"Application of Nocrystals Embedded in High-k Gate Dielectric to Improve Programming Speed and Retention Period"

Akeed A Pavel, University of Missouri-Columbia

June 9th (Wednesday), 12:00pm
HFH 4164

The relentless effort to scale down floating gate based non-volatile memory (NVM) structures is being challenged by the degradation of the insulation property of thin SiO2 layers, which are conventionally used as the gate dielectric material. Making the oxide layer thicker, on the other hand, reduces the tunneling current through it, and adversely affects the programming speed. Improving programming speed without affecting the data retention capability has therefore instigated significant research effort in recent time and replacement of SiO2 by other materials with higher dielectric constants (high-k) and lower conduction band offset with Si have been proposed.

With the recent advancements in fabrication technology, metal nanocrystals (NCs) have emerged as the most promising candidate for the discrete charge storage nodes in the floating-gate-based non-volatile memories. Metal NCs, compared to Si NCs have stronger influence on enhancing electric field in gate stack, which improves the programming speed. Furthermore, due to their dissimilar workfunctions compared to Si, metal NCs produce asymmetric potential barrier in favor of retaining the stored charge for longer period of time. Although integration of metal NCs with high-k dielectrics seems to be an obvious choice, recent experimental investigation demonstrated that the efficacy of metal NCs may deteriorate when embedded in certain high-k dielectrics through the effect of Fermi level pinning at the metal NC/dielectric interface.

This work is devoted to the modeling of charge transport in metal NC based NVMs to optimize the combination of high-k/metal NCs for maximum programming speed and retention time. A two-step computational scheme has been developed to quantify the quantum mechanical tunneling currents in a NC based memory structure. First the potential profile in the dielectric is generated considering the geometry and material properties of the NCs and the dielectric. The tunneling currents are then computed within a self consistent process, which is based on the Green’s function formalism and the concept of quantum mechanical wave impedance. Along with quantifying the physical phenomena that makes metal NCs superior in improving programming speed and retention time, the model provides an effective guideline for choosing metal NC/high-k dielectric material combination by taking the effect of Fermi level pinning into account.

About Akeed A Pavel:

Akeed A. Pavel has been pursuing PhD degree in Electrical Engineering at the University of Missouri-Columbia, since spring 2006. His general research interest includes physics and modeling of novel semiconductor devices like metal nanocrystal non-volatile memory, nanowire transistor, and photovoltaic devices.

Hosted by: Professor Dmitri Strukov