1990 C R burch prize-joint award: Ion beam synthesis of epitaxial CoSi 2 layers

Abstract

Epitaxial CoSi 2 have been layers fabricated by implanting 200 keV cobalt ions, to doses between 1.5×10 17 and 7×10 17 Co + cm −2, into (100), (110) and (111) single crystal silicon substrates. The experimental results after implantation are in close agreement with those obtained from computer simulations using single crystal targets. The as implanted structures for the lower doses (1.5×10 17→3×10 17 Co + cm −2) are silicon rich and consist of a mixture of A- and B-type CoSi 2 precipitates interwoven by silicon. For the medium doses (4×10 17→5×10 17 Co + cm −2) an epitaxial layer of CoSi 2 is formed with any excess cobalt being incorporated in small CoSi inclusions, at the peak of the implanted distribution. For the highest doses (6×10 17→7×10 17 Co + cm −2) preferential sputtering of silicon at the surface of the silicide causes this region to become increasingly rich in cobalt. To accomodate this cobalt excess, grains of CoSi form above the epitaxial CoSi 2 layer. As the dose is increased up to 5×10 17 Co + cm −2 the crystallinity of the synthesised layer improves and there is a corresponding decrease in resistivity, after which any further increase in dose causes the crystallinity to deteriorate again and the resistivity to increase. These changes can be correlated with the value of x in CoSi x , i.e. when the layer is either rich in cobalt or silicon (2>×>2) the resistivity increases and there is a corresponding deterioration in crystal quality. For all doses, furnace annealing (at 1000°C, 30 min) causes a decrease in resistivity and a corresponding improvement in crystallinity as the ratio of Co:Si approaches the value for the stoichiometric compound (CoSi 2). After rapid thermal annealing (RTA) at 1000°C for 5 s. the resistivity and crystal quality of the synthesised layer for the medium dose specimen (5×10 17 cm −2) are close to those achieved by conventional furnace annealing. For the higher and lower doses the crystal quality is slightly worse and the resistivity slightly higher than those for analogous samples which have been furnace annealed.