Damir Asoli

InP Nanowire Array Solar Cells Achieving 13.8% Efficiency by Exceeding the Ray Optics Limit

Improving Nanowire Photovoltaics In principle, solar cells based on arrays of nanowires made from compound inorganic semiconductors, such as indium phosphide (InP), should decrease materials and fabrication costs compared with planar junctions. In practice, device efficiencies tend to be low because of poor light absorption and increased rates of unproductive charge recombination in the surface region. Wallentin et al. (p. 1057, published online 17 January) now report that arrays of p-i-n InP nanowires (that switch from positive to negative doping), grown to millimeter lengths, can be optimized by varying the nanowire diameter and length of the n-doped segment. Efficiencies as high as 13.8% were achieved, which are comparable to the best planar InP photovoltaics. Nanowire solar cells were fabricated that exhibit high photocurrents and low surface recombination. Photovoltaics based on nanowire arrays could reduce cost and materials consumption compared with planar devices but have exhibited low efficiency of light absorption and carrier collection. We fabricated a variety of millimeter-sized arrays of p-type/intrinsic/n-type (p-i-n) doped InP nanowires and found that the nanowire diameter and the length of the top n-segment were critical for cell performance. Efficiencies up to 13.8% (comparable to the record planar InP cell) were achieved by using resonant light trapping in 180-nanometer-diameter nanowires that only covered 12% of the surface. The share of sunlight converted into photocurrent (71%) was six times the limit in a simple ray optics description. Furthermore, the highest open-circuit voltage of 0.906 volt exceeds that of its planar counterpart, despite about 30 times higher surface-to-volume ratio of the nanowire cell.

Absorption of light in InP nanowire arrays

An understanding of the absorption of light is essential for efficient photovoltaic and photodetection applications with III–V nanowire arrays. Here, we correlate experiments with modeling and verify experimentally the predicted absorption of light in InP nanowire arrays for varying nanowire diameter and length. We find that 2,000 nm long nanowires in a pitch of 400 nm can absorb 94% of the incident light with energy above the band gap and, as a consequence, light which in a simple ray-optics description would be travelling between the nanowires can be efficiently absorbed by the nanowires. Our measurements demonstrate that the absorption for long nanowires is limited by insertion reflection losses when light is coupled from the air top-region into the array. These reflection losses can be reduced by introducing a smaller diameter to the nanowire-part closest to the air top-region. For nanowire arrays with such a nanowire morphology modulation, we find that the absorptance increases monotonously with increasing diameter of the rest of the nanowire.

Study of photocurrent generation in InP nanowire-based p(+)-i-n(+) photodetectors

We report on electrical and optical properties of p+-i-n+ photodetectors/solar cells based on square millimeter arrays of InP nanowires (NWs) grown on InP substrates. The study includes a sample series where the p+-segment length was varied between 0 and 250 nm, as well as solar cells with 9.3% efficiency with similar design. The electrical data for all devices display clear rectifying behavior with an ideality factor between 1.8 and 2.5 at 300 K. From spectrally resolved photocurrent measurements, we conclude that the photocurrent generation process depends strongly on the p+-segment length. Without a p+-segment, photogenerated carriers funneled from the substrate into the NWs contribute strongly to the photocurrent. Adding a p+-segment decouples the substrate and shifts the depletion region, and collection of photogenerated carriers, to the NWs, in agreement with theoretical modeling. In optimized solar cells, clear spectral signatures of interband transitions in the zinc blende and wurtzite InP layers of the mixed-phase i-segments are observed. Complementary electroluminescence, transmission electron microscopy (TEM), as well as measurements of the dependence of the photocurrent on angle of incidence and polarization, support our interpretations.

Processing and Characterization of Nanowire Arrays for Photodetectors

We present a fabrication scheme of contacting arrays of vertically standing nanowires (NW) for LEDs (Duan et al. Nature 409:66–69, 2001), photodetectors (Wang et al. Science (NY) 293:1455–1457, 2001) or solar cell applications (Wallentin et al. Science (NY) 339:1057–1060, 2013). Samples were prepared by depositing Au films using nano-imprint lithography (Ma rtensson et al. Nano Lett 4:699–702, 2004) which are used as catalysts for NW growth in a low-pressure metal organic vapour phase epitaxy system where III-V precursors and dopant gases are flown at elevated temperatures which lead to the formation of NWs with different segments (Borgstrom et al. Nano Res 3:264–270, 2010). An insulating SiO2 layer is then deposited and etched from the top segments of the NWs followed by sputtering of a transparent top conducting oxide and opening up 1 × 1 mm2 device areas through a UV lithography step and etching of the top contact from non-device areas. A second UV lithography step was subsequently carried out to open up smaller windows on the ITO squares for bond pad definition, followed by metallization and lift-off; and the substrate is used as back contact. We also report on the electrical and optical properties of near-infrared \(p^{+} - i - n^{+}\) photodetectors/solar cells based on square millimeter ensembles of InP nanowires grown on InP substrates. The study includes a sample series where the p +-segment length was varied between 0 and 250 nm, as well as solar cell samples with 9.3 % efficiency with similar design. The NWs have a complex modulated crystal structure of alternating wurtzite and zincblende segments, a polytypism that depends on dopant type. The electrical data for all samples display excellent rectifying behavior with an ideality factor of about 2 at 300 K. From spectrally resolved photocurrent measurements, we conclude that the photocurrent generation process depends strongly on the p +-segment length. Without p +-segment in the NWs, photogenerated carriers funneled from the substrate into the NWs contribute significantly to the photocurrent. Adding a p +-segment shifts the depletion region up into the i-region of the NWs reducing the substrate contribution to photocurrent while strongly improving the collections of carriers generated in the NWs, in agreement with theoretical modeling (Fig. 48.1).