Structure and magnetism of transition-metal implanted dilute magnetic semiconductors

The discovery of a dilute magnetic semiconductor (DMS) in which ferromagnetism is carrier-mediated and persists above room temperature is a critical step towards the development of semiconductor-based spintronics. Among the many types of DMS materials which have been investigated, the current research interest can be narrowed down to two main classes of materials: (1) narrow-gap III-V semiconductors, mostly GaAs and InAs, doped with Mn; (2) wide-gap oxides and nitrides doped with 3d transition metals, mostly Mn- and Co-doped ZnO and Mn-doped GaN. With a number of interesting functionalities deriving from the carrier-mediated ferromagnetism and demonstrated in various proof-of-concept devices, Mn-doped GaAs has become, among DMS materials, one of the best candidates for technological application. However, despite major developments over the last 15 years, the maximum Curie temperature (185 K) remains well below room temperature. On the other hand, wide-gap DMS materials appear to exhibit ferromagnetic behavior at and above room temperature, but there is no consensus regarding the origin (intrinsic or not) of such ferromagnetism. The major challenges facing the DMS field can thus be summarized as: (i) in narrow-gap DMS, to increase the Curie temperature above the current record values; (ii) in wide-gap DMS, to understand the origin of the often observed high-temperature ferromagnetism and identify the conditions under which it can be reproduced. In Mn-doped GaAs, the narrow-gap DMS with the highest Curie temperature reported so far (T$_{C}$ = 185 K), a further increase in T$_{C}$ depends on a better understanding and control over the balance between substitutional and interstitial Mn impurities. Here, we studied the lattice location of Mn in GaAs, and to what extent thermal annealing can be used as a means to remove the detrimental interstitial fraction. We determined that the interstitial Mn fraction occupies tetrahedral interstitial sites with As nearest neighbors, where it is stable up to 400$^{\circ}$C, with an activation energy for diffusion of 1.7–2.3 eV. Substitutional Mn (in Ga sites) becomes mobile at higher temperatures (∼700$^{\circ}$C) with an activation energy of ∼3 eV. Being difficult to reconcile with the general belief that interstitial Mn is the donor defect that out-diffuses at ∼200$^{\circ}$C, these findings motivate a careful reassessment of the structural effects of low temperature annealing. First, placing the diffusion of interstitial Mn well above 200$^{\circ}$C reopens the problem of identifying the donor defect that diffuses at low temperatures. Second, and more importantly, it suggests new strategies and prospects for achieving pure substitutional doping and higher Curie temperatures in Mn-doped GaAs. In transition-metal doped ZnO and GaN, the emerging picture of defectrelated ferromagnetism indicates that high-temperature magnetic order may be associated with structural disorder, i.e. lattice defects. Here, we searched for room-temperature ferromagnetism in ZnO and GaN implanted with 3d transition-metals (Cr, Mn, Fe, Co) for a varying degree of lattice disorder, by varying the implantation and thermal annealing conditions. However, none of the prepared wide-gap DMS systems showed measurable room-temperature ferromagnetism. Among these systems, we focused on selected impurity-host combinations, performing detailed studies of their structural and magnetic properties, in terms of the lattice location and the magnetic interactions between transition-metal impurities, as well as their interplay with other lattice defects. We determined that, in addition to the majority substituting for the host cations (Zn in ZnO and Ga in GaN), significant fractions of the Mn and Co impurities (∼ 20 – 30%) occupy substitutional anion sites (O in ZnO and N in GaN), which are virtually unaffected by thermal annealing up to 900$^{\circ}$C. Although anion site fractions have never been identified before in ZnO and only for Mn in GaN, a number of arguments can be put forward which suggest that anion-site Co and Mn defects may indeed have low formation energies, particularly under a high concentration of anion vacancies (or equivalently, under O or N deficient conditions). Regarding the magnetism of the wide-gap DMS, we focused on two representative impurity-host combinations: Cr-implanted GaN and Fe-implanted ZnO. With Cr-implanted GaN, we aimed at studying the magnetic interactions between Cr moments, by varying the Cr atomic concentrations from the very dilute regime to well above the percolation threshold. With Fe-implanted ZnO, we aimed at a comprehensive description of the material’s structure in terms of phase pureness and lattice location of the Fe impurities, as well as a detailed modeling of the paramagnetism observed with a varying density of lattice defects created by ion implantation. For both systems, we have shown that the magnetic interactions between the transition-metal moments are antiferromagnetic. Our findings support the view of a growing fraction of the DMS community that in general the transition-metal impurities in wide-gap semiconductors do not order ferromagnetically at high temperatures. The often observed high-temperature ferromagnetism can, alternatively, be attributed to non-intrinsic sources, such as a non-uniform distribution of the magnetic impurities, magnetic contamination or setup-related artifacts.