Engineering topics
Publications and source records attributed to Sah, C. T..
Conceptual transistor embedded in photovoltaic diode promises to increase efficiency to more than 20 percent. Solar-cell transistor has front-surface contact, rear contact, and floating emitter. Variety of other contact and junction configurations possible, but do not offer ease of fabrication in combination with high performance.
The factors which may limit current crystalline silicon solar cells to less than 20 percent efficiency at AM 1 are investigated together with the factors which may limit the ultimate efficiency achievable. It was found that base recombination at residual defect and impurity recombination centers was the likely cause of the 20-percent efficiency barrier. Suggestions for design changes that would cut the losses due to recombinations are presented.
Minimum recombination and low injection level are essential for high efficiency. Twenty percent AM1 efficiency requires a dark recombination current density of 2 x 10 to the minus 13th power A/sq cm and a recombination center density of less than 10 to the 10th power /cu cm. Recombination mechanisms at thirteen locations in a conventional single crystalline silicon cell design are reviewed. Three additional recombination locations are described at grain boundaries in polycrystalline cells. Material perfection and fabrication process optimization requirements for high efficiency are outlined. Innovative device designs to reduce recombination in the bulk and interfaces of single crystalline cells and in the grain boundary of polycrystalline cells are reviewed.
The activation behavior of the minority-carrier mobility and diffusivity in heavily doped (about 10 to the 20th per cu cm) Si(Si:As) was investigated in the temperature range, 20 - 350 K. Experimental results indicate that hole transitions between the valence band and localized shallow states give rise to the observed behavior. The activation energy is about 10 meV, which suggests that the localized states originate from band tails but does not rule out trapping at boron atoms in the compensated n(+) region.
Loss mechanisms in high-efficiency solar cells were discussed. Fundamental limitations and practical solutions were stressed. Present cell efficiency is limited by many recombination sites: emitter, base, contacts, and oxide/silicon interface. Use of polysilicon passivation was suggested. After reduction of these losses, a 25% efficient cell could be built. A floating emitter cell design was shown that had the potential of low recombination losses.
The dominant recombination phenomena which limit the highest efficiency attainable in silicon solar cells under terrestrial sunlight are reviewed. The ultimate achievable efficiency is limited by the two intrinsic recombination mechanisms, the interband Auger recombination and interband Radiative recombination, both of which occur in the entire cell body but principally in the base layer. It is suggested that an optimum (26%) cell design is one with lowly doped 50 to 100 micron thick base, a perfect BSF, and zero extrinsic recombination such as the thermal mechanism at recombination centers the Shockley-Read-Hall process (SRH) in the bulk, on the surface and at the interfaces. The importance of recombination at the interfaces of a high-efficiency cell is demonstrated by the ohmic contact on the back surface whose interface recombination velocity is infinite. The importance of surface and interface recombination is demonstrated by representing the auger and radiative recombination losses by effective recombination velocities. It is demonstrated that the three highest efficiency cells may all be limited by the SRH recombination losses at recombination centers in the base layer.
A study was conducted to identify loss mechanisms in high efficiency silicon solar cells. The following were considered: (1) recombination loss mechanisms; (2) high efficiency cells; (3) very high efficiency cells; and (4) ultra high efficiency cells.
The performance improvements obtainable from extending the traditionally thin back-surface-field (BSF) layer deep into the base of silicon solar cells under terrestrial solar illumination (AM1) are analyzed. This extended BSF cell is also known as the back-drift-field cell. About 100 silicon cells were analyzed, each with a different emitter or base dopant impurity distribution whose selection was based on physically anticipated improvements. The four principal performance parameters (the open-circuit voltage, the short-circuit current, the fill factor, and the maximum efficiency) are computed using a FORTRAN program, called Circuit Technique for Semiconductor-device Analysis, CTSA, which numerically solves the six Shockley Equations under AM1 solar illumination at 88.92 mW/cm, at an optimum cell thickness of 50 um. The results show that very significant performance improvements can be realized by extending the BSF layer thickness from 2 um (18% efficiency) to 40 um (20% efficiency).
Material imperfections, impurity clusters and fabrication defects across the back-surface-field junction can degrade the performance of high-efficiency solar cells. The degradation from defects appearing on the circumference of a solar cell is analyzed using a two-region developed perimeter device model. The width of the defective perimeter region is characterized by the range or the distance-of-influence of the defective edge and this width is about two diffusion lengths. The defective edge is characterized by a surface recombination velocity. Family of theoretical curves and numerical examples are presented to show that significant reduction of open-circuit voltage can occur in high-efficiency cells which are thin compared with the diffusion length. In one example, the degradation is decreased from 135 mV to 75 mV when the cell size is increased from 10 to 100 times the diffusion length in a thin cell whose thickness is 1% of the diffusion length.
The significance of including factors such as the base resistivity loss for solar cells thicker than 100 microns and emitter and BSF layer recombination for thin cells in predicting the fill factor and efficiency of solar cells is demonstrated analytically. A model for a solar cell is devised with the inclusion of the dopant impurity concentration profile, variation of the electron and hole mobility with dopant concentration, the concentration and thermal capture and emission rates of the recombination center, device temperature, the AM1 spectra and the Si absorption coefficient. Device equations were solved by means of the transmission line technique. The analytical results were compared with those of low-level theory for cell performance. Significant differences in predictions of the fill factor resulted, and inaccuracies in the low-level approximations are discussed.
The degradation of solar cell performance due to bulk defects distributed across the back-surface field junction is analyzed in terms of a three-region developed-perimeter model. Families of curves are computed and their physical significance is discussed in detail with reference to three parameters used to characterize the defects: defect area, defect density, and defect surface recombination velocity. A reduction in the open-circuit voltage due to the presence of a defect is expressed as a function of the defect area, density, cell thickness, and defect surface recombination velocity. Numerical examples are presented to illustrate the importance of the particular defect parameters.
Design principles suggested here aim toward high conversion efficiency (greater than 15 percent) in polysilicon cells. The principles seek to decrease the liabilities of both intragranular and grain-boundary-surface defects. The advantages of a phosphorus atom concentration gradient in a thin (less than 50 microns) base of a p(+)/n(x)/n(+) drift-field solar cell, which produces favorable gradients in chemical potential, minority-carrier mobility and diffusivity, and recombination lifetime (via phosphorus gettering) are suggested. The degrading effects of grain boundaries are reduced by these three gradients and by substituting atoms (P, H, F or Li) for vacancies on the grain-boundary surface. From recent experiments comes support for the benefits of P diffusion down grain boundaries and, for quasi-grain-boundary-free and related structures. New analytic solutions for the n(x)-base include the effect of a power-law dependence between P concentration and lifetime. These provide an upper-bound estimate on the open circuit voltage. Finite-difference numerical solutions of the six Shockley equations furnish complete information about all solar-cell parameters and add insight concerning design.
The effect of defects across the back-surface-field junction on the performance of high efficiency and thin solar cells, using a developed-perimeter device model for the three-dimensional defects is investigated. Significant degradation of open-circuit voltage can occur even if there are only a few defects distributed in the bulk of the solar cell. Two features in the thickness dependences of the fill factor and efficiency in impurity-doped back-surface-field solar cells are discovered in the exact numerical solution which are associated with the high injection level effect in the base and not predicted by the low-level analytical theory. What are believed to be the most accurate recombination parameters at the Ti center to date are also given and a theory is developed which is capable of distinguishing an acceptor-like deep level from a donor-like deep level using the measured values of the thermal emission and capture cross sections.
The effect of silicon film thickness on the energy conversion efficiency of a back surface field solar cell is investigated. A computer-aided design study on the dependence of efficiency peaks on the concentrations of the recombination and dopant impurities is presented. The illuminated current voltage characteristics of over 100 cell designs were obtained using the transmission line circuit model to numerically solve the Shockley Equations. Using an AM1 efficiency of 17% as a target value, it is shown that the efficiency versus thickness dependence has a broad maximum which varies less than 1% over more than three-to-one range of cell thickness from 30 to 100 microns. Optical reflecting back surface will give only a slight improvement of AM1 efficiency, about 0.7%, in this thickness range. The sensitive dependence of efficiency on patchiness across the back surface field, low high junction in thin cells is noted.
Zinc is a major residue impurity in the preparation of solar grade silicon material by the zinc vapor reduction of silicon tetrachloride. It was found that in order to get a 17 percent AMl cell efficiency, the concentration of the zinc recombination centers in the base region of silicon solar cells must be less than 4 x 10 to the 11th power Zn/cu cm in the p-base n+/p/p+ cell and 7 x 10 to the 11th power Zn/cu cm in the n-base p+/n/n+ cell for a base dopant impurity concentration of 5 x 10 to the 14th power atoms/cu cm. If the base dopant impurity concentration is increased by a factor of 10 to 5 x 10 to the 15th power atoms/cu cm, then the maximum allowable zinc concentration is increased by a factor of about two for a 17 percent AMl efficiency. The thermal equilibrium electron and hole recombination and generation rates at the double acceptor zinc cancers were obtained from previous high field measurements as well as new measurements at zero field. The rates were used in the exact d.c. circuit model to compute the projections.
Numerical solutions were obtained from the exact one dimensional transmission line circuit model to study the following effects on the terrestrial performance of silicon solar cells: interband Auger recombination; surface recombination at the contact interfaces; enhanced metallic impurity solubility; diffusion profiles; and defect-impurity recombination centers. Thermal recombination parameters of titanium impurity in silicon were estimated from recent experimental data. Based on those parameters, computer model calculations showed that titanium concentration must be kept below 6x10 to the 12th power Ti/cu cm in order to achieve 16% AM1 efficiency in a silicon solar cell of 250 micrometers thick and 1.5 ohm-cm resistivity.
The high-low (H-L) junction in the emitter region that defines the high-low-junction emitter (HLE) solar cell suppresses the dark emitter recombination current J sub E so that the base recombination current dominates in determining the open-circuit voltage. For silicon cells this enables the achievement of considerably larger values of open-circuit voltage than those achievable in conventional structures. This paper describes experiments that demonstrate the achievement of J sub E suppression (to less than 5 x 10 to the -14th A/sq cm) and large open-circuit voltage (640 mV) in HLE test cells of two distinct types. In the first type (the diffused HLE structure) impurity diffusion forms the H-L junction in the emitter; in the second type (the oxide-charge-induced HLE structure) the H-L junction is formed in emitter material of relatively low doping concentration by an oxide-charge-induced electron accumulation layer.
The greatest gains in solar energy conversion efficiency of p-n-junction silicon solar cells come from increasing the open-circuit voltage V sub OC; it is important to understand and characterize the material parameters that limit the V sub OC. Strong experimental evidence exists to support the assertion that either an anomalously large minority carrier charge storage or an anomalously small minority carrier lifetime in the quasi-neutral emitter region limits the open circuit voltage. A method is presented for measuring charge storage and effective lifetime. Static and transient measurements are analyzed using physical models of the solar cell characteristics. This analysis yields the emitter charge storage and life-time, which then can be related to the various physical mechanisms, such as energy band gap shrinkage, that have been proposed earlier as responsible for limiting V sub OC.