RBS/simulated annealing analysis of iron-cobalt silicides

Abstract

Iron and cobalt were implanted into silicon, to total doses of 5 × 10 17 /cm 2, leading to a ternary silicide layer. Different implantation energies and annealing procedures were used, leading to distinct depth distributions of the Co and Fe. The samples were analysed with Rutherford backscattering. The signals of the neighbouring elements iron and cobalt overlap due to their similar mass, leading to complex spectra requiring difficult and time-consuming deconvolution using traditional analysis packages. The combinatorial optimisation simulated annealing algorithm was used to analyse the data. The only human input required is the elements present in the sample, without any further intervention necessary. The algorithm is able to effectively deconvolute the Fe and Co signals in about 2 min using a Pentium processor running at 200 MHz, and individual depth profiles for the elements are obtained. The spectra are fitted very closely, and the results obtained were confirmed by separate X-ray photoelectron spectroscopy (XPS) and Cross sectional transmission electron microscopy (XTEM) experiments on the samples.