Search

With remarkable electrical and optical switching properties induced at low power and near room temperature (68 °C), vanadium dioxide (VO2) has sparked rising interest in unconventional computing among the phase-change materials research community. The scalability and the potential to compute beyond the von Neumann model make VO2 especially appealing for implementation in oscillating neural networks for artificial intelligence applications, to solve constraint satisfaction problems, and for pattern recognition. Its integration into large networks of oscillators on a Silicon platform still poses challenges associated with the stabilization in the correct oxidation state and the ability to fabricate a structure with predictable electrical behavior showing very low variability. In this work, the role played by the different annealing parameters applied by three methods (slow thermal annealing, flash annealing, and rapid thermal annealing), following the vanadium oxide atomic layer deposition, on the formation of VO2 grains is studied and an optimal substrate stack configuration that minimizes variability between devices is proposed. Material and electrical characterizations are performed on the different films and a step-by-step recipe to build reproducible VO2-based oscillators is presented, which is argued to be made possible thanks to the introduction of a hafnium oxide (HfO2) layer between the silicon substrate and the vanadium oxide layer. Up to seven nearly identical VO2-based devices are contacted simultaneously to create a network of oscillators, paving the way for large-scale implementation of VO2 oscillating neural networks. [ABSTRACT FROM AUTHOR]

Analog memory presents a promising solution in the face of the growing demand for energy-efficient artificial intelligence (AI) at the edge. In this study, we demonstrate efficient deep neural network transfer learning utilizing hardware and algorithm co-optimization in an analog resistive random-access memory (ReRAM) array. For the first time, we illustrate that in open-loop deep neural network (DNN) transfer learning for image classification tasks, convergence rates can be accelerated by approximately 3.5 times through the utilization of cooptimized analog ReRAM hardware and the hardware-aware Tiki-Taka v2 (TTv2) algorithm. A simulation based on statistical 14 nm CMOS ReRAM array data provides insights into the performance of transfer learning on larger network workloads, exhibiting notable improvement over conventional training with random initialization. This study shows that analog DNN transfer learning using an optimized ReRAM array can achieve faster convergence with a smaller dataset compared to training from scratch, thus augmenting AI capability at the edge. [ABSTRACT FROM AUTHOR]

In this study, we present a significant advance in the growth of epitaxial GeTe-Sb2Te3 alloys on Si(111)-(√3 × √3)R30°-Sb reconstructed surface by means of Molecular Beam Epitaxy. By employing X-ray diffraction and Raman spectroscopy, we show how phase, composition, and vacancy ordering can be tailored by tuning the growth parameters such as Ge and Te elemental fluxes as well as the substrate temperature. The effect of each parameter on the GeSbTe composition and phase is also discussed. A very surprising finding is that GeSbTe alloys are more ordered when grown at lower substrate temperatures. In addition, it was possible to fabricate ordered GeSbTe even for substrate temperatures as low as 120 °C. In situ high-energy electron diffraction is employed to monitor the crystallinity and surface roughness of GeSbTe films at different stages of growth. Thus, we identify the deposition parameter ranges whereby high structural quality GeSbTe with flat surfaces can be obtained. [ABSTRACT FROM AUTHOR]

The in‐memory computing paradigm aims at overcoming the intrinsic inefficiencies of Von‐Neumann computers by reducing the data‐transport per arithmetic operation. Crossbar arrays of multilevel memristive devices enable efficient calculations of matrix‐vector‐multiplications, an operation extensively called on in artificial intelligence (AI) tasks. Resistive random‐access memories (ReRAMs) are promising candidate devices for such applications. However, they generally exhibit large stochasticity and device‐to‐device variability. The integration of a sub‐stoichiometric metal‐oxide within the ReRAM stack can improve the resistive switching graduality and stochasticity. To this purpose, a conductive TaOx layer is developed and stacked on HfO2 between TiN electrodes, to create a complementary metal‐oxide‐semiconductor‐compatible ReRAM structure. This device shows accumulative conductance updates in both directions, as required for training neural networks. Moreover, by reducing the TaOx thickness and by increasing its resistivity, the device resistive states increase, as required for reduced power consumption. An electric field‐driven TaOx oxidation/reduction is responsible for the ReRAM switching. To demonstrate the potential of the optimized TaOx/HfO2 devices, the training of a fully‐connected neural network on the Modified National Institute of Standards and Technology database dataset is simulated and benchmarked against a full precision digital implementation. [ABSTRACT FROM AUTHOR]

A high degree of vacancy ordering is obtained by annealing amorphous GeTe-Sb2Te3 (GST) alloys deposited on a crystalline substrate, which acts as a template for the crystallization. Under annealing the material evolves from amorphous to disordered rock-salt, to ordered rocksalt with vacancies arranged into (111) oriented layers, and finally converts into the stable trigonal phase. The role of the interface in respect to the formation of an ordered crystalline phase is studied by comparing the transformation stages of crystalline GST with and without a capping layer. The capping layer offers another crystallization interface, which harms the overall crystalline quality. [ABSTRACT FROM AUTHOR]

In the present work, GeTe and Sb2Te3 van der Waals bonded superlattices epitaxially grown by molecular beam epitaxy are investigated. These structures are grown on passivated Si substrates, resulting in one single epitaxial domain and its twinned domain, both sharing the same out-of-plane orientation. Supported by X-ray diffraction and Raman spectroscopy, attention is called to the thermodynamically driven tendency of GeTe and Sb2Te3 to intermix into a Ge-Sb-Te (GST) alloy at the interfaces. A growth model is proposed to explain how these GST structures are formed. [ABSTRACT FROM AUTHOR]

Time-dependent X-ray absorption spectroscopy at the Ge L3edge is used to study the optical switching dynamics of epitaxial phase change materials grown on Si(111) membranes. Small differences between the static absorption spectra of the crystalline and laser-amorphized phase of GeTe are observed. Furthermore, a new approach for pump-probe studies on phase change materials is presented. However, no time-dependent change in the X-ray absorption > 0.1% has been detected so far. This is mainly attributed to sample deterioration at 3 kHz pump-probe rates. [ABSTRACT FROM AUTHOR]

The extraordinary electronic and optical properties of the crystal-to-amorphous transition in phase-change materials have led to important developments in memory applications. A promising outlook is offered by nanoscaling such phase-change structures. Following this research line, we study the interband optical transmission spectra of nanoscaled GeTe/Sb2Te3 chalcogenide superlattice films. We determine, for films with varying stacking sequence and growth methods, the density and scattering time of the free carriers, and the characteristics of the valence-to-conduction transition. It is found that the free carrier density decreases with increasing GeTe content, for sublayer thicknesses below ~3 nm. A simple band model analysis suggests that GeTe and Sb2Te3 layers mix, forming a standard GeSbTe alloy buffer layer. We show that it is possible to control the electronic transport properties of the films by properly choosing the deposition layer thickness, and we derive a model for arbitrary film stacks. [ABSTRACT FROM AUTHOR]

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)/HfO x 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 TaO x /HfO x 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 TaO x 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 TaO x 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 TaO x 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 TaO x /HfO x ReRAM devices, increasing the physical understanding and providing the equations necessary for circuit simulations incorporating this technology.