Characterization and Optimization of Radiation at Nano Scale: Applications in Solar Cell Design

High energy needs and environmental concerns associated with fossil fuels have raised the demand for efficient and clean alternatives of power generation. Solar cell technology is one of the most promising options of reliable renewable power sources despite high costs. Thin film solar cells offer the potential for reduction in the cost per kilowatt-hour due to the lower material usage. Nevertheless, most thin film solar cells suffer from low efficiency, though advancements in the science of near field radiation have led to substantial improvements in their optical efficiency. Many design challenges remain to be overcome for the wide-scale commercialization of thin film solar cells. In this dissertation, a numerical study is conducted for optical, optoelectrical and scattering performance enhancement of subwavelength optical devices (i.e., thin film solar cells and light trapping nanoparticles). The proposed design framework of thin film solar cells is based on learning based optimization and characterization methods, which utilize approximations of time consuming simulations. Additionally, a free form nanoparticle design procedure using evolutionary shape optimization is detailed. The background of thin film solar cells and a comprehensive literature review of the thin film solar cell design approaches are provided in Chapters 2 and 3, respectively. The optical enhancement of thin film solar cells using nanoparticles with different shapes is studied in Chapter 4. In Chapter 5, an approximate formulation for optoelectrical efficiency of thin film solar cells is developed to accelerate the design optimization. The learning based design methodology that is introduced in Chapter 5 is further improved in Chapter 6 using a knowledge transfer concept (also known as transfer learning). In this chapter, multiple sets of material combinations are optimized and compared with each other in terms of their optoelectrical efficiencies. In Chapter 7, nanoparticles are designed for maximum scattering, which is desired for enhanced optical performance, using a nonparametric evolutionary design method. In Chapter 8, a predictive model for scattering of arbitrarily shaped nanoparticles using descriptive geometric features is proposed. Overall, this dissertation has led to significant contributions in the field of thin film solar cell design. The results show that the computational burden of the thin film solar cell design can be overcome significantly without sacrificing accuracy. Furthermore, the design methods developed for this dissertation can easily be transferred to other engineering areas involving repetitive, time consuming simulations for design optimization, such as other photonic design problems and integrated circuit design