Abstract

N-type microcrystalline silicon oxide thin films (n- c- :H) have been deposited by VHF-PECVD (40 MHz) with reactant gas mixtures of / and . N- c- thin films exhibiting low refractive index value ( ), and medium/high conductivity (  S/cm) are suitable to be used as an “n-type reflector” in micromorph tandem solar cells. Transmission electron microscopy (TEM) results show that microstructures of n- c- :H thin films contain nanocrystalline Si particles, which are randomly embedded in the a- matrix. This specific microstructure provides n- c- :H thin films excellent optoelectronic properties; therefore, n- c- :H thin films are appropriate candidates for “n-type reflector” structures in Si tandem solar cells.

1. Introduction

Microcrystalline hydrogenated silicon oxide (μc- :H) thin films have been widely investigated due to their outstanding optoelectronic properties, such as wideband gap (low light absorption in the visible range) and high photo/dark conductivity [13]. By adjusting plasma-enhanced chemical vapor deposition (PECVD) processing of n-type μc- :H (n-μc- :H) thin films, a wide range of optical band gap and electrical conductivity can be obtained, which primarily depends on the amount of oxygen incorporated inside and their relative microstructures. Therefore, this material draws much attention and has been used in many photovoltaic applications [4, 5]. One of the optical applications in photovoltaics is that it can be inserted between a-Si/μc-Si micromorph tandem solar cells as an intermediate reflector structure (IRS) due to its relatively low refractive index [4]. The purpose of applying intermediate reflectors in Si thin-film tandem solar cells is that a thin a-Si top cell is needed to minimize its inherent light-induced degradation of amorphous material. Therefore, more light has to be reflected back from the intermediate reflector and then absorbed in a-Si top cells to enhance their short-circuit current densities ( ). In this paper, we first present optoelectronic properties of n-μc- :H thin films and their microstructural features were characterized by transmission electron microscopy (TEM) techniques. Secondly, a-Si/μc-Si tandem solar cells with IRS were modified to “n-type reflector” structures (NRS) by incorporating specific n-μc- :H thin films, which simplifies the process complexity of modern a-Si/μc-Si tandem solar cells. As shown in Figure 1, the “n-type reflector” (NR) can efficiently reflect incident visible light (500~800 nm, primarily) backward and also forms the n-type active layer for a-Si top cells, which ideally creates a tunneling junction with bottom μc-Si p-layers. This n-p tunneling junction must provide good carrier recombination paths (for electrons from top cells and holes from bottom cells) and maintains a low series resistance for a-Si/μc-Si tandem solar cells. Finally, cell parameters and respective external quantum efficiency (EQE) spectrum were also present to verify the adaptation of NRS.

2. Experimental

Silicon micromorph tandem solar cells were fabricated by using very high-frequency plasma-enhanced chemical vapor deposition (VHF-PECVD, 40 MHz) at 200°C to deposit Si thin films and pulsed dc magnetron sputtering for electrodes. The intermediate reflector/layer and n-type reflector (n-μc-SiOx:H thin films) were also deposited by the VHF-PECVD technique employing reactant gas mixtures of H2/SiH4, PH3, and CO2 (PH3/SiH4 = 0.025). The final cell structures are “Asahi glass substrate (U type)/a-Si top cell (~300 nm, w/or w/o Si n-layer)/n-μc- :H layer/μc-Si bottom cell (~3 μm)/ZnO:Ga/Ag”.

n-μc- :H thin films were deposited on bare glass (Corning Eagle 2000) to determine their optical and electrical properties. Crystalline volume fraction (crystallinity; μc %) of n-μc-SiOx:H thin films is deduced from integral intensities of deconvoluted Gaussian bands in Raman spectroscopy (excitation at 514.5 nm of an Ar+ laser). From the empirical equation, μc , where is the integral intensity of the Gaussian band with its peak near 480 cm−1 for a-Si, and is the sum of integral intensities from two Gaussian bands with peaks at 500~510 cm−1 and 515~518 cm−1 for crystalline Si. Transmittance and reflectance spectra of n-μc- :H thin films were measured by UV-VIS-NIR photospectrometer (Perkin-Elmer λ750) within a spectral range from 300 to 1200 nm, and then optically simulated to calculate the relative refractive index, and values. The dark conductivities (σ) of n-μc- :H thin films were measured by the four-point probe method with a pair of Ag coplanar electrodes on their surface under vacuum <10−3 torr at room temperature.

Microstructures of n-μc-SiOx:H thin films were characterized by transmission electron microscopy (JEOL, JEM-2100F; 200 keV). I-V tests of silicon micromorph tandem solar cells ( ) were performed under AM1.5 illumination (1 Sun) at room temperature, and EQE spectrum of each subcell was measured to determine their .

In Raman spectroscopy, the peak at ~480 cm−1 represents the transverse optical (TO) mode in a-Si:H and have a broad full width at half maximum (FWHM). On the other hand, the Raman peak of ideal crystalline Si is located around 520 cm−1, which is the position of the TO mode in crystalline Si. However, disorder and/or the size of Si crystallites may make a deviation of the peak position from 520 cm−1, and it has been widely discussed in research [610]. Afterward, Xia et al. [11] and Islam and Kumar [12] further incorporated the influence of Si nanocrystallite size distributions into the Raman analysis and demonstrated that extra Raman peaks exist with their peak positions shifting to lower frequency. In this paper, we also included a Gaussian band with Raman peaks at  cm−1 to calculate crystalline volume fractions of n-μc- :H thin films. The intermediate lower frequency band is associated with Raman scattering from nanocrystal boundaries and/or disorder crystalline Si arrangements [11, 13, 14]. Therefore, we estimated the crystalline volume fraction of n-μc- :H thin films from the empirical relation: the sum of integrated intensities from two Gaussian Raman bands with peak positions at ~516 cm−1 and  cm−1; = the integrated intensity of the Gaussian Raman band with its peak near 480 cm−1.

3. Results and Discussion

Increasing the hydrogen dilution of n-μc- :H thin films depositions enhances the crystallinity themselves. As shown in Figure 2(a), there is a very low crystalline volume fraction of thin films (~amorphous, μc% ~0) for [CO2]/[SiH4] (R = [CO2]/[SiH4]: ratios of CO2 flow rates to silane flow rates) = 2.33~3.33 when the hydrogen gas flow rate ([H2]) that is fixed at 600 sccm during the CVD process. However, by increasing hydrogen dilution during the deposition of μc- :H thin film, a great enhancement of crystallinity is obtained in Figure 2(b). For example, crystallinity of n-μc- :H thin films deposited with [H2] = 900 sccm and reaches near 40%, as compared with that ~0% when they are deposited with [H2] = 600 sccm and . In addition, crystallinity of n-μc- :H thin films also strongly depends on the amount of oxygen incorporated inside the μc-SiOx:H thin films. As shown in Figure 2(a), while values continue decreasing, from 3.33 to 1.33, a narrower Raman band with its peak at ~516 cm−1 gradually appears, which shows an increase of crystalline volume fraction of Si crystallites in the n-μc- :H thin films with lower amounts of oxygen.

Figure 3 shows the refractive index (n600 nm), dark conductivity, and crystalline volume fraction of n-μc- :H thin films, which are deposited with varied reactant flow rate ratios of [CO2]/[SiH4] and [H2]. As the hydrogen flow rate is fixed to 600 sccm, n-μc- :H thin films deposited with R < 2 show an abrupt increase of dark conductivity, from 10−10 S/cm to 10−6 S/cm, which accompanies with a boost of crystalline volume fraction of thin films (μc% = 20~30%). On the other hand, the refractive index of n-μc-SiOx:H thin films varies gradually, increasing from 1.8 to 2.1, during the whole gas flow range of [CO2]/[SiH4]. Another set of data in this figure (open symbols) show that as the hydrogen flow rate increases to 900 sccm with , crystalline volume fraction of n-μc- :H thin films rises to a high value ~40%, and a great enhancement of dark conductivity of thin films is also obtained (>2 order). However, when the hydrogen gas flow increases from 600 to 900 sccm with same R values, the change of n600 nm, is quite small (Δ ). Optical band gap (E04) of n-μc-SiOx:H thin films (derived from transmittance/reflectance spectrum), deposited with fixed [H2] = 600 sccm, changes from 2.48 to 2.78 eV (as ), which shows high optical band gap energy (R dependent) and low optical absorption of n-μc- :H thin films within the visible light spectrum.

Figures 4(a) and 4(b) show the TEM bright field and dark field images of n-μc-SiOx:H thin films deposited on glass substrates. The inset of Figure 4(b) shows the selective area diffraction (SAD) pattern of n-μc- :H thin films, and it is characterized by TEM crystallographic analysis, which identifies a Si crystalline phase, for example, Si (111), (311), and (400) for the first three rings, existing inside n-μc- :H thin films. The dark field image of Figure 4(b) shows that the distribution of Si nanocrystallites in the n-μc- :H thin films represents a random type and also clusters of nanocrystallites exist in the oxygen-rich amorphous matrix. As for high-resolution TEM (HRTEM) images in Figure 4(d), we demonstrated Si nano-crystallites with their sizes 2~10 nm inside the n-type reflector layer of a-Si/μc-Si tandem cells; similar results are obtained as n-μc- :H thin films were deposited on bare glass substrates (not shown).

He et al. [15] investigated the conduction mechanism of hydrogenated nanocrystalline Si films and concluded that the high conductivity of nanocrystalline Si films results mainly from the electrical conduction between Si crystallites. As shown in Figures 3 and 4, electrical conductivity of n-μc- :H thin films depends on the density of the randomly dispersed nanocrystalline Si phase, which is proportional to their crystalline volume fractions; in addition, the doping efficiency of crystalline Si is usually higher than the amorphous :H phase, which may enhance the conductivity of n-μc-SiOx:H thin films further as compared with that of n-a- :H thin films. Also, the refractive index, n600nm, of n-a- :H thin films strongly depends on the amount of a- :H phase, which lower the values of n600nm to ~2. All these unique optoelectronic characteristics discussed above make n-μc- :H have thin films high electrical conductivity and transparency; thus they are suitable to be applied in NRS of thin film solar cells.

Three types of a-Si/μc-Si tandem cells were measured under AM1.5 (100 mW/cm2) illumination at room temperature, which are standard a-Si/μc-Si tandem cells (without IRS and/or NRS), and cells with IRS and NRS. N-μc- :H thin films with excellent optoelectronic properties (σ ~ 6 × 10−8 S/cm; μc% ~ 40%; t ~ 50 nm; n600 nm ~ 2; [PH3]/[SiH4] = 0.043) are used as light reflectors in IRS and NRS tandem solar cells. The ability of light reflection from IRS or NRS is proportional to the product of (Δ ), where Δ is the variance in index of refraction between the reflector and the surrounding Si layers, and is the thickness of the reflector. In our case, we need the thickness of n-μc- :H thin film (n600 nm~ 2) >40 nm to be an effective light reflector in the wavelength region 500 to 800 nm. Table 1 displays I-V parameters of these a-Si/μc-Si tandem cells. As in our standard cells, the of a-Si top cells does not match well with that of μc-Si bottom cells (top-cell current limited); thus, the net and energy conversion efficiency of tandem cells are relatively low. Conventional tandem cells with IRS will have more light reflected (500~800 nm, Figure 5) from the intermediate reflector (n-μc- :H thin films) to the a-Si top cells; therefore, more light can be absorbed in top cells to improve the total of tandem cells. Due to the high conductivity and excellent optical properties of n-μc- :H thin films (low absorption and low n600nm), we simplify cell processing with IRS to “n-type reflector” structures between a-Si/μc-Si tandem solar cells, which can take the advantage of eliminating the Si n-layer processing of a-Si top cells. It is demonstrated in Table 1 and Figure 5. Comparable cell parameters are achievable for tandem cells with NRS and IRS (Δ ≦ 0.02 V; Δ ~ 0.15 mA/cm2; ΔF.F.~1.8%), and from EQE analyses, as shown in Figure 5, NRS tandem cells also function well in current matching as some portion of incident light between 500 and 800 nm can be efficiently reflected from the n-type reflector to a-Si top cells. Consequently, both NRS and IRS tandem cells accompany with a decline in QE of μc-Si bottom cells within the wavelength range due to more light being reflected toward a-Si top cells. Nevertheless, n-μc- :H thin films with dark conductivity lower than 10−10 S/cm are not suitable to be used as NRS in our a-Si/μc-Si tandem cells, which usually show high series resistance and low fill factor (not shown). More detailed analysis about the effects of optoelectronic properties of n-μc- :H thin films (or multilayer reflector structures) on cell parameters are still undergoing.

4. Conclusions

In conclusion, it is present that microstructures of n-μc- :H thin films contain nanocrystalline Si particles (2~10 nm), which are randomly embedded in the a-SiOx matrix. This specific microstructure provides n-μc- :H thin films sufficient dark conductivity and low refractive index, which are suitable as efficient light reflectors in a-Si/μc-Si tandem solar cells. In n-type reflector structures of tandem solar cells, the n-μc- :H thin film is functioned as an activated doping layer of the top cell and simultaneously as an intermediate light reflector in a-Si/μc-Si tandem solar cells, which takes the advantage of omitting a processing step of the Si n-type layer during tandem cell fabrications. Also, we have demonstrated that silicon thin-film tandem solar cells with μc- :H-based NRS have comparable cell parameters to those with conventional intermediate reflector structures, which shows good carrier conduction at the n-μc- /p-μc-Si tunneling junction. All these results would create alternative opportunities to reduce the fabrication cost of thin film solar cells/modules.

Acknowledgment

The authors would like to thank the support from the Bureau of Energy, Ministry of Economic Affairs, Taiwan, under Grant no. A455DR1110.