Substitutional cobalt in ZnO has a weak preference for antiferromagnetic ordering. Stabilization of ferromagnetism is achieved through n-type doping, which can be understood through a band coupling model. However, the description of the transition to a ferromagnetic ground state varies within different levels of band theory; issues arise due to the density functional theory underestimation of the band gap of ZnO, and the relative position of the nominally unfilled Co t 2d states. We examine these limitations, including approaches to overcome them, and explain the contradictions in previous studies, which drastically overestimate the doping threshold for magnetic ordering. DOI: 10.1103/PhysRevLett.100.256401 PACS numbers: 71.20.Nr, 71.15.Mb, 75.10.ÿb, 75.50.Pp The potential to simultaneously tune both charge and spin in solid state materials has lead to great interest in the field of spintronics [1]. The ultimate aim is a controllable room temperature semiconducting ferromagnet, which could be used in magnetoelectric and magnetotransport devices. Intrinsically magnetic semiconductors generally possess relatively low Curie temperatures (T C ); however, it has been proposed that incorporating transition metal or rare earth ions into a nonmagnetic semiconductor host lattice, forming a dilute magnetic semiconductor (DMS), may help raise T C above room temperature [2,3].In DMS materials, magnetism is influenced and controlled by the presence of charge carriers in the form of holes (GaAs:Mn) or electrons (GaN:Gd). There is still great debate over the fundamental mechanism behind the origin of the observed high temperature magnetic alignment in many of these systems. The situation is not helped by large variation in both experimental and theoretical studies [4 -8].Cobalt doped ZnO has become a focus of attention due to its reported high T C of up to 700 K and, most promisingly, the reversible cycling of FM ordering [9]. Coupled with existing optical and electrical properties of ZnO, the addition of controllable magnetism could make it a technologically essential material. It has been reported that at Co concentrations of less than 10%, the solubility is good and Co occupies a substitutional Zn site in a 2 charge state [9,10]. B . The splitting between filled e d and empty minority t 2d states for tetrahedral Co is typically on the order of 1.5-2 eV [12]. In a wide gap semiconductor such as ZnO (E g 3:4 eV), this would place the minority spin Co t 2d states within the band gap, as has been reported from both optical and magnetic measurements for ZnO:Co [13][14][15]. Surprisingly, this is not the case in previous theoretical studies [5][6][7][8].From consideration of Co 3d band coupling, FM interactions between occupied majority t 2d states results in no net energy gain due to filled bonding and antibonding combinations, while AFM interactions can result in a small energy gain (Fig. 1). Partial occupation of the empty minority t 2d states will result in a transition to FM ordering through the energy gained from the fi...