SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
Using a fairly comprehensive model, researchers have done a parametric variation study of the InP n+p homojunction solar cell for AMO, 25 C operation. The results of this study are presented. These results indicate that an efficiency of about 25 percent should be realistically possible in a shallow homojunction InP solar cell with near-optimum design.
The main points of discussion, conclusions and recommendations of a workshop on InP materials and cell fabrication are given. The importance of assessing the quality of p-Inp crystals supplied by different vendors, back contacts to solar cells, junction formation, energy conversion efficiency, testing for radiation resistance, and future develpments were among the topics discussed.
Very shallow n(+) layers have been obtained in InP by using gallium sulfide as a source for sulfur diffusion, and chemically vapor-deposited SiO2 as a cap. Diffusions were carried out from 585 to 725 C in an open-tube system with a nitrogen ambient. The doping profile of sulfur in InP is estimated to be of the complementary error function type with a surface concentration of 5.6 x 10 to the 18th/cc and a diffusion constant of 1.1 x 10 to the -14th sq cm/s at 670 C. Diodes made on n(+)-p junctions obtained by this diffusion technique show ideality factors close to unity and saturation current densities as low as 3.4 x 10 to the -15th A/sq cm, signifying the presence of a defect-free junction. These diffusions, with junction depths in the 400-700 A range, are ideal for solar cell applications.
Photoluminescence (PL) measurements have been performed on InP samples in situ during various surface treatments including chemical etching, wet anodization, and low-pressure chemical vapor deposition. It was found, in agreement with previously published results, that the magnitude of the PL signal varies markedly with surface treatment due presumably to changes in either surface-state density, and/or surface potential. In an attempt to assess the effectiveness of this noninvasive method as a tool for characterizing and monitoring the progressive development of a semiconductor surface during processing, a number of experiments on InP have been performed. The results indicate that although some uncertainty may exist in assigning a mechanism for the PL change in any given experiment, the general trend appears to be that surface degradation results in a reduced signal. As a result, process steps which enhance the PL intensity are likely to be beneficial in the preparation of a high-quality interface.
A new technique for tracing the organometallic vapor epitaxial growth is reported. The pyrolysis of PH3, alone and in the presence of trimethylindium (TMIn), and of TMIn alone was studied by conducting the epitaxial growth of InP in D2 as the carrier gas, tracing growth reactions by mass spectrometric analysis of the product molecules. The TMIn alone pyrolyzes mostly homogeneously in the gas phase, while the PH3 pyrolysis is completely heterogeneous at the InP surface. Adding TMIn to PH3 results in a dramatic decrease in the pyrolysis temperature. PH3 molecules which interact with TMIn in the gas phase pyrolyze at temperatures as low as 250, and those decomposing without TMIn interaction pyrolyze at temperatures approximately 200 C higher. Similarly, the presence of PH3 lowers the TMIn pyrolysis temperature by at least 50 C. TMIn alone in D2 produces mainly CH3D molecules. For high PH3:TMIn ratios, CH4 is the only carbon-containing reaction product.
Although InP-based solar cells have considerable potential for space applications, it is necessary to improve efficiencies to around the level of GaAs or Si cells before their excellent radiation resistance can be regarded as a dominant advantage. The authors concentrate on indium-tin-oxide/InP cells, presenting data relating to reduction of the contact resistance of the rear surface metallization, reduction of reflectance losses by choosing indium-tin-oxide deposition conditions to give specific optical properties, and reduction of losses associated with the grid. Simultaneous optimization of all of these has led to improved values of Jsc. For devices of approximately 1 cm2 in area, the largest Jsc achieved to date is 28.1 mA/aq cm (AM1.5, SERI/NASA direct normal spectrum, 25 C, total area, 100 mW/sq cm). For this particular cell, the equivalent AM0 value of Jsc was 34.6 mA/sq cm, which appears to be the largest reported for any InP-based cell.
A survey was conducted on past studies of hole traps in InP. An experiment was designed to evaluate hole traps in Zn-doped InP after fabrication, after electron irradiation and after annealing using deep level transient spectroscopy. Data similar to that of Yamaguchi was seen with observation of both radiation-induced hole and electron traps at E sub A=0.45 eV and 0.03 eV, respectively. Both traps are altered by annealing. It is also shown that trap parameters for surface-barrier devices are influenced by many factors such as bias voltage, which probes traps at different depths below the surface. These devices require great care in data evaluation.
Experimental results are presented for InP surfaces treated by using red phosphorus as a source to create an excess overpressure of phosphorus during annealing and prior to silicon dioxide deposition. The surface has been probed by in situ photoluminescence, noncontacting remote gate C-V, and conventional high-frequency and quasi-static C-V methods. A study has also been made of the surface of sulfurized InP following heating in aqueous (NH4)2S(x). MISFETs fabricated using the benefits of these surface treatments show high transconductances and stabilities approaching those of thermal SiO2/Si with less than 5-percent variation in drain current over a 12-hr period.
InP films were grown by MOCVD on Si GaAs substrates (as well as on InP substrates, included as controls), and were used to fabricate solar cells, using the Spitzer et al. (1987) technique. Contact to the substrate was made with Al-Ti-Pd-Ag to the Si wafers and with Au-Zn alloy to the GaAs wafers, while contract to the front was made with Cr-Au-Ag. Air mass zero efficiencies were found to be 7.1 percent for Si-substrate cells and 9.4 percent for GaAs-substrate cells.
An attempt is made to improve device efficiencies by depositing indium tin oxide onto epitaxially grown p-InP on p(+)-InP substrates. This leads to a reduction in the device series resistance, high-quality reproducible surfaces, and an improvement in the transport properties of the base layer. Moreover, many of the facets associated with badly characterized bulk liquid encapsulated Czochralski substrates used in previous investigations are removed in this way.
Indium phosphide solar cells with very thin n-type emitters have been made by both ion implantation and metal-organic chemical vapor deposition. Air mass zero efficiencies as high as 18.8 percent (NASA measurement) have been achieved. The best cells, which were those made by ion implantation, show an open-circuit voltage of 873 mV, short-circuit current of 35.7 mA/sq cm, and fill factor of 0.829. Improvements are anticipated in all three of these parameters. Internal quantum efficiency peaks at over 90 percent in the red end of the spectrum, but drops to 54 percent in the blue end. Other cells have achieved 74 percent in the blue end. A preliminary investigation of InP solar cells on foreign substrates has been carried out. Although problems have been encountered with doping of the InP by the substrate, cells of 7.1 percent efficiency on silicon and cells of 9.4 percent, efficiency on GaAs have been made.
Hybrid indium-tin-oxide (ITO)/InP solar cells are discussed. The cells are constructed by dc magnetron sputter deposition of ITO onto high-quality InP films grown by atmospheric pressure metal-organic vapor-phase epitaxy (APMOVPE). A record efficiency of 18.9 percent, measured under standard Solar Energy Research Institute reporting conditions, has been obtained. The p-InP surface is shown to be type converted, principally by the ITO, but with the extent of conversion being modified by the nature of the sputtering gas. The deposition process, in itself, is not responsible for the type conversion. Dark currents have been suppressed by more than three orders of magnitude by the addition of hydrogen to the sputtering gas during deposition of a thin (5 nm) interface layer. Without this layer, and using only the more usual argon/oxygen mixture, the devices had poorer efficiencies and were unstable. A discussion of associated quantum efficiencies and capacitance/voltage measurements is also presented from which it is concluded that further improvements in efficiency will result from better control over the type-conversion process.
While dc magnetron sputter deposition of indium tin oxide leads to the formation of a buried homojunction in single crystal p-type InP, the mechanism of type conversion of the InP surface is not apparent. In view of the recent achievement of nearly 17-percent global efficiencies for cells fabricated solely by sputter deposition of In2O3, it is presently surmised that tin may not be an essential element in type conversion. A variety of electrical and optical techniques are presently used to evaluate the changes at both indium tin oxide/InP and indium oxide/InP interfaces. Such mechanisms as the passivation of acceptors by hydrogen, and sputter damage, are found to occur simultaneously.
Indium phosphide solar cells have a higher resistance to electron irradiation than Si or GaAs cells of comparable junction depth. As a result, there is much interest in the use of this material for space applications. Cells of this material were made in bulk InP by a number of techniques, including ion implantation, direct diffusion in sealed ampoules, and by open tube diffusion. However, it is generally considered that the epitaxial approach will be superior to all of these techniques. The epitaxy of InP is considerably more difficult than that of gallium arsenide, for a number of reasons. Perhaps the most important is the fact that the native oxides of Indium are extremely difficult to remove, as compared to that of Gallium. In addition, thermal treatments for the desorption of these oxides often result in the formation of phosphorus vacancies and free indium on the surface. Thus, inadequate sample preparation before epitaxy, poor reactor cleaning procedures, or poor transition procedures between the growth of successive layers, all give rise to trap phenomena and to high interface recombination velocities. Moreover, the lifetime of the grown material is dominated by the occurrence of native defects, so that it is a strong function of growth parameters. These problems are of special interest to the fabrication of solar cells, where long life-time, combined with the absence of traps, is highly desirable. A study of this problem is described using a non-invasive diagnostic technique which was developed.
The purpose was to demonstrate the possibility of fabricating thermally diffused p(+)n InP solar cells having high open-circuit voltage without sacrificing the short circuit current. The p(+)n junctions were formed by closed-ampoule diffusion of Cd through a 3 to 5 nm thick anodic or chemical phosphorus-rich oxide cap layer grown on n-InP:S Czochralski LEC grown substrates. For solar cells made by thermal diffusion the p(+)n configuration is expected to have a higher efficiency than the n(+)p configuration. It is predicted that the AM0, BOL efficiencies approaching 19 percent should be readily achieved providing that good ohmic front contacts could be realized on the p(+) emitters of thickness lower than 1 micron.
Specific contact resistivities for as-fabricated Au contacts on n-p InP solar cells are typically in the 10(exp -3) ohm/sq cm range, but contact resistivities in the 10(exp -6) ohm/sq cm range can be obtained if the cells are heat treated at 400 C for a few minutes. This heat treatment, however, results in a dramatic drop in the open circuit voltage of the cell due to excessive dissolution of the emitter into the metallization. It was found that low values of contact resistivity can be secured without the accompanying drop in the open circuit voltage by adding Ga and In in the Au metallization. It is shown that Au contacts containing as little as 1 percent atomic Ga can suppress the reaction that takes place at the metal-InP interface during heat treatment, while exhibiting contact resistivity values in the low 10(exp -5) ohm/sq cm. Detailed explanations for the observed superior thermal stability of these contacts are presented.
Formation of a textured or grooved front surface on a solar cell can increase the efficiency in several ways, including enhanced absorption and light trapping. In III-IV materials the (111) plane is chemically different form the (1'1'1') plane, and both etching and epitaxial deposition behave differently on these surfaces. The current state of profile etching in InP is summarized. Data are presented on novel geometries attainable as a function of etchant temperature and composition, substrate orientation and carrier concentration, and the oxide thickness between the substrate and the photoresist. Depending on dopant concentration, the same etchant can produce either anisotropic or isotropic grooves. V-grooved solar cells were manufactured on InP, and the improved optical absorption was demonstrated. Preferred parameters for various applications are listed and discussed.
The apparently unrelated phenomena of temperature dependency, carrier removal and photoluminescence are shown to be affected by the high dislocation densities present in heteroepitaxial InP solar cells. Using homoepitaxial InP cells as a baseline, it is found that the relatively high dislocation densities present in heteroepitaxial InP/GaAs cells lead to increased volumes of dVoc/dt and carrier removal rate and substantial decreases in photoluminescence spectral intensities. With respect to dVoc/dt, the observed effect is attributed to the tendency of dislocations to reduce Voc. Although the basic cause for the observed increase in carrier removal rate is unclear, it is speculated that the decreased photoluminescence intensity is attributable to defect levels introduced by dislocations in the heteroepitaxial cells.