In this work the phenomenon of second phase conductive filaments is explored, with a focus on device applications. The second phase is often characterized by a significant increase in electrical conductivity, e.g. the result of the transition from an amorphous to a crystalline state in chalcogenide materials or VO2 used in modern electronic devices. A filament is formed when a region consisting solely of the second phase spans the volume of a host material. Such filaments may either be persistent or reversible depending on material parameters and operating regime underlying their formation.
When two opposing sides of the host material have electrical contacts, a conductive filament between the two will act as shunt. The presence of a shunt drastically reduces the device resistance. In memory applications a shunt that forms unintentionally will lead to data loss, while at other times it is necessary to form a shunt in order to produce a detectable change in the state of the system.
Theories are developed to describe the processes behind both types of filamentation. For each an underlying theoretical framework is presented, and then analytical methods are used to derive key results in terms of material parameters and device geometry. In doing so there is a focus on thin-film devices with inter-electrode distances that can be as small as tens of nanometers. Numerical simulations are also employed to substantiate the analytical results. Lastly, a comparison is made showing agreement with the available experimental data.
The theories presented here are rather general in nature. In order to provide examples of practical applications and to compare with experimental data, the discussion is usually developed within the context of chalcogenide glass switches. Such switches have found applications as memory devices, of which there are two distinct types: phase change memory (PCM) and threshold switches (TS).
Along with the work on second phase formation discussed above, a survey of conduction mechanisms in bulk chalcogenide glasses is also presented. In addition to the existing models of conduction, our consideration here proposes two new mechanisms: percolation conduction in a potential relief created by second phase particles, and pinhole conduction channels through very thin structures. Many of the results are equally applicable to other disordered systems. The motivation behind the review is to summarize the established physics of charge transport in bulk chalcogenide systems that can exist in parallel with the second phase filament or can precede the filament formation in chalcogenide glasses. We point out potential shortcomings in our current level of understanding and suggest specific relationships that may be obtained from future experiments which can be used to indicate which mechanism(s) are dominant.