A proper approach to characterize retention-after-cycling in 3D-Flash devices

We propose a procedure to evaluate retention-after-cycling in 3D-Flash devices. Proper comparison of retention transients requires the initial charging level to be as close as possible, but P/E cycling results in serious I_D-V_G degradation, preventing a consistent extraction of the threshold voltage. We introduce a test where a relaxation phase is added after cycling, consisting in baking samples for 24 hours at 200°C. This relaxation appears to anneal interface traps and to remove locally accumulated charge, restoring similar shape of I_D-V_G curves before and after cycling, hence allowing a proper comparison of retention of fresh and stressed devices.