Low-Cost High-Efficiency Solar Cells with Wafer Bonding and Plasmonic Technologies

Citation

Tanabe, Katsuaki (2008) Low-Cost High-Efficiency Solar Cells with Wafer Bonding and Plasmonic Technologies. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/4W2B-RD63. https://resolver.caltech.edu/CaltechETD:etd-05272008-123439

Abstract

III-V compound multijunction solar cells enable ultrahigh efficiency performance in designs where subcells with high material quality and high internal quantum efficiency can be employed. However the optimal multijunction cell bandgap sequence cannot be achieved using lattice-matched compound semiconductor materials. Most current compound semiconductor solar cell design approaches are focused on either lattice-matched designs or metamorphic growth (i.e., growth with dislocations to accommodate subcell lattice mismatch), which inevitably results in less design flexibility or lower material quality than is desirable. An alternative approach is to employ direct bonded interconnects between subcells of a multijunction cell, which enables dislocation-free active regions by confining the defect network needed for lattice mismatch accommodation to tunnel junction interfaces.

We fabricated for the first time a direct-bond interconnected multijunction solar cell, a two-terminal monolithic GaAs/InGaAs dual-junction cell, to demonstrate a proof-of-principle for the viability of direct wafer bonding for solar cell applications. The bonded interface is a metal-free n⁺GaAs/n⁺InP tunnel junction with highly conductive Ohmic contact suitable for solar cell applications overcoming the 4% lattice mismatch. The quantum efficiency spectrum for the bonded cell was quite similar to that for each of unbonded GaAs and InGaAs subcells. The bonded dual-junction cell open-circuit voltage was equal to the sum of the unbonded subcell open-circuit voltages, which indicates that the bonding process does not degrade the cell material quality since any generated crystal defects that act as recombination centers would reduce the open-circuit voltage. Also, the bonded interface has no significant carrier recombination rate to reduce the open circuit voltage.

Such a wafer bonding approach can also be applied to other photovoltaic heterojunctions where lattice mismatch accommodation is also a challenge, such as the InGaP/GaAs/InGaAsP/InGaAs four-junction tandem cell by bonding a GaAs-based lattice-matched InGaP/GaAs subcell to an InP-based lattice-matched InGaAsP/InGaAs subcell. Simple considerations suggest that for such a cell the currently-reported interfacial resistance of 0.12 Ohm-cm² would result in a negligible decrease in overall cell efficiency of ~0.02%, under 1-sun illumination.

Engineered substrates consisting of thin films of InP on Si handle substrates (InP/Si substrates or epitaxial templates) have the potential to significantly reduce the cost and weight of compound semiconductor solar cells relative to those fabricated on bulk InP substrates. InGaAs solar cells on InP have superior performance to Ge cells at photon energies greater than 0.7 eV and the current record efficiency cell for 1 sun illumination was achieved using an InGaP/GaAs/InGaAs triple junction cell design with an InGaAs bottom cell. Thermophotovoltaic (TPV) cells from the InGaAsP-family of III-V materials grown epitaxially on InP substrates would also benefit from such an InP/Si substrate. Additionally, a proposed four-junction solar cell fabricated by joining subcells of InGaAs and InGaAsP grown on InP with subcells of GaAs and AlInGaP grown on GaAs through a wafer-bonded interconnect would enable the independent selection of the subcell band gaps from well developed materials grown on lattice matched substrates. Substitution of InP/Si substrates for bulk InP in the fabrication of such a four-junction solar cell could significantly reduce the substrate cost since the current prices for commercial InP substrates are much higher than those for Si substrates by two orders of magnitude. Direct heteroepitaxial growth of InP thin films on Si substrates has not produced the low dislocation-density high quality layers required for active InGaAs/InP in optoelectronic devices due to the ~8% lattice mismatch between InP and Si.

We successfully fabricated InP/Si substrates by He implantation of InP prior to bonding to a thermally oxidized Si substrate and annealing to exfoliate an InP thin film. The thickness of the exfoliated InP films was only 900 nm, which means hundreds of the InP/Si substrates could be prepared from a single InP wafer in principle. The photovoltaic current-voltage characteristics of the In0.53Ga0.47As cells fabricated on the wafer-bonded InP/Si substrates were comparable to those synthesized on commercially available epi-ready InP substrates, and had a ~20% higher short-circuit current which we attribute to the high reflectivity of the InP/SiO²/Si bonding interface. This work provides an initial demonstration of wafer-bonded InP/Si substrates as an alternative to bulk InP substrates for solar cell applications.

Metallic nanostructures can manipulate light paths by surface plasmons and can dramatically increase the optical path length in thin active photovoltaic layers to enhance photon absorption. This effect has potential for cost and weight reduction with thinned layers and also for efficiency enhancement associated with increased carrier excitation level in the absorber layer.

We have observed photocurrent enhancements up to 260% at 900 nm for a GaAs cell with a dense array of Ag nanoparticles with 150 nm diameter and 20 nm height deposited through porous alumina membranes by thermal evaporation on top of the cell, relative to reference GaAs cells with no metal nanoparticle array. This dramatic photocurrent enhancement is attributed to the effect of metal nanoparticles to scatter the incident light into photovoltaic layers with a wide range of angles to increase the optical path length in the absorber layer.

GaAs solar cells with metallic structures at the bottom of the photovoltaic active layers, not only at the top, using semiconductor-metal direct bonding have been fabricated. These metallic back structures could incouple the incident light into surface plasmon mode propagating at the semiconductor/metal interface to increase the optical path, as well as simply act as back reflector, and we have observed significantly increased short-circuit current relative to reference cells without these metal components.

Thesis Files