Investigation of diffusional intermixing in Si/Co/Ta system by Secondary Neutral Mass Spectrometry A. Lakatos a, * , G. Erdelyi a , A. Makovec a , G.A. Langer a , A. Csik b , K. Vad b , D.L. Beke a a University of Debrecen, Department of Solid State Physics, H-4010 Debrecen, P.O. Box 2, Hungary b Institute of Nuclear Research, Hungarian Academy of Sciences (ATOMKI), H-4001 Debrecen, P.O. Box 51, Hungary abstract Low temperature analysis of diffusion and intermixing of CoeSi systems are very important in appli- cations for microelectronics and Ultra Large Scale Integration (ULSI). In this communication a compre- hensive report has been given on degradation and diffusion processes in the Si(substrate)/Co(150 nm)/ Ta(10 nm) system. The samples were prepared by DC magnetron sputtering and were annealed in argon ambient at several temperatures ranging from 400 to 623 K for various times. The composition of the samples was investigated by Secondary Neutral Mass Spectrometry (SNMS). The degradation/inter- mixing starts with fast (grain boundary (GB)) diffusion of the Si into the Co layer. After some incubation time Si atoms appear and spread over the Co/Ta interface. This amount of Si accumulated at the Co/Ta interface acts as a reservoir for back-diffusion into the Co layer from the Co/Ta interface through the slower grain boundaries. At higher temperatures the formation of a CoeSi phase was detected at the Co/ Si and Co/Ta interface. Three different diffusion coefficients were calculated from the SNMS concentra- tionedepth profiles using “Central-gradient” (CG) and “First-appearance” methods. The observed intermixing was interpreted as a mixture of different “C-type” grain boundary diffusion processes. Furthermore, the experimental results are also compared with computer simulations modelling the grain-boundary diffusion through different grain-boundary paths. From the SNMS profiles measured at different temperatures the activation energy of the GB interdiffusion coefficients was deduced using the “CG method”. 1. Introduction Investigations of the properties of silicon/transition metal contacts are particularly crucial. Indeed transitional metals and their silicides are commonly used as interconnects, Schottky contacts and ohmic contacts [1e8] in the technology of different electronic devices. In most cases the investigations are focusing on the phase formation kinetics. In early stages of the intermixing, the atomic transport along grain boundaries (GBs) of the metal film certainly plays an important role but it is difficult to observe because the experimental methods are not sensitive enough to map the changes in the concentration profiles. We have investigated a Si/Co/ Ta system, where the Tawas used as a cap layer. In order to study the initial steps of the Si transport into the cobalt layer, we annealed our samples at 473, 553, 583, 603 and 623 K for various times. Although there are no data in the literature for Si GB diffusion in Co, other GB diffusion results suggested [9] that a considerable atomic transport along Co GBs should take place in the above temperature interval. The time evolution of the concentration profiles were followed by Secondary Neutral Mass Spectrometry. 2. Experimental The samples were prepared by DC magnetron sputtering at room temperature. The base pressure in the sputtering chamber was lower than 2 Â 10 À5 Pa. Disk-shaped Si, Co, Ta targets with diameter of 2 inches were used as sputtering sources. The 10 nm thin Ta layer was used as a cap layer in order to prevent the Co/Si bi- layer from oxidation. After removing the native Silicon oxide layer by HF acid, the metal layers were deposited by sputtering in Ar (99.999%) pressure of 5 Â 10 À1 Pa (under dynamic flow) and the sputtering power was 40 W. The sputtering rates were calculated from the layer thickness measured by AMBIOS XP-1 profilometer. The heat treatments were carried out in argon flow (99,999%) at temperatures ranging from 473 to 623 K. At 583 K isothermal * Corresponding author. E-mail address: alakatos@dragon.unideb.hu (A. Lakatos). doi:10.1016/j.vacuum.2011.07.017 Vacuum 86 (2012) 724e728