ICMCTF2004 Session H3-1: Thin Films for Next Generation Devices

As a semiconductor material, silicon carbide (SiC) offers great advantages to overcome the physical device limitations in Si devices set by its materials properties. For example, Si devices are generally used below its maximum junction temperature of ~200°C and confined to blocking voltages of a few kilo-Volts. Due to its large band gap, SiC possesses a very high breakdown field and low intrinsic carrier concentration, which accordingly makes high voltage and high temperature operation possible. SiC is also suitable for high frequency device applications, because of the high saturation drift velocity and low permittivity. So far, dielectrics including AlN and TiO₂ have been investigated on SiC, which shows that the best performance was obtained when those deposited insulators are integrated with SiO₂. In ferroelectrics, the spontaneous polarization can be switched by an externally applied electric field, and thus are attractive for nonvolatile memory and sensor applications.

In this work, a brief introduction to the materials and process technology of SiC including metal contacts to SiC, will be given. Particularly, the fundamental physics research and technology development for the successful realization of SiC diodes and transistors including BJTs and JFETs, will be described. The motivation for SiC Ferroelectric-FET will be then provided as well as the first experimental results from the prototype devices, which operated up to 300°C with memory function retained up to 200°C.

Recently the optical activity in the UV range and the resonant tunneling (RT) behaviour of ZnO based quantum wells has been demonstrated [1]. The wells were 6nm and 8nm with emission at 345.55 nm and 348.22 nm respectively after excitation by a N2 laser and I-V characteristics showed 2 negative differential resistance peaks. Here we investigate the correlation between the optical and transport properties of this system by evaluating the localization of the energy levels in the quantum well. The wavefunctions of the levels spread out not only in the ZnMgO barrier material but also in the outer ZnO material and therefore a textbook model will not be sufficient. This system is large for ab-initio methods, hence we use an effective mass approach. The study shows that there are 3 distinct levels in the 6 nm well at 24.8, 97.6, and 211.0meV. The 2 peaks in the I-V can be explained by observing that the 211meV level is two orders of magnitude less localized than the other two levels. Similarly the levels in valence band are derived at 4.1, 16.0, and 32.6meV, and the allowed optical transitions are identified. Since the system is excited by a N2 laser, the de-excitation from the 211 level to the 32.6 below Ev level (only ?n=0 transitions are allowed) gives 3.54eV in close agreement with the experimentally observed 3.58eV. The level above Ec that gives the optical activity is "absent" in the transport data due to its poor localization, which can be explained if one examines the spread of the wavefunction in the entire system. Furthermore, we observed that although the degree of localization has been significantly reduced by decreasing the barrier width (as expected), the positioning of the levels in the wells remains almost unaffected. Hence by changing the barrier width b one can change the transport properties of the system without affecting optical properties which depend almost entirely on well width a.

[1] A. Iliadis et al SPIE Vol.4650, pp67-74, 2002.