1.1 ORIGINS AND ALUMINIUM METALLISATION

Paul Totta, IBM Fellow, wrote a marvellous report in 1968 on metallisation [1] that sets the stage for the present, over three decades later. The introduction reads [1]:

The design trend in monolithic integrated circuits has moved rapidly in the direction of using smaller, faster devices and circuits in ever-increasing numbers on a single silicon chip. To further shorten the electrical path between active elements, much of the wiring, which had formerly been done on modules or printed circuit cards, is now in a second or third metallisation level on the chip. One of the present design limitations for the miniaturization trend is the device thin film metallurgy. The extent of reduction of thin film conductors is, in turn, restricted partly by intrinsic metal properties such as electromigration capability or conductivity, and also by processing limitations such as the ability to shape the film by photolithography and subtractive etching.

In this case, aluminium (Al) won over Ag- and Cu-based systems. Even when Al fell out of favour due to electromigration problems, the addition of Cu, Cu-doped Al , fixed the electromigration problem.

The use of Al as metallisation in contact with Si, ran into problems due to pit formation in the Si following heat treatment at temperatures between 400 and 450 C.

1.2 PIT FORMATION AND Al SPIKES

To make electrical connections to the Si devices, a layer (usually 1 2 ?m thick) of metal is deposited on a patterned Si wafer.