Threshold voltage in heterogeneous chalcogenide semiconductors

Abstract

The change from high to low impedance of an amorphous semiconductor switch takes place at a critical field identified by the threshold voltage. The presence of conductive regions in the switching path reduces this threshold voltage. This change has been modelled by calculating the minimum resistive path length obtained when randomly placed uniform conductive spheres grow in a resistive matrix. The computed results are rather similar to those given by simply placing the spheres on a square or cubic lattice and are consistent with electrical observations in switching experiments.

Conduction and switching in InxSe1−x

“…The structure of the material in the second case could be considered as a heterogeneous semiconductor in which crystalline regions were randomly arranged in the amorphous matrix. The small variations in the high-resistance state for the different compositions could be attributed to the fact that below a critical concentration of the conductive phase, the effective resistivity remained high, in good agreement with Stevenson's model [5]. The crystallites were assumed to grow symmetrically a t a uniform rate from randomly situated sites, and to have a negligible resistivity compared to the matrix.…”

“…The change from high to low resistance during switching usually takes place at a critical field identified by the threshold voltage. The presence of the conductive regions in the switching path reduces the value of the threshold voltage [5]. It is now well accepted that a unique mechanism probably does not exist [6].…”

Some aspects of reliability in amorphous chalcogenide memory switches

“…After the first 100 or so set/reset cycles the scatter of the lock-on times became much wider, with a corresponding decrease in the reliability. In terms of an earlier model (Steventon 1974) this data implies the reset current has caused an increase in the variation of the local filament temperature. The use of trains of reset pulses also alters the lock-on time.…”

“…The distribution of the lock-on time is generally log-normal with a median which decreases with increasing set current (Steventon 1974). A typical distribution is shown in figure 3.…”

“…Tellurium-rich crystallites begin to separate in the filament with some consequent redistribution of conductivity and temperature, and completion of a continuous crystalline path between the electrodes is often shown by a small impedance discontinuity at the 'lock-on' time, which indicates the onset of a stable low resistance condition. The growth of the filament has been treated earlier (Steventon 1974, Steventon andBond 1974). Continued application of the set pulse will allow the filament to grow to a diameter determined by the set current.…”