New Percolation Model for Understanding Nonvolatile Resistance-switching Memories

Abstract

Unipolar resistance-switching (URS) phenomena refer to reversible resistance changes between two metastable resistance states driven by external voltages with the same polarities. Since the early 1960s, URS phenomena have been observed in numerous material systems, including oxides, sulfides, semiconductors, and organics. Recently, there has been a flurry of investigations into URS phenomena due to their inherent scientific interest and potential for nonvolatile memory applications. In spite of extensive efforts, the basic operation mechanisms of URS still remain to be elucidated. One of the reasons is that URS usually occurs in very disordered material systems, where defects and associated inhomogeneities can play important roles. In addition, most proposed models are applicable only at a qualitative level, so using them to understand some experimental observations is not easy. In this article, we explain a new kind of percolation model for URS phenomena. The model is material independent and can be used to demonstrate some experimental observations on a quantitative level. This percolation model, called the random circuit-breaker (RCB) network model, allows reversible changes between two metastable resistance states. This model can describe the formation of conducting paths due to the dielectric breakdown process and can make quantitative predictions especially for scaling behaviors. We explain that the collective behavior of conducting channels plays an important role in most aspects of unipolar RS, including voltage-/current-driven RS and the wide distribution of switching voltages, i.e., set and reset voltages, and abnormal and multi-level reset behaviors. More importantly, we present the experimentally observed universal scaling behaviors in URS and explain them by using a scaling theory based on the RCB network model. Percolation approaches may become a fundamental basis for understanding URS and may provide an engineering technique for the future developments.