First-Principles Study on the Ferromagnetism and Curie Temperature of Mn-Doped AlX and InX (X ¼ N, P, As, and Sb) Kazunori SATO  , Peter H. DEDERICHS 1 , and Hiroshi KATAYAMA-YOSHIDA The Institute of Scientific and Industrial Research, Osaka University, 8-1 Mihogaoka, Ibaraki, Osaka 567-0047 1 Institut fuer Festkoerperforschung, Forschungszentrum Juelich, D-52425 Juelich, Germany (Received September 11, 2006; accepted December 20, 2006; published February 13, 2007) We investigate the electronic structure and magnetic properties of AlN-, AlP-, AlAs-, AlSb-, InN-, InP-, InAs-, and InSb-based dilute magnetic semiconductors (DMS) with Mn impurities from first- principles. The electronic structure of DMS is calculated by using the Korringa–Kohn–Rostoker coherent potential approximation (KKR-CPA) method in connection with the local density approximation (LDA) and the LDA+U method. Describing the magnetic properties by a classical Heisenberg model, effective exchange interactions are calculated by applying magnetic force theorem for two impurities embedded in the CPA medium. With the calculated exchange interactions, T C is estimated by using the mean field approximation, the random phase approximation and the Monte Carlo simulation. It is found that the p–d exchange model [Dietl et al.: Science 287 (2000) 1019] is adequate for a limited class of DMS and insufficient to describe the ferromagnetism in wide gap semiconductor based DMS such as (Ga,Mn)N and the presently investigated (Al,Mn)N and (In,Mn)N. KEYWORDS: dilute magnetic semiconductor, ferromagnetism, Curie temperature, first-principles calculation, Monte Carlo simulation, magnetic percolation, p–d exchange, double exchange, super exchange DOI: 10.1143/JPSJ.76.024717 1. Introduction Dilute magnetic semiconductors (DMS) are materials in which the charge degree of freedom is strongly coupled to the magnetic degree of freedom. Due to the carrier mediated nature of the ferromagnetism in DMS such as (Ga,Mn)As, the magnetic state can be controlled by the carrier concen- tration. This characteristic feature of DMS is indispensable for semiconductor spintronics and the DMS are widely investigated as spintronic materials. 1) One of the most important issues in semiconductor spintronics is to fabricate ferromagnetic DMS with Curie temperature (T C ) higher than room temperature. So far, a lot of effort is devoted for this purpose. 1) In particular, many theories were proposed to explain the ferromagnetism in DMS and to propose a materials design for high-T C DMS. Basically, there are two different approaches to deal with the ferromagnetism of DMS. One is to use a model Hamiltonian 2–4) and the other is first-principles electronic structure calculations. 5–7) In the model approach, firstly the DMS is described by using a simplified Hamiltonian, then approximate and some cases, exact solutions are calculated for the desired physical properties of the system. Therefore, the derived conclusions are very precise and give a clear physical meaning, as far as the assumption for the model Hamiltonian and the approx- imations used are justified and the model parameters are experimentally known. However, once the target material is out of this framework, the results become unreliable. This means that the model approach inherently has difficulty in predicting physical properties of absolutely new materials of which we have no information on the electronic structure and relevant mechanism. On the other hand, in the first-principles approach the electronic structure of the proposed material is calculated normally based on the density functional theory (DFT) 8) by using the local density approximation (LDA). 9) Therefore the calculation does not require any experimental values and we do not have to assume specific mechanism which governs the material properties. Rather, we extract the mechanism from calculated electronic structure of the proposed materi- als, i.e., we can use the first-principles approach heuristi- cally. Apparently, this feature is ideal for the design of new materials. Of course, the first-principles approach is not perfect. We always have to be careful how acceptable the electronic structure calculated by the LDA, and in principle the DFT predicts only the ground state properties. So far, both above approaches have played important roles in the materials design for high-T C DMS. For example, Dietl et al. proposed p–d exchange model for the magnetism in DMS. 2) They succeeded to explain magnetic properties of (Ga,Mn)As such as Curie temperature, magnetic anisotropy, magnetic circular dichroism and so on. Encouraged by this success, they proposed the materials design for ferromag- netic DMS based on the p–d exchange model. They predicted Curie temperatures of several new DMS systems by using the mean field approximation (MFA) and gave predictions of high-T C DMS such as (Ga,Mn)N and p-type (Zn,Mn)O. However, we should notice that Dietl’s approach to describe DMS is justified only when (i) the energy position of Mn 3d level is deeper than the host valence band resulting in a Mn 2þ (d 5 ) localized moment and a weekly bound hole and (ii) the exchange interactions between Mn impurities are sufficiently long ranged. By the condition (i), the p–d exchange mechanism is justified and by the condition (ii) the MFA used to evaluate the Curie temper- ature is reasonably justified. In the case of (Ga,Mn)As, it is experimentally known that the Mn impurity forms the acceptor level with moderate binding energy E B ¼ 110 meV, and (Ga,Mn)As shows ferromagnetism for low Mn concen- trations of 5%, 1) therefore, the above two conditions are considered to be satisfied. However, the appropriateness of the model for wide gap DMS such as GaN-, ZnO-, AlN-, and  E-mail: ksato@cmp.sanken.osaka-u.ac.jp Journal of the Physical Society of Japan Vol. 76, No. 2, February, 2007, 024717 #2007 The Physical Society of Japan 024717-1