Are dangling bond centers important interface traps in 4H-SiC metal oxide semiconductor field effect transistors?

Abstract

Silicon carbide (SiC) based metal-oxide-semiconductor field-effect transistors (MOSFETs) have great promise in high power and high temperature applications. Unfortunately, effective channel mobilities remain disappointingly low, typically about 30 cm2/Vs. A major contributor to the disappointing effective channel mobilities is the presence of substantial densities of interface traps at the SiC/SiO2 interface. Many investigators have invoked silicon or carbon dangling bonds to be the dominating source of these interface defects, but very little, if any, direct experimental evidence exists to support this assumption in the SiC/SiO2 system. Cantin et al. [Phys. Rev. Lett. 92, 1 (2004)] have used conventional electron paramagnetic resonance measurements on porous oxidized SiC structures to measure the g tensor for the SiC/SiO2 interface carbon dangling bond. These results provide a particularly straightforward means to search for the presence of carbon dangling bonds in fully processed SiC MOSFETs using electrically detected magnetic resonance. Additionally, simple theory provides guidance to search for silicon dangling bond defects. In this study, we utilize K band electrically detected magnetic resonance via spin dependent charge pumping measurements in which almost all of the SiC band gap at the SiC/SiO2 interface is accessed. Although quite high signal to noise measurements are achieved, we are unable to detect any trace of the carbon dangling bond spectra. However, in very poor quality p-channel devices, we observe a spectrum which could be consistent with silicon dangling bonds. Other defect centers are clearly present and we conclude that these other centers dominate the interface trap density of states.

Electrically detected magnetic resonance of carbon dangling bonds at the Si-face 4H-SiC/SiO2 interface

“…3b interface. [16][17][18][20][21][22]34 However, those studies reported on an isotropic g-factor in the range of g ≈ 2.0023 − 2.0031. 17,20 It is not clear at this point why there is such a discrepancy between the observed g-factors.…”

“…[16][17][18][20][21][22] A recent study by M.A. Anders et al ruled out the presence of dangling bond defects at the SiC/SiO 2 interface 34. However, their argumentation was based on the absence of resonance lines additional to the dominant defect spectrum which was assigned to the V Si based on its isotropic g-factor.…”

Recombination defects at the 4H-SiC/SiO2 interface investigated with electrically detected magnetic resonance and ab initio calculations

“…8,16,37,38 Several EDMR studies performed at interfaces produced by thermal oxidation linked the observed defect to the V Si center, based on the isotropic nature of the g-factor. 11,39,40 The study of Cochrane et al 11 resolved the HF structure of the V Si using the adiabatic fast passage EDMR technique. The fast passage technique for EPR measurements was pioneered by Hyde and Weger in the 1960s and 1970s and relates to a scan rate of sufficient speed that the electron spins are not able to relax between modulation sweep cycles.…”

“…11,41,42 However, while this study identified the V Si as the defect observed by fast passage EDMR, the other referenced studies used a conventional detection scheme (without fast passage) and showed a dramatically different spectrum. 39,40 This is suggestive of a time component to the measurement resulting in different defects probed as a result of a) j.cottom.12@ucl.ac.uk 0021-8979/2018/124(4)/045302/10/$30.00…”

“…A recent study attempted to rule out the presence of dangling bond defects at the SiC/SiO 2 interface. 40 This conclusion is based upon the assumption that the observed defect is the V Si defect and not the dangling bond and neglects contributions from more than one (similar) defect. In addition, there is also the device generation challenge that needs to be explored.…”