Grating back reflector design

A sketch of the structure used for the design of the grating back reflector is shown in Fig. 1(a). It consists of a n–i–p stack of a-Si:H sandwiched between two layers of indium tin oxide (ITO), with a periodically patterned silver back reflector. The simulation considered a semi-infinite silver substrate; it was assumed that all the light incidents on the metal layer was reflected from or absorbed by the metal. The grating depth is denoted with h, the lattice period with Λ, and the ITO fraction with b/Λ. The grating back reflector design was determined based on rigorous coupled wave analysis (RCWA) simulations^[ Reference Bozzola, Liscidini and Andreani ^{4 ,} Reference Liscidini, Gerace, Andreani and Sipe ^{18 ,} Reference Andreani, Bozzola, Kowalczewski and Liscidini ^{19 ]} that made use of experimentally determined optical constants of the individual layers, which have been obtained using spectroscopic ellipsometry. A comparison of the simulated and experimentally measured reflectance of a flat n–i–p a-Si:H cell with front and back ITO contacts shows good agreement for energies <2.4 eV and confirms the validity of the theoretical approach used to describe the optical properties of the system (see Supplementary Material).

Figure 1.

(a) Cell structure for the RCWA simulations. (b) Simulated absorption enhancement in a-Si:H cell with 1D silver grating with height of 60 nm, 500 nm period, and fill fraction of 60%.

In the design of the periodic patterned substrate, we considered a one-dimensional (1D) grating to reduce computation time; we also assumed that the n–i–p layers and contact layers were flat. From previous studies,^[ Reference Bozzola, Liscidini and Andreani ^{4 ]} we know that optimized 1D and 2D gratings for light trapping are characterized by similar values of the parameters h, Λ, and b. Our calculations lead to a conservative estimate in the absorption, for they do not account for any light that is scattered due to roughness contributed by the underlying porous a-Si:H layer and the grating structure transferred to the n–i–p layers. Finally, it is well known that 2D gratings can lead to larger absorption enhancement than 1D gratings due to the opening of more scattering channels.^[ Reference Bozzola, Liscidini and Andreani ^{4 ]}

The absorption in the intrinsic layer of the cell with the grating back reflector was maximized by varying the parameters h, Λ, and b/Λ. The numerical strategy adopted in this work allows one to calculate the absorption solely in the intrinsic layer, which gives the relevant contribution to the photocurrent, while also taking into account losses in the p- and n-doped layers, which are described by their complex dielectric functions derived by ellipsometric measurements.^[ Reference Zanotto, Liscidini and Andreani ^{20 ]} The RCWA calculations suggested that a 1D grating with a period of 500 nm, height of 60 nm, and fill fraction of 60% should provide a 16% increase in the short-circuit current with respect to a cell with a flat metal back reflector. The estimated absorption enhancement is shown in Fig. 1(b). The absorption peaks between 1.8 and 2 eV are due to the coupling of light in the guided modes supported by the high-index silicon layers in the cell stack.

Porous a-Si:H back reflector fabrication

The a-Si:H solar cell was fabricated on a glass substrate (Corning EAGLE XG), which was first rinsed with isopropyl alcohol, ethanol, and methanol to remove any contaminants adhering to the surface. A 100 nm SiO₂ layer was deposited on the glass substrate using plasma-enhanced chemical vapor deposition (PECVD) to promote the adhesion of the a-Si:H layer used to form the back reflector. Figure 2 shows a schematic illustrating the two approaches for back reflector fabrication. It was found that samples prepared without the thin SiO₂ layer delaminated during the a-Si:H porosification process. The a-Si:H layer used for the back reflector formation was a lightly p-doped 280 nm thick a-Si:H layer deposited via PECVD with 10% silane (SiH₄) diluted with He, 2% trimethylboron [TMB, B(CH₃)₃] diluted with hydrogen (H₂), and H₂. The band gap of the a-Si:H film was estimated to be 2 eV based on using the Tauc method to fit spectroscopic ellipsometry data.

Figure 2.

Schematic illustrating the fabrication of the two back reflector designs implemented in porous a-Si:H.

To porosify the a-Si:H layer to achieve a roughened surface that promotes light trapping, MAC etching was carried out. Extensive work has been done to characterize MAC etching of silicon and other semiconductor materials, including several studies on MAC-etched silicon nanowires for enhanced light trapping and carrier extraction in solar cells.^[ Reference Huang, Geyer, Werner, de Boor and Gösele ^{21 –} Reference Li ^{23 ]} However, very few studies have been conducted on MAC etching of amorphous silicon.^[ Reference Omampuliyur, Bhuiyan, Han, Jing, Li, Fitzgerald, Thompson and Choi ^{24 –} Reference Douani, Piret, Hadjersi, Chazalviel and Solomon ^{26 ]} In one of these studies, Douani et al. carefully investigated the formation of hydrogenated amorphous silicon–carbon alloy nanowires using a hydroflouric acid (HF)/AgNO₃ solution and reported the dependence of the nanowire morphology on substrate doping, etchant concentration, and etching time.^[ Reference Douani, Piret, Hadjersi, Chazalviel and Solomon ^{26 ]} In our work, MAC etching of a-Si:H was carried out with a discontinuous 5 nm layer of thermally evaporated gold acting as the catalyst. Three parameters of the MAC etch were explored in detail, as explained in Supplementary Material: the molar concentrations of [HF] and hydrogen peroxide [H₂O₂] in the etchant, the time delay between carrying out the a-Si:H deposition and the MAC etch, and the MAC etch duration. As the ratio of [HF]:[H₂O₂] was decreased, the morphology of the MAC etched a-Si:H layer changed from a dense, spike-like structure to grass-like to sparse short stumps. The grass-like morphology was identified to be most favorable as a textured surface to encourage light trapping and, consequently, a MAC etch composition of 49% HF, 30% H₂O₂, and deionized (DI) water in the ratio of 1:1:4 was selected. We found that if the MAC etch was performed immediately after the a-Si:H and gold depositions, it did not result in a very porous surface. On the contrary, if the etch was performed after 24 to 48 h, the desired textured, porous substrate was obtained. We hypothesize that this difference in behavior is due to the hygroscopic nature of a-Si:H, which absorbs moisture upon exposure to air, thus impacting the nature of the MAC etch. Finally, we investigated the MAC etch duration. An etch time of 45 s resulted in a 150 nm porous layer, as shown in Fig. 3. Longer etch times did not increase the thickness of the porous layer significantly, and instead the etch began to progress along weak pathways in the a-Si:H and undercut at the a-Si:H and SiO₂ interface. To complete the formation of the textured back reflector, the porous a-Si:H layer was coated with a 5 nm thick chromium adhesion layer followed by a 100 nm thick layer of silver.

Figure 3.

Cross-sectional SEM image of porous a-Si:H obtained from a 45 s MAC etch.

To form a cost-effective patterned back reflector, we further processed a MAC-etched porous a-Si:H layer by imprinting it with a 2D grating pattern—with the dimensions suggested by the RCWA simulations (Λ = 500 nm, b/Λ = 60%, and h = 50–60 nm)—using DIPS. The 2D grating was symmetric and employed the same grating pitch and fill factor (FF) in both dimensions. The 2D grating was selected for fabrication due to its predicted improved light scattering ability compared with a 1D grating. DIPS was performed by pressing a grating patterned silicon stamp into a porous a-Si:H layer using a hydraulic press.^[ Reference Ryckman, Liscidini, Sipe and Weiss ^{17 ]} A pressure of 170–180 psi was applied to a 4.5 × 4.5 mm² stamp. Prior to performing the imprintation, the surface of the silicon stamp was oxidized at 800 °C for 30 min and silanized overnight in an evacuated desiccator containing an open glass vial with 70 µl of Triphenylcyclopropenylium Hexafluoroantimonate (TCPF) in order to minimize adhesion of the porous a-Si:H to the stamp.^[ Reference Mares, Fain, Beavers, Duvall and Weiss ^{27 ]} Representative scanning electron microscopy (SEM) and atomic force microscopy (AFM; Digital instruments Nanoscope III) images of a 2D DIPS grating in porous a-Si:H are shown in Fig. 4. The AFM scan shows an RMS roughness of 13 nm. After the imprinting, the porous a-Si:H grating was coated with a 5 nm thick chromium adhesion layer followed by a 100 nm thick layer of silver, completing the metallic grating back reflector.

Figure 4.

Images of a 2D grating imprinted on porous a-Si:H using DIPS are shown: (a) top view SEM image, (b) AFM image, and (c) AFM line scan.

Fabrication of n–i–p layers of a-Si:H solar cell and ITO contacts

The remainder of the a-Si:H cells was fabricated on top of the back reflectors beginning with the back contact. A 60 nm thick ITO layer was sputtered using an Angstrom Åmod RF magnetron sputter tool. The layer was deposited at 200 °C using an argon plasma with 1% O₂. This method of deposition resulted in a layer with a resistivity of 6.7 × 10⁻⁴ Ω-cm, as measured by a hand held four-point probe meter (SRM-232-1000). The sample was then transferred to the PECVD chamber for n–i–p a-Si:H depositions; during these depositions, a small area in the corner of the sample was covered with a piece of silicon that acted as a mask and allowed access to the bottom ITO contact for measurements on the completed solar cell. The n, i, and p-layers were deposited sequentially at 200 °C and 600 mT for all layers. The n-layer was doped with phosphorous using 1% phosphine (PH₃) diluted in He as the dopant gas, while the p-layer was formed using TMB during the deposition. Detailed recipes for the n–i–p layers are included in Supplementary Material. A clean was performed in between the n- and i-layer depositions to prevent PH₃ from contaminating the intrinsic region of the device. The thicknesses of the p, i, and n-layers were 12, 180, and 25 nm, respectively. The cells were completed by sputter depositing a 60 nm thick ITO layer as the top contact. The same recipe as detailed previously for the bottom ITO contact was followed at a lower temperature of 175 °C in order to reduce the impact of high-temperature processing on the underlying a-Si:H device layers, resulting in a resistivity of 7.5 × 10⁻⁴ Ω-cm. A stainless steel contact mask was used during the deposition to pattern the ITO with 0.07 cm² circular cells. Figure 5 shows SEM images of completed a-Si:H solar cells with the textured, MAC-etched porous a-Si:H back reflector and the imprinted porous a-Si:H grating back reflector. The random roughness introduced by the MAC etch is clearly shown and is distinct from the more regular grating pattern.

Figure 5.

Cross-sectional SEM images of a-Si:H solar cells with (a) textured porous a-Si:H back reflector and (b) 500 nm pitch, 60% fill fraction porous a-Si:H grating back reflector For imaging purposes, these cells were fabricated on a silicon substrate instead of glass, but were processed at the same time as the cells on glass substrates that were electrically characterized.

Electrical characterization of solar cell performance

A LabView controlled Keithley 2400 source meter was used to perform the light and dark I–V measurements to characterize the solar cell performance. The light source was a 100 W Xenon ozone-free lamp with an AM1.5G filter. Prior to the measurement, the intensity of the light source was calibrated to maintain one sun conditions, which is equivalent to an irradiance of 100 mW/cm², by using a standard c-Si reference cell with an integrated KG5 window. Performance of the a-Si:H solar cells with the textured porous a-Si:H back reflector and imprinted a-Si:H grating back reflector was compared with a reference cell fabricated on a flat silver back reflector without the porous a-Si:H layer. As shown in Fig. 6 and Table I, an approximately 30% enhancement in both short-circuit current (J _sc) and overall efficiency was observed in the a-Si:H cells with the porous a-Si:H back reflectors. The textured porous a-Si:H back reflector and the imprinted grating back reflector both led to a similar performance enhancement. The reported enhancements are average values measured across the four best cells on each sample. As discussed earlier, the measured enhancement is greater than that predicted by 1D simulations that did not account for absorption arising from the rough texture of the porous a-Si:H and the transferring of the grating pattern to all the layers of the cell. It should be noted that, as expected, both the texturing and grating approaches lead to improved optical performance as quantified by the increase in J _sc, while the electrical performance quantified by the open-circuit voltage (V _oc) and FF remain largely unchanged.

Figure 6.

I–V characteristics obtained from a-Si:H n–i–p solar cells with (a) metal-coated porous a-Si:H back reflector with (black) and without (red) imprinted grating and (b) flat metal back reflector. A 30% increase in J _sc is observed due to the addition of the porous a-Si:H layer with either the randomly textured or periodic grating back reflector.

Table I.

I–V measurement results.

It is not entirely surprising that the cells with the textured porous a-Si:H back reflector and grating back reflector showed little difference in their I–V characteristics, for it is known that the use of rough substrates can lead to absorption enhancements that are comparable with those determined by perfectly periodic structures.^[ Reference Kowalczewski, Liscidini and Andreani ^{28 ]} Prior studies have extensively investigated the effect of grating design on the light-trapping properties of amorphous silicon thin-film solar cells as well as the tradeoff between optimizing the optical and electrical properties of amorphous silicon solar cells through the use of random versus periodic nanostructures.^[ Reference Bozzola, Liscidini and Andreani ^{4 ,} Reference Battaglia, Hsu, Söederströem, Escarré, Haug, Charrière, Boccard, Despeisse, Alexander, Cantoni, Cui and Ballif ^{29 ]} Our work is significant in its experimental demonstration of comparable absorption enhancements from random texturing and periodic gratings on the same amorphous silicon platform. Further enhancement of amorphous silicon solar cell performance may be achieved by following two different routes. In a first approach, the MAC etch conditions could be further refined to achieve a morphology with a more isotropic absorption enhancement approaching the Lambertian limit to light trapping.^[ Reference Green ^{30 ]} A second strategy may be the optimization of the 2D grating parameters including micro-roughness on a sub-wavelength scale or introducing a controlled degree of disorder. In both routes, electrical transport issues will need to be considered being mindful of the aforementioned trade-off between optical and electrical properties and the impact on overall cell efficiency.