Remarkable room-temperature ferromagnetism was observed in undoped TiO 2 , HfO 2 , and In 2 O 3 thin films. The magnetic moment is rather modest in the case of In 2 O 3 films on MgO substrates ͑while on Al 2 O 3 substrates, it is negative showing diamagnetism͒ when the magnetic field was applied parallel to the film plane. In contrast, it is very large in the other two cases ͑about 20 and 30 emu/ cm 3 for 200-nm-thick TiO 2 and HfO 2 films, respectively͒. Since bulk TiO 2 , HfO 2 , and In 2 O 3 are clearly diamagnetic, and moreover, there are no contaminations in any substrate, we must assume that the thin film form, which might create necessary defects or oxygen vacancies, would be the reason for undoped semiconducting or insulating oxides to become ferromagnetic at room temperature.

Room-temperature ferromagnetism (FM) has been observed in laser-ablated ZnO thin films. The FM in this type of compound does not stem from oxygen vacancies as in the case of TiO 2 and HfO 2 films, but from defects on Zn sites. Magnetization of very thin films is much larger than that of the thicker films, showing that defects must be located mostly at the surface and/or the interface between the film and the substrate. Results on Fe-doped ZnO and Mn-doped ZnO films reveal clearly that the metal-transition doping does not play any essential role in introducing the magnetism in ZnO.Since 2000, many research groups have tried to look for room-temperature ferromagnetic compounds, which can be obtained by doping transition-metals into semiconducting and insulating oxides such as ZnO, TiO 2 , SnO 2 , In 2 O 3 and HfO 2 , and the main cause was supposed to be the doping [1-5]. However, very recently, reports of Coey's group about the magnetism observed in HfO 2 thin films [6, 7], our findings for ferromagnetism (FM) in laser-ablated HfO 2 , TiO 2 and In 2 O 3 thin films [8], and results on TiO 2 films of Yoon et al [9] have raised a serious question if a transition-metal doping indeed plays any key role in introducing FM in those oxides. In fact, results on TiO 2 and HfO 2 systems suggest that magnetism originates from oxygen vacancies and/or defects [8,10]. As for ZnO system, some groups have also reported that defects could also tune FM [11,12], and a perfect crystallinity might destroy the FM, or having more oxygen could degrade the magnetic ordering [13]. Since ZnO is a semiconductor that has been widely used in spintronics applications, it is important to know whether the undoped ZnO can be ferromagnetic or not. Additionally, we need to re-judge carefully the role that a transition-metal indeed can play in introducing FM in this specific system. In order