No abstract

The grain size and surface morphology of sputtered Au films are studied by x-ray diffraction and atomic force microscope. For as-deposited samples the grain growth mechanism is consistent with the two-dimensional (2D) theory, which gives relatively low diffusion coefficient during deposition. The annealing process demonstrates the secondary grain growth mechanism in which the thickness dependence of grain boundary energy plays a key role. The surface roughness increases with the increase of grain size.

The superconducting transition temperature and grain size of dc sputtered Nb films are systematically investigated. The results show that the superconductivity is closely related to the grain size, rather than to the scattering strength of electrons or the surface layer proximity effect of the films. dc sputtering, superconductivity, Nb film, grain size effect, localization effect, proximity effect It almost becomes a common sense that the superconductivity of a thin film is progressively destroyed when the thickness of the film reduces to well below the coherence length, yet the underlying mechanism of the phenomenon is still a standing problem, in spite of the continuous experimental efforts on different systems [1] . In the present work, a series of dc sputtered high quality Nb films are investigated. The results show a close correlation between the superconducting transition and the grain size of the films.Nb films are easily contaminated by reactive gases, especially oxygen, during the depositing process. Two special measures are taken to ensure the high quality of the samples. First, the magnetron sputtering system is prevacuumed to better than 2×10 −7 Pa. Then the pump is switched off and a built-in Non-Evaporable Getter (NEG) 1) is activated. Thanks to the very low leakage of the system and the usefulness of NEG in selective absorption of oxygen, the system can be kept in a nearly "oxygen free" situation in the whole sputtering process. Secondly, a "static" Ar atmosphere is established for sputtering. In contrast to the usual way where a flowing gas is used, the present method is proved to be useful in avoiding contamination by flowing gas and getting a stable sputtering rate. Experiments and resultsAn Nb target with a purity of 99.999% is used. The film substrates are high-polished Si covered by a SiO 2 layer of about 500 nm. The substrates are precleaned in an ultrasonic cleaner by sequentially using acetone, alcohol and distilled water. Before sputtering, the substrates are further baked in-situ to 150℃ for about 7-8 h to degas the surfaces.Superconducting transition temperature, normal state resistivity and its temperature dependence, and XRD are measured for Nb films of thicknesses from 740 to 2.5 nm. The high quality of the samples is demonstrated by comparing their resistivities with MBE films [2] and sputtering samples in UHV base vacuum [3,4] (Figure 1). We see that the resistivities of all these samples can roughly be normalized to follow a single thickness dependence, except for the data of the e-beam films reported in ref. [5]. The superconducting transition width ΔT c gives the information of how the homogeneity of the film is. The results are plotted in Figure 2. In most cases, ΔT c of our samples are narrower than that of the MBE samples [2] .