Crested Tunnel Barriers for Fast Scalable Nonvolatile Memories

摘要

The main goal of Stony Brook work will be to move as fast as possible toward the demonstration of first practical crested barriers based on a combination of plasma-grown A1203 layers with either thermally-grown aluminum oxide or silicon dioxide (see Fig. 7 and its discussion above). Listed below are the most urgent steps we plan to make in this direction: 1) Ways to combine plasma-formed (Cr19-type) layers with thermally-oxidized (Cr05-type) layers will be explored. One possible way is to deposit a thin (-2-nm) aluminum layer on the thermally-grown oxide, and plasma-oxidize it thoroughly. If this way will not give a sufficiently smooth aluminum film (e.g., because of poor wetting), other methods of the second layer deposition will be studied. 2. ) Effects of higher oxygen exposure and post-annealing on Cr30-type layers will be studied. Either of these procedure (or their combination) may lead to films combining high barrier of annealed Cr12 layers with larger thickness. 3) Rapid post-annealing at higher temperatures (up to 500tC 2) of Cr05-type films will be explored, in order to increase the barrier height even further, while sustaining the barrier homogeneity and reproducibility. 4) Chemical and physical structure of various aluminum oxide layers and their interfaces will be understood in more detail using such methods as high-resolution TEM, scanning TEM contrast imaging, and energy-loss spectroscopy, all available through our collaboration with Dr. Yimei Zhu of the Brookhaven National Laboratory (at no cost for the AFOSR).