SiC CMOS and memory devices for high-temperature integrated circuits

Sammanfattning: High-temperature electronics find use in extreme environments, like data logging in downhole drilling for geothermal energy production, inside of high-temperature turbines, industrial gas sensors and space electronics. The simplest systems use a sensor and a transmitter, but more advance electronic systems would additionally require a microcontroller with memory. Silicon carbide (4H-SiC) integrated circuits target high-temperature electronics, although the current integration level is low due to immature process technology and non-volatile memory has not been demonstrated. SiC CMOS would allow highly dense integrated circuits for microcontrollers and random access memory (RAM). Ferroelectric capacitors could serve as high-temperature non-volatile memory devices.In this work, significant efforts have been taken to develop a SiC CMOS process and ferroelectric capacitors. SiC CMOS is challenging and mostly unexplored technology. A recessed channel transistor design was investigated. Several key challenges in the SiC CMOS process was identified, leading to a polyoxide-based field oxide, a deposited gate-dielectric process, reproducible Ni-Al semi-salicide contacts to p-type SiC, and a high-temperature CMP enabled two-level TiW-based metallisation. Self-aligned cobalt silicide contacts were investigated, and was found to produce low-resistance ohmic contactsto n-type SiC. Inverters and ring oscillators that operate at 200 °C were achieved in this recessed channel SiC CMOS process. It was found that steam-treating the gate oxide interface produced both NMOS and PMOS transistors that could be used for circuits. However, the reliability suffered due to poor PMOS performance. Wafer-level statistical measurements of interface trap density was performed on NMOS transistors treated by steam, dry oxygen and nitrided by nitrous oxide. A deposition and etch process for ferroelectric capacitors, using vanadium-doped bismuth titanate as ferroelectric material, was developed. High-temperature operation was demonstrated, and several scalability challenges for the etched process was identified.The implication of this thesis is that while operational recessed channel SiC CMOS was demonstrated at high temperature, more promising technologies like ion implanted bulk transistors should be investigated instead, due to the numerous difficulties in optimising both NMOS and PMOS with this recessed channel design. The presented recessed channel process technology can be used to fabricate short channel length NMOS-logic. Ferroelectric capacitors is a good candidate for high-temperature non-volatile memory applications, although more work is needed in the CMOS integration.