ArticlePublisher preview available
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Metal–oxide–semiconductor junctions are the building blocks of modern electronics and can provide a variety of functionalities, from memory to computing. The technology, however, faces constraints in terms of further miniaturization and compatibility with post–von Neumann computing architectures. Manipulation of structural—rather than electronic—states could provide a path to ultrascaled low-power functional devices, but the electrical control of such states is challenging. Here we report electronically accessible long-lived structural states in vanadium dioxide that can provide a scheme for data storage and processing. The states can be arbitrarily manipulated on short timescales and tracked beyond 10,000 s after excitation, exhibiting features similar to glasses. In two-terminal devices with channel lengths down to 50 nm, sub-nanosecond electrical excitation can occur with an energy consumption as small as 100 fJ. These glass-like functional devices could outperform conventional metal–oxide–semiconductor electronics in terms of speed, energy consumption and miniaturization, as well as provide a route to neuromorphic computation and multilevel memories. Electronically accessible states in vanadium dioxide can be arbitrarily manipulated on short timescales and tracked beyond 10,000 s after excitation.
Tracing the state dynamics of VO2 switches with incubation time a, Schematic of the ultrafast time-domain experimental setup. The SEM image shows the VO2 switch. Scale bar, 5 μm. The devices investigated had lengths varying from 50 nm to 3 μm. b, Transient conductance of the VO2 channel corresponding to different relaxation times T, as well as the very first switching cycle. The incubation time (tinc) and conductance of the insulating state (Gins) were studied. c, Incubation time versus relaxation time. Here tinc is a logarithmic function of T. The error bars are smaller than the symbol dimension. d, Gins versus relaxation time. After ~1 s, variations in conductance are no longer detectable. e, Incubation time versus increased conductance ((Gins−Ḡ)/Gins\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$(G_{{{{\mathrm{ins}}}}} - \bar G)/G_{{{{\mathrm{ins}}}}}$$\end{document}, where Ḡ\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\bar G$$\end{document} is the average over Gins for measurements corresponding to 10 s relaxation) for different relaxation times for a 100-nm-long device, showing the reproducibility of the results in nano-devices. The inset illustrates the fast relaxation of temperature and resistance, as well as the slow dynamics of tinc. f, Monitored incubation time for 5,000 measurements with two different relaxation times (10 and 100 ms), showing that the effect is reversible and consistent.
… 
Evidence of glass-like dynamics a, Illustration of the excitation signal of a VO2 switch for measurements under laser light. Every second, two identical pulses with time separation T are applied to the device and incubation times are monitored (tincref\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$t_{{{{\mathrm{inc}}}}}^{{{{\mathrm{ref}}}}}$$\end{document} and tinc). b, Relative change in the incubation time (tinc−tincref)/tincref\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\left( {(t_{{{{\mathrm{inc}}}}} - t_{{{{\mathrm{inc}}}}}^{{{{\mathrm{ref}}}}})/t_{{{{\mathrm{inc}}}}}^{{{{\mathrm{ref}}}}}} \right)$$\end{document} versus increased Gins in a 200-nm-long channel VO2 switch. Measurements under a CW 532 nm laser light with a power density of ~100 W cm⁻² show a considerably higher conductivity; however, the memory effect is unchanged. The experiment shows that the memory effect is likely structural. c, Relative change in incubation time after T = 10 ms with respect to the reference pulses with relaxation time of T = 1 s at different temperatures. The more pronounced change in tinc at higher temperatures indicates faster relaxation (inset). d, Transient current density of the VO2 switch corresponding to two consecutive triggering events with Vset = 290 mV at a chuck temperature of 55 °C. The results show a notable change in tinc.
… 
This content is subject to copyright. Terms and conditions apply.
A preview of the PDF is not available
... Neuromorphic computing represents a computational paradigm that aims to emulate characteristics of information processing in the brain. Mott semiconductors offer a powerful platform to realize neuromorphic hardware via electrically driven conductance transitions (1)(2)(3). For instance, a single vanadium dioxide (VO 2 ) device connected to a capacitor can serve as an oscillatory artificial neuron in a more compact manner (4-7) compared to traditional silicon (Si) complementary metal-oxide semiconductor layouts. ...
Article
Full-text available
The cointegration of artificial neuronal and synaptic devices with homotypic materials and structures can greatly simplify the fabrication of neuromorphic hardware. We demonstrate experimental realization of vanadium dioxide (VO2) artificial neurons and synapses on the same substrate through selective area carrier doping. By locally configuring pairs of catalytic and inert electrodes that enable nanoscale control over carrier density, volatility or nonvolatility can be appropriately assigned to each two-terminal Mott memory device per lithographic design, and both neuron- and synapse-like devices are successfully integrated on a single chip. Feedforward excitation and inhibition neural motifs are demonstrated at hardware level, followed by simulation of network-level handwritten digit and fashion product recognition tasks with experimental characteristics. Spatially selective electron doping opens up previously unidentified avenues for integration of emerging correlated semiconductors in electronic device technologies.
Article
Inspired by the neurosynaptic frameworks in the human brain, neuromorphic computing is expected to overcome the bottleneck of traditional von Neumann architecture and be used in artificial intelligence. Here, we predict a class of potential candidate materials, MCu3X4 (M = V, Nb, Ta;X=S,Se,Te), for neuromorphic computing applications through first-principles calculations based on density-functional theory. We find that when MCu3X4 are inserted with Li atom, the systems would transform from semiconductors to metals due to the considerable electron filling [approximately 0.8 electrons per formula unit (f.u.)] and still maintain well-structural stability. Meanwhile, the inserted Li atom also has a low diffusion barrier (approximately 0.6 eV/f.u.), which ensures the feasibility to control the insertion and extraction of Li by gate voltage. These results establish that the system can achieve the reversible switching between two stable memory states, i.e., high or low resistance state, indicating that it could potentially be used to design synaptic transistor to enable neuromorphic computing. Our work provides inspiration for advancing the search of candidate materials related to neuromorphic computing from the perspective of theoretical calculations.
Article
Full-text available
In-memory computing is a highly efficient approach for breaking the bottleneck of von Neumann architectures, i.e., reducing redundant latency and energy consumption during the data transfer between the physically separated memory and processing units. Herein we have designed a in-memory computing device, a van der Waals ferroelectric semiconductor (InSe) based metal-oxide-ferroelectric semiconductor field-effect transistor (MOfeS-FET). This MOfeS-FET integrates memory and logic functions in the same material, in which the out-of-plane (OOP) ferroelectric polarization in InSe is used for data storage and the semiconducting property is used for the logic computation. The MOfeS-FET shows a long retention time with high on/off ratios (>106), high program/erase (P/E) ratios (103), and stable cyclic endurance. Moreover, inverter, programmable NAND, and NOR Boolean logic operations with nonvolatile storage of the results have all been demonstrated using our approach. These findings highlight the potential of van der Waals ferroelectric semiconductor-based MOfeS-FETs in the in-memory computing and their potential of achieving size scaling beyond Moore's law.
Article
An insulator–metal transition (IMT) is an emergent characteristic of quantum materials, which have a great amount promise for applications, such as memories, optical switches, and analog brain functions. This is due to their ability to switch between two well-defined states. Thus, the characterization of the state-switching process is essential for the application of these materials. For vanadium dioxide ([Formula: see text]), the phase transition can be determined from temperature, magnetic field, and dielectric constant. In this paper, we propose a diamond quantum sensing approach based on nitrogen-vacancy centers for analyzing phase transitions. By using lock-in-based optically detected magnetic resonance and Rabi measurement protocols, temperature and magnetic field can reflect local IMT information of the circuit, and microwave can determine IMT information of an electrical isolation region. Our multifunctional quantum sensor exhibits local, nondestructive, and integrated measurements, which are useful for reliability testing in IMT technology applications.
Article
Full-text available
The evolution of electronics has largely relied on downscaling to meet the continuous needs for faster and highly integrated devices¹. As the channel length is reduced, however, classic electronic devices face fundamental issues that hinder exploiting materials to their full potential and, ultimately, further miniaturization². For example, the carrier injection through tunnelling junctions dominates the channel resistance³, whereas the high parasitic capacitances drastically limit the maximum operating frequency⁴. In addition, these ultra-scaled devices can only hold a few volts due to the extremely high electric fields, which limits their maximum delivered power5,6. Here we challenge such traditional limitations and propose the concept of electronic metadevices, in which the microscopic manipulation of radiofrequency fields results in extraordinary electronic properties. The devices operate on the basis of electrostatic control of collective electromagnetic interactions at deep subwavelength scales, as an alternative to controlling the flow of electrons in traditional devices, such as diodes and transistors. This enables a new class of electronic devices with cutoff frequency figure-of-merit well beyond ten terahertz, record high conductance values, extremely high breakdown voltages and picosecond switching speeds. This work sets the stage for the next generation of ultrafast semiconductor devices and presents a new paradigm that potentially bridges the gap between electronics and optics.
Article
Vanadium dioxide (VO2)-based smart windows show excellent promise for energy-saving and have been extensively researched. However, for the glass industry-compatible magnetron sputtering process, VO2 films are difficult to obtain and have homogeneous crystalline state, leaving them lacking the ideal solar modulation (ΔTsol) and sensitivity (narrow hysteresis loop). More importantly, the instability of VO2 hinders its commercialization. Multilayer structures have been repeatedly investigated to solve these problems. Unfortunately, the mediocre thermochromic properties as well as the complex and expensive manufacturing steps still hinder its commercialization. In this work, we prepared gradient variation oxygen-content vanadium-oxygen composite films (V2O3/VO2/V2O5, VOgv) with enhanced crystallinity and excellent durability by one-step continuous sputtering. According to optical measurements, the ΔTsol of the VOgv films was significantly increased by 145% (from 6.85 to 16.80%) compared to VO2 films, and the width of the hysteresis loop was reduced by 67% (from 19.34 to 6.36 °C), while the VOgv films exhibited a wider preparation window. The accelerated tests have shown that the film has an equivalent service life of approximately 20 years. We exploited the intrinsic similarity in properties of homologous compounds of vanadium oxide and simplified the preparation process, which is supposed to break the existing application bottlenecks and increase the commercializing possibility of VO2-based thermochromic smart windows.
Preprint
Full-text available
Three condensed matter findings, difficult to understand in the standard framework and providing support for the TGD view of dark matter, are discussed. The first discovery is that the antiprotons in hybrid matter-antimatter atoms do not seem to interact with the surrounding matter. A possible TGD based explanation is that the antiproton is dark and resides at a magnetic flux tube. The second finding might be called invisible magnetic fields. In TGD their identification could be as magnetic fields associated with monopole flux tubes, which are not possible in Maxwellian electrodynamics. The third finding is that vanadium-oxide VO2 is able to have memories about electric currents and can be said to learn. Dark valence electrons, proposed to explain strange disappearance of valence electrons in transition metals induced by heating, could explain the finding and also why the conductivity of VO2 is 10 times higher than expected on the basis of the Wiedemann Franz law.
Article
Full-text available
Vanadium dioxide (VO2) is widely studied for its prominent insulator–metal transition (IMT) near room temperature, with potential applications in novel memory devices and brain-inspired neuromorphic computing. We report on the fabrication of in-plane VO2 metal–insulator–metal structures and reproducible switching measurements in these two-terminal devices. Resistive switching can be achieved by applying voltage or current bias, which creates Joule heating in the device and triggers the IMT. We analyze the current/voltage-induced resistive switching characteristics, including a pronounced intermediate state in the reset from the low to the high resistance state. Controllable switching behavior is demonstrated between multiple resistance levels over several orders of magnitude, allowing for multibit operation. This multi-level operation of the VO2-bridge devices results from exploiting sub-hysteresis loops by Joule heating.
Article
Full-text available
Significance Emerging neuromorphic computing with resistive switching devices is one of the promising technologies toward hardware-based artificial intelligence. VO 2 has been demonstrated as a great candidate to emulate the spiking neurons because of the nature of its room-temperature metal–insulator transition and resistive switching. However, the fundamental understanding of the switching stochasticity in this strongly correlated material remains unaddressed. In this work, the inherent electrical and structural stochasticity in a VO 2 /TiO 2 device has been unambiguously revealed by combining in situ transmission electron microscopy experiments and ex situ resistive switching measurement on the same device. We conclude that the randomly oriented monoclinic domains in insulating VO 2 between each resistive switching is the key factor governing the stochasticity behavior.
Article
Full-text available
To circumvent the von Neumann bottleneck, substantial progress has been made towards in-memory computing with synaptic devices. However, compact nanodevices implementing non-linear activation functions are required for efficient full-hardware implementation of deep neural networks. Here, we present an energy-efficient and compact Mott activation neuron based on vanadium dioxide and its successful integration with a conductive bridge random access memory (CBRAM) crossbar array in hardware. The Mott activation neuron implements the rectified linear unit function in the analogue domain. The neuron devices consume substantially less energy and occupy two orders of magnitude smaller area than those of analogue complementary metal–oxide semiconductor implementations. The LeNet-5 network with Mott activation neurons achieves 98.38% accuracy on the MNIST dataset, close to the ideal software accuracy. We perform large-scale image edge detection using the Mott activation neurons integrated with a CBRAM crossbar array. Our findings provide a solution towards large-scale, highly parallel and energy-efficient in-memory computing systems for neural networks.
Article
Full-text available
The prototypical metal-insulator transition in VO2 at 340 K is from a high-temperature rutile phase to a low-temperature monoclinic phase. The lower symmetry of the monoclinic structure removes the degeneracy of the two equivalent directions of the tetragonal structure, giving rise to twin domains. Since formation of domain walls require energy most needle-like monoclinic single crystal are single-domain. The mixed metal-insulator state in self-heated needle-like single crystals exhibits various domain patterns, the most remarkable being static insulating triangular domains embedded in the metal and narrow insulating domains sliding along the metallic background in the direction of the electric current. Reported here are results obtained for some rare needle-like twinned VO2 single crystals. Such sample revealed a unique feature: joint static triangular twins emit sliding twin domains, first overlapping and later disjoining. Dark and bright twins and dim metallic background were seen for optimal orientation under a microscope, due to polarization by reflection.
Article
Full-text available
The broad applications of ultrawide-band signals and terahertz waves in quantum measurements1,2, imaging and sensing techniques3,4, advanced biological treatments⁵, and very-high-data-rate communications⁶ have drawn extensive attention to ultrafast electronics. In such applications, high-speed operation of electronic switches is challenging, especially when high-amplitude output signals are required⁷. For instance, although field-effect and bipolar junction devices have good controllability and robust performance, their relatively large output capacitance with respect to their ON-state current substantially limits their switching speed⁸. Here we demonstrate a novel on-chip, all-electronic device based on a nanoscale plasma (nanoplasma) that enables picosecond switching of electric signals with a wide range of power levels. The very high electric field in the small volume of the nanoplasma leads to ultrafast electron transfer, resulting in extremely short time responses. We achieved an ultrafast switching speed, higher than 10 volts per picosecond, which is about two orders of magnitude larger than that of field-effect transistors and more than ten times faster than that of conventional electronic switches. We measured extremely short rise times down to five picoseconds, which were limited by the employed measurement set-up. By integrating these devices with dipole antennas, high-power terahertz signals with a power–frequency trade-off of 600 milliwatts terahertz squared were emitted, much greater than that achieved by the state of the art in compact solid-state electronics. The ease of integration and the compactness of the nanoplasma switches could enable their implementation in several fields, such as imaging, sensing, communications and biomedical applications.
Article
The results of first principles electronic structure calculations for the metallic rutile and the insulating monoclinic phase of vanadium dioxide are presented. In addition, the insulating phase is investigated for the first time. The density functional calculations allow for a consistent understanding of all three phases. In the rutile phase metallic conductivity is carried by metal orbitals, which fall into the one‐dimensional band, and the isotropically dispersing bands. Hybridization of both types of bands is weak. In the phase splitting of the band due to metal‐metal dimerization and upshift of the bands due to increased p‐d overlap lead to an effective separation of both types of bands. Despite incomplete opening of the optical band gap due to the shortcomings of the local density approximation, the metal‐insulator transition can be understood as a Peierls‐like instability of the band in an embedding background of electrons. In the phase, the metal‐insulator transition arises as a combined embedded Peierls‐like and antiferromagnetic instability. The results for VO2 fit into the general scenario of an instability of the rutile‐type transition‐metal dioxides at the beginning of the d series towards dimerization or antiferromagnetic ordering within the characteristic metal chains. This scenario was successfully applied before to MoO2 and NbO2. In the compounds, the and bands can be completely separated, which leads to the observed metal‐insulator transitions.
Article
Probabilistic computing is a paradigm in which data are not represented by stable bits, but rather by the probability of a metastable bit to be in a particular state. The development of this technology has been hindered by the availability of hardware capable of generating stochastic and tunable sequences of "1s" and "0s". The options are currently limited to complex CMOS circuitry and, recently, magnetic tunnel junctions. Here, we demonstrate that metal-insulator transitions can also be used for this purpose. We use an electrical pump/probe protocol and take advantage of the stochastic relaxation dynamics in VO2 to induce random metallization events. A simple latch circuit converts the metallization sequence into a random stream of 1s and 0s. The resetting pulse in between probes decorrelates successive events, providing a true stochastic digital sequence.
Article
Watching a metal filament grow Resistive switching is a process in which the electrical resistance of a sample changes abruptly in response to a voltage pulse, often by orders of magnitude. This process is at the heart of many neuromorphic computing approaches but visualizing it in both space and time is tricky. del Valle et al . monitored the resistive switching in three different vanadium oxide compounds by measuring time- and space-resolved optical reflectivity (see the Perspective by Hilgenkamp and Gao). A characteristic conducting filament was quickly nucleated on the inhomogeneities in the sample and then propagated due to Joule heating. —JS
Article
Control of the metal‐insulator phase transition is vital for emerging neuromorphic and memristive technologies. The ability to alter the electrically driven transition between volatile and non‐volatile states is particularly important for quantum‐materials‐based emulation of neurons and synapses. The major challenge of this implementation is to understand and control the nanoscale mechanisms behind these two fundamental switching modalities. Here, in situ X‐ray nanoimaging is used to follow the evolution of the nanostructure and disorder in the archetypal Mott insulator VO2 during an electrically driven transition. Our findings demonstrate selective and reversible stabilization of either the insulating or metallic phases achieved by manipulating the defect concentration. This mechanism enables us to alter the local switching response between volatile and persistent regimes and demonstrates a new possibility for nanoscale control of the resistive switching in Mott materials.
Article
We investigate THz conductivity dynamics in NdNiO3 and EuNiO3 ultrathin films (15 u.c., ∼ 5.7 nm thick) following a photoinduced thermal quench into the metallic state and reveal a clear contrast between first- and second-order dynamics. While in EuNiO3 the conductivity recovers exponentially, in NdNiO3 the recovery is non- exponential and slower than a simple thermal model. Crucially, it is consistent with first-order dynamics and well-described by a 2d Avrami model, with supercooling leading to metastable phase coexistence on the nano- to mesoscopic scale. This novel observation is a fundamentally dynamic manifestation of the first order character of the insulator-to-metal transition, which the nanoscale thickness of our films and their fast cooling rate uniquely enable us to detect. The large transients seen in our films are promising for fast electronic (and magnetic) switching applications.