The aim of this work is to study the impact of silicon nitride deposition/treatment technologies on charge trap (CT) nonvolatile memory performances. The authors have found that the technology modifies the charge trapping behavior with a one to one correlation between write/erase and charge retention characteristics. In particular, they used rapid thermal chemical vapor deposition techniques to obtain films with different compositions, but they were not able to improve CT performances with respect to standard low pressure chemical vapor deposition (LPCVD). Besides, an in situ steam generated treatment applied to standard LPCVD silicon nitride modifies the film properties inducing a lower programming efficiency, but improving charge retention characteristics.

Aim of this work is the study the crystallization treatment of Alumina used as blocking oxide in Charge Trap memories. We have found that thermal treatment before gate deposition is able to crystallize Alumina giving better erase and write efficiency than any treatment after gate deposition, without degradation of the memory stack.

In this work a systematic comparison of TXRF, SR-TXRF, ToF-SIMS, lifetime and DLTS data of iron concentration is carried out. SR-TXRF is considered as a reference for the other techniques. Good correlations are generally obtained, though, as expected, SR-TXRF exhibits the maximum sensitivity among surface techniques. Lifetime measurements (both Elymat and SPV) show a very good correlation with SR-TXRF data. The Elymat techniques shows the best sensitivity at very low concentrations. A method is developed to elaborate Elymat data in order to quantify iron concentration when more contaminants are simultaneously present.