Lattice location of transition metals in silicon by means of emission channeling

The behavior of transition metals (TMs) in silicon is a subject that has been studied extensively during the last six decades. Their unintentional introduction during the Si production, crystal growth and device manufacturing have made them difficult contaminants to avoid. Once in silicon they easily form deep levels, either when in the isolated form or when forming precipitates. One important effect is the reduction of efficiency of silicon-based devices, being dramatic, in particular, in photovoltaic applications. One way to avoid such effects is by engineering the location of the TM: some TM complexes or lattice sites of the isolated form do not introduce any level in the silicon bandgap. Which point defects lead to such passivation is still under debate. Another way is to mitigate the reduction of efficiency by reducing the dangling bonds of TMs with hydrogen. The most important and commonly used procedures to diminish the unwanted effects of the introduced deep levels are, nevertheless, based on the so-called gettering techniques, where TMs are forced to move away from the active area of devices. Although the macroscopic effects of all the gettering techniques are well known, the related microscopic mechanisms are still poorly understood, namely the TM complexes involved in gettering into defective regions, into p-type layers and by phosphorus diffusion. TMs can also be used as magnetic dopants in semiconductors. So far, only binary semiconductors have been intensively addressed, after the discovery of dilute ferromagnetism of (Ga,Mn)As. However, the maximum found Curie temperature was only 185 K, which is too low for practical applications. Other semiconductors have hence been recently suggested, such as silicon. One important feature is the lattice sites occupied by TMs, and related fractions. It is currently well established that the incorporation of TMs on substitutional sites leads to thermally stable point defects with none-zero magnetic moment. One way to obtain substitutional TMs is by ion implantation. Which fraction is incorporated on S sites is, however, still unknown. All these issues can be explored by investigating the lattice sites of implanted 3d TMs. In particular, different complexes will correspond to different lattice sites. Here, we have investigated the lattice location and thermal stability of the implanted 3d TM probes $^{56}$Mn, $^{59}$Fe, $^{61}$Co and $^{65}$Ni in both lightly and heavily doped n- and p-type Si by means of emission channeling experiments. In all cases we identified ideal substitutional (S) sites, displaced bond-centered (near-BC) sites and displaced tetrahedral interstitial (near-T) sites. The dependence of the near-BC sites on annealing temperature was similar for all investigated TMs, hence the related complexes may be formed irrespective of the TM nature. Two main origins are suggested: involving divacancies where the TM occupies BC sites (between two vacant sites) and involving fourfold coordinated vacancy clusters with the TM occupying a position near the ideal BC site. While the first case cannot explain the observed displacement, the second case definitely does. The observed near-BC sites may hence correspond to a combination of different complexes where these two types of complexes may participate. Since the thermal stability of near-BC sites increases from n$^-$ to n$^+$-type Si, it is suggested that the related multivacancy complexes may participate in P-diffusion gettering. The complexes involved in the observed near-T sites seem to be more diversified, as the related fraction and displacement change significantly with annealing temperature. One has found that implanted TMs occupy near-T sites around the related peak concentration (at Rp from the surface), after low temperature anneals, while for high annealing temperatures TMs tend to occupy near-T sites at midway between the peak concentration and the surface (at Rp/2 from the surface), rich in vacancies. Moreover, the near-T fraction of Mn, Fe and Co increases from n$^-$ to p$^+$-type silicon, which might be due to the TM pairing with the electric dopant boron. This fact agrees with the positive charge nature of interstitial Mn, Fe and Co, while, for instance, the charge state of Ni is neutral in p-type Si which might have prevented the increase of its near-T fraction. It is therefore confirmed that the related pairing mechanism is driven by Coulomb interactions. The observed displacement also matches with theoretical predictions. Some of the results obtained in this thesis were also compared to Mössbauer spectroscopy investigations from literature. In particular, it was found that, in n-type Si, the near-T fraction of $^{ 61}$Co increases in the annealing temperature range where a Mössbauer doublet component appears, attributed to Co-dimers. A comparison to emission channeling investigations on iron is, however, at variance with such a conclusion. The attribution of near-T sites to the beginning of clustering is hence most likely incorrect. The observation of ideal S sites might be due to the trapping of TMs into single vacancies produced during implantation. The particular cases of Mn and Co are important as they are the main TM candidates for magnetic dopants. While ~60% of Co can occupy ideal S sites, no more than ~30% of Mn is in the S form. Hence, although the magnetic moment of Co is smaller than that of Mn, Co is easier incorporated on S sites, for fluences of ~$10^{12}$ cm$^{-2}$. The question that this work raises is whether the S sites actually originate from the substitutional form of Co.