Reliability and Variability of 1S1R OxRAM-OTS for High Density Crossbar Integration

Robayo, D. Alfaro; Deleruyelle, Damien; Vianello, Elisa; Castellani, N.; Ciampolini, L.; Giraud, B.; Cagli, C.; Ghibaudo, G.; Nowak, E.; Molas, G.; Sassine, Gilbert; Lopez, J. Minguet; Grenouillet, L.; Verdy, Anthonin; Navarro, G.; Bernard, M.; Esmanhotto, Eduardo; Carabasse, C.

doi:10.1109/iedm19573.2019.8993439

Cited by 3 publications

(9 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This leads a targeted RRAM programming voltage of ~1.5 V to be compatible with logic CMOS to reach the best bitcell density. For more advanced nodes (1× nm nodes), the memory could be integrated in a 1S1R configuration with a backend selector [58,59] to reduce the bitcell size and target higher capacities than the 1T1R architecture allows [60].…”

Section: Filamentary Memorymentioning

confidence: 99%

Advances in Emerging Memory Technologies: From Data Storage to Artificial Intelligence

Molas

Nowak

2021

Applied Sciences

View full text Add to dashboard Cite

This paper presents an overview of emerging memory technologies. It begins with the presentation of stand-alone and embedded memory technology evolution, since the appearance of Flash memory in the 1980s. Then, the progress of emerging memory technologies (based on filamentary, phase change, magnetic, and ferroelectric mechanisms) is presented with a review of the major demonstrations in the literature. The potential of these technologies for storage applications addressing various markets and products is discussed. Finally, we discuss how the rise of artificial intelligence and bio-inspired circuits offers an opportunity for emerging memory technology and shifts the application from pure data storage to storage and computing tasks, and also enlarges the range of required specifications at the device level due to the exponential number of new systems and architectures.

show abstract

Section: Filamentary Memorymentioning

confidence: 99%

Advances in Emerging Memory Technologies: From Data Storage to Artificial Intelligence

Molas

Nowak

2021

Applied Sciences

View full text Add to dashboard Cite

show abstract

“…The Read Voltage Margin variability being the most critical parameter for reliable 1S1R operation [26], both the 1S1R stack and the applied programming conditions must be engineered to guarantee satisfactory 1S1R Read Voltage Margin and reduce its variability while avoiding device early degradation [27]. Fig.…”

Section: S1r Programming and Reading Capabilitiesmentioning

confidence: 99%

OxRAM + OTS optimization for binarized neural network hardware implementation

Lopez¹,

Hirtzlin²,

Dampfhoffer³

et al. 2021

Semicond. Sci. Technol.

View full text Add to dashboard Cite

Low-power memristive devices embedded on graphics or central processing units logic core are a very promising non-von-Neumann approach to improve significantly the speed and power consumption of deep learning accelerators, enhancing their deployment on embedded systems. Among various non-ideal emerging neuromorphic memory devices, synaptic weight hardware implementation using resistive random-access memories (RRAMs) within 1T1R architectures promises high performance on low precision binarized neural networks (BNN). Taking advantage of the RRAM capabilities and allowing to substantially improve the density thanks to the ovonic threshold selector (OTS) selector, this work proposes to replace the standard 1T1R architecture with a denser 1S1R crossbar system, where an HfO2-based resistive oxide memory (OxRAM) is co-integrated with a Ge-Se-Sb-N-based OTS. In this context, an extensive experimental study is performed to optimize the 1S1R stack and programming conditions for extended read window margin and endurance characteristics. Focusing on the standard machine learning MNIST image recognition task, we perform offline training simulations in order to define the constraints on the devices during the training process. A very promising bit error rate of ∼10−3 is demonstrated together with 1S1R 104 error-free programming endurance characteristics, fulfilling the requirements for the application of interest. Based on this simulation and experimental study, BNN figures of merit (system footprint, number of weight updates, accuracy, inference speed, electrical consumption per image classification and tolerance to errors) are optimized by engineering the number of learnable parameters of the system. Altogether, an inherent BNN resilience to 1S1R parasitic bit errors is demonstrated.

show abstract

“…Chalcogenide glasses serve as the core materials in state-of-the-art three-dimensional (3D) stacking phase-change random-access memory with 1 S 1 R integration scheme. − Reversible structural transformation between amorphous and crystalline phases in phase-change materials − enables the encoding of digital information in the resistor ( R ) cell. To access or isolate the R cell, the selector ( S ) unit in connection needs to be turned on or off accordingly, , for which reversible and volatile ovonic threshold switching (OTS) in good glass formers, for example, Ge 7 Si 18 As 35 Te 40 and Ge 40 Se 60 , creates several orders of magnitude contrast in electrical conductance, whereas maintaining glassy phases without inducing any observable crystallization. − An excellent S unit should provide a large driving current ( I on ) to trigger phase transitions in the R cell and a minimized sneak current ( I off ) to enlarge the programming nonlinearity . A proper OTS voltage ( V th ) and a low hold-on voltage ( V h ) are of necessity to enhance programming reliability and integrating capacity, , when a half-selected scheme is usually adopted to bias the dense cross-point array.…”

Section: Introductionmentioning

confidence: 99%

“…In this regard, multicomponent OTS materials containing rich cross-link elements like Ge, Si, and As are not an ideal option. Although they possess superior thermal stability and good endurance, the small I on , high V th , and sluggish switching speed would be unfavorable. − , By comparison, binary OTS materials with B/C/Si/Ge/Al-doped Te-rich compositions − exhibit faster switching speed and larger I on , but the increased I off at V th /2 and the resultant small nonlinearity (<∼10 5 ), as well as the narrower programming window (<∼0.2 to 0.3 V) due to close V h and V th , , all wait for further improvements. Among the binary OTS tellurides, amorphous C n Te m and B x Te y have the best and comparable thermal stabilities, , while the B x Te y -based S unit with a bottom electrode contact (BEC) of ∼ (30 nm) 2 showed an order of magnitude lower I off (∼1 nA) than that of C n Te m -based one (∼10 nA). , It is clear that the electrical performances of the B x Te-based S units still need to be optimized, but most importantly, the underlying mechanism of not only the B x Te y but also other binary tellurides is still unclear.…”

Section: Introductionmentioning

confidence: 99%

“…Although they possess superior thermal stability and good endurance, the small I on , high V th , and sluggish switching speed would be unfavorable. [11][12][13]15 By comparison, binary OTS materials with B/C/Si/Ge/Al-doped Te-rich compositions 18−21 exhibit faster switching speed and larger I on , but the increased I off at V th /2 and the resultant small nonlinearity (<∼10 5 ), as well as the narrower programming window (<∼0.2 to 0.3 V) due to close V h and V th , 18,21 all wait for further improvements. Among the binary OTS tellurides, amorphous C n Te m and B x Te y have the best and comparable thermal stabilities, 21,22 while the B x Te y -based S unit with a bottom electrode contact (BEC) of ∼ (30 nm) 2 showed an order of magnitude lower I off (∼1 nA) than that of C n Te m -based one (∼10 nA).…”

Section: ■ Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Electronic Excitation-Induced Semiconductor-Metal Transitions Enabling Ovonic Threshold Switching in Boron Telluride Glasses

Liu

Gong

et al. 2023

Chem. Mater.

View full text Add to dashboard Cite

Ovonic threshold switching in chalcogenide glasses is a crucial physical phenomenon behind state-of-the-art memory chip technologies. Binary tellurides are one of the emerging candidates that deliver excellent properties, notably the large driving current and fast accessing speed, for volatile selector applications; however, the underlying switching mechanism remains poorly understood. To tackle this issue, we targeted the prototypical boron tellurides, further elevating the selector performances. Via ab initio simulating the electronic excitation process in the amorphous boron tellurides, we observed reversible semiconductor-metal transitions that can reflect the switching principles of the selectors. The transient switching in conductance originates from fine structural tunings, namely, the changes of constraint surrounding the big boron clusters and conversions between covalent and hypervalent bonding schemes throughout the entire tellurium network. Our work provides much-needed atomistic insight into the ovonic threshold switching mechanisms that may enlighten the material design to enable superior selector devices.

show abstract

Reliability and Variability of 1S1R OxRAM-OTS for High Density Crossbar Integration

Cited by 3 publications

References 4 publications

Advances in Emerging Memory Technologies: From Data Storage to Artificial Intelligence

Advances in Emerging Memory Technologies: From Data Storage to Artificial Intelligence

OxRAM + OTS optimization for binarized neural network hardware implementation

Electronic Excitation-Induced Semiconductor-Metal Transitions Enabling Ovonic Threshold Switching in Boron Telluride Glasses

Contact Info

Product

Resources

About