No abstract

From the viewpoint of application in power electronics, SiC possesses the greatest advantage of having SiO2 as its native oxide. Unfortunately, the usual thermal oxidation produces an unacceptably high density of interface states, with a complex energy distribution. Deep states are assumed to be caused by carbon excess at the interface, while the slow electron traps, called NIT, with especially high density near the conduction band of 4H-SiC (which would be the best polytype for power devices), are expected to originate from oxide defects near the interface. Unlike the case of the Si/SiO2 interface, simple hydrogen passivation does not help to reduce the high trap density. A possible passivation method for both deep states and NIT is post-oxidation annealing or oxidation in the presence of NO or N2O molecules. Here we present systematic and sophisticated theoretical calculations on a model of the 4H-SiC/SiO2 interface, in order to establish the main reaction routes and the most important defects that are created during dry oxidation, and may give rise to the observed interface traps. We also investigate the effect of nitrogen in suppressing them.

Large-scale and gap error free calculations of the electronic structure of vacancies in 4H-SiC have been carried out using a hybrid density functional (HSE06) and an accurate charge correction scheme. Based on the results the carbon vacancy is proposed to be responsible for the Z 1/2 and EH 6/7 DLTS centers.

The carbon vacancy (V C ) has been suggested by different studies to be involved in the Z 1 /Z 2 defect-a carrier lifetime killer in SiC. However, the correlation between the Z 1 /Z 2 deep level with V C is not possible since only the negative carbon vacancy (V − C ) at the hexagonal site, V − C (h), with unclear negative-U behaviors was identified by electron paramagnetic resonance (EPR). Using freestanding n-type 4H -SiC epilayers irradiated with low energy (250 keV) electrons at room temperature to introduce mainly V C and defects in the C sublattice, we observed the strong EPR signals of V − C (h) and another S = 1/2 center. Electron paramagnetic resonance experiments show a negative-U behavior of the two centers and their similar symmetry lowering from C 3v to C 1h at low temperatures. Comparing the 29 Si and 13 C ligand hyperfine constants observed by EPR and first principles calculations, the new center is identified as V − C (k). The negative-U behavior is further confirmed by large scale density functional theory supercell calculations using different charge correction schemes. The results support the identification of the lifetime limiting Z 1 /Z 2 defect to be related to acceptor states of the carbon vacancy.

No abstract

No abstract

In a recent rewiew article we have reported—among other things—on preliminary computational studies regarding the effect of nitridation on the SiC/SiO2 interface, in case of n-type doping. We have since discovered a few numerical errors in the paper, which also influence some of our statements. Here we present the correct results and their consequences. 1. Introduction In our review we reported energies of reactions occurring at the SiC/SiO2 interface upon nitridation. Our computational approach was to calculate the energy of incoming and outgoing molecules in a bulk supercell of SiO2, and that of interface defects before or after the reaction in a slab model of the 4H-SiC/SiO2 interface. Incoming molecules in bulk SiO2 were assumed to be neutral, but the interface was coupled to a reservoir of electrons with energy corresponding to the Fermi level of the semiconductor. The latter was determined by the temperature of nitridation (~1100 °C) and the usual doping of the channel region (1015–1016 cm-3) of a MOS (metal-oxide-semiconductor) transistor with an n-type channel, to be 1.9–2.2 eV, with respect to the valence band edge of SiC. Using the calculated total energies, energies of exchange reactions between the models have been calculated, adding a band off-set correction for each electron transfer between the models (to account for the error introduced by using the pseudopotential approximation in two different models). Unfortunately, some of the results were erroneously given. Here we present the correct values, and briefly state the consequences. 2. Corrected results (A) The energy gain on reaction (1) of the original paper is equal to 3EF − 2.7 eV, which gives 3.0–3.9 eV in the Fermi-energy range considered (not, as erroneously reported 1.0–1.3 eV). As a consequence, this reaction is, in fact, dominating over reaction (2) of the original paper with an energy gain of EF + 0.7 eV, i.e., 2.6–2.9 eV in the chosen Fermi-level range. Therefore, nitrogen insertion into an interstitial position at the interface is more likely, than as a carbon substitutional. (B) The energy gain on reactions (8) of the original paper is, in fact, 2EF + 0.8 eV (not EF + 0.8 eV, as given), leading to as much as 4.4–5.2 eV in the chosen EF range. (C) The energy gains on reactions (10)--(11) of the original paper are, in fact, EF + 2.1 eV, and 3EF − 1.3 eV, respectively. (The latter was given erroneously as EF − 1.3 eV.) Consequently, in the chosen EF range the latter is (with 4.4–5.3 eV) somewhat even more favourable, than the former (with 4.0–4.3 eV). This indicates, together with (1), a pronounced nitrogen incorporation. (D) Finally, the correct energy gain on reaction (14) of the original paper is 2EF + 0.8 eV, i.e., 4.6–5.0 eV in the chosen range, underlining the possible importance of atomic oxygen. The full equations (1), (2), (8), (10), (11) and (14) are displayed in the attached PDF file. We apologize for our mistakes which, however, do not affect the conclusions of the paper, which are not mentioned here.

Preliminary results of a systematic theoretical study on the reactions of NO with a model 4H-SiC/SiO2 interface are presented. We show, that nitridation is a complex process, in which the balance between various mechanisms depends on doping and temperature. For weakly doped (1015-16 cm-3) n-type SiC, the crucial effect is an additional oxidation without creation of excess carbon at the interface.