Ovonic threshold switching combines the phase-change memory element with an OTS selector device, both based on a chalcogenide material.
Chalcogenide phase-change materials can be quickly and reversibly switched between an amorphous and a crystalline phase with very different optical and electrical properties. This unconventional property combination led to their use first in rewritable optical storage products and currently in advanced non-volatile resistive memories.
In addition, the unique phenomenon called ovonic threshold switching (OTS) has recently led to a major technological breakthrough in the field of memories. Indeed, it is the basis of the advent of high-density 3D non-volatile resistive memories, thanks to the combination of a phase-change memory (PCM) element and an OTS selector device, both based on a chalcogenide material (Figure 1).
When the applied voltage exceeds Vth, the OTS material experiences a spectacular drop of resistivity, enabling high current flow (Figure 2). The latter permits programming and reading the selected memory cell within the 3D cross-point network without undesired programming of adjacent cells.
This is done by limiting the leakage current to the Ileak value using a Vth/2 polarization strategy. When this high voltage is removed, the OTS material recovers its highly-resistive state. These new memory architectures pave the way for the realization of storage class memory (SCM) and neuromorphic circuits inspired by human brain function.
However, more than 50 years after its discovery by Stanford Ovshinsky in 1968, the origin of the OTS effect had still not been explained or fully understood. We now have the explanation. In retrospect, it’s apparent that it was a unique non-linear conductivity behavior observed in some chalcogenide glasses during the application of high electric fields.
New OTS materials
CEA-Leti scientists, in collaboration with Jean-Yves Raty from Liège University in Belgium and within the framework of an agreement between CEA and the university, have made it possible to finally elucidate the physical switching mechanism of OTS. This unique result is derived from an in-depth study of new OTS materials based on Ge-Sb-Se-N (GSSN) glassy thin films.
The key was coupling advanced experiments of electrical, optical, and X-ray absorption spectroscopy measurements with an innovative method for the simulation of ab initio molecular dynamics (AIMD). The XAS experiments were performed at the European synchrotron radiation facility (ESRF) on the Italian LISA beamline. This work has been the subject of several publications in top scientific journals, including a recent article in the American journal Science Advances.
Despite their wide adoption over the years in PCMs and other devices, the basic understanding of some specific behaviors of chalcogenide materials, which has a major impact on device performance, remains the domain of science. This quest for understanding is best illustrated by the longstanding debate about and search for an explanation of the unique contrast of electronic properties between the amorphous and crystalline phases of phase-change materials.
The origin of the latter has been very recently revisited by using modern ab initio simulation approaches that go beyond the usual one-electron description obtained with the density functional theory (DFT). Indeed, phase-change materials are in a class of materials that show particular properties that were previously attributed to resonant bonding.
How OTS works
The corresponding materials, called incipient metals, and their bonding, which is known as metavalent bonding, differ significantly from covalent materials or materials with resonant bonding, such as graphene or benzene. Again, in the vitreous state of chalcogenide materials, some specific and uncommon behaviors have also remained unclear since their discovery 50 years ago.
While the sub-threshold conduction mechanisms in chalcogenide glasses are now well understood, the underlying physical mechanism involved in OTS is still much debated. The OTS mechanism is now revealed to involve subtle atomic rearrangements during the application of high electric fields in thin layers of OTS chalcogenide glasses.
These are manifested by the alignment of certain chemical bonds and the appearance of local atomic structural patterns. That’s reminiscent of the new metavalent bond (MVB) recently described in the crystalline phase of chalcogenide phase-change materials.
These MVBs are thus responsible for a huge change in the electronic state density (eDoS) accompanied by a dramatic decrease in the resistivity of the material and the appearance of a metallic behavior (Figure 3). Such a fundamental understanding now allows scientists to define design rules to further optimize the OTS properties of amorphous chalcogenides.
The huge increase of conductivity upon electric field application observed at threshold switching in OTS materials can be explained by the strong delocalization of the electronic states around EF wave functions—shown as blue isocurves. Although the global topology of the network is conserved upon excitation, important changes can be observed on the proportion of quasi-aligned bonds giving rise to structural motifs reminiscent to MVBs, as illustrated by the small molecules drawn at the top of the figure.
These MVBs are similar to those found in the crystalline state of chalcogenide phase-change materials. They differ in that they occur and remain stable only during application of a high-electric field in the OTS glasses, in contrast with phase-change materials for which they are easily stabilized in the crystalline state.
This novel OTS model, for the first time, establishes the common link between chalcogenide materials belonging to the phase-change materials family—used in PCM—and those belonging to the OTS family, used as selectors. In both systems, the MVB mechanism is responsible for the unique properties that led to the recent breakthrough of non-volatile resistive memories through 3D integration. The main difference between PCM and OTS materials lies in their ability to stabilize the MVB mechanism and the energy barrier to crystallization.
This article was originally published on EDN.
Pierre Noé is a research scientist and senior expert at CEA-Leti. His field of expertise is chalcogenide materials science for electronic and photonic applications.