DOI: 10.1002/adma.200602222 Control of Defects and Magnetic Properties in Colloidal HfO 2 Nanorods** By Einat Tirosh and Gil Markovich* It has been known for a long time that not only open-shell transition metal impurities, but also open-shell point defects may have a magnetic moment. [1–3] Recently, Sawatzky and co- workers suggested that such magnetic moments may interact and lead to collective magnetism in CaO due to Ca vacan- cies, [4] but so far without experimental proof. More recently, Coey and co-workers have observed ferromagnetism in thin HfO 2 films produced by pulsed laser deposition, [5] suggesting that it originated in point defects in the oxide lattice. Hong et al. have also detected ferromagnetic behavior in similar films. [6] Conversely, Rao et al. did not observe ferromagne- tism in undoped HfO 2 films also prepared by pulsed laser de- position under various conditions, [7] and Abraham et al. have shown that while HfO 2 films grown by metalorganic chemical vapor deposition were purely diamagnetic, magnetism could develop as a result of metallic contamination of the samples. [8] A recent theoretical paper explained the factors at play; [9] to see defect-induced magnetism one needs to fulfill the fol- lowing conditions: i) The equilibrium defect charge configura- tion needs to have a magnetic moment (some stable charge states may have a closed shell, e.g., doubly charged oxygen vacancies). ii) The density of such defects has to exceed the percolation threshold, i.e., there needs to be a defect state connectivity across the crystal. This requires knowing the “in- teraction radius” of each defect, and the percolation threshold for the lattice type considered. For CaO the computed vacancy interaction radius was the 4th nearest neighbor with the percolation threshold of the NaCl lattice. Accordingly, the estimated vacancy concentra- tion required for obtaining ferromagnetism in CaO was ca. 10 21 vacancies per cm 3 , or about 5 % Ca vacancies. This ex- ceeded by 3 orders of magnitude the computed equilibrium concentration of vacancies, calling for strong nonequilibrium effects that would be needed to achieve ferromagnetic cou- pling in CaO. Vacancies in HfO 2 may have a similar interaction radius and the HfO 2 lattice has a higher percolation threshold than the CaO lattice, so an even higher vacancy concentration would probably be needed to obtain ferromagnetism in HfO 2 . [10] This explains why the observation of ferromagnetism in HfO 2 has been elusive, as a very large concentration of va- cancies, far exceeding equilibrium densities, may be needed. Thus, near-equilibrium growth methods are inappropriate for this purpose, [8] but far-from-equilibrium preparation methods could produce crystals with the high concentration of defects required. Colloidal nanocrystal growth techniques are ideally suited for the introduction of large defect concentrations due to the relatively low synthesis temperatures and the large nanocrys- tal surface area which is inherently defective. Colloidal nano- crystal syntheses may offer a variety of surface capping agents or treatments to control the density of surface defects, un- paired electrons etc.. Protective organic monolayers that are usually adsorbed to these surfaces in the process of the nano- crystal synthesis may have a wide variety of interactions with metal or oxygen ions at the surfaces and were shown to affect photoluminescence properties of semiconductor nanocrys- tals [11] as well as other properties such as ferromagnetism in magnetic oxides. [12] In this work we have synthesized HfO 2 nanorods by the method of Tang et al. [13] (synthesis I) and then modified the conditions of this synthesis (synthesis II). By modifying the synthesis conditions we have obtained control over the defect concentrations in the HfO 2 nanorods and were thus able to obtain ferromagnetic hafnium oxide nanorods. In both synthesis procedures HfO 2 nanorods were formed, as shown in the transmission electron microscopy (TEM) im- ages in Figure 1. The nanorods produced by synthesis I were (9.2 ± 1.4) nm in length and (3.5 ± 0.8) nm in diameter (Fig. 1a), while those prepared by synthesis II had a larger aspect ratio and were (10.5 ± 1.3) nm in length and (2.6 ± 0.4) nm in diameter (Fig. 1b). The X-ray diffraction (XRD) patterns confirmed the mono- clinic crystalline phase of the nanorods produced in both syn- theses (see Supporting Information). The lattice parameters of both synthesis products were larger than the bulk HfO 2 val- ues. High resolution TEM (HRTEM) revealed larger variations in lattice spacing and orientations within the particles pro- duced by synthesis II, which indicated that they were more defective than the particles formed by synthesis I. Figure 2 displays representative HRTEM images of a nanorod from COMMUNICATION 2608 © 2007 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim Adv. Mater. 2007, 19, 2608–2612 – [*] Dr. G. Markovich, E. Tirosh School of Chemistry Beverly and Raymond Sackler Faculty of Exact Sciences Tel Aviv University Tel Aviv 69978 (Israel) E-mail: gilmar@post.tau.ac.il [**] The authors gratefully acknowledge the help of Dr. Alex Zunger throughout this work and thank Dr. Yossi Lereah for his help with HRTEM analysis. This research was supported by The Israel Science Foundation grant no. 779/06 and the James Frank program. Sup- porting information is available online from Wiley InterScience or from the authors.