,36) suggesting many trap states are created on the p-a-Si:H surface by the B₂H₆ plasma treatment. Hence, the heavily B-doped layer would lead to efficient tunneling of holes across i-a-Si:H by enhancement of the upward band bending, resulting in hole accumulation at the TCO/p-a-Si:H heterointerface. The accumulated holes can readily recombine with electrons in TCO via the increased trap states in the heavily B-doped layer. These results reduce the ρ_c of the heterostructures with heavily doped p-a-Si:H, as observed in Fig. 3(c). Note that the ρ_c obtained by TLM measurements includes the contact resistance contributed by the tunneling from c-Si to p-a-Si:H and the exchange between holes in p-a-Si:H. Therefore, the reduced ρ_c would stem from enhanced tunneling processes at the IWO/p-a-Si:H/i-a-Si:H/c-Si heterointerfaces, leading to increased FF and thus improved conversion efficiency.

In conclusion, we investigated the impact of the B₂H₆ plasma treatment on the structural and electrical properties of p-a-Si:H/i-a-Si:H/c-Si heterostructures and the performance of the SHJ solar cells. The B₂H₆ plasma treatment is demonstrated by continuous B₂H₆ flows with RF power after depositing the p-a-Si:H layer. The SIMS measurements revealed an increase in B concentration at the p-a-Si:H surface by employing the B₂H₆ plasma treatment. Furthermore, specific contact resistance is decreased by about one-third owing to the B₂H₆ plasma treatment. These results indicate that B is heavily doped at the p-a-Si:H surface. No degradation of passivation performance is observed in the presence of the heavily B-doped layer. The η of SHJ solar cells with the heavily B-doped layer is improved by a higher FF than that without the heavily B-doped layer due to decreased R_s and increased R_p. Numerical simulations reveal that the upward band bending at the TCO/p-a-Si:H heterojunction is enhanced by the heavily B-doped layer. The enhanced upward band bending is responsible for the decrease in R_s and increases in R_p owing to facilitated tunneling holes from c-Si in the p/i-a-Si:H layers and the TCO/p-a-Si:H heterointerface. We propose the B₂H₆ plasma treatment, adaptable to the conventional fabrication procedures of SHJ solar cells. This process can improve TCO/p-a-Si:H contact properties without sacrificing its passivation performance and consequently produce high-performance SHJ solar cells.

This work was supported by the New Energy and Industrial Technology Development Organization (NEDO).