This project aims to develop scientific foundations for a continuous process to produce crystalline Silicon wafers from high purity poly-Silicon. In this process Silicon is floated over a two-layered molten substrate to form a very thin (less than 0.3 mm) Silicon sheet, which solidifies to produce crystalline Silicon wafers suitable for solar cells. The production cost will be low relative to expensive wire saw processes since this process is continuous and does not incur Silicon loss.
Intellectual Merit: In this research the PIs will study the stabilization and control of a freezing front on a molten substrate. Systems of this type may exhibit Mullins-Sekerka instability and active control is needed to stabilize the process. They will develop multi-scale process models capable of representing the instabilities present in the freezing front. The models will be matched to a physical system using experimental data. Control methods will be developed to stabilize the freezing front using a thermodynamics based approach to passivity based control. The focus is on the application of the methods to Silicon for making solar cells. A number of new measurement techniques will be tested for the Silicon solidification problem and process parameters needed for process design, scale-up and control will be determined. A thin layer of molten Silicon will be slowly poured on a high-density liquid used as a substrate in micro scale experiments. The physical properties of the substrate will be tuned so that the molten Silicon can be continually cooled and withdrawn in the form a continuous sheet of single-/multi-crystalline Silicon. They will show that the use of molten (liquid) substrate forming three liquid-liquid layer prevents the crystal imperfections, dislocations and grain boundaries present in current continuous, horizontal wafering processes. They will also demonstrate that limited purification can be experienced. The simultaneous goal is to develop multi-scale mathematical models to compare the micro-scale experimental results on movement and flow of the molten Silicon over the liquid substrate. Model predictions will be compared with macro-scale experiments.
Broad Impact: Solar energy has so far not measured up to its potential due to the high cost of producing high purity Silicon and Silicon wafers. Significant progress has been made in developing cheaper processes for making high purity poly-silicon in fluid bed reactors. Very limited progress has been made in finding alternatives to the expensive band-saw process for wafering, however. The ideas described here may contribute towards solving this problem. This three layer process draws inspiration from the Pilkington glass process which revolutionized the glass industry. Preliminary experiments show that it is feasible to produce silicon wafers in small scale using a similar idea. The most important broader impacts of this research are expected to be found in the area of alternative energy. It is also expected that the research will lead to new methods for multi-scale modeling and stabilization and control of solidification fronts. These problems turn up in a number of application areas, including the drying of paints, film processing and coating.
The objective of this research project was to verify system components experimentally and to develop integrated, multi-scale models for the Horizontal Ribbon Growth (HRG) process for making silicon wafers. The HRG process can reduce capital and operating costs by a factor of 3 or more relative to current silicon wafer production technology due to the elimination of Kerf losses, continuous production and low energy usage. The main application domain is in the area of solar cells and the process concept promises to reduce the cost of solar electricity below coal parity at about $0.30 per kWp in the foreseeable future (about 5 years). The HRG process for continuous production of silicon wafers is motivated by how a sheet of ice freezes on water. In the HRG process a molten bath of silicon is cooled from the top, below its freezing point at 1414 degC. Solid silicon is less dense than liquid silicon (by about 4%) so that the sheet floats like ice floats on liquid water. The solid sheet is continuously removed, liquid silicon is continuously replenished at the same rate to maintain the mass balance. Heating and cooling is carefully controlled to maintain a steady state energy balance with sufficient temperature gradient to stabilize the process and maintain high production rate. The production rate is not limited by the rate of solidification since the direction of crystallization is perpendicular to the process flow. This property distinguishes the HRG process from many current technologies including the Czochralski and Edge Film Defined Growth (EFG) processes. Furthermore, directional solidification on a liquid substrate produces mono-crystalline sheet of high quality by moving impurities to the melt. The PhD research program supported by the National Science Foundation enabled us to demonstrate in experiments and computer simulation that The continuous process can produce mono-crystalline wafers of silicon (Si<111> at present). Silicon losses were estimated to be less than 12 %. We also developed a green texturization process to reduce reflectivity of the as-produced wafers. The new texturization process uses water-based (green) chemicals to reduce the reflectance of Czochralski wafers to 10% of incident light while the reflectance of our wafers was reduced to about 8%. The silicon losses were estimated to be lower than 4%. We completed solidification models to show that a wide range of impurities, including iron and aluminum, are removed by directional solidification. This opens up for the possibility of using less pure raw-materials than previously thought possible. We are exsploring the possibility of using Kaolin as a start material for an integrated system. We developed a computational fluid dynamics model and a physical water model to study the stability of control systems and develop estimates for the feasible operating range, the so-called process envelope. The models demonstrate the feasibility of the HRG concept at an industrial scale as we show that sheets with a thickness of 0.180 mm can be produced at a rate of at least 150 mm per min. We developed detailed cost models to shown that price of silicon wafers can be reduced by a factor of three relative to current technology. We predict that cost of a solar cell produced in HRG process will be about $0.30 per Wp. Several papers, conference proceedings and one patent application disclose various aspects of the Carnegie Mellon University NSF Final Report proposed process. We made several important contacts with industry represtnatives and academics interested in the HRG process. We were also able to install a pilot plant system at CMU that we will use for further studies of the integrated system. The system will be operational in the summer of 2013. It represents an investment of about $1M. We exepct to be able to verify the complete process within a time frame of 3 years.
