Finite Element and Imaging Approaches to Analyze Multiscale Electrothermal Phenomena

Amirkoushyar Ziabari, Purdue University

Abstract

Electrothermal effects are crucial in the design and optimization of electronic devices. Thermoreflectance (TR) imaging enables transient thermal characterization at submicron to centimeter scales. Typically, finite element methods (FEM) are used to calculate the temperature profile in devices and ICs with complex geometry. By comparing theory and experiment, important material parameters and device characteristics are extracted. In this work we combine TR and FEM with image blurring/reconstruction techniques to improve electrothermal characterization of micron and nanoscale devices. ^ We present an ultrafast yet highly accurate technique, based on image blurring and FEM, for thermal modeling of 3D ICs that include thermal vias. Modeling shows the impact of thermal vias placement on the reduction of maximum temperature in different layers of 3D IC. Next, we experimentally investigate high field non-equilibrium in electron gas and its impact on thermoelectric cooling. When there is a large temperature gradient or a high current density, thermoelectric coefficients can become non-linear. We provide the first detailed experimental study of non-linear Peltier coefficient in low-doped InGaAs semiconductor. A hybrid analytical-numerical model is developed to extract current-dependent Peltier coefficient from TR thermal images of nonlinear InGaAs microrefrigerators at room and cryogenic temperatures. Finally, we combine FEM, image deconvolution techniques and TR to accurately reconstruct thermal images of devices with spatial resolution below visible light diffraction limit. We present full thermal distribution around submicron size heat sources on InGaAs thin films on InP substrate both in steady-state and in transient regime indicating emergence of non-diffusive heat transport.^

Degree

Ph.D.

Advisors

Ali Shakouri, Purdue University.

Subject Area

Electrical engineering|Nanotechnology

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS