A test for lateral nonuniformities in MOS devices using only capacitance curves

Abstract

Capacitance vs voltage ( C- V) curves of metal-oxide-semiconductor (MOS) devices are affected both by interface states and by lateral nonuniformities (such as barrier-height or fixed-charge variations across planes parallel to the interface plane). If the doping profile of the device is known, then it is shown that from C- V curves alone one can distinguish between nonuniformity effects and interface state effects. Two ways to make this distinction are described and tried experimentally. One method uses the quasi-static C- V curve to determine the interface-potential, ψ, and the high-frequency C- V curve to determine the depletion width, w. Then this experimental w-ψ relation is compared with the w-ψ relation expected for the known doping profile. Any discrepancy can be due only to nonuniformities. The second (preferred) approach is to use the quasi-static and high-frequency C- V curve to determine an apparent doping profile corrected for interface-states. This apparent profile is then compared with the known profile. Again, only nonuniformities can cause the two profiles to differ. Our method of detecting nonuniformities is also useful in assessing whether it is the interface state density or the interface uniformity which is altered during a given fabrication or aging process performed upon an MOS device. If a quasi-static and a high-frequency C- V curve are measured both before and after the process under study, then “before” and “after” apparent doping profiles (or w-ψ relations) can be compared. Any difference between the “before” and “after” profiles can be interpreted in terms of an effect of the process upon device uniformity. Experimental measurements on devices known to be nonuniform show the method to be viable at least for gross nonuniformities.