Search

Abstract The pressure‐driven Mott‐transition in Chromium doped V2O3 films is investigated by direct electrical measurements on polycrystalline films with thicknesses down to 10 nm, and doping concentrations of 2%, 5%, and 15%. A change in resistivity of nearly two orders of magnitude is found for 2% doping. A simulation model based on a scaling law description of the phase transition and percolative behavior in a resistor lattice is developed. This is used to show that despite significant deviations in the film structure from single crystals, the transition behavior is very similar. Finally, the influence of the variability between grains on the characteristics of scaled devices is investigated and found to allow for scaling down to at least 50 nm device width.

Metal-organic frameworks (MOFs) are an emerging class of nanomaterials with immense biomedical potential for their unique interactions with biological and organic materials. In this work, we select two candidate two-dimensional (2D) MOF systems based on Fe3+ and Ni2+ metal centers and 2-aminoterephthalate acid ligand (Fe-MOF and Ni-MOF) and evaluate their performance as an active interface for study of cell-interactions. 2D Fe-MOF and Ni-MOF were synthesized onto hydroxyl-modified gold and glass substrates using a layer-by-layer liquid-phase-epitaxy (LbL-LPE) growth at room temperature and used as active substrates (Fe-MOF/glass, Fe-MOF/Au, Ni-MOF/glass and Ni-MOF/Au, respectively) for MTT cell-proliferation and reactive oxygen species tests using the PC-12 cell-line in order to investigate the biocompatibility. Immunostaining and morphological analyses of PC-12 cells on MOF interfaces suggested a stronger cell-substrate interaction in comparison to glass and were further characterized using the Electrical Cell-substrate Impedance Sensing (ECIS) technique, here for the first time, employed to study cell attachment, spreading and proliferation on 2D Fe-MOF. The 2D Fe-MOF showed superior long-term stability in cell culture medium by recording impedance over 24 h, crucial to monitor cell-dynamics at a solid-liquid interface. A significant increase of interfacial impedance was observed in ECIS, due to PC-12 cells adhering onto 2D Fe-MOF, which was also confirmed by the focused ion beam etching followed by scanning electron microscopy. Our novel findings, therefore, suggest 2D MOFs as highly suitable platform for the study of cell-related interactions using electrical techniques and potentially pave the way for future use of MOFs for bioelectronics and biosensor applications.

Abstract HfO2 is one of the most common memristive materials and it is widely accepted that oxygen vacancies are prerequisite to reduce the forming voltage of the respective memristive devices. Here, a series of six oxygen engineered substoichiometric HfO2 − x thin films with varying oxygen deficiency is investigated by conductive atomic force microscopy (c‐AFM) and the switching process of substoichiometric films is observed on the nanoscale. X‐ray diffractometry (XRD) exhibits a phase transition from stoichiometric, monoclinic HfO2 toward oxygen deficient, rhombohedral HfO1.7. The conductance of HfO2 − x is increasing with increasing oxygen deficiency, which is consistent with the increasing prevalence of the highly conductive rhombohedral phase. Simultaneously, c‐AFM reveals significant local conductivity differences between grains and grain boundaries, regardless of the level of oxygen deficiency. Single grains of highly oxygen deficient samples are formed at significant lower voltages. The mean forming voltage is reduced from (7.0 ± 0.6) V for HfO2 to (1.9 ± 0.8) V for HfO1.7. Resistive switching on the nanoscale is established for single grains for the highest deficient thin film samples. The final resistance state is thereby dependent on the initial conductivity of the grains. These studies offer valuable insights into the switching behavior of memristive polycrystalline HfO2.

Abstract Memristive devices with valence change mechanism (VCM) show promise for neuromorphic data processing, although emulation of synaptic behavior with analog weight updates remains a challenge. Standard filamentary and area‐dependent resistive switching exhibit characteristic differences in the transition from the high to low resistance state, which is either abrupt with inherently high variability or gradual and allows quasi‐analog operation. In this study, the two switching modes are clearly correlated to differences in the microstructure and electronic structure for Pt/Al2O3/TiOx/Cr/Pt devices made from amorphous layers of 1.2 nm Al2O3 and 7 nm TiOx by atomic layer deposition. For the filamentary mode, operando spectromicroscopy experiments identify a localized region of ≈50 nm in diameter of reduced titania surrounded by crystalline rutile‐like TiO2, highlighting the importance of Joule heating for this mode. In contrast, both oxide layers remain in their amorphous state for the interfacial mode, which proves that device temperature during switching stays below 670 K, which is the TiO2 crystallization temperature. The analysis of the electronic conduction behavior confirms that the interfacial switching occurs by modulating the effective tunnel barrier width due to accumulation and depletion of oxygen vacancies at the Al2O3/TiOx interface. The results are transferable to other bilayer stacks.

Abstract Metal-free chemical vapor deposition (CVD) of single-layer graphene (SLG) on c-plane sapphire has recently been demonstrated for wafer diameters of up to 300 mm, and the high quality of the SLG layers is generally characterized by integral methods. By applying a comprehensive analysis approach, distinct interactions at the graphene-sapphire interface and local variations caused by the substrate topography are revealed. Regions near the sapphire step edges show tiny wrinkles with a height of about 0.2 nm, framed by delaminated graphene as identified by the typical Dirac cone of free graphene. In contrast, adsorption of CVD SLG on the hydroxyl-terminated α-Al2O3 (0001) terraces results in a superstructure with a periodicity of (2.66 ± 0.03) nm. Weak hydrogen bonds formed between the hydroxylated sapphire surface and the π-electron system of SLG result in a clean interface. The charge injection induces a band gap in the adsorbed graphene layer of about (73 ± 3) meV at the Dirac point. The good agreement with the predictions of a theoretical analysis underlines the potential of this hybrid system for emerging electronic applications.

Resistively switching electrochemical metallization memory cells are gaining huge interest since they are seen as promising candidates and basic building blocks for future computation-in-memory applications. However, especially filamentary-based memristive devices suffer from inherent variability, originating from their stochastic switching behavior. A variability-aware compact model of electrochemical metallization memory cells is presented in this study and verified by showing a fit to experimental data. It is an extension of the deterministic model. Since this extension consists of several different features allowing for a realistic variability-aware fit, it depicts a unique model comprising physics-based, stochastically and experimentally originating variabilities and reproduces them well. In addition, a physics-based model parameter study is executed, which enables a comprehensive view into the device physics and presents guidelines for the compact model fitting procedure.

Abstract The proton conductivity of Y‐doped BaZrO3‐based ceramic (BZY) films is highly dependent on the preparation process, especially thermal treatment. In order to further improve the electrochemical properties of BZY thin films, a novel sintering method was introduced to enhance the densification of such refractory materials. BaZr0.9Y0.1O3‐δ (BZY10) was used as a composition in this case study and the concept is realized using chemical solution deposition (CSD) derived porous cathode and anode films as auxiliary layers underneath the BZY10 films, which are CSD derived as well. After co‐sintering at 1200°C for 4 h, BZY10 films with a high degree of densification were attained, accompanied by average grain sizes of 300–350 nm. The grain sizes are comparable to common results found in BZY10 pellets sintered at 1600°C for more than 6 h. However, the densification conditions are much better. For the porous NiO/BZY10 composite films, the in‐plane conductivity is enhanced to about 2 × 10−4 S/cm at 600°C, which is quite close to that (10−3 S/cm) of optimized BZY bulk materials. The compatible stack structures and the remarkably low sintering temperature are applicable to the fabrication of proton‐conducting solid oxide fuel and electrolyzer cells.

In this work, we experimentally characterize the endurance of 2 Mbit resistive switching random access memories (ReRAMs) from a 16 MBit test-chip. Here, very rare failure events where the memory cells become stuck in the low-resistive state (LRS) are observed. As this failure mechanism is the limiting one concerning the endurance of this ReRAM implementation, extensive investigations are conducted and presented. The experimental findings are detailed via a voltage divider model, illustrating why memory cells can become stuck in the LRS. It is proposed, that an insufficient voltage dropping over the cell due to an unfavorable combination of cell- and transistor resistances is responsible for stuck-at-LRS bits. Furthermore, predictions for the origin of these suboptimal combinations are given. Additionally, a one-dimensional Kinetic Monte Carlo (KMC) model that allows a statistical investigation of large numbers of cells with regard to rare random events has been developed. Here, our proposed explanation for the observed failure mechanism is fortified by the simulation and evaluation of the switching process of the memory. All simulations are in very good agreement with the experimental data. Finally, based on our findings, we give suggestions for the improvement of switching algorithms.

The proliferation of machine learning algorithms in everyday applications such as image recognition or language translation has increased the pressure to adapt underlying computing architectures towards these algorithms. Application specific integrated circuits (ASICs) such as the Tensor Processing Units by Google, Hanguang by Alibaba or Inferentia by Amazon Web Services were designed specifically for machine learning algorithms and have been able to outperform CPU based solutions by great margins during training and inference. As newer generations of chips allow handling of and computation on more and more data, the size of neural networks has dramatically increased, while the challenges they are trying to solve have become more complex. Neuromorphic computing tries to take inspiration from biological information processing systems, aiming to further improve the efficiency with which these networks can be trained or the inference can be performed. Enhancing neuromorphic computing architectures with memristive devices as non-volatile storage elements could potentially allow for even higher energy efficiencies. Their ability to mimic synaptic plasticity dynamics brings neuromorphic architectures closer to the biological role models. So far, memristive devices are mainly investigated for the emulation of the weights of neural networks during training and inference as their non-volatility would enable both processes in the same location without data transfer. In this paper, we explore realisations of different synapses build from memristive ReRAM devices, based on the Valence Change Mechanism. These synapses are the 1R synapse, the NR synapse and the 1T1R synapse. For the 1R synapse, we propose three dynamical regimes and explore their performance through different synapse criteria. For the NR synapse, we discuss how the same dynamical regimes can be addressed in a more reliable way. We also show experimental results measured on ZrOx devices to support our simulation based claims. For the 1T1R synapse, we explore the trade offs between the connection direction of the ReRAM device and the transistor. For all three synapse concepts we discuss the impact of device-to-device and cycle-to-cycle variability. Additionally, the impact of the stimulation mode on the observed behavior is discussed.

Abstract Thin films of the prototypical charge transfer insulator nickel oxide appear to be a promising material for novel nanoelectronic devices. The fabrication of the material is challenging, however, and mostly a p‐type semiconducting phase is reported. Here, the results of a factorial experiment are presented that allow optimization of the properties of thin films deposited using sputtering. A cluster analysis is performed, and four main types of films are found. Among them, the desired insulating phase is identified. From this material, nanoscale devices are fabricated, which demonstrate that the results carry over to relevant length scales. Initial switching results are reported.

By imitating the synaptic connectivity and plasticity of the brain, emerging electronic nanodevices offer new opportunities as the building blocks of neuromorphic systems. One challenge for large-scale simulations of computational architectures based on emerging devices is to accurately capture device response, hysteresis, noise, and the covariance structure in the temporal domain as well as between the different device parameters. We address this challenge with a high throughput generative model for synaptic arrays that is based on a recently available type of electrical measurement data for resistive memory cells. We map this real-world data onto a vector autoregressive stochastic process to accurately reproduce the device parameters and their cross-correlation structure. While closely matching the measured data, our model is still very fast; we provide parallelized implementations for both CPUs and GPUs and demonstrate array sizes above one billion cells and throughputs exceeding one hundred million weight updates per second, above the pixel rate of a 30 frames/s 4K video stream.

Abstract The inevitable variability within electronic devices causes strict constraints on operation, reliability and scalability of the circuit design. However, when a compromise arises among the different performance metrics, area, time and energy, variability then loosens the tight requirements and allows for further savings in an alternative design scope. To that end, unconventional computing approaches are revived in the form of approximate computing, particularly tuned for resource-constrained mobile computing. In this paper, a proof-of-concept of the approximate computing paradigm using memristors is demonstrated. Stochastic memristors are used as the main building block of probabilistic logic gates. As will be shown in this paper, the stochasticity of memristors’ switching characteristics is tightly bound to the supply voltage and hence to power consumption. By scaling of the supply voltage to appropriate levels stochasticity gets increased. In order to guide the design process of approximate circuits based on memristors a realistic device model needs to be elaborated with explicit emphasis of the probabilistic switching behavior. Theoretical formulation, probabilistic analysis, and simulation of the underlying logic circuits and operations are introduced. Moreover, the expected output behavior is verified with the experimental measurements of valence change memory cells. Hence, it is shown how the precision of the output is varied for the sake of the attainable gains at different levels of available design metrics. This approach represents the first proposition along with physical verification and mapping to real devices that combines stochastic memristors into unconventional computing approaches.

Memristive devices based on the Valence Change Mechanism (VCM) are promising devices for storage class memory, neuromorphic computing and logic-in-memory (LIM) applications. They are suited for such a wide range of applications, due to their possibility for extreme dense integration, low power consumption and multilevel capabilities. Through LIM concepts, Boolean logic operations can be performed directly in memory. In many of these concepts, the resistance state of the device is interpreted as the logical input and output of the logic function, which is why these concepts are called ‘stateful’ logic. Most of the proposed ideas, however, are defined based on only binary switching VCM devices and neglect their multi-level capabilities. Extending LIM concepts towards multinary logic, e.g. a ternary logic, would increase the data density inside the memory array and reduce the number of devices required to perform a certain operation. In this work, we discuss two possibilities of realizing a ternary logic based on the analog switching in the RESET or in the SET direction. For both directions we verify the logic functionality by showing the basic operations of implication, negation and false operation, which together form a functionally complete logic. Additionally, for both switching directions, we discuss a 41-Trit ( ${\approx}$ 64-Bit) addition. For all investigations the physics-based compact model JART VCM v1b is used, which has been verified on the RESET direction multilevel properties of the TaO $_{\mathrm{x}}$ devices.

Brain-inspired computing proposes a set of algorithmic principles that hold promise for advancing artificial intelligence. They endow systems with self learning capabilities, efficient energy usage, and high storage capacity. A core concept that lies at the heart of brain computation is sequence learning and prediction. This form of computation is essential for almost all our daily tasks such as movement generation, perception, and language. Understanding how the brain performs such a computation is not only important to advance neuroscience, but also to pave the way to new technological brain-inspired applications. A previously developed spiking neural network implementation of sequence prediction and recall learns complex, high-order sequences in an unsupervised manner by local, biologically inspired plasticity rules. An emerging type of hardware that may efficiently run this type of algorithm is neuromorphic hardware. It emulates the way the brain processes information and maps neurons and synapses directly into a physical substrate. Memristive devices have been identified as potential synaptic elements in neuromorphic hardware. In particular, redox-induced resistive random access memories (ReRAM) devices stand out at many aspects. They permit scalability, are energy efficient and fast, and can implement biological plasticity rules. In this work, we study the feasibility of using ReRAM devices as a replacement of the biological synapses in the sequence learning model. We implement and simulate the model including the ReRAM plasticity using the neural network simulator NEST. We investigate two types of ReRAM memristive devices: (i) a gradual, analog switching device, and (ii) an abrupt, binary switching device. We study the effect of different device properties on the performance characteristics of the sequence learning model, and demonstrate that, in contrast to many other artificial neural networks, this architecture is resilient with respect to changes in the on-off ratio and the conductance resolution, device variability, and device failure.

Machine learning models for sequence learning and processing often suffer from high energy consumption and require large amounts of training data. The brain presents more efficient solutions to how these types of tasks can be solved. While this has inspired the conception of novel brain-inspired algorithms, their realizations remain constrained to conventional von-Neumann machines. Therefore, the potential power efficiency of the algorithm cannot be exploited due to the inherent memory bottleneck of the computing architecture. Therefore, we present in this paper a dedicated hardware implementation of a biologically plausible version of the Temporal Memory component of the Hierarchical Temporal Memory concept. Our implementation is built on a memristive crossbar array and is the result of a hardware-algorithm co-design process. Rather than using the memristive devices solely for data storage, our approach leverages their specific switching dynamics to propose a formulation of the peripheral circuitry, resulting in a more efficient design. By combining a brain-like algorithm with emerging non-volatile memristive device technology we strive for maximum energy efficiency. We present simulation results on the training of complex high-order sequences and discuss how the system is able to predict in a context-dependent manner. Finally, we investigate the energy consumption during the training and conclude with a discussion of scaling prospects.

With the arrival of the Internet of Things (IoT) and the challenges arising from Big Data, neuromorphic chip concepts are seen as key solutions for coping with the massive amount of unstructured data streams by moving the computation closer to the sensors, the so-called “edge computing.” Augmenting these chips with emerging memory technologies enables these edge devices with non-volatile and adaptive properties which are desirable for low power and online learning operations. However, an energy- and area-efficient realization of these systems requires disruptive hardware changes. Memristor-based solutions for these concepts are in the focus of research and industry due to their low-power and high-density online learning potential. Specifically, the filamentary-type valence change mechanism (VCM memories) have shown to be a promising candidate In consequence, physical models capturing a broad spectrum of experimentally observed features such as the pronounced cycle-to-cycle (c2c) and device-to-device (d2d) variability are required for accurate evaluation of the proposed concepts. In this study, we present an in-depth experimental analysis of d2d and c2c variability of filamentary-type bipolar switching HfO2/TiOx nano-sized crossbar devices and match the experimentally observed variabilities to our physically motivated JART VCM compact model. Based on this approach, we evaluate the concept of parallel operation of devices as a synapse both experimentally and theoretically. These parallel synapses form a synaptic array which is at the core of neuromorphic chips. We exploit the c2c variability of these devices for stochastic online learning which has shown to increase the effective bit precision of the devices. Finally, we demonstrate that stochastic switching features for a pattern classification task that can be employed in an online learning neural network.

Memristors have become an emerging technology capable in emulating human brain information processing, but understanding and controlling the switching mechanism remains elusive. Here, Milano et al. combine memristive and neuromorphic functionalities in a single crystalline nanowire model system.

This work studies a computation in‐memory concept for binary multiply‐accumulate operations based on complementary resistive switches (CRS). By exploiting the in‐memory boolean exclusive OR (XOR) operation of single CRS devices, the Hamming Distance (HD) can be calculated if the center electrodes of multiple CRS cells are connected. This HD is linearly encoded in the voltage drop of the common electrode, and from it the result of a binary multiply‐accumulate operation can be calculated. A small‐scale demonstration is experimentally realized and the feasibility of the in‐memory computation concept is confirmed. A simulation study identifies the low resistance state (LRS) variability as the main reason for the variations in the output voltage. The application as a potential hardware accelerator for the inference step of binary neural networks is investigated. Therefore, a 1‐layer fully connected neural network is trained on a binarized version of the MNIST data set and the inference step of the test data set is simulated. The concept achieves a prediction accuracy of approximately 86%.

An introduction to the APL Materials Special Issue on “Emerging Materials in Neuromorphic Computing,” by the guest editors.

Metal exsolution is a dynamic process driven under a reducing atmosphere and at elevated temperatures that results in the self-assembly of nanoparticles at the surface of complex perovskite catalysts. The nanoparticle characteristics of metal exsolution catalysts can be subject to considerable inhomogeneity, and the anisotropic surface properties of ceramic oxides have been identified to have a major influence on the exsolution behavior. We systematically reveal the orientation-dependent anisotropy of the exsolution behavior of Ni in SrTi _0.9 Nb _0.05 Ni _0.05 O _3−δ using multi-faceted epitaxial thin films that represent a material system with properties in between functional ceramics and single-crystalline perovskite thin film model systems. Using an approach of combined orientation mapping and surface imaging we study the exsolution behavior with particular focus on the initial exsolution response, i.e. after short annealing times. We find orientation-specific variations in the surface morphology of the thin film facets. In the as-prepared state, surface reconstructions cause the formation of patterned surface structures for all thin film facets apart from (001) surfaces, which exhibit a plain surface morphology as well as an enhanced exsolution response. Surface reconstructions and their inherent energy landscape may hence cause an additional energy barrier for the exsolution reaction that results in orientation-dependent differences in the exsolution kinetics.

Resistive switching in metal oxides is related to the migration of donor defects. Here Baeumer et al. use in operandoX-ray spectromicroscopy to quantify the doping locally and show that small local variations in the donor concentration result in large variations in the device resistance.

Disordered and uniform (2√3 × 2√3)R30° superstructures of fullerenes on the Au(111) surface have been studied using scanning tunneling microscopy and spectroscopy. It is shown that the deposition and growth process of a fullerene monolayer on the Au(111) surface determine the resulting superstructure. The supply of thermal energy is of importance for the activation of a Au vacancy forming process and thus, one criterion for the selection of the respective superstructure. However, here it is depicted that a vacancy–adatom pair can be formed even at room temperature. This latter process results in C60 molecules that appear slightly more bright in scanning tunnelling microscopy images and are identified in disordered (2√3 x 2√3)R30° superstructures based on a detailed structure analysis. In addition, these slightly more bright C60 molecules form uniform (2√3 x 2√3)R30° superstructures, which exhibit intermolecular interactions, likely mediated by Au adatoms. Thus, vacancy–adatom pairs forming at room temperature directly affect the resulting C60 superstructure. Differential conductivity spectra reveal a lifting of the degeneracy of the LUMO and LUMO+1 orbitals in the uniform (2√3 x 2√3)R30° superstructure and in addition, hybrid fullerene–Au(111) surface states suggest partly covalent interactions.

In this study, we investigated the electronic surface structure of donor-doped strontium titanate. Homoepitaxial 0.5 wt. % donor-doped SrTiO3 thin films were analyzed by in situ near ambient pressure X-ray photoelectron spectroscopy at a temperature of 770 K and oxygen pressures up to 5 mbar. Upon exposure to an oxygen atmosphere at elevated temperatures, we observed a rigid binding energy shift of up to 0.6 eV towards lower binding energies with respect to vacuum conditions for all SrTiO3 core level peaks and the valence band maximum with increasing oxygen pressure. The rigid shift is attributed to a relative shift of the Fermi energy towards the valence band concomitant with a negative charge accumulation at the surface, resulting in a compensating electron depletion layer in the near surface region. Charge trapping effects solely based on carbon contaminants are unlikely due to their irreversible desorption under the given experimental conditions. In addition, simple reoxygenation of oxygen vacancies can be ruled out as the high niobium dopant concentration dominates the electronic properties of the material. Instead, the negative surface charge may be provided by the formation of cation vacancies or the formation of charged oxygen adsorbates at the surface. Our results clearly indicate a pO2-dependent surface space charge formation in donor-doped SrTiO3 in oxidizing conditions.

BaTiO3 – Ag composite nanopowders were synthesized via microemulsion mediated synthesis through the hydrolytic decomposition of mixed metal alkoxide solutions as precursor for the BaTiO3 and the reduction of silver nitrate in the presence of polyvinylpyrrolidone (PVP) as source for the Ag nanoparticles. The X-ray diffraction (XRD) patterns indicate that BaTiO3 and Ag phases were successfully synthesized in the composite powders. Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) show that the synthesized BaTiO3 nanoparticles were aggregates of nanosized primary particles as small as 10 nm in diameter and the average particle size of nanocrystalline Ag was about 100 nm. Calcination and sintering studies reveal that there exists a difference in the sintering behaviour of BaTiO3 and Ag in the composite nanopowders. Thermogravimetric analysis (TGA) shows weight losses due to the burnout of organic residues arising from the synthesis, the release of water from the surface and separation of hydroxyl ions from the lattice of BaTiO3 nanoparticles. A dilatometric study of BaTiO3-Ag composite confi rmed a strong difference in the shrinkage behaviour compared to that of the pure BaTiO3 obtained by microemulsion mediated synthesis.

Eu1−xLaxTiO3−δ (x = 0, 0.3, 0.5) films were deposited in a p(Ar(96%)/H2(4%)) = 4 × 10−4 mbar atmosphere on (LaAlO3)0.3-(Sr2AlTaO6)0.7 vicinal substrates (0.1°). Reflection high-energy electron diffraction oscillation characteristics of a layer-by-layer growth mode were observed for stoichiometric and Ti-rich films and the laser fluence suited to deposit stoichiometric films was identified to be 1.25 J/cm2 independent of the La content. The variety of resulting film compositions follows the general trend of Eu-enrichment for low laser and Ti-enrichment for high laser fluence. X-ray diffraction confirms that all the films are compressively strained with a general trend of an increase of c-axis elongation for non-stoichiometric films. The surfaces of non-stoichiometric films have an increased roughness, the highest sheet resistances, exhibit the presence of islands, and are Eu3+ rich for films deposited at low laser fluence.