Impact of probe placement on high frequency measurements of on-chip interconnects

Experimental data showing the impact of probe placement on the extracted per unit length R, L, and C for deep submicron cross-sectional copper interconnects are discussed. We present data for high frequency measurements up to 40 GHz that show the precise location of microwave probes on 50 micron square pads can have significant effects on the extracted per unit length R, L, and C parameters at frequencies above about 20 GHz. The effect appears to be primarily due to length offset errors from the different landing placement of the probes. Variations in landing position as small as 20 microns introduce enough phase error to produce significant effects for a 2000 micron long interconnect test structure. This is particularly problematic since most microwave probes require skating distances on the order often microns to ensure good contact.