8:00 am – 9:30 am
Part I: Geoffrey W. Burr
Memory technology is rapidly evolving, as new nonvolatile memories (NVM) such as Phase Change Memory (PCM) and Resistive RAM (RRAM) emerge. Such memories have begun to enable Storage-Class Memory (SCM) by combining the high performance and robustness of solid-state memory with the long-term retention and low cost of conventional storage.
Simultaneously, as we seek to continue to improve computing systems, attention is turning to non-Von Neumann algorithms, including computing architectures motivated by the human brain. The same large NVM arrays that have enabled initial SCM products can also be used in non-Von Neumann neuromorphic computational schemes, with device conductance serving as synaptic “weight.” This allows the all-important multiply-accumulate operation within these algorithms to be performed efficiently at the weight data.
I will discuss these two broad classes of applications and their various sub-classes. The sub-classes of SCM will include Memory-class SCM and Storage-class SCM applications; the sub-classes for Deep Learning Accelerators include Forward Inference-only applications (e.g., ex situ training) and in situ/on-chip Training. I will describe how the various application needs drive the target device specifications for the ideal PCM devices that would be needed to enable each of these applications.
9:30 am – 10:00 am BREAK
10:00 am – 11:30 am
Part II: Hidekazu Tanaka
Basics and Applications of Electronic Phase-Change Oxides
Phase-change materials enable rapid switching between different structural phases, resulting electric properties switching. Some classes of materials are interesting on switching between different electronic/spin phases itself, such as Mott insulator-metal transition, and their electronic phase change would produce new classes of devices. This part of the tutorial will focus on electronic phase-change phenomena on transition-metal oxides. Topics included will be:
- Physics and materials science on structural and metal/insulator transition of VO2 as a prototype material
- Brief review of phase-change phenomena on transition-metal oxides (vanadate, manganite, nickelate, ferrite, ruthenate, etc.)
- External field induced electronic phase-change phenomena
- Switching/Memristive/Biology-inspired/Photonic devices based on electronic phase-change materials
12:00 pm – 1:30 pm BREAK
1:30 pm – 3:00 pm
Part III: Paul Fons, Alex Kolobov
Phase-Change Material Properties, Mechanisms and iPCM
This tutorial will introduce the attendee to the properties of phase-change materials and the relationship between structure and material properties. In the first part of the tutorial, an overview will be presented of the historical development of phase-change materials and the harnessing of the large property differences between the crystalline and amorphous states for data recording.
Emphasis will be placed on the experimental determination of alloy properties and their relationship to structure. A brief review of the kinetics of phase-change materials will then follow, including a discussion of the strong deviation from Arrhenius behavior of Ge-Sb-Te alloys that allows for switching between amorphous and crystalline phase on nanosecond timescales, but still offering years of data retention. The short timescale of the transition also opened the possibility of using ab initiodensity functional theory methods (DFT) to follow the both the transition from the liquid to glass phase as well as the crystallization process. We discuss some of the important findings of the DFT results including structural, electronic, optical and dynamical quantities and the origin of the strong property contrasts observed between amorphous and crystalline phases. The final topic of the tutorial is a discussion of interfacial phase-change memory (iPCM), a new form of phase-change material based upon a short wavelength superlattice of GeTe and Sb2Te3 that allows for a significant reduction in current compared to conventional Ge-Sb-Te alloys. A summary of the electrical and structural properties of iPCM will be presented along with several proposed models for the iPCM switching process.
- Geoffrey W. Burr, IBM Almaden Research Center
- Hidekazu Tanaka, The Institute of Scientific and Industrial Research (ISIR), Osaka University
- Paul Fons, National Institute of Advanced Industrial Science and Technology (AIST)
- Alexander V. Kolobov, National Institute of Advanced Industrial Science and Technology (AIST)