Bismuth- and aluminum-substituted Iron Garnet Film, Deposited by rf Magnetron-sputtering Using a Novel Target

Abstract

Bismuth-substituted iron garnet ®lms are promising magneto-optical (MO) recording materials for the next generation because of their high corrosion resistance and large magneto-optical effects in the short visible wavelength [1]. They have been extensively prepared by rf magnetron-sputtering from ceramic targets of constituent oxides. Preparation of the ceramic target involves a complicated process: mixing raw oxide powders by ball milling, ®ring at 800 8C for several hours, grounding synthesized powders with a mill, pressing the powders into a disk of a certain diameter and thickness, and further sintering at around 1000 8C for several hours. The ceramic target for sputtering is 2± 4 inches in diameter and 3±4 mm in thickness [2±5]. Preparing a target with this diameter and thickness by this ceramic process uses a lot of materials and is time-consuming. Furthermore, for sputter-depositing, single-phase garnet ®lms, the composition of the target should be determined through times of adjustment, each of which is ®nished by a ceramic process. Accordingly, the target preparation is quite troublesome. In fact, even after hundreds of sputtering processes, the target is negligibly consumed. Therefore, from the experimental point of view, the ceramic target for sputtering is not economical. In this letter, a novel target for sputter-depositing bismuthand aluminumsubstituted iron garnet (BiAl:DyIG) ®lm is presented. Compared with the conventional ceramic target, the novel target is readily prepared with quite small amounts of raw materials. In the experiments, a metal-chelate, sol-gel process was used to prepare the powders for target. According to the nominal composition of Bi2:2Dy1:2 Fe3:9Al0:7O12, stochiometric amounts of Bi2O3 and Dy2O3 were dissolved in hot nitric acid. Measured amounts of Fe(NO3)3(1:6M) and Al(NO3)3(1M) solution were added and stirred to form a multicomponent nitrate solution, about one ®fth of which was reserved in a separate beaker. To the remaining solution, citric acid (CA) was added to chelate, the metallic cations in the ratio of one CA formula for each matallic cation present in the solution. To avoid precipitation during the subsequent heating process, the pH of the solution was kept under 2. The resulting solution was heated on a hot plate to remove the solvent (water). Along with evaporation of the solvent, the solution gradually turned into a gel. By heating strongly, the gel turned into a black foam and ®nally, after self-combustion with liberating vapors such as H2O, CO2, NOx, a brown powder precursor was obtained. After grounding, the powder precursor was ®red at 700 8C for 1 h to obtain ®ne powders with a nominal composition of Bi2:2Dy1:2Fe3:9Al0:7O12. On the other hand, the reserved multicomponent nitrate solution was heated at 80 8C for the gradual removal of water. Due to the hydrolysis of the metal cations (Bi3‡, Dy3‡, Fe3‡, Al3‡) present in the solution, a viscous sol could be formed when the volume of the solution was half reduced. Subsequently, 5 g of sol-gel-derived Bi2:2 Dy1:2Fe3:9Al0:7O12 powders and about 5 ml viscous sol were mixed to form a paste, which was manually coated on a stainless iron wafer having a diameter of 90 mm. This coating was then dried at 150 8C for 2 h. Finally, the dried coating ®rmly adhering to the iron wafer was used as a target for sputter-depositing garnet ®lms. In the rf-sputtering process, the base pressure was 1 3 10y4 Pa, the Ar gas pressure was 0.2 Pa, the substrate temperature was 400 8C, and the rf power was 100 W. After a 20-min pre-sputtering period, the shutter was opened for ®lm deposition on the substrates of quartz slides and silicon wafers for 1h . The as-sputtering amorphous ®lm was then ®red at 700 8C for 10 min. The ®lm thickness was 360 nm, measured by a DEKTAK thickness meter. The crystal phase present in the ®red ®lm was determined by powder X-ray diffraction (XRD) using Cu-Ka radiation. Fig. 1 shows the XRD pattern of the ®red ®lm. All the peaks are ascribed to garnet phase, which means that the novel target presented in this letter for sputter-depositing garnet ®lm is feasible in the experiment. Fig. 2 illustrates a typical scan electron microscopy (SEM) of the garnet ®lm, which reveals that the ®lm is smooth, dense, crack-free, and has appreciable ®ne grain. The MO rotation spectrum of the garent ®lm shown in Fig. 3 was measured by an advanced MO spectroscopy [6]. There is a maximum MO rotation around a wavelength of 540 nm, which is bene®cial to MO recording in the short visible wavelength. For