GaSb solar cells grown on GaAs via interfacial misfit arrays for use in the III-Sb multi-junction cell

Abstract

Growth of GaSb with low threading dislocation density directly on GaAs may be possible with the strategic strain relaxation of interfacial misfit arrays. This creates an opportunity for a multijunction solar cell with access to a wide range of well-developed direct bandgap materials. Multijunction cells with a single layer of GaSb/GaAs interfacial misfit arrays could achieve higher efficiency than state-of-the-art inverted metamorphic multi-junction cells while forgoing the need for costly compositionally graded buffer layers. To develop this technology, GaSb single junction cells were grown via molecular beam epitaxy on both GaSb and GaAs substrates to compare homoepitaxial and heteroepitaxial GaSb device results. The GaSb-on-GaSb cell had an AM1.5g efficiency of 5.5% and a 44-sun AM1.5d efficiency of 8.9%. The GaSb-on-GaAs cell was 1.0% efficient under AM1.5g and 4.5% at 44 suns. The lower performance of the heteroepitaxial cell was due to low minority carrier Shockley-Read-Hall lifetimes and bulk shunting caused by defects related to the mismatched growth. A physics-based device simulator was used to create an inverted triple-junction GaInP/GaAs/GaSb model. The model predicted that, with current GaSb-on-GaAs material quality, the not-current-matched, proof-of-concept cell would provide 0.5% absolute efficiency gain over a tandem GaInP/GaAs cell at 1 sun and 2.5% gain at 44 suns, indicating that the effectiveness of the GaSb junction was a function of concentration.

GaSb-based solar cells for multi-junction integration on Si substrates

“…Nevertheless, these parameters are close to recently reported progress on GaSb-on-GaAs cells. In particular, the measured JSC and FF are in between the values reported in [31] and [33]. Both references used front/back contact configurations.…”

“…Table I shows a comparison of the results obtained in both measurement sites. The best control solar cell demonstrated a 1-sun efficiency (η) of 5.90%, indicating a reasonably high material quality [31]. The cell had a 280 mV open-circuit voltage (VOC), a 36 mA/cm² short-circuit current density (JSC) and a 58.9% fill factor (FF).…”

“…It seems that this partially compensated for the much larger lattice-mismatch (12% vs 8% in the case of GaSb-on-GaAs cells). The use of a thicker buffer (twice or more as thick as the ones used in [31] and [33]) also likely contributed to the lattice-mismatch compensation. However, the VOC is lower in both cases, suggesting a still higher TDD in the active zone.…”

“…The optical generation was calculated with a transfer-matrix based method, involving optical properties (n,k) taken from literature (GaSb: [34]), or from experimental ellipsometry measurements (SiNx:H). The bandgap and effective density values were taken from the literature [31], [35]. The mobility values were taken from measurements made during the doping calibration stage, fitted to a simple impurity-dependent model [36].…”

“…High bandgap materials such as AlGaAsSb or AlInAsSb alloys, lattice-matched to GaSb, could provide a well-suited window layer. Optimization of the emitter thickness could also improve high-energy collection, as we have used a thick emitter compared to standard values for photovoltaic applications ( [26], [31]). The difference between the EQE and IQE curves further shows that the ARC layer thickness could be improved.…”