An experimental method allowing quantifying and localizing failed cells of an EEPROM CAST after a retention test.

Data retention is one of the main issues of non volatile memory reliability. The CAST (cell array stress test) is a structure of several thousand memory cells connected in parallel, used to study and to qualify data retention time of EEPROM and flash cells, respectively in erased state and programmed state. In manufacturing, CAST results are interesting to compare different process recipes in order to improve data retention, but there is no simple method to extract the threshold voltage(VT) distribution of the cells which compose the CAST. The drain current(ID) measured on a CAST is the sum of all drain currents of all cells connected in parallel. A hump was observed in the ID-VCG characteristic of a stressed CAST. This hump is caused by extrinsic cells which have lower VT and contribute of increasing the CAST drain current at low VCG. In this paper we proposed a new methodology, based only on experimental data, which allows us to localize and to quantify the extrinsic cells which compose the hump of the ID-VCG characteristic.