First-principles-based investigation of kinetic mechanism of SiC(0001) dry oxidation including defect generation and passivation

Abstract

Effect of gate oxidation method on electrical properties of metal-oxide-semiconductor field-effect transistors fabricated on 4H-SiC C (0001) face Appl.The key stages of the dry oxidation of the SiC͑0001͒ surface are analyzed based on first-principles calculations. It is found that an abrupt SiC/ SiO 2 interface model results in a large activation barrier of oxygen penetration to the silicon carbide, and thus the penetration is probably the rate-limiting step for the entire dry-oxidation process. The subsequent reactions of SiC oxidation after oxygen penetration are investigated, and it is found that CO release is competing with carbon dimer formation. These dimers probably are responsible for near-interface traps in the silica layer generated during SiC oxidation. The possible passivation reactions of a carbon dimer defect by active species, such as O 2 , NO, and H 2 are investigated. It is found that an oxygen molecule can break a Si-C bond via dissociation in the triplet state and finally can produce two CO molecules from the carbon dimer defect. The NO molecule can easily break a Si-C bond of a carbon dimer defect and form cyano groups -CN, which can finally recombine to form a C 2 N 2 molecule. This molecule can hardly diffuse in silica matrix, and it is suggested that it is further oxidized by an NO molecule to CO and N 2 molecules. It is suggested that the process of passivation by O 2 and NO molecules is restricted by the incorporation of these molecules in small voids near the carbon defect. Based on the calculated results, a simple kinetic mechanism of dry SiC oxidation is proposed and kinetic modeling of the oxidation process is performed. It is found that in the framework of this mechanism, the carbon defect density should weakly depend on temperature.

Competing atomic and molecular mechanisms of thermal oxidation—SiC versus Si

“…Furthermore, an explanation of why the more diffusive O 2 does not contribute to the oxidation of Si-face SiC is needed. Gavrikov et al reported a barrier of 3.5 eV for O 2 diffusing through the last layer of SiO 2 on Si-face SiC, [22] which could have suggested that O 2 cannot be incorporated. However, they reported an even higher barrier of 4.5 eV for O i and concluded that O i is irrelevant for Si-face oxidation.…”

Structural and electronic properties of the transition layer at the SiO2/4H-SiC interface

“…Such defective structures have not yet been considered in the previous theoretical studies. 20,21,24,27,30 Apart from that, our non-abrupt interface model also include excited configurations, such as fivefold-coordinated Si atoms (SiO 5 ) and threefold-coordinated O atoms. The SiO 5 center does not mean the O enrichment of the interface because a fivefold-coordinated silicon event is accompanied by the generation of threefold-coordinated oxygen atoms.…”

“…20,21 Gavrikov et al subsequently presented a kinetic mechanism of dry SiC oxidation based on the first-principles investigation. 24 They suggested that the carbon dimer generated during SiC oxidation probably is responsible for the NITs. However, these defects could not fully account for the TDRC and PST measurements.…”

“…Unfortunately, due to the limitation of formed SiO 2 film to the diffusion of carbon oxide, a small portion of carbon atoms still remain in the oxide and at the interface. [24][25][26][27][28][29][30] The residual carbon atoms may contribute to the formation of carbon clusters, as observed by atomic force microscopy (AFM) 13 and combined transmission electron microscopy (TEM)/electron energy loss spectroscopy (EELS). 14,15 Much theoretical work has been performed to understand the atomic-scale nature of carbon-related defects at the SiO 2 /SiC interface.…”

Si and C emission into the oxide layer during the oxidation of silicon carbide and its influence on the oxidation rate

“…[17][18][19] It has been reported that C injection into the SiO 2 layer can be attributed to the interface state over a wide energy range, 20 and the injected C interstitials create C dimers/di-interstitials in the SiC layer, which gives rise to the interface state near the conduction band edge. 21,22 Furthermore, it has been assumed in the development of bipolar devices that C vacancies, which give rise to deep-level states called Z 1/2 centers, are filled with C interstitials that diffuse into the SiC layer during oxidation. [17][18][19] It is believed that such atomic emissions during oxidation can help explain device-killing defects.…”