Tungsten as the gate material

In the discussion below we limit ourselves to pure metal gates although alternatives have been proposed such as a tungsten/poly-silicon stack [Wong and Saraswat201]. It is helpful to keep the process flow as depicted in figure 8.1 in mind. 

Metal gates have already been proposed studied and used as early as 1960. Issues which were encountered include  

We will briefly discuss the solutions proposed for the issues mentioned above. 

The stability of the metal-oxide interface The main problem here is the physical and chemical stability during heat treatment (anneal or oxidation). The metals Ta,Ti, Mo and W have been studied by Kobayashi et al.200. Only the W/Si02 interface was stable for anneals up to 1200°C. Krusin-Elbaum et al.202 came to a like wise conclusion that only the W/Si02 interface is acceptable for gate applications in a comparison of Mo/SiOz and W/Si02 interfaces. This experimental result is supported by thermodynamic calculations regarding the W/Si02 interface stability. 

Mobile ion contamination Whereas tungsten gates fabricated using sputter deposition need ultra pure targets [Yamamoto et al.199], CVD-W (using 99.5% purity WF6) showed very low (0.01 ppm) mobile ion contamination [Kobayashi et al.200]. 

---