A novel top cell material for multi-junction solar cells
Much research effort is being invested in developing more efficient photovoltaics to convert solar energy into electricity. The best performing photovoltaics thus far are multi-junction solar cells, which combine several different semiconductor material layers. Each layer is tailored to convert a narrower portion of the solar spectrum with greater efficiency. Usually, the widest ‘bandgap’ material (highest eV) is placed at the top of a multilayer structure (the 'top junction') where the incoming light first hits. However, these multiple layers must not only have the correct bandgap energy, but also be compatible with one another in terms of substrate, thermal expansion, and fabrication temperatures. To address these requirements, III–V semiconductors (crystalline materials using elements from the III and V columns of the periodic table) are most often chosen.
Presently employed III–V multi-junction solar cells are grown on substrates either of gallium arsenide (GaAs) or germanium, and have achieved over 40% energy conversion efficiency.1Germanium and GaAs have very similar crystal lattice constants (i.e., the dimension of unit cells in a crystal lattice) and so do the materials that can be successfully grown on top of these substrates. However, the semiconductor alloys within this class may not have the ideal bandgap energies.
Indium phosphide (InP) is another III–V compound semiconductor substrate, but with a larger lattice constant, which makes it possible to use other materials for solar energy conversion. Thus far, there has been no published work on InP-based multi-junction cells, even though there are materials perfectly lattice-matched to InP for the middle and bottom junctions with the desired bandgaps. The problem is primarily the lack of a suitably wide (greater than 1.6eV) bandgap top junction material.
Aluminum gallium arsenide antimonide (Al1−xGaxAsySb1−y) is a candidate for the top junction, but we have found no experimental studies on Al1−xGaxAsySb1−y solar cells. One reason for this may be that a miscibility gap exists for Al1−xGaxAsySb1−y when the composition values are set to match the lattice constant of InP and have a bandgap greater than 1.6eV.2, 3 The miscibility gap makes it exceedingly difficult to physically manifest the material of that specific composition in a homogenous way. We have used a digital alloy technique with molecular beam epitaxy (MBE) to overcome the miscibility gap issue and successfully grow Al1−xGaxAsySb1−y material for solar cells. In this method, alternating thin layers of AlGaSb and AlGaAs are grown and repeated, with the end result being a material that has the properties of a homogenously grown alloy. 4–6
For a triple junction solar cell under AM0 illumination (i.e., the solar spectrum above earth's atmosphere), the ideal top junction would have a bandgap energy of 1.9eV. We first modeled an ideal junction.7 For Al0.75Ga0.25As0.56Sb0.44, the indirect bandgap is ∼1.72eV and the direct bandgap is ∼1.95eV. After approximating the amount of current expected to be generated from the solar absorption by the top junction, we calculated the bandgaps for the remaining two junctions for optimal performance. These values are ∼1.30eV and ∼0.82eV for the middle and bottom junctions, respectively, forming the device shown in Figure 1. Based on our modeling, the theoretical efficiency for a device of this design under AM0 is 45.1%. As a reference, an ideal triple-junction with direct bandgaps of 1.95, 1.3, and 0.82eV under AM0 will have efficiency of 48.6% as listed.7
To demonstrate the feasibility of AlGaAsSb as the top junction material, we made (by MBE using the digital alloy technique) and characterized a single-junction test sample, which we used to make several cells. The previously mentioned Al0.75Ga0.25As0.56Sb0.44 alloy serves as the absorber layer. Figure 2 plots a typical measured current-voltage characteristic for a fabricated cell. The typical observed energy conversion efficiency was 7.46% for cells without an anti-reflection (AR) coating. The highest measured efficiency is 8.9% for one cell with a smaller device area. Higher efficiency is expected with AR coating on cell surface since more incident light will be transmitted into absorber layer. The electrical characterization results, with the soft and rounded corner of the current-voltage curve, show the dark current (current that flows when there is no incident light) is still high and the fill factor (i.e., the maximum power divided by the product of the open-circuit voltage and the short-circuit current) is low as seen in Figure 2, partially due to the unoptimized growth conditions for AlGaAsSb. Given these growth conditions, this initial experimental result of a first Al1−xGaxAsySb1−y solar cell device is very encouraging. We are now working to further improve the layer growth and design of the triple-junction solar cell in this material system.
Yiqiao Chen received a PhD in electrical engineering from Columbia University, NY, in 2005. He then joined SVT Associates, where he is currently working on MBE growth and physics of semiconductor materials and devices based on III–V and II–VI compounds.