Atomic Layer Deposition of Ovonic Non-linear Switches for the Advancement of Microelectronic Industry Roadmap

The advent of Artificial Intelligence and Machine Learning will impose such a tall demand on the performance and energy efficiency for memory storage that the current Siliconbased technology will reach its fundamental scaling limits before the end of this decade. Thus, there is an urgent need for innovative memory technologies, which will be enabled by novel functional materials coupled with Atomic Precision manufacturing. More importantly, the future manufacturing for our most advanced memory chips should be carried out domestically within the US borders to ensuring national security. Phase Change Memory is an emerging nonvolatile technology that has demonstrated the potential for the advancement of microelectronic industry roadmap for memory technology. However, the current phase change materials and its associate fabrication techniques fall short in realizing the desired scalability, speed, and energy efficiency. Radiation Monitoring Device RMD Inc. is proposing to address this problem with the introduction of a new phase change material and its fabrication with greater atomic precision than the incumbent stateofthe art materials. During the Phase I, RMD was remarkable successful in demonstrating the growth of the novel phase change material by atomic layer deposition. The new material exhibited phase change properties at higher energy efficiency, faster speed and greater endurance than the stateoftheart materials that are currently used in the commercial devices. The growth of the novel phase change material was shown in the deep trenches measuring 20:1 aspect ratio, which demonstrated its efficacy for 3dimensional vertical integration for the first time. During Phase II, RMD will fabricate and demonstrate phase change memory devices with welldefined geometries enabled by electron beam lithography. We will demonstrate a novel device design where both the selector as well as memory resistor can be fabricated in single lithographic masking step and a single atomic layer deposition tool. This could radically reduce the cost of fabrication for both nonvolatile memory and neural network architectures. The proposed research plan, if successful, will reduce the cost of fabrication for phase change memory dramatically and simultaneously provide a huge performance boost to device scalability, endurance, and energy efficiency. The atomic precision manufacturing of phase change materials will enable inmemory computing and neuromorphic computing, which will further reduce the latency and energy consumption for data intensive computations in applications such as machine learning and artificial intelligence.