Thermal Transient Measurement and Dimension-dependent Modeling of Self-heated Advanced Devices

In the era of 3D device, the self-heating effect brings higher temperature to device, and significantly affects the electrical performance of device. Accurate thermal modeling is required to optimize the device structure and circuit design. In this paper, a fifth-order thermal RC network is developed to describe the transient heating process based on the transient thermal simulation of 14-nm FinFET technology. Moreover, a size-dependent dynamic thermal model including fin width, fin height, extension length and materials of the source and drain extension regions, and the thickness of the shallow trench isolation (STI) is developed to estimate the peak temperature at given frequency. The parameters are randomly selected to verify the proposed models, and the average mean relative error of the dimension-dependent model is about 0.42 %, the root mean square error is about 2.33 K.