Physical Electronics of Multi-Dielectric Microstructures

这项研究关注的是基础物理，利用量子力学隧穿载流子进入深能级的多介质微结构的电子和材料特性，存储陷阱，以创建非易失性半导体存储器件。正在研究的特殊器件是多晶硅栅阻挡氧化物-氮化硅-隧道氧化物-硅(SONOS)多晶硅-介电微结构。陷阱位于氮化硅中，本研究探索了电荷注入的物理机制，运输和存储是在“规模化”的设备结构中完成的。这些陷阱的起源和控制是该计划的一个重要目标，也是检查电子和空穴传输的一个重要方面。界面界面采用新的电荷分离技术。尤其是，我们建议研究一种新的SONOS器件结构，该结构使用多晶硅存储区域来增加低编程时的电荷存储电压。这些SONOS结构显示了低电压编程的巨大前景，并且这项研究的结果将影响诸如高密度、非易失性存储器、专用集成电路、晶片规模集成和神经网络。尤其是低功耗，电可重编程Sonos模拟电导是一种用于神经网络实施的吸引人的“可修改”突触。

This research is concerned with the basic physical, electronic and material properties of multi- dielectric microstructures which use quantum mechanical tunneling of carriers into deep-level, storage traps to create a nonvolatile semiconductor memory device. The particular device under investigation is the polySilicon gate-blocking Oxide-silicon Nitride- tunnel Oxide-Silicon (SONOS) multi- dielectric microstructure. The traps are located in the silicon nitride and this research explores the physical mechanisms by which charge injection, transport and storage is accomplished in 'scale' device structures. The origin and control of these traps is an important objective in the program as well as the examination of electron and hole transport across interfacial boundaries with novel charge-separation techniques. In particular, we propose to investigate a new SONOS device structure which uses a polysilicon storage region for increased charge storage at low programming voltages. These SONOS structures show great promise for low voltage programming and the results of this research will impact such areas as high density, nonvolatile memories, application specific integrated circuits, wafer scale integration and neural networks. In particular, the low power, electrically reprogrammable SONOS analog conductance is an attractive 'modifiable' synapse for neural network implementation.