Thousands of conductance levels in memristors integrated on CMOS

Abstract

Neural networks based on memristive devices [1][2][3] have shown potential in substantially improving throughput and energy efficiency for machine learning [4] and artificial intelligence [5], especially in edge applications. [6][7][8][9][10][11][12][13][14][15][16][17][18][19] Because training a neural network model from scratch is very costly, it is impractical to do it individually on billions of memristive neural networks distributed at the edge. A practical approach would be to download the synaptic weights obtained from the cloud training and program them directly into memristors for the commercialization of edge applications (Figure 1a). Some posttuning in memristor conductance to adapt local situations may follow afterward or during applications. Therefore, a critical requirement on memristors for neural network applications is a high-precision programming ability to guarantee uniform and accurate performance across a massive number of memristive networks. [20][21][22][23][24][25][26] That translates into the requirement of many distinguishable conductance levels on each memristive device, not just lab-made devices but more importantly, devices fabricated in foundries. High precision memristors also benefit other neural network applications, such as training and scientific computing. [23,27] Here we report over 2048 conductance levels, the largest number among all types of memories ever reported, achieved with memristors in fully integrated chips with 256 ´ 256 memristor arrays monolithically integrated on CMOS circuits in a standard foundry. We have unearthed the underlying physics that previously limited the number of achievable conductance levels in memristors and developed electrical operation protocols to circumvent such limitations. These results reveal insights into the fundamental understanding of the microscopic picture of memristive switching and provide approaches to enable high-precision memristors for various applications.Memristive switching devices are known for their relatively large dynamical range of conductance, which can potentially lead to a large number of discrete conductance levels. However, the highest number reported to date has been no more than two hundred. [20]

Quantum Transport Properties of Nanosized Ta2O5 Resistive Switches: Variable Transmission Atomic Synapses for Neuromorphic Electronics

“…36 The inset of Figure 1b shows the low-bias conductance values in the LCS and HCS of the hysteretic I(V) characteristics demonstrating fine analogue tunability in the 4G 0 −7G 0 range. Recently, thousands of stable conductance levels were demonstrated in metal oxide-based memristors in the range of 0.5G 0 −50G 0 , 52 which is a standard operation regime for redox-based filamentary RS devices. 53−59 Our measurements on several RS junctions focused on this typical conductance regime for filamentary devices, ≈1G 0 −15G 0 , where the quantum transport properties can be compared with those of atomic-sized pure tantalum nanowires.…”

Recent Advances and Future Prospects for Memristive Materials, Devices, and Systems

“…This lack of infrastructure has slowed progress in the field, particularly in comparison with the rapid advancements in digital circuit technology. Recently memristors, nonvolatile analog memory devices, have been utilized to design analog circuits including analog content-addressable memories (CAM) and field-programmable analog arrays (FPAAs), because of their fast-switching speed and multiple data storage up to thousands of levels. ,− Such memristor-based analogue circuits can reduce the footprint required for analogue circuitry while providing faster programming speeds and more accurate performance with tunability.…”

Controllability of the Conductive Filament in Porous SiOx Memristors by Humidity-Mediated Silver Ion Migration

“…Oxide-based memristors have large potential in future neuromorphic devices due to their simple structure, fast switching speed, and compatibility with complementary metal oxide semiconductor processing. 1,2 Generally, the properties of oxide-based memristors mainly depend on resistance switching between the two conductors from the high-resistance state (HRS) to the low-resistance state (LRS) under electrical control. 3,4 The switching effects in oxide-based memristors are attributable to the nanoscale conductive filaments (CF), which are formed/annihilated by controlling the thermal or external electric field generation.…”

“…Oxide-based memristors have large potential in future neuromorphic devices due to their simple structure, fast switching speed, and compatibility with complementary metal oxide semiconductor processing. , Generally, the properties of oxide-based memristors mainly depend on resistance switching between the two conductors from the high-resistance state (HRS) to the low-resistance state (LRS) under electrical control. , The switching effects in oxide-based memristors are attributable to the nanoscale conductive filaments (CF), which are formed/annihilated by controlling the thermal or external electric field generation. , The memristive switching mechanism relies on electrochemical effects and nanoionic processes involving dissolution of metal atoms from an electrochemically active electrode. The resulting migration of metal ions in an insulating matrix forms a metallic conductive bridge that is responsible for the change of the device resistance .…”