Scott E. Sills

A copper ReRAM cell for Storage Class Memory applications

Hybrid memory systems that incorporate Storage Class Memory (SCM) as non-volatile cache or DRAM data backup are expected to bolster system efficiency and cost because SCM promises higher density than DRAM cache and higher speed than the storage I/F. This paper demonstrates a Cu-based resistive random access memory (ReRAM) cell that meets the SCM performance specifications for a 16Gb ReRAM with 200MB/s write and 1GB/s read [1].

Voltage-Controlled Cycling Endurance of HfO x -Based Resistive-Switching Memory

Resistive-switching memory (RRAM) based on metal oxide is currently considered as a possible candidate for future nonvolatile storage and storage-class memory. To explore possible applications of RRAM, the switching variability and the cycling endurance are key issues that must be carefully understood. To this purpose, we studied the switching variability and the endurance in pulsed regime for HfOx-based RRAM. We found that the resistance window, the set/reset variability, and the endurance are all controlled by the maximum voltage Vstop, which is applied during the negative-reset operation. We demonstrate that the endurance failure is triggered by a negative-set event, where the resistance suddenly decreases during the reset. Cycling endurance is studied as a function of time, compliance current and Vstop, allowing to develop an Arrhenius-law model, which is capable of predicting device lifetime under various conditions.

Performance comparison of O-based and Cu-based ReRAM for high-density applications

The resistive switching memory (ReRAM) landscape encompasses several cell technology options. Filamentary systems that employ oxygen ion motion (O-ReRAM) or metal ion motion (M-ReRAM) and systems that employ uniform oxygen ion motion are being widely studied as potential candidates for next generation of non-volatile memory systems (NVM). While comparisons between different systems have been made at single-cell level, enabling a future NVM technology mandates an evaluation of a statistically significant population of bits. This paper presents an array-level comparison of two filamentary systems: O-ReRAM and Cu ion based M-ReRAM. The key factors for enabling a manufacturable product are compared, such as read window, noise, variability, endurance and retention.

Process integration of a 27nm, 16Gb Cu ReRAM

A 27nm 16Gb Cu based NV Re-RAM chip has been demonstrated. Novel process introduction to enable this technology include a Damascene Cell, Line-SAC Digit Lines filled with Cu, exhumed-silicided array contacts, raised epitaxial arrays, and high-drive buried access devices.

Pulsed cycling operation and endurance failure of metal-oxide resistive (RRAM)

Oxide-based resistive memory (RRAM) is under scrutiny for possible use for non-volatile storage and storage-class memory (SCM) complementing DRAM and SRAM. For SCM applications, set/reset times, variability and endurance are key concerns, which must be carefully understood to explore potential applications of RRAM. To that purpose we studied pulsed operation and endurance of oxide RRAM. We show that (i) resistance window (RW) is controlled by the negative voltage Vstop applied during reset, (ii) failure at high Vstop is due to negative set, causing filament overgrowth and RW collapse and (iii) endurance is independent of the pulse-width, which supports an Arrhenius model for endurance failure.

Postcycling Degradation in Metal-Oxide Bipolar Resistive Switching Memory

Resistive switching memory (RRAM) features many optimal properties for future memory applications that make RRAM a strong candidate for storage-class memory and embedded nonvolatile memory. This paper addresses the cyclinginduced degradation of RRAM devices based on a HfO2 switching layer. We show that the cycling degradation results in the decrease of several RRAM parameters, such as the resistance of the low-resistance state, the set voltage Vset, the reset voltage Vreset, and others. The degradation with cycling is further attributed to enhanced ion mobility due to defect generation within the active filament area in the RRAM device. A distributed-energy model is developed to simulate the degradation kinetics and support our physical interpretation. This paper provides an efficient methodology to predict device degradation after any arbitrary number of cycles and allows for wear leveling in memory array.

Challenges for high-density 16Gb ReRAM with 27nm technology

Enabling a high-density ReRAM product requires: developing a cell that meets a stringent bit error rate, BER, at low program current, integrating the cell without material damage, and providing a high-drive selector at scaled nodes. We discuss ReRAM performance under these constraints and present a 16Gb, 27nm ReRAM capable of 105 cycles with BER < 7×10−5.

Understanding pulsed-cycling variability and endurance in HfO x RRAM

Resistive switching memory (RRAM) devices based on metal oxides are receiving strong interest for future high-density stand-alone memories and storage class memories. To explore possible applications of RRAM, the set/reset variability and cycling endurance must be addressed and understood. This work shows a comprehensive study of pulsed-operated variability and endurance in HfOx-RRAM. We analysed the dependence of switching variability on operation current, voltage and pulse-width, providing guidelines to optimize set/reset distributions in oxide RRAM. The impact of pulse-amplitude and pulse-width on cycling endurance is then discussed. The results are explained by a new physics-based model for endurance controlled by defect injection from the bottom electrode during reset.

Cycling-induced degradation of metal-oxide resistive switching memory (RRAM)

Resistive switching memory (RRAM) is raising interest for future storage-class memory (SCM) and embedded applications due to high speed operation, low power and non-volatile behavior. While cycling endurance is currently well understood, the impact of cycling on switching and reliability is still a matter of concern. To that purpose we study the cycling-induced degradation of HfOx RRAM in this work. We show that the resistance of the low-resistance state (LRS), the set voltage Vset and the reset voltage Vreset decrease with cycling, which we attribute to defect generation causing enhanced ion mobility. The degradation kinetics is modelled by an Arrhenius-driven distributed-energy model. Our study allows to predict set/reset voltages after any arbitrary number of cycles and for any set/reset cycling condition.

Molecular Mobility and Interfacial Dynamics in Organic Nano-electromechanical Systems (NEMS)

The underpinnings of material properties in nanoscopic polymer systems are reviewed in light of the relaxation modes available for molecular motion. When motion is altered due to the presence of constraints, a rich variety of material and transport behaviors become apparent. On the one hand, external constraints imposed by interfaces and system boundaries generate structural and dynamical anisotropies that can propagate over device-relevant length scales. On the other hand, the ability to cater relaxation behavior through molecularly-engineered internal constraints offers a path to optimize material properties in nanoscopic systems. These two aspects are highlighted throughout the review, and their technological implications are discussed for MEMS/NEMS operations, including frictional and mechanical loading of polymer thin films. In this regard, material performance attributes and feedback for molecular designs are drawn from perturbation techniques that provide access to the energetic signatures and characteristic scales of molecular relaxation. In particular, the importance of the operating time scale in nanoscopic devices is emphasized with examples of both quasi-static and dynamic operations. Material responses have been found to change significantly when the drive velocity or loading rate was either comparable to or in excess of the characteristic molecular frequencies. Correspondingly, intrinsic molecular modes were found to either couple with the external mechanical disturbance, thus establishing channels for energy transport, or to remain passive, leading to apparent material stiffening. Findings such as these suggest that comprehensive investigations of the spatial distribution of molecular relaxation spectra in confined systems are necessary in order to match (or mismatch) molecular response times with system operating times. Along this line, this review provides examples for cognitive engineering of functional materials, with molecular structures that are tailored to system constraints and employed in nanoscale device technologies.