Before the aforementioned work by Bockstedte and coworkers106 almost no work was devoted to migra­tion properties of point defects in SiC. We should, however, cite previous preliminary works by the same group,111,112 a work on the mechanisms of formation of antisite pairs,113 and a work on vacancy migration published in 2003.114 The comprehensive study of migration barriers in Bockstedte et a/.106 showed, first of all, that vacancies have much higher migration energies than those of interstitials: higher than 3 eV for the former in the neutral state, around 1 eV for the latter (0.5 for IC, 1.4 for ISi). Another

remarkable finding is the strong variation of the migration energy with the charge state; indeed, the migration energy for the carbon vacancy is raised by almost 2 eV going from the neutral to the 2+ charged state, whereas the silicon vacancy finds its migration barrier reduced by 1 eV when its charge goes from neutral to 2—. Interstitials are reported to have their lowest migration barriers in the neutral state, except for the ISTiC configuration, which is expected to have an almost zero energy barrier of migration in the 2+ and 3+ charge states. Such large changes in the migration energies of defects with their charge should induce tremendous variations in their kinetic behavior under different charge states.

The energy barriers of recombinations of close interstitial vacancy pairs have also been tack — led.115-117 It appears that the energetic landscape for the recombination of Frenkel pairs is extremely complex. One should distinguish the regular recom­bination of an homo interstitial-vacancy pairs from those of hetero interstitial-vacancy pairs, which leads to the formation of an antisite. Recent works tend to suggest that the latter may, in certain con­ditions, have a lower energy than the recombination of a regular Frenkel pair. A kinetic bias for the formation of antisites, preliminary to decomposition, may thus be active in SiC under irradiation.118

Calculations of threshold displacement energies from first-principles molecular dynamics29 have also been reported. Their results show that this quantity is strongly anisotropic, and they found average values (38 eV for Si and 19 eV for C) that are in agreement with currently accepted values (coming from experi­mental evaluations that are, however, largely dis­persed). These calculations prove that available CPU power is now large enough to calculate TDE from ab initio molecular dynamics. This is good news as empirical potentials are basically not reliable in the prediction of TDE.

1.08.5.1.1 Defect complexes

Several defect complexes have been studied by first-principles calculations in silicon carbide. The identification of EPR signals, deep level transient spectroscopy (DLTS), or photoluminescence (PL) experiments based on calculated properties have been attempted for some of them. Crucial to these identifications is the reliability of the predictions of charge transition levels (for the position of DLTS peaks) and of annealing temperatures, through more or less complicated mechanisms.

One of the first, and simplest, defect complex iden­tified through comparison of theory and experiment was the VC-CSi coming from the annealing of silicon vacancies in 6H-SiC, as previously mentioned. More complex antisite defects or antisite complexes119,120 as well as divacancy complexes121-123 were called upon for the attribution of PL or EPR peaks.

Various kinds of carbon clusters were studied in detail theoretically.124-127 The cited works deal with the stability, electrical properties, and local vibrational modes (LVM) of several structures. It was shown that the aggregation of carbon interstitials with carbon antisites can lead to various bound configurations. In particular, two, three, or even four carbon atoms can substitute one silicon atom forming very stable structures. The binding energy of these structures is high: from 3.9 to 5 eV, according to the charge state, for the (C2)Si, and further energy is gained when adding further carbon atoms. Silicon clusters did not raise as much interest as carbon ones; however, a recent work107 deals with the stability and dynamics of such silicon clusters (see Figure 12).