1. Introduction

Much attention has been paid to the $\mu$c-Si for promising applications in thin-film solar cells^[1]. In contrast to amorphous silicon, $\mu$c-Si exhibits promoted light absorption in the infrared region of the solar spectrum. An absorber layer with low defect density is necessary to improve efficiency of $\mu$c-Si solar cells^[2]. Defect states in $\mu$c-Si will promote carrier recombination, resulting in a decrease of carrier collection and solar cell efficiency. Therefore synthesis of $\mu$c-Si films with low defect density is required.

The adjustment of deposition parameters in the initial stage of $\mu$c-Si deposition has been used to improve film quality. The decrease of a thick incubation layer (with high defect density^[3]) is obtained using initial profiling of silane (SiH$_{4}$) flow^{[4, 5]}, hydrogen (H$_{2}$) flow^{[4, 6]} and the VHF power density^[4]. Also gradual SiH$_{4}$ introduction and slow power application has been used to prepare $\mu$c-Si with a low dangling-bond defect density^[7]. It has also been reported that HPT of the absorber layer in the n-i-p solar cell can decrease defect density at the i/p interface^[8]. Therefore the initial growth is very important to decrease defect density and HPT may be a good method to adjust the $\mu$c-Si film quality.

In this paper, we propose a new approach, which is hydrogen plasma treatment on an as deposited Si:H layer at the initial stage of $\mu$c-Si growth to decrease defect density. An effectively improvement of solar cell performance is obtained using absorber layer with HPT.

2. Experimental details

$\mu$c-Si thin films were prepared by very high frequency (60 MHz) plasma enhanced chemical vapor deposition (VHF-PECVD) using SiH$_{4}$ and H$_{2}$ gas at a temperature of 220 ℃. The process pressure for film deposition was 133 Pa and for HPT was 117 Pa. For the sample with HPT, a thin initial silicon film was first deposited for 5 min and then it was treated by hydrogen plasma for 2 min, after which $\mu$c-Si film was continually deposited for 45 min. The compared sample without HPT was directly deposited for 50 min. The defect density in $\mu$c-Si was characterized by the sub-band gap absorption, which was determined from the constant photocurrent method (CPM). To obtain detailed information on bonded hydrogen, we performed IR spectroscopy measurements between 400 and 4000 cm$^{-1}$ with a Niconet560 FTIR spectrometer. Raman spectroscopy was carried out to study structure properties by using a micro-Raman spectrometer (Renishaw-RM2000) operating with the 633 nm line of a helium-neon laser.

Single junction solar cells were deposited with the structure: indium tin oxide (ITO)/p/i/n/stainless steel and the area was 0.07 cm$^{2}$. The I-V characteristics of the solar cells were measured under AM 1.5 illumination using a Xeon lamp solar simulator. Quantum efficiency (QE) was also measured for the cells.

3. Results and discussion

The low absorption ($\alpha$) of $\mu$c-Si films with and without plasma treatment at the initial stage of growth is shown in Fig. 1. It is reported that the value of the optical absorption coefficient in 0.8-1.0 eV scales roughly with the density of dangling bonds^[9]. We estimated the defect density in $\mu$c-Si from the sub-band gap absorption coefficient. Figure 1 shows that hydrogen plasma treatment causes a decrease in absorption coefficient in 0.8-1.0 eV, which reveals less density of defects with HPT than its counterpart without HPT. Lower defect density would reduce the carriers' recombination and has potential to improve carriers' collection.

Figure 2 presents the infrared (IR) spectroscopy (500-750 cm$^{-1}$) in the $\mu$c-Si. The hydrogen content $C_{\rm H}$ in μc-Si films can be evaluated by numerical integration of the Si-H wagging mode at around 630 cm$^{-1}$ through Eq. (1).

$\begin{equation} C_{\rm H}=\frac{A}{N_{\rm Si}}\int \frac{\alpha (\omega)}{\omega}{\rm d} \omega, \end{equation}$

(1)

where A is the proportionality constant and the value used for $\mu$c-Si is 2.1 × 10$^{19}$ cm^-2[10], and $N_{\rm Si}$ is the atomic density of silicon atoms, which is taken as 5 × 10$^{22}$ cm^-3[10]. The result shown in Table 1 exhibits an increase in $C_{\rm H}$ from 8.7% to 11.6% with HPT. The increased 2.9% $C_{\rm H}$ reveals that more H atoms are incorporated in Si:H matrix. We suggest that the decreased defect density indicated in Fig. 1 is due to the reduction of dangling bonds, which is passivated by increased H atoms.

The grain boundary fraction ($f_{\rm GB}$) deduced from Raman spectroscopy is summarized in the second column of Table 1. The Raman spectra were deconvoluted in integrated crystalline, $I_{\rm C}$ (520 cm$^{-1}$), amorphous, $I_{\rm A}$ (480 cm$^{-1}$) and grain boundary, $I_{\rm GB}$ (510 cm$^{-1}$) peaks. The grain boundary fraction, $f_{\rm GB}$, was calculated from Eq. (2)^[11].

$\begin{equation} X_{\rm GB}=I_{\rm GB}/(I_{\rm C}+I_{\rm GB}+Y I_{\rm A}), \end{equation}$

(2)

where Y is a constant and a value of 0.9 was chosen for Y^[12]. It shows a rise of $f_{\rm GB}$ from 30.2% to 37.8% by HPT. Considering the decreased defect density, we suggest that the increased $f_{\rm GB}$ might be due to the filling of intergranular space by amorphous tissue, which increases the amorphous/crystalline boundary fraction. A low defect density would be obtained at grain boundary for a good passivation by amorphous tissue^[13].

IR spectroscopy in the range of the hydride stretching modes (1850-2230 cm$^{-1}$) of $\mu$c-Si is shown in Fig. 3. The Si-H middle stretching mode (MSM) at around 2040 cm$^{-1}$ corresponds to Si-H in a platelet-like configuration^[14] in the amorphous/crystalline interfaces^[15]. The increase of MSM hydrides with HPT is indicated by the arrow in Fig. 3. This reveals that HPT helps to form a hydrogen-dense amorphous tissue around the small crystalline grains^[16], which is consistent with the increase of $f_{\rm GB}$ deduced from Raman spectra. The increased $C_{\rm H}$ (calculated in Fig. 3) is proposed to be incorporated within the amorphous tissue which fills the pores between crystalline grains and decreases dangling bonds. Thus the reduction of defect density by HPT is ascribed to the better grain boundary passivation.

Finally, parameters of $\mu$c-Si solar cells fabricated using microcrystalline i-layer deposited with and without HPT during the initial stage of growth are shown in Table 2. The $J_{\rm sc}$ increases from 14.49 to 17.15 mA/cm$^{2}$, which is 18% higher than its counterpart. Figure 4 presents quantum efficiency (QE) of the two cells, an obvious improvement of response in long wavelength range (500-850 nm) is displayed and this is in agreement with the enhancement of $J_{\rm sc}$. We propose that HPT decreases defect density in the absorber layer, thus reduces carrier recombination, which promotes carrier collection and increases the QE.

4. Conclusion

The present study indicates lower defect density in $\mu$c-Si deposited with HPT during the initial stage of growth compared with its counterpart. This is due to the formation of hydrogen-denser amorphous tissue around the small crystalline grains which passivates dangling bonds in the crystal surface. Single junction solar cells have been fabricated using an i-layer with and without HPT. As defect density in the i-layer with HPT decreases, QE in long wavelength range is improved and the value of $J_{\rm sc}$ and FF increased by 18% and 19.4% respectively. The results show that HPT during the initial stage of $\mu$c-Si formation can be an effective approach to improve the performance of $\mu$c-Si solar cell.