Measurement of electrical self-heating for thermal characterization of MEMS and NEMS structures

Robert A Sayer, Purdue University


Self-heating (Joule heating) is typically an unwanted effect of electrical power dissipation in electronic devices. In micro- and nano-electromechanical systems (MEMS and NEMS), the heating effect is especially pronounced due to the high power densities that are carried in the active regions of these devices. Although temperature rises are typically undesirable because of its negative impact on electrical performance, mechanical integrity and user comfort, self-heating can be exploited in the micro- and nano-scale structures to measure thermal properties of component materials. This dissertation describes four such self-heating techniques for thermal property measurement and develops new data analysis procedures to extend the efficacy of these methods. Due to its wide spread use, the 3 omega method is discussed, and drawbacks of the technique are presented. A modified data analysis method for the steady-state DC thermal conductivity technique is developed, enabling measurements to be performed on devices with non-ideal boundary conditions. The new data analysis procedure has been applied to polycrystalline silicon, an important building block of many MEMS and NEMS devices. Self-heating of a radio frequency (RF) MEMS capacitive switch has been measured with infrared (IR) thermography. A thermal model that includes radiation and convection is presented, and the finite volume method is used to fit thermal conductivity to the model. Thermal expansion (and deformation) during self-heating is shown to have a profound effect on device performance, based upon both simulations and analytical calculation. Noise thermometry is discussed as a method for the measurement of both temperature and thermal resistance (based upon a thermal model that was developed). 1/f noise magnitudes in MEMS and NEMS are very large and can overwhelm the desired shot and Johnson noise signals. A data analysis procedure that accounts for 1/f noise is presented. Stochastic modeling of noise is employed to demonstrate the limits and accuracy of the method. Furthermore, a RF noise measurement system is set up for measuring noise in micro- and nanoscale devices. Measurements on vertical carbon nanotube (CNT) devices and RF MEMS capacitive switches are presented. Fabrication of carbon based (CNT and graphene) devices is discussed. The fabrication procedures are compatible with wafer scale processing for large-scale integration. Single-walled CNTs (SWCNTs) are grown in a porous anodic alumina (PAA) template. Processes for obtaining sub-micron channel lengths without lithography, creating improved contacts to SWCNTs in the PAA template, and for fabricating surround gates for field effect devices is presented. Field effect and suspended graphene devices are fabricated from few layer graphene films grown using microwave plasma chemical vapor deposition.




Fisher, Purdue University.

Subject Area

Mechanical engineering|Nanoscience

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