1. Introduction

Textures produced by anisotropic alkaline etching, with a typical size of several microns, have good light-trapping properties in crystalline silicon (c-Si) solar cells [1], but antireflection coatings are still necessary in order to get high conversion efficiency. Nano-structured c-Si (with cones, wires, or holes on the wafer surface) [2–9] has received steadily increasing interest as possible architectures for high efficiency and low cost solar cells, because they exhibit very low reflection especially in the visible region. Many approaches were developed to prepare these nanostructures [10,11], e.g., reactive ion etching, irradiating silicon with fs-laser pulses and metal-assisted chemical etching. However, the recombination loss associated with the dramatically increased surface area limits the efficiency of real cells [7].

To reduce the material cost and improve the efficiency further, designs of solar cells based on c-Si film of several microns thick have also been explored. Two typical models are silicon nanowire array and nanohole array in silicon [12–15]. But it is difficult to get high efficiency from such nano-structured c-Si film, due to the increased recombination loss and contact resistance arising from nano-structures [16]. A better design would be nano-holes or cones on a solid c-Si film [17,18]. However, in principle, there should be a somehow pre-existing silicon thin film in order to realize these designs experimentally. By etching a silicon wafer down to several microns one can obtain a free-standing c-Si film, and Wang et al. demonstrated a solar cell with nanocones on such a c-Si film [19]. Silicon-on-insulator (SOI) wafers can also provide thin c-Si films, and solar cells with nanocones or nanopyramids based on expensive SOI wafers were reported very recently [20,21]. On the other hand, c-Si thin films full of straight pores of micrometers in diameter (“macropores”) can be detached from a silicon substrate by electrochemical etching (the remaining substrate could be used to produce macroporous c-Si films continuously [22]). Recently solar cells based on macroporous c-Si films were demonstrated by Ernst et al. [23,24], but lacking efficient light-trapping structures.

In this paper, we present a kind of “one-step strategy” in which the c-Si thin film itself and the antireflection/light-trapping structures are fabricated simultaneously and directly from a silicon wafer, rather than relying on a pre-existing c-Si film. We first present the comparison of absorptivity, minority charge carrier lifetime, and photovoltaic characteristics between macroporous black silicon and silicon nanowires (both with substrates) in Section 2, and then show in Section 3 the macroporous black silicon film (detached from the substrate) and analyze its optical property in view of thin film c-Si photovoltaic applications.

2. Macroporous black silicon

The macroporous black silicon was fabricated by exposing an anodically biased n-type <100> silicon wafer (500 μm thick, Czochralski, bulk resistivity 2~6 Ω·cm) to dilute hydrofluoric acid (5 wt % HF), as described previously [25]. A hexagonal pattern of inverted pyramidal pits on the exposed surface, defined by photolithography, served as the initial seeds for pore growth (interpore distance 6 µm). The back side of the wafer was highly doped with phosphorus for ohmic contacts (by rapid thermal processing at 1000°C for 120 s with spin-on dopant P509 from Filmtronics). For a given interpore distance and fixed etching conditions, the pore diameter is determined by the current density flowing across the Si/HF interface, so that the diameter can be modulated by tuning the current density via backside illumination. Here the pore shape was chosen intuitively with some arbitrariness. As the cross-sectional SEM image shown in Fig. 1(a), blade-like tapered silicon walls at the entrance of the pore and the kinks inside the pore serve as antireflection and light-trapping structures, respectively.

 

Fig. 1 Cross-sectional SEM images of (a) macroporous black silicon and (b) silicon nanowires. The etch depth is around 17 μm for both cases.

Download Full Size | PDF

Optical absorptivity is a key factor to be considered before evaluating the photoelectrical conversion performance of solar cells. Reflection and transmission were measured by a spectrophotometer (Perkin Elmer Lambda 950) with a 150 mm integrating sphere, and the spectra are shown in Figs. 2(a)–2(b). For comparison, samples of silicon nanowires (with diameter in the range of 20–200 nm, filling fraction ~50%), as shown in Fig. 1(b), were also prepared, by metal-assisted chemical etching [26]. Figure 2(a) shows that the reflectance of both the macroporous black silicon and silicon nanowires is below 5% in the range of 300 nm to 1000 nm. It is clear from Fig. 2(b) that the macroporous black silicon absorbs more energy than the silicon nanowires for wavelengths above 800 nm, indicating the excellent light trapping capability of the macroporous black silicon especially in the near infrared region. For wavelengths below 500 nm the nanowires may absorb a little bit more, but the corresponding photon number seen from the solar spectra is relatively small, and therefore this part would not hinder the fact that the macroporous black silicon has overall higher absorptivity than the silicon nanowires (as is confirmed by integrating the product of the measured absorptance and AM 1.5G photon flux).

 

Fig. 2 (a,b) Reflectance, transmittance, and absorptance (A = 1 - R - T) spectra for a planar silicon, silicon nanowires, and macroporous black silicon. (c) Lifetime of minority carriers of the planar silicon, silicon nanowires, and macroporous black silicon.

Download Full Size | PDF

Lifetime of minority carriers is another key factor for solar cells. To track the effect from surface structures, the lifetime of different silicon structures were measured by microwave photoconductivity decay (µ-PCD) technique (Semilab WT-1200). To be fair, all samples were highly doped on the backside. The number shown in Fig. 2(c) was an average of multiple measurements. The lifetime does not decline much after electrochemical etching (the lifetime is 45.72 µs for the planar silicon, while it is 43.31 µs for the macroporous black silicon), however, the sharp decline of the lifetime for silicon nanowires (24.5 µs) reveals that the surface of silicon nanowires has large amount of lattice defects.

The different silicon samples were assembled into photoelectrochemical (PEC) solar cells for further comparison. It is a simple way to measure the photovoltaic characteristics by using the semiconductor/liquid junction, which provides a conformal contact to the structured silicon and allows assessment of the trade-offs between increased surface area and decreased carrier collection distances in such systems [27]. The electrolyte used here was consisting of 8.6 M hydrogen bromide (HBr) and 0.05 M bromine (Br₂) [28]. We chose such an aqueous solution because of its easy handling. Since silicon is susceptible to oxidization in water under anodic bias, platinum nanoparticles (size 5-10 nm) were deposited on the surface of silicon, by electroless deposition in a bath consisting of 2 vol% HF solution and 1 mM H₂PtCl₆ for 25 minutes, as described in Ref [28]. Then the Pt-loaded samples were assembled into PEC cells with a Pt circle as the counter electrode in a Teflon cell with a quartz window. The photocurrent of the cells was measured by using a sourcemeter (Keithley 2420) under irradiation from a Class AAA solar simulator (Newport 94023A) which was used to simulate AM 1.5 illumination (100 mW/cm²), and the intensity of the light source was calibrated with a reference cell (Newport 91150V). The incident photon-to-current conversion efficiency (IPCE) was measured by using the IPCE Measurement Kit from Newport (QEPVSI-b).

Figure 3(a) shows that the macroporous black silicon has the highest short-circuit photocurrent density (J_sc) of 28.9 mA/cm² and conversion efficiency (η) of 8.2%, while the silicon nanowires and the planar silicon only get J_sc = 19.9 mA/cm² (η = 5.2%) and J_sc = 11.9 mA/cm² (η = 3.4%), respectively (comparable to those reported in the literature [28]). The corresponding IPCE spectra are shown in Fig. 3(b). In agreement with the optical absorption measurement, the macroporous black silicon has the highest IPCE for wavelengths above 600 nm (light with wavelength shorter than 550 nm is absorbed by the Br₂/HBr electrolyte as indicated by Fig. 3(c)). IPCE is mainly affected by three fundamental processes, including charge generation, transport within the material, and collection at the electrode/electrolyte interface. The charge generation efficiency is closely related to the amount of light absorbed by the semiconductor, and the charge transport efficiency is related to the recombination rate. The excellent light-trapping mechanisms especially for long wavelength light and low recombination loss render a high J_sc for the macroporous black silicon.

 

Fig. 3 (a) Photocurrent densities versus voltage (J - V curves) for PEC cells of the planar silicon, silicon nanowires, and macroporous black silicon. (b) IPCE spectra of the planar silicon, silicon nanowires, and macroporous black silicon. (c) The transmittance spectrum of Br₂/HBr electrolyte in a quartz cuvette of 1 mm thick. The shorter cut-off wavelength in (c) than in (b) is due to the smaller thickness of the electrolyte in (c).

Download Full Size | PDF

3. Macroporous black silicon film

After the macroporous black silicon structures were fabricated, the diameter of the nethermost pores was enlarged by increasing the applied etching current, until the adjacent pores got connected, and then the film can be detached from the silicon substrate, as illustrated in Fig. 4(a). Figure 4(b) shows the SEM image of a macroporous black silicon film with thickness of about 56 µm, and Fig. 4(c) shows the measured absorptance, with the data given by the Yablonovitch limit [18] (A_Yablonovitch = 1 – 1/(1 + 4n²αd_eq), α = 4πk/λ, d_eq the equivalent silicon thickness) for reference. The c-Si filling fraction is estimated to be 0.45 (from several different SEM images) when calculating d_eq, and the extinction coefficient k of silicon is from Ref [29]. It is clear from Fig. 4(c) that the absorptance of the detached film in the near infrared region is close to the Yablonovitch limit. Intuitively, some light can escape through the pores, especially for the visible region, but in practice this part can be reflected back into silicon by, e.g., the back electrode at the bottom. The integrated J_sc from the measured absorptance is 39.95 mA/cm², close to 41.17 mA/cm² calculated from the Yablonovitch limit. Note that no back reflectors were used in our measurement, in contrast to Ref [30]. with a similar equivalent thickness.

 

Fig. 4 (a) Schematic illustration of the detachment of a macroporous black silicon film from the silicon substrate. (b) Cross-sectional SEM image of a black macroporous silicon film. (c) Absorptance spectrum of the macroporous black silicon film and the Yablonovitch limit of an equivalent solid silicon film.

Download Full Size | PDF

To help understand the role of different structures in light absorption enhancement, we built a model as shown in Fig. 5(a). In order to release the burden to computer memory, the model has smaller interpore distance and smaller thickness than in Fig. 4(b), but it is chosen to mimic a real structure shown in Fig. 5(b) which was fabricated several years ago. Calculations were performed with the finite-difference time-domain method (using the code MEEP [31]). The unit cell is a × a × 20 μm³ (a is the lattice constant), treated with perfectly matched layers (PML) on the top and bottom, and periodic boundary conditions in the lateral directions. The excitation is a normal incident plane-wave pulse placed close to the top PML. The dielectric function of c-Si (only simulate the range of 1.2−4.0 eV) was fitted from the experimental data in [29]. The calculated absorptance is plotted in Fig. 5(c), together with that of a non-porous (solid) planar film and a film with straight nanoholes (with the c-Si filling fraction 0.5 and the lattice constant 500 nm as in [14]), with the same physical thickness (8.25 μm). We see that the macroporous black silicon film has the highest absorptance over the whole wavelength range, even though the macroporous film has the smallest c-Si filling fraction (0.46) among these three films. We also performed calculations for macroporous films with straight pores (without kinks) in the middle part (keeping the blade-like structures on the top and the bottom unchanged) and with different c-Si filling fraction, and found that the kinks inside the pore really helps to enhance light absorption. Another advantage of the macroporous silicon film over a nano-structured film is that the former has a much smaller surface area. The total surface area over the projected area is about 8.85 for the macroporous film, whereas it is 42.36 for the nanohole film. The increased carrier recombination that occurs at or near a textured surface is a direct consequence of the increased surface area, and thus a smaller total surface area is undoubtedly beneficial. We further calculated the case when we scaled the macroporous structure down to a lattice constant of 500 nm (keeping the profile along the pore axis unchanged), though it is difficult to produce such structures by electrochemical etching. The absorptance is shown in Fig. 5(c) as well. In the visible region, the absorptance of the scaled-down structure is close to 100%, mainly due to the much sharper tapers on the top surface. However, in the near infrared region, the absorptance of the scaled-down structure is lower than that of the original macroporous structure. This confirms that the interpore distance does influence the absorption of light with photon energy close to the band edge of silicon. Efficient light-trapping mechanisms especially for long wavelength light are necessary to achieve a high short-circuit current density for a thin-film solar cell. Dielectric structures with feature size comparable with the wavelength can scatter strongly the incident light and thus enhance the absorption.

 

Fig. 5 (a) Schematic drawing of a macroporous black silicon film, consisting of truncated cones and ellipsoids of air in a silicon film of thickness 8.25 μm (the dimensions are shown in the axial cross section on the right). The pore shape is chosen to mimic an experimentally realized structure shown in (b). (b) Cross-sectional SEM image of a black macroporous silicon sample. The pores were arranged in a square pattern with a lattice constant of a = 2 μm. (c) Calculated absorptance spectra for a non-porous planar film (gray dashed), a film with straight nanoholes (red square), and the macroporous black silicon film (blue dot), while keeping the same silicon film thickness of 8.25 μm. The c-Si filling fraction and the lattice constant for the nanohole film are 0.5 and 500 nm, respectively. For the macroporous film, the lattice constant is 2 μm and the c-Si filling faction is about 0.46. We also show the absorptance by a film scaled down from the macroporous film to a lattice constant of 500 nm (magenta solid).

Download Full Size | PDF

4. Conclusion

We have demonstrated a black silicon film with modulated macropores produced by electrochemical etching. Here the through pores were modulated also to function as antireflection and light-trapping elements. We have shown that the collaborative effects from the blade-like tapered silicon walls and the micrometer-scale modulation inside the pore render the macroporous black silicon superior in light harvesting, and the latter also helps for light absorption enhancement especially in the near infrared region. The macroporous silicon film has much smaller surface area than a film with pure nano-holes. This is important since the enhancement in optical absorption should overwhelm the increase in surface recombination loss. The present macroporous black silicon film is thus believed to be promising as a thin active layer for a high efficiency silicon solar cell.