Ge-Sb-Te ALD process
The current PCRAM technology generally adopts a planar structured (so called T-shaped) cell. However, further scaling is difficult with a T-shaped cell due to the extremely low heating efficiency (only ~ 2 %) of this structure. Plugging the PC material into a small contact hole (called the confined cell structure) was suggested as the most efficient method to improve the cell efficiency and increase the integration density. However, the confined cell structure requires the PC material to be deposited with highly conformal, reproducible, and rapid growth properties, which can usually be accomplished by chemical vapor deposition (CVD) and atomic layer deposition (ALD). However, the ALD of a PC material (most typically the Ge-Sb-Te ternary material) is difficult because of the strong bond energy between the metal ions (Ge, Sb and Te) and C as well as their covalent nature in most common alkyl-based metal-organic precursors.
A genuine thermal ALD of Ge-Sb-Te phase changing material layers at low temperatures (~ 70 oC), of which the composition lies on the GeTe2 – Sb2Te3 tie line, is realized using silyl-Te and alkoxy-Ge and alkoxy-Sb without the use of any reaction gas. The strong affinity between the silyl group in the Te-precursor and oxygen ions in the Ge- and Sb-precursors provide a fluent thermal ALD reaction route to the system. This chemistry-specific ALD process was quite robust against process variations resulting in highly conformal, smooth, and reproducible film growth over a contact hole structure with an extreme geometry.
Although the research group developed ALD of Ge-Sb-Te ternary material, there were limitations in composition which lies on GeTe2-Sb2Te3 tie line. For various composition, novel Ge(II) precursors/silyl-Sb precursor was suggested. As a result, stoichiometric GeTe and Sb-rich SbxTey ALD were also developed. Thus, large area in Ge-Sb-Te ternary phase diagram can be covered by combining these ALD processes with previously developed processes.
PCRAM Cell Fabrication
In order to evaluate/improve the electrical properties of chalcogenide materials for PCRAM application, we design and fabricate various cell geometries with e-beam lithography. The advantages of e-beam lithography include high resolution of ~20nm and versatile pattern formations. We also investigate different doping materials which can be applied to ALD and enhance PCRAM device performances.
Ovonic Threshold Switch for Memory Selecotr Device
D. Adler, et al., Rev. Mod. Phys., 1978. D. C. Kau, et al., IEDM 2009
3D vertical cross-point array is a promising approach for enabling next generation ultra-high density nonvolatile memory. However, the implementation of resistive memory (e.g., RRAM and PCRAM) into dense cross-point memory array must be addressed by a selector device with a strong nonlinear IV characteristic in order to suppress sneak current paths.
Here, we focus on developing an ovonic threshold switch (OTS) which is based on chalcogenide materials for memory selector application. Since 3D vertical architecture requires a deposition method with exceptional conformity on high aspect ratio, we are interested in using atomic layer deposition to achieve this goal.
Resistive Switching Behavior or Chalcogenide without Phase Change
Among the chalcogenide materials, alloys and compounds of Ge, Sb and Te are one of the major materials for phase change memory (PCRAM.) The applications such as a non-volatile memory entirely depend on the electrical resistivity difference between crystalline and amorphous phase. This phase change effect and consequent resistivity change by polarity-independent electric pulse require high current density, which leads to high power consumption issue that has been an obstacle for commercialization of PCRAM.
Chalcogenides play an important role as the electrolytes in the electrochemical metallization (ECM) cell. Because the operation of the ECM cell does not depend on the (local) melting and subsequent amorphization of the chalcogenides, it does not require an excessively high reset switching from a low-resistance state (LRS) to a high-resistance state (HRS) current, which was the most significant problem in PCRAM. Nevertheless, the adoption of Ag or Cu as the active metallic element could pose a problem such as contamination issue. According to this concern, recent studies reported on the chalcogenide-based ReRAM cell, wherein the conducting filament is comprised of non-Cu element inside the chalcogenide alloys as a solid electrolyte.
Our research group focuses on polarity-dependent resistive switching in amorphous Ge2Sb2Te5 (GST) thin film by formation and rupture of Te-rich conducting filament, which does not involve any phase change and thus can overcome demerits of conventional PCRAM.