Demonstrating the Instability of SiC Ohmic Contacts and Drain Terminal Metallization Schemes Aged at 300 °C

The long-term thermal stability of the drain contacts of three different commercially available SiC MOSFET devices has been determined at a storage temperature of 300 °C. Existing literature suggests that, at this temperature, the nickel silicides associated with ohmic contact creation should be stable, but this was found not to be the case. Our TEM and EDX work revealed silicide phase transformations, further silicide growth and severe thermally-driven degradation of the drain contact metallization stack on top of the silicide layers. We attribute this instability and growth of the silicides to the high storage temperature and large supply of nickel atoms available from the metal stack. The nickel atoms diffuse and decompose the original silicides to enable the formation of a new low temperature Ni32Si12 phase, and slowly decompose the SiC substrate to form additional nickel silicide.

Atomic Transport in Nano-Crystalline Silicides Studied by In Situ Auger Electron Spectroscopy: Interfacial Reaction Effect

Silicide growth via reaction between a metallic film and a Si substrate has been well documented. In general, atomic transport kinetic during the growth of silicides is considered to be the same as during equilibrium diffusion, despite the reaction and its possible injection of point-defects in the two phases on each side of the interface. To date, the main studies aiming to investigate atomic transport during silicide growth used immobile markers in order to determine which element diffuses the fastest during growth and in which proportion. The quantitative measurements of effective diffusion coefficients during growth was also performed using Deal-and-Groove-type of models, however, these effective coefficients are in general not in agreement with the interdiffusion coefficients calculated using the equilibrium diffusion coefficients measured during diffusion experiments. In general, atomic transport kinetic measurements during growth and without growth are performed using different types of samples for experimental reasons. In this paper, we discuss the possible use of ultrahigh vacuum in situ Auger electron spectroscopy in order to measure the effective diffusion coefficient during growth, as well as the equilibrium self-diffusion coefficients, in the same samples, in the same experimental conditions. The first results on the Pd-Si system show that atomic transport during Pd2Si growth is several orders of magnitude faster than at equilibrium without interfacial reaction.

From Contact to Diffusion Controlled Growth of Nickel Silicides in Silicon Nanowires

Semiconducting nanowires (NW) are implemented as the active channel of field effect transistor (FET) with linear and Schottky barrier source and drain contacts. Thermally activated axial intrusion of nickel silicides into the silicon NW from pre-patterned Ni reservoirs is used in the formation of nickel silicide/silicon contacts in SiNW FETs. In the present work, the kinetics of nickel silicide axial growth in SiNWs was analyzed in the framework of the model taking into account the balance between transition of Ni atoms from the Ni reservoir to the NW surface, diffusion transport of these Ni atoms from the contact area to the interfaces between different silicides and nickel silicide/Si interface, and corresponding reactions of Ni atoms with Si and the nickel silicides formed. Simultaneous growth of mono-and nickel rich silicide was described for different kinetic and geometrical parameters of the system. Critical parameters for transition from the linear to the parabolic dependences were introduced. The model was applied to the experimental results on nickel silicide growth in SiNWs of 25÷50 nm in diameters in a temperature range of 300÷440C°. The silicide intrusions were obtained by annealing of SiNWs with pre-patterned Ni electrodes in a rapid thermal annealing machine under nitrogen atmosphere for different temperatures and times up to 120 s. In most cases the intrusions consisted of two nickel silicides, Ni-rich and mono-silicide NiSi, as was confirmed by TEM and measuring the electrical resistance of the SiNW after full silicidation. The total intrusion length, L, and particular silicide lengths, showed various time dependences, from a linear (with low growth rates (1÷4nm/s)) to a square root, diffusion-type dependence (with higher rates (10÷15 nm/s)). This behavior is well described by the model developed.