Back to Table of Contents

46713-B10
Exploring Defect Level Position and Occupation in Metamorphic Heterostructures for Multi-junction Photovoltaic Converters

Timothy H. Gfroerer, Davidson College

Technological improvement of the conversion efficiency in state-of-the-art high efficiency solar cells depends, in part, on the development and refinement of high-bandgap alloys that can be grown on GaAs substrates.  The bandgap in alloys like GaAsP and GaInP can be tuned to match design parameters that capitalize on the power available in the high-energy portion of the solar spectrum.  However, the atomic spacing in these crystals differs from that of the underlying GaAs, so a higher concentration of crystalline defects is expected in GaAsP and GaInP epilayers.  These defects impair solar cell performance by providing new pathways for photo-excited charge carriers to recombine rather than generate electricity.

At Davidson College, we are using three different experimental techniques to characterize defects in GaAsP and GaInP.  In the first, we use spatially-integrated measurements of photoluminescence spectra and radiative efficiency to investigate collective properties of defect-related transitions.  In the second, we use transient capacitance measurements to explore transport into and out of defect-related states.  And in the third, which is a new experiment for us in 2008, we use a digital CCD camera to obtain photoluminescence images showing the defect distribution and carrier depletion via recombination at defects.

I.  Photoluminescence spectra from 1.75eV GaAsP are shown in Figure 1.  The broad defect-related (D-R) peak below the band-to-band (B-B) emission indicates that radiative recombination is occurring through relatively shallow defect levels in this alloy.  At higher temperatures these transitions become nonradiative and, with increasing temperature, the defect states become thermally depleted.  Analysis of the radiative efficiency (the radiative divided by the total recombination rate) provides further evidence that a broad band of shallow states exists in this material.  Running a rigorous computational algorithm to minimize the error between experimental measurements and theoretical predictions of radiative efficiency vs. photo-excitation produces the results shown in Figure 2.  The inset graph shows the computer-generated density of states (DOS) distribution between the valence and conduction band edges (E_v and E_c).  We find that a high density of defect levels just below E_c is required to obtain the best fit.

II. Transient capacitance measurements on diodes during and after the application of a bias pulse can be used to monitor the capture and emission of carriers into and out of defect-related traps.  Representative transients on devices incorporating 1.75eV GaAsP are shown in Figure 3.  The capture and escape response deviates strongly from conventional exponential behavior.  Such non-exponential behavior is often attributed to the presence of a broad band of defect energies, with shallow levels emptying quickly at early times and slower deeper levels contributing at later times.  However, with non-exponential behavior occurring during both the introduction and exclusion of carriers from the depletion region, we assign the behavior to anomalous transport rather than the filling and thermally-activated emptying of defect states.  For example, if carriers are hopping between defect sites, quick motion would be expected between nearby sites, while slower motion would occur during more distant hops.  We fit the temperature-dependent response to the stretched exponential function shown in the graph, holding the stretching parameter d and amplitude A fixed.  Arrhenius plots of the resulting capture and emission rates (k) are linear and yield comparable capture and escape activation energies (see Figure 4), providing further evidence for the transport-limited interpretation noted above.  Indeed, we also observe a thermally-activated, stretched-exponential response in 2.0eV GaInP, suggesting that the simultaneous incorporation of Gallium and Phosphorous may be responsible for this effect.

III. This summer, we undertook a new experiment to image the photoluminescence generated by these materials using a high-sensitivity CCD camera.  We observed striking spatial inhomogeneity in the luminescence brightness, even in nominally lattice-matched GaAs.  We also noted that the size of the defect-related features depends strongly on the laser excitation intensity (see Figure 5).   At low density, carriers diffuse more readily to defective regions rather than recombining radiatively, producing larger effective “dead” areas.  We model the behavior by assigning a large defect recombination rate to the defect position and allowing diffusion to control the density of carriers in adjacent regions (see Figure 6).  The results are consistent with our observations, but a more sophisticated model will be required to achieve better quantitative agreement with experiment.  At this stage, we can conclude that (1) if the defect recombination rate is non-uniform throughout the sample, diffusion will contribute to a reduction in net efficiency and (2) even for high-quality semiconductors with few defects, diffusion can lead to significant defect recombination at low excitation intensity.  Preliminary images from GaAsP grown lattice-mismatched on GaAs suggest that a hypothetical network of defective pixels will be required to reconcile our diffusion-limited model with experiment.  Since the defect concentration in these structures is much higher, we expect photo-excited carriers to find defects more readily, raising the photo-excitation intensity threshold for good solar cell performance.

Back to top