Search

Direct growth of GaP on Si enables the integration of III–V and Si optoelectronic devices for a wide variety of applications, and has therefore been the subject of much research for many decades. Most of this effort has been directed toward overcoming the two main technical challenges: 1) removing atmospheric oxygen and carbon contamination and 2) establishing the conditions needed for “APD-free” III–V epitaxy ( i.e ., without antiphase domains). We have developed an OMVPE process for APD-free GaP growth on Si which overcomes both challenges by using AsH 3 to clean and prepare the Si surface in situ at a relatively low temperature. This process is based upon a very brief “AsH 3 -cleaning” step which simultaneously removes atmospheric contamination (thereby eliminating the need for Si regrowth) and creates a single-domain As-terminated Si surface. Here we discuss the key process steps using results from a suite of analysis tools.

Antiphase-domain (APD) free growth of GaP on Si has been achieved on Si surfaces prepared in situ by etching with AsH3. The pre-nucleation AsH3 etching removes O and C contaminants at a relatively low temperature, and creates a single-domain arsenic-terminated Si surface. The As-As dimer rows are all parallel to the step edges, and subsequent GaP growth by MOCVD retains this dimerization orientation. Both LEED and TEM indicate that the resulting epilayer is APD-free, and could thereby serve as a template for III–V/Si multijunction solar cells.

We present resonant Raman scattering (RRS) studies on single-variant, spontaneously ordered GaInP{sub 2} alloys in backscattering geometries on the (001) growth surface and on a (110) cleaved surface. A systematic study of RRS in the (110) backscattering geometry enables us to resolve the controversy regarding the assignment of the ordering-induced phonon mode at 205 cm-1 in this scattering geometry, and shows that this phonon mode is a forbidden longitudinal mode in this geometry but appears in the Raman spectra due to Froehlich scattering. (c) 2000 The American Physical Society.

GaInP/GaAs tandem solar cells have achieved new record efficiencies, specifically 25.7% under air-mass 0 (AMO) illumination, 29.5% under AM 1.5 global (AM1.5G) illumination, and 30.2% at 140-180x concentration under AM 1.5 direct (AM1.5D) illumination. These values are the highest two-terminal efficiencies achieved by any solar cell under these illumination conditions. The monolithic, series-connected design of the tandem cells allows them to be substituted for silicon or gallium arsenide cells in photovoltaic panel systems with minimal design changes. The advantages of using GaInP/GaAs tandem solar cells in space and terrestrial applications are discussed primarily in terms of the reduction in balance-of-system costs that accrues when using a higher efficiency cell. The new efficiency values represent a significant improvement over previous efficiencies for this materials system, and we identify grid design, back interface passivation, and top interface passivation as the three key factors leading to this improvement. In producing the high-efficiency cells, we have addressed nondestructive diagnostics and materials growth reproducibility as well as peak cell performance. >

A unique aspect of the inverted metamorphic multijunction (IMM) solar cell is the bandgap tunability of each junction, creating extremely flexible device designs. The optimal structure has subcell photocurrents that are matched for a given spectrum. However, the subcell photocurrents depend on the cell operating temperature, and therefore, the bandgaps need to be optimized for a certain range of operating conditions. In addition, imperfect material quality results in a loss of voltage and current that depends on the cell bandgap and thickness. In this case, an iterative process of multijunction design and subcell characterization is necessary to determine the optimal design. We compare two different three-junction devices to demonstrate the effect of bandgap selection and lattice-mismatched material quality on device performance at different temperatures. The triple-junction (3J)-IMM design with two lattice-mismatched junctions of perfect material quality (2MMJ) is theoretically optimal at room temperature but experimentally performs similarly to a simpler design with one mismatched junction (1MMJ) at higher temperature because of material quality tradeoffs and the temperature dependence of the designs. Significant progress in the growth, processing, and measurement has led to a 1MMJ design with (42.6 ± 2.1)% peak efficiency at 327 suns and (40.9 ± 2.0)% efficiency at 1093 suns under the direct spectrum.

State-of-the-art multijunction cell designs have the potential for significant improvement before going to higher number of junctions. For example, the V oc can be substantially increased if the photon recycling taking place in the junctions is enhanced. This has already been demonstrated (by Alta Devices) for a GaAs single-junction cell. For this, the loss of re-emitted photons by absorption in the underlying layers or substrate must be minimized. Selective back surface reflectors are needed for this purpose. In this work, different architectures of semiconductor distributed Bragg reflectors (DBR) are assessed as the appropriate choice for application in monolithic multijunction solar cells. Since the photon re-emission in the photon recycling process is spatially isotropic, the effect of the incident angle on the reflectance spectrum is of central importance. In addition, the DBR structure must be designed taking into account its integration into the monolithic multijunction solar cells, concerning series resistance, growth economics, and other issues. We analyze the tradeoffs in DBR design complexity with all these requirements to determine if such a reflector is suitable to improve multijunction solar cells.

High quality, direct-bandgap solar cells emit significant luminescence at their band-edge when forced to operate in forward bias, thereby creating a possible source of photocurrent in lower bandgap junctions of a multijunction cell. We study the effects of luminescent coupling on the measurement of the subcell photocurrents for a series-connected III–V multijunction solar cell. We describe a technique that uses a set of LEDs and a Xenon-lamp white-light source to accurately determine the subcell photocurrents under a reference spectrum, taking the luminescent coupling current into account. The technique quantifies the luminescent coupling efficiencies and compensates for any spectral overlap between the LEDs and the other junctions. Since quantum efficiency curves are used in the adjustment of the simulator spectrum, we also show how to correct those curves to remove the effects of luminescent coupling.

Four different band gap combinations of triple-junction inverted metamorphic solar cells are characterized as a function of temperature and concentration up to 120°C and ∼1000 suns. We demonstrate that the standard 1.82/1.40/1.00 eV combination is an excellent choice for typical operating conditions of 1000 suns and 75°C. Improved metal grids and thermal management in such a cell has achieved 42.6% efficiency at 327 suns and 40.9% at 1093 suns at 25°C.

External quantum efficiency of semiconductor photonic devices is directly measured by wavelength-dependent laser-induced temperature change (scanning laser calorimetry) with very high accuracy. Maximum efficiency is attained at an optimum photo-excitation level that can be determined with an independent measurement of power-dependent photoluminescence. Differential power-dependent photoluminescence measurement is used to quickly screen the sample quality before processing.

We have modeled a Ga 0.5 In 0.5 P/GaAs/Ge triple junction solar cell, including two Esaki diodes, to analyze how performance changes with illumination intensity. As has been observed experimentally, fill factor (FF) is the primary aspect of performance that limits efficiency at high concentration levels. The FF decreases because of series resistance and barriers created by the GaAs/GaInP back surface field. By adjusting carrier concentration or the material associated with the back surface field (BSF) and the carrier concentration in the absorber, FF losses can be reduced and efficiency enhanced at high concentration levels.

The spontaneous long-range-order-induced splitting of the valence band of ${\mathrm{Ga}}_{0.5}$${\mathrm{In}}_{0.5}$P is studied by polarized near-gap-excited photoluminescence and photoluminescence excitation. The results allow a direct determination of the size of the valence-band splitting and the relative ordering of the resultant states. Values obtained for the splitting are compared with recent calculations.

We present here a new approach to tandem cell design that offers near-optimum subcell bandgaps, as well as other special advantages related to cell fabrication, operation, and cost reduction. Monolithic, ultra-thin GaInP/GaAs/GaInAs triple-bandgap tandem solar cells use this new approach, which involves inverted epitaxial growth, handle mounting, and parent substrate removal. The optimal ~1-eV bottom subcell in the tandem affords an ~300 mV increase in the tandem voltage output when compared to conventional Ge-based, triple-junction tandem cells, leading to a potential relative performance improvement of 10-12% over the current state of the art. Recent performance results and advanced design options are discussed.

Polarized photoluminescence, photoluminescence excitation, piezomodulated reflectivity, and resonance Raman techniques have been employed to probe the electronic structure of GaInP2 alloys exhibiting Cu–Pt ordering. The ordering induced band gap reduction and the crystal field splitting of the valence bands have been studied for samples exhibiting various degrees of ordering. Our studies provide evidence for a distribution of order parameters in spontaneously ordered GaInP2, in contrast to earlier works, which assume that uniformly ordered domains are embedded in a perfectly disordered matrix. Furthermore, the band gap reduction and the crystal field splitting obtained from our experiments are in good agreement with a theoretical model relating these quantities to the long range order parameter in the sample.