Ilya V. Karpov

A stackable cross point Phase Change Memory

A novel scalable and stackable nonvolatile memory technology suitable for building fast and dense memory devices is discussed. The memory cell is built by layering a storage element and a selector. The storage element is a Phase Change Memory (PCM) cell [1] and the selector is an Ovonic Threshold Switch (OTS) [2]. The vertically integrated memory cell of one PCM and one OTS (PCMS) is embedded in a true cross point array. Arrays are stacked on top of CMOS circuits for decoding, sensing and logic functions. A RESET speed of 9 nsec and endurance of 106 cycles are achieved. One volt of dynamic range delineating SET vs. RESET is also demonstrated.

Fundamental drift of parameters in chalcogenide phase change memory

We present the data on temporal (t) drift of parameters in chalcogenide phase change memory that significantly complement the earlier published results. The threshold voltage Vth and the amorphous state resistance R are shown to drift as ΔVth∝v ln t and R∝tα in broad intervals spanning up to nine decades in time; the drift coefficient v depends on glass parameters and temperature, but does not depend on device thickness. We have demonstrated that drift saturates at long enough times that can be shorten with temperature increase. All available data on drift dynamics are fully consistent with the classical double-well-potential model, which gives simple analytical expressions for the observed temporal dependencies including numerical parameters.

Nucleation switching in phase change memory

The authors propose a simple physical model of threshold switching in phase change memory cells based on the field induced nucleation of conductive cylindrical crystallites. The model is solved analytically and leads to a number of predictions including correlations between the threshold voltage Vth and material parameters, such as the nucleation barrier and radius, amorphous layer thickness, as well as Vth versus temperature and switching delay time. The authors have carried out verifying experiments, and good agreement is achieved.

Electrical conduction in chalcogenide glasses of phase change memory

Amorphous chalcogenides have been extensively studied over the last half century due to their application in rewritable optical data storage and in non-volatile phase change memory devices. Yet, the nature of the observed non-ohmic conduction in these glasses is still under debate. In this review, we consolidate and expand the current state of knowledge related to dc conduction in these materials. An overview of the pertinent experimental data is followed by a review of the physics of localized states that are peculiar to chalcogenide glasses. We then describe and evaluate twelve relevant transport mechanisms with conductivities that depend exponentially on the electric field. The discussed mechanisms include various forms of Poole-Frenkel ionization, Schottky emission, hopping conduction, field-induced delocalization of tail states, space-charge-limited current, field emission, percolation band conduction, and transport through crystalline inclusions. Most of the candidates provide more or less satisfact...

Phase Change Materials

Phase change materials can be switched rapidly and repeatedly between amorphous and crystalline phases, which differ distinctly in their optical and electrical properties. This combination of properties is utilized to store information in rewritable optical storage media and in emerging phase change memory technology. This article describes the physical properties of phase change materials such as Ge 2 Sb 2 Te 5 and relates these properties to specific structural and bonding characteristics. Electrical conduction and switching, which are relevant for phase change memory operation, are explained from a physical perspective. Phase change memory device integration and technology development are discussed, including aspects of access device selection and integration.

A unified model of nucleation switching

A unified field-induced nucleation model provides a common mechanism for switching in chalcogenide phase change memory and related devices of arbitrary thickness. We employ the model to derive equations for the threshold and holding voltages in terms of material parameters and device thickness, which are in excellent agreement with previous measurements and our data.

SET to RESET Programming in Phase Change Memories

Experimental data on details of SET (crystalline) to RESET (amorphous) transition are presented for Ge2Sb2Te 5 (GST) nonvolatile memory cell. It is shown that the main source of heat for a SET to RESET transition is the GST bulk and interface regions instead of the contacting electrode. A small-contact-area electrode is used primarily to supply current into and minimize heat loss from the chalcogenide. Increasing bottom contact resistivity offers a scaling path for RESET current with no electrical penalties

The Role of Interfaces in Damascene Phase-Change Memory

Phase change memory (PCM) research has largely focused on bulk properties to evaluate cell efficiency. Now both electrical and thermal interface resistances are characterized and shown to be critical for understanding power in a novel damascene-GST cell. Interfaces reduce reset power 20% and reset current 40% and allow reset current to scale faster than it would without interfaces.

Crystal nucleation in glasses of phase change memory

We propose a theory of field induced crystal nucleation in disordered glass structure applicable to chalcogenide phase change memory. In the region of symmetry breaking strong electric fields, the nucleation is dominated by cylinder shaped particles with bias dependent nucleation barriers. Statistical fluctuations in microscopic structure of a glass translate into probabilistic distributions of induction times and threshold voltages having respectively log-normal and normal shape. These distributions are exponentially sensitive to the applied voltage, temperature, and material parameters.

Conductive path formation in glasses of phase change memory

We present a model of data retention for phase change memory devices in which the active medium is a thin layer of chalcogenide glass. Data retention capability is compromised when a crystalline path is spontaneously formed in the glassy host, essentially shunting the device. We determine the probability and statistics of device failure for systems in which the crystalline volume fraction is below the critical volume fraction of percolation theory. In that regime, we show that rectilinear crystalline path formation is favored and we determine the criteria for when such paths dominate over the typical percolation cluster scenario. Our analytical approach, based on modeling the formation of such paths in terms of a half-space random walk, leads to closed form expressions that relate data retention characteristics to device parameters. The model is used to examine the effects of device geometry, temperature, and external fields. The temporal statistics of device reliability are also considered for several fa...