This research entails the first comprehensive and systematic study on a heat-driven, self-cooling application based on the thermoelectric generation effect. The system was studied using the first and second laws of thermodynamics to provide a solid and basic understanding of the physical principles governing the system. Multiphysics equations that relate heat transfer, fluid dynamics and thermoelectric generation are derived. The equations are developed with increasing complexity, from the basic Carnot heat engine to externally and internally irreversible engines. A computational algorithm to systematically use the fundamental equations has been presented and computer code is implemented based on the algorithm. Experiments were conducted to analyze the geometric and system parameters affecting the application of thermoelectric based self-cooling in devices. Experimental results show that for the highest heat input studied, the temperature of the device has been reduced by 20-40% as compared to the natural convection case. In addition, it has been found that in the self-cooling cases studied, convection thermal resistance could account for up to 60% of the total thermal resistance. A general numerical methodology was developed to predict steady as well as transient thermal and electrical behavior of a thermoelectric generation-based self-cooling system. The methodology is implemented by using equation modeling capabilities to capture the thermo-electric coupled interaction in TEG elements, enabling the simulation of major heating effects as well as temperature and spatial dependent properties. An alternative methodology was also presented, which integrates specialized ANSI-C code to integrate thermoelectric effects, temperature-dependent properties and transient boundary conditions. It has been shown that the computational model is able to predict the experimental data with good accuracy (within 5% error). A parametric study has been done using the model to study the effect of heat sink geometry on device temperature and power produced by TEG arrays. In addition, a dynamic model suited for integration in control systems is developed. Therefore, the study has shown the potential for a heat driven self-cooling system and provides a comprehensive set of tools for analysis and design of thermoelectric generation.

This dissertation presents to our knowledge the first demonstration of a quantum well (QW) laser monolithically integrated with internal optical pump based on a light emitting diode (LED). The LED with high efficiency is operated in a thermoelectrophotonic (TEP) regime for which it can absorb both its own emitted light and heat. The LED optical pump can reduce internal optical loss in the QW laser, and enables monolithically integrated TEP heat pumps to the semiconductor laser. The design, growth and fabrication processes of the laser chip are discussed, and its experimental data is presented. In order to further increase the TEP laser efficiency the development of QDs as the active region for TEP edge emitting laser (EEL) is studied. The usage of QD as TEP laser's active region is significant in terms of its low threshold current density, low internal optical loss and high reliability, which are mainly due to low transparency in QD laser. The crystal growth of self-organized QDs in molecular beam epitaxial (MBE) system and characterization of QDs are mentioned. The design, growth, processing and fabrication of a QD laser structure are detailed. The characteristics of laser devices with different cavity length are reported. QD active regions with different amount of material are grown to improve the active region performance. Theoretical calculations based on material parameters and semiconductor physics indicate that with proper design, the combination of high efficiency LED in TEP regime with a QD laser can result in the integrated laser chip power conversion efficiency exceeding unity.