A new process-variation-immunity method for extracting capacitance coupling coefficients in flash memory cells

Abstract

Overestimation of capacitance coupling coefficients in flash memory cells is encountered in the subthreshold slope method. By means of a two-parameters subthreshold current model I-D = I-o exp[q(V-GB - nV(SB))/nkT], a mathematical formulation of the subthreshold swing ratio in the subthreshold slope method is constructed to isolate the measurement errors caused by process variations from the errors traditionally caused by bulk capacitance coupling. To minimize the effect of process variations, a new method is developed based on the model. In this method, the control gate voltage shift due to weak body effect is measured in flash memory cells in subthreshold, while the corresponding slope factor n is adequately deduced from threshold voltage versus source-to-substrate bias measurement in dummy devices. The corrected capacitance coupling coefficients show large improvements compared to the design values, and the updated errors are found to be close to that caused solely by bulk capacitance coupling. The method is also fast since only a small source-to-substrate bias of 0.1 V is needed for implementation of weak body effect, and thereby it can be used as an in-line monitor of capacitance coupling coefficients.