Material to system-level benchmarking of CMOS-integrated RRAM with ultra-fast switching for low power on-chip learning

Oxygen tracer diffusion in amorphous hafnia films for resistive memory

The oxygen diffusion rate in hafnia (HfO2)-based resistive memory plays a pivotal role in enabling nonvolatile data retention. However, the information retention times obtained in HfO2 resistive memory devices are many times higher than the expected values obtained from oxygen diffusion measurements in HfO2 materials. In this study, we resolve this discrepancy by conducting oxygen isotope tracer diffusion measurements in amorphous hafnia (a-HfO2) thin films. Our results show that the oxygen tracer diffusion in amorphous HfO2 films is orders of magnitude lower than that of previous measurements on monoclinic hafnia (m-HfO2) pellets. Moreover, oxygen tracer diffusion is much lower in denser a-HfO2 films deposited by atomic layer deposition (ALD) than in less dense a-HfO2 films deposited by sputtering. The ALD films yield similar oxygen diffusion times as experimentally measured device retention times, reconciling this discrepancy between oxygen diffusion and retention time measurements. More broadly, our work shows how processing conditions can be used to control oxygen transport characteristics in amorphous materials without long-range crystal order.

Analytical modelling of the transport in analog filamentary conductive-metal-oxide/HfOx ReRAM devices

The recent co-optimization of memristive technologies and programming algorithms enabled neural networks training with in-memory computing systems. In this context, novel analog filamentary conductive-metal-oxide (CMO)/HfOx redox-based resistive switching memory (ReRAM) represents a key technology. Despite device performance enhancements reported in literature, the underlying mechanism behind resistive switching is not fully understood. This work presents the first physics-based analytical model of the current transport and of the resistive switching in these devices. As a case study, analog TaOx/HfOx ReRAM devices are considered. The current transport is explained by a trap-to-trap tunneling process, and the resistive switching by a modulation of the defect density within the sub-band of the TaOx that behaves as electric field and temperature confinement layer. The local temperature and electric field distributions are derived from the solution of the electric and heat transport equations in a 3D finite element ReRAM model. The intermediate resistive states are described as a gradual modulation of the TaOx defect density, which results in a variation of its electrical conductivity. The drift-dynamics of ions during the resistive switching is analytically described, allowing the estimation of defect migration energies in the TaOx layer. Moreover, the role of the electro-thermal properties of the CMO layer is unveiled. The proposed analytical model accurately describes the experimental switching characteristic of analog TaOx/HfOx ReRAM devices, increasing the physical understanding and providing the equations necessary for circuit simulations incorporating this technology.

Enhanced Computational Study with Experimental Correlation on I-V Characteristics of Tantalum Oxide (TaOx) Memristor Devices in a 1T1R Configuration

Memristors, non-volatile switching memory platform, has recently attracted significant interest, offering unique potential to enable the realization of human brain-like neuromorphic computing efficiency. Memristors also demonstrate excellent temperature tolerance, long-term durability, and high tunability with nanosecond pulses, making them highly attractive for neuromorphic computing applications. To better understand the material processing, microstructure, and property relationship of switching mechanisms in memristor devices, computational methodologies, and tools are developed to predict the I-V characteristics of memristor devices based on tantalum oxide (TaOx) resistive random-access memory (ReRAM) integrated with an n-channel metal-oxide-semiconductor (NMOS) transistor. A multiphysics model based on coupled partial differential equations for electrical and thermal transport phenomena is solved for the high- and low-resistance states during the formation, growth, and destruction of a conducting filament through SET and RESET stages. These stages effectively represent the migration of oxygen vacancies within an oxide exchange layer. A series of parametric studies and energy minimization calculations are conducted to determine probable ranges for key material and model parameters accounting for the experimental data. The computational model successfully predicted the measured I-V curves across various gate voltages applied to the NMOS transistor in the one transistor one resistance (1T1R) configuration.

Analog Resistive Switching Devices for Training Deep Neural Networks with the Novel Tiki-Taka Algorithm

A critical bottleneck for the training of large neural networks (NNs) is communication with off-chip memory. A promising mitigation effort consists of integrating crossbar arrays of analogue memories in the Back-End-Of-Line, to store the NN parameters and efficiently perform the required synaptic operations. The "Tiki-Taka" algorithm was developed to facilitate NN training in the presence of device nonidealities. However, so far, a resistive switching device exhibiting all the fundamental Tiki-Taka requirements, which are many programmable states, a centered symmetry point, and low programming noise, was not yet demonstrated. Here, a complementary metal-oxide semiconductor (CMOS)-compatible resistive random access memory (RRAM), showing more than 30 programmable states with low noise and a symmetry point with only 5% skew from the center, is presented for the first time. These results enable generalization of Tiki-Taka training from small fully connected networks to larger long-/short-term-memory types of NN.