Using nanowires to enhance light trapping in solar cells

The photovoltaic (PV) industry reduced the cost of silicon solar panels by 50% between 2006 and 2011, thanks to modifications that improved panel efficiency.^{1, 2} Moreover, the cost per watt of solar power is now below $0.75, and prices per kilowatt-hour are fast approaching parity with electricity available on the grid.³ It is possible to further increase panel efficiency using thin-film silicon (Si) that incorporates nanometer-sized features (nanostructures) to improve light trapping, thus achieving greater efficiency without requiring more of the active material. Random texturing or unevenly distributed pyramid structures within the silicon can enhance light gathering up to the theoretical limit described by Yablonovitch.^4,5 However, we can overcome this limit using concepts based on wave optics.^{6, 7}

Solar cells featuring nanowires with p-n junctions constructed in a radial direction have high light-trapping efficiency⁸ and enhanced charge collection. This is because radial charge separation requires a shorter carrier diffusion length than for planar junction arrangements.^{9, 10} Individual nanowires exhibit high absorption by coupling incoming light into their localized resonances.¹¹ By changing the distance between nanowires without changing any other parameter, we showed that this coupling does not require periodic arrangement.^{11, 12} For the best performance, we needed to carefully adjust the nanowire diameters and the silicon (the volume fraction). We confirmed that vertical nanowire arrays have high tolerances to changes in geometry without causing a drastic reduction in device performance (see Figure 1). This is important for many low-cost fabrication processes, which cannot guarantee exact geometrical dimensions for every nanowire in the array. We also showed that optimal performance does not depend heavily on whether the arrangements are square or hexagonal (see Figure 1), given the same volume fraction, and that the light absorption improves logarithmically with nanowire heights (see Figure 2).

Figure 1. Ultimate short-circuit current density (J_sc) as a function of nanowire diameter and the distance between nanowire centers. Results for nanowires organized in square periodic arrays (left). Results for hexagonal periodic arrays (right). The length of the wires was 5μm in both cases. mA: Milliampere.

Figure 2. Maximum values of J_sc as a function of nanowire length. Maxima were determined from all values for diameters between 100 and 700nm and all pitches between 200 and 700nm. The two curves in the plot are for periodic square and hexagonal vertical crystalline silicon nanowire arrays, respectively.

Furthermore, we improved the performance of already optimized arrays by combining two types of nanowires with different diameters in closely packed hexagonal periodic arrangements. This created a cross section of wires much larger than the geometric dimensions of a single nanowire, with the benefit that wires with smaller diameters are more efficient in trapping light at short wavelengths, while larger wires absorb better at longer wavelengths.¹³ We optimized dual-diameter nanowire arrays for fixed distances between centers of the neighboring nanowires. The results indicate that combining wires of two different diameters is superior to the standard single-diameter configuration (see Figure 3). The highest short-circuit current-density value obtained from calculations on a 5μm-long dual-diameter array is close to the value for an optimized 7.5μm single-diameter array, representing a 33% reduction in material used.

Figure 3. Dependence of ultimate J_sc values on two different diameters of vertical nanowires in a closely packed hexagonal grid periodic array. The distance between centers of the closest nanowires is 450nm.

We fabricated radial junction solar cells using hydrogenated amorphous silicon (aSi:H) as an active material (see Figure 4). First, we prepared p-doped crystalline silicon nanowire cores using a low-temperature plasma-enhanced vapor-liquid-solid process.¹⁴ Next, we deposited a 100nm-thick intrinsic aSi:H coating, followed by a thin coating of n-doped aSi:H, and sputtered a transparent indium tin oxide electrode on top of the nanowires.⁹ Figure 5 shows intermediate steps in scanning electron microscopy images.

Figure 4. Schematic structure of radial junction devices prepared using plasma-enhanced vapor-liquid-solid growth followed by deposition of all subsequent layers of the P-I-N stack. Cg: Corning glass. ZnO:Al: Aluminum-doped zinc oxide. p-SiNW: p-doped crystalline silicon nanowire. i-aSi:H: Intrinsic hydrogenated amorphous silicon. n-aSi:H: n-doped amorphous silicon. ITO: Indium tin oxide.

Figure 5. Scanning electron microscopy images of radial junction devices during different fabrication stages. VLS: Vapor-liquid-solid. c-SiNW: Crystalline silicon nanowire.

Strong light scattering, combined with coupling of light into resonances inside the nanowires, enhances the absorption. Samples prepared with different nanowire densities (from 110 to 370 million per cm²) achieved energy conversion efficiencies of up to 8.1%, with the highest-performing cell reaching an open-circuit voltage of 0.80V, a short-circuit current of 16.1mA/cm², and a fill factor of 0.628.¹⁵ Figure 6 shows an example of density-voltage characteristics for a solar cell with energy conversion efficiency of greater than 8%. Relatively high short-circuit current (J_sc) values measured on radial junction-based solar cells are mainly due to the efficient light scattering between nanowires (see Figure 7).

Figure 6. Current density as a function of the voltage for a solar cell with measured 8% energy conversion efficiency (η). Measured values of open-circuit voltage (V_oc), J_sc, and fill factor (FF) are 0.794V, 16.23mA/cm², and 0.623, respectively.

Figure 7. Spectral absorptance inside nanowire-based solar cells with densities of about 110, 260, and 370 million nanowires per cm², compared with thin amorphous silicon films with thicknesses of 100, 250, and 500nm.

Our future work will focus on further improving device performance using extensive numerical modeling and by optimizing the cell's transparent conductive coating (TCO). We plan to further increase J_sc density by switching from amorphous silicon to micro-crystalline for the active absorber layer. Using atomic layer deposition techniques, we aim to achieve improved uniformity of the TCO layer.