Blooming and pruning: learning from mistakes with memristive synapses.

Blooming and pruning is one of the most important developmental mechanisms of the biological brain in the first years of life, enabling it to adapt its network structure to the demands of the environment. The mechanism is thought to be fundamental for the development of cognitive skills. Inspired by this, Chialvo and Bak proposed in 1999 a learning scheme that learns from mistakes by eliminating from the initial surplus of synaptic connections those that lead to an undesirable outcome. Here, this idea is implemented in a neuromorphic circuit scheme using CMOS integrated HfO2-based memristive devices. The implemented two-layer neural network learns in a self-organized manner without positive reinforcement and exploits the inherent variability of the memristive devices. This approach provides hardware, local, and energy-efficient learning. A combined experimental and simulation-based parameter study is presented to find the relevant system and device parameters leading to a compact and robust memristive neuromorphic circuit that can handle association tasks.

TiN/Ti/HfO2/TiN memristive devices for neuromorphic computing: from synaptic plasticity to stochastic resonance.

We characterize TiN/Ti/HfO2/TiN memristive devices for neuromorphic computing. We analyze different features that allow the devices to mimic biological synapses and present the models to reproduce analytically some of the data measured. In particular, we have measured the spike timing dependent plasticity behavior in our devices and later on we have modeled it. The spike timing dependent plasticity model was implemented as the learning rule of a spiking neural network that was trained to recognize the MNIST dataset. Variability is implemented and its influence on the network recognition accuracy is considered accounting for the number of neurons in the network and the number of training epochs. Finally, stochastic resonance is studied as another synaptic feature. It is shown that this effect is important and greatly depends on the noise statistical characteristics.

Chemical Vapor Deposition Growth of Graphene on 200 mm Ge(110)/Si Wafers and Ab Initio Analysis of Differences in Growth Mechanisms on Ge(110) and Ge(001).

For the fabrication of modern graphene devices, uniform growth of high-quality monolayer graphene on wafer scale is important. This work reports on the growth of large-scale graphene on semiconducting 8 inch Ge(110)/Si wafers by chemical vapor deposition and a DFT analysis of the growth process. Good graphene quality is indicated by the small FWHM (32 cm-1) of the Raman 2D band, low intensity ratio of the Raman D and G bands (0.06), and homogeneous SEM images and is confirmed by Hall measurements: high mobility (2700 cm2/Vs) and low sheet resistance (800 Ω/sq). In contrast to Ge(001), Ge(110) does not undergo faceting during the growth. We argue that Ge(001) roughens as a result of vacancy accumulation at pinned steps, easy motion of bonded graphene edges across (107) facets, and low energy cost to expand Ge area by surface vicinals, but on Ge(110), these mechanisms do not work due to different surface geometries and complex reconstruction.

Strongly enhanced sensitivities of CMOS compatible plasmonic titanium nitride nanohole arrays for refractive index sensing under oblique incidence.

Titanium nitride (TiN) is a complementary metal-oxide-semiconductor (CMOS) compatible material with large potential for the fabrication of plasmonic structures suited for device integration. However, the comparatively large optical losses can be detrimental for application. This work reports a CMOS compatible TiN nanohole array (NHA) on top of a multilayer stack for potential use in integrated refractive index sensing with high sensitivities at wavelengths between 800 and 1500 nm. The stack, consisting of the TiN NHA on a silicon dioxide (SiO2) layer with Si as substrate (TiN NHA/SiO2/Si), is prepared using an industrial CMOS compatible process. The TiN NHA/SiO2/Si shows Fano resonances in reflectance spectra under oblique excitation, which are well reproduced by simulation using both finite difference time domain (FDTD) and rigorous coupled-wave analysis (RCWA) methods. The sensitivities derived from spectroscopic characterizations increase with the increasing incident angle and match well with the simulated sensitivities. Our systematic simulation-based investigation of the sensitivity of the TiN NHA/SiO2/Si stack under varied conditions reveals that very large sensitivities up to 2305 nm per refractive index unit (nm RIU-1) are predicted when the refractive index of superstrate is similar to that of the SiO2 layer. We analyze in detail how the interplay between plasmonic and photonic resonances such as surface plasmon polaritons (SPPs), localized surface plasmon resonances (LSPRs), Rayleigh Anomalies (RAs), and photonic microcavity modes (Fabry-Pérot resonances) contributes to this result. This work not only reveals the tunability of TiN nanostructures for plasmonic applications but also paves the way to explore efficient devices for sensing in broad conditions.

Towards the Growth of Hexagonal Boron Nitride on Ge(001)/Si Substrates by Chemical Vapor Deposition.

The growth of hexagonal boron nitride (hBN) on epitaxial Ge(001)/Si substrates via high-vacuum chemical vapor deposition from borazine is investigated for the first time in a systematic manner. The influences of the process pressure and growth temperature in the range of 10-7-10-3 mbar and 900-980 °C, respectively, are evaluated with respect to morphology, growth rate, and crystalline quality of the hBN films. At 900 °C, nanocrystalline hBN films with a lateral crystallite size of ~2-3 nm are obtained and confirmed by high-resolution transmission electron microscopy images. X-ray photoelectron spectroscopy confirms an atomic N:B ratio of 1 ± 0.1. A three-dimensional growth mode is observed by atomic force microscopy. Increasing the process pressure in the reactor mainly affects the growth rate, with only slight effects on crystalline quality and none on the principle growth mode. Growth of hBN at 980 °C increases the average crystallite size and leads to the formation of 3-10 well-oriented, vertically stacked layers of hBN on the Ge surface. Exploratory ab initio density functional theory simulations indicate that hBN edges are saturated by hydrogen, and it is proposed that partial de-saturation by H radicals produced on hot parts of the set-up is responsible for the growth.

An energy-efficient in-memory computing architecture for survival data analysis based on resistive switching memories.

One of the objectives fostered in medical science is the so-called precision medicine, which requires the analysis of a large amount of survival data from patients to deeply understand treatment options. Tools like machine learning (ML) and deep neural networks are becoming a de-facto standard. Nowadays, computing facilities based on the Von Neumann architecture are devoted to these tasks, yet rapidly hitting a bottleneck in performance and energy efficiency. The in-memory computing (IMC) architecture emerged as a revolutionary approach to overcome that issue. In this work, we propose an IMC architecture based on resistive switching memory (RRAM) crossbar arrays to provide a convenient primitive for matrix-vector multiplication in a single computational step. This opens massive performance improvement in the acceleration of a neural network that is frequently used in survival analysis of biomedical records, namely the DeepSurv. We explored how the synaptic weights mapping strategy and the programming algorithms developed to counter RRAM non-idealities expose a performance/energy trade-off. Finally, we discussed how this application is tailored for the IMC architecture rather than being executed on commodity systems.

Novel Graphene Adjustable-Barrier Transistor with Ultra-High Current Gain.

A graphene-based three-terminal barristor device was proposed to overcome the low on/off ratios and insufficient current saturation of conventional graphene field-effect transistors. In this study, we fabricated and analyzed a novel graphene-based transistor, which resembles the structure of the barristor but uses a different operating condition. This new device, termed graphene adjustable-barriers transistor (GABT), utilizes a semiconductor-based gate rather than a metal-insulator gate structure to modulate the device currents. The key feature of the device is the two graphene-semiconductor Schottky barriers with different heights that are controlled simultaneously by the gate voltage. Due to the asymmetry of the barriers, the drain current exceeds the gate current by several orders of magnitude. Thus, the GABT can be considered an amplifier with an alterable current gain. In this work, a silicon-graphene-germanium GABT with an ultra-high current gain (ID/IG up to 8 × 106) was fabricated, and the device functionality was demonstrated. Additionally, a capacitance model is applied to predict the theoretical device performance resulting in an on-off ratio above 106, a swing of 87 mV/dec, and a drive current of about 1 × 106 A/cm2.

Activity of AC electrokinetically immobilized horseradish peroxidase.

Dielectrophoresis (DEP) is an AC electrokinetic effect mainly used to manipulate cells. Smaller particles, like virions, antibodies, enzymes, and even dye molecules can be immobilized by DEP as well. In principle, it was shown that enzymes are active after immobilization by DEP, but no quantification of the retained activity was reported so far. In this study, the activity of the enzyme horseradish peroxidase (HRP) is quantified after immobilization by DEP. For this, HRP is immobilized on regular arrays of titanium nitride ring electrodes of 500 nm diameter and 20 nm widths. The activity of HRP on the electrode chip is measured with a limit of detection of 60 fg HRP by observing the enzymatic turnover of Amplex Red and H2 O2 to fluorescent resorufin by fluorescence microscopy. The initial activity of the permanently immobilized HRP equals up to 45% of the activity that can be expected for an ideal monolayer of HRP molecules on all electrodes of the array. Localization of the immobilizate on the electrodes is accomplished by staining with the fluorescent product of the enzyme reaction. The high residual activity of enzymes after AC field induced immobilization shows the method's suitability for biosensing and research applications.

Toward controlling the Al2O3/ZnO interface properties by in situ ALD preparation.

An Al2O3/ZnO heterojunction was grown on a Si single crystal substrate by subsequent thermal and plasma-assisted atomic layer deposition (ALD) in situ. The band offsets of the heterointerface were then studied by consecutive removal of the layers by argon sputtering, followed by in situ X-ray photoelectron spectroscopy. The valence band maximum and conduction band minimum of Al2O3 are found to be 1.1 eV below and 2.3 eV above those of ZnO, resulting in a type-I staggered heterojunction. An apparent reduction of ZnO to elemental Zn in the interface region was detected in the Zn 2p core level and Zn L3MM Auger spectra. This suggests an interface formation different from previous models. The reduction of ZnO to Zn in the interface region accompanied by the creation of oxygen vacancies in ZnO results in an upward band bending at the interface. Therefore, this study suggests that interfacial properties such as the band bending as well as the valence and conduction band offsets should be in situ controllable to a certain extent by careful selection of the process parameters.

AC electrokinetic immobilization of influenza virus.

The use of alternating current (AC) electrokinetic forces, like dielectrophoresis and AC electroosmosis, as a simple and fast method to immobilize sub-micrometer objects onto nanoelectrode arrays is presented. Due to its medical relevance, the influenza virus is chosen as a model organism. One of the outstanding features is that the immobilization of viral material to the electrodes can be achieved permanently, allowing subsequent handling independently from the electrical setup. Thus, by using merely electric fields, we demonstrate that the need of prior chemical surface modification could become obsolete. The accumulation of viral material over time is observed by fluorescence microscopy. The influences of side effects like electrothermal fluid flow, causing a fluid motion above the electrodes and causing an intensity gradient within the electrode array, are discussed. Due to the improved resolution by combining fluorescence microscopy with deconvolution, it is shown that the viral material is mainly drawn to the electrode edge and to a lesser extent to the electrode surface. Finally, areas of application for this functionalization technique are presented.

Characterization and Separation of Live and Dead Yeast Cells Using CMOS-Based DEP Microfluidics.

This study aims at developing a miniaturized CMOS integrated silicon-based microfluidic system, compatible with a standard CMOS process, to enable the characterization, and separation of live and dead yeast cells (as model bio-particle organisms) in a cell mixture using the DEP technique. DEP offers excellent benefits in terms of cost, operational power, and especially easy electrode integration with the CMOS architecture, and requiring label-free sample preparation. This can increase the likeliness of using DEP in practical settings. In this work the DEP force was generated using an interdigitated electrode arrays (IDEs) placed on the bottom of a CMOS-based silicon microfluidic channel. This system was primarily used for the immobilization of yeast cells using DEP. This study validated the system for cell separation applications based on the distinct responses of live and dead cells and their surrounding media. The findings confirmed the device's capability for efficient, rapid and selective cell separation. The viability of this CMOS embedded microfluidic for dielectrophoretic cell manipulation applications and compatibility of the dielectrophoretic structure with CMOS production line and electronics, enabling its future commercially mass production.

Adaptation Strategies for Personalized Gait Neuroprosthetics.

Personalization of gait neuroprosthetics is paramount to ensure their efficacy for users, who experience severe limitations in mobility without an assistive device. Our goal is to develop assistive devices that collaborate with and are tailored to their users, while allowing them to use as much of their existing capabilities as possible. Currently, personalization of devices is challenging, and technological advances are required to achieve this goal. Therefore, this paper presents an overview of challenges and research directions regarding an interface with the peripheral nervous system, an interface with the central nervous system, and the requirements of interface computing architectures. The interface should be modular and adaptable, such that it can provide assistance where it is needed. Novel data processing technology should be developed to allow for real-time processing while accounting for signal variations in the human. Personalized biomechanical models and simulation techniques should be developed to predict assisted walking motions and interactions between the user and the device. Furthermore, the advantages of interfacing with both the brain and the spinal cord or the periphery should be further explored. Technological advances of interface computing architecture should focus on learning on the chip to achieve further personalization. Furthermore, energy consumption should be low to allow for longer use of the neuroprosthesis. In-memory processing combined with resistive random access memory is a promising technology for both. This paper discusses the aforementioned aspects to highlight new directions for future research in gait neuroprosthetics.

Neuromorphic on-chip recognition of saliva samples of COPD and healthy controls using memristive devices.

Chronic Obstructive Pulmonary Disease (COPD) is a life-threatening lung disease, affecting millions of people worldwide. Implementation of Machine Learning (ML) techniques is crucial for the effective management of COPD in home-care environments. However, shortcomings of cloud-based ML tools in terms of data safety and energy efficiency limit their integration with low-power medical devices. To address this, energy efficient neuromorphic platforms can be used for the hardware-based implementation of ML methods. Therefore, a memristive neuromorphic platform is presented in this paper for the on-chip recognition of saliva samples of COPD patients and healthy controls. Results of its performance evaluations showed that the digital neuromorphic chip is capable of recognizing unseen COPD samples with accuracy and sensitivity values of 89% and 86%, respectively. Integration of this technology into personalized healthcare devices will enable the better management of chronic diseases such as COPD.

Analogue pattern recognition with stochastic switching binary CMOS-integrated memristive devices.

Biological neural networks outperform current computer technology in terms of power consumption and computing speed while performing associative tasks, such as pattern recognition. The analogue and massive parallel in-memory computing in biology differs strongly from conventional transistor electronics that rely on the von Neumann architecture. Therefore, novel bio-inspired computing architectures have been attracting a lot of attention in the field of neuromorphic computing. Here, memristive devices, which serve as non-volatile resistive memory, are employed to emulate the plastic behaviour of biological synapses. In particular, CMOS integrated resistive random access memory (RRAM) devices are promising candidates to extend conventional CMOS technology to neuromorphic systems. However, dealing with the inherent stochasticity of resistive switching can be challenging for network performance. In this work, the probabilistic switching is exploited to emulate stochastic plasticity with fully CMOS integrated binary RRAM devices. Two different RRAM technologies with different device variabilities are investigated in detail, and their potential applications in stochastic artificial neural networks (StochANNs) capable of solving MNIST pattern recognition tasks is examined. A mixed-signal implementation with hardware synapses and software neurons combined with numerical simulations shows that the proposed concept of stochastic computing is able to process analogue data with binary memory cells.

Dielectrophoretic Immobilization of Yeast Cells Using CMOS Integrated Microfluidics.

This paper presents a dielectrophoretic system for the immobilization and separation of live and dead cells. Dielectrophoresis (DEP) is a promising and efficient investigation technique for the development of novel lab-on-a-chip devices, which characterizes cells or particles based on their intrinsic and physical properties. Using this method, specific cells can be isolated from their medium carrier or the mixture of cell suspensions (e.g., separation of viable cells from non-viable cells). Main advantages of this method, which makes it favorable for disease (blood) analysis and diagnostic applications are, the preservation of the cell properties during measurements, label-free cell identification, and low set up cost. In this study, we validated the capability of complementary metal-oxide-semiconductor (CMOS) integrated microfluidic devices for the manipulation and characterization of live and dead yeast cells using dielectrophoretic forces. This approach successfully trapped live yeast cells and purified them from dead cells. Numerical simulations based on a two-layer model for yeast cells flowing in the channel were used to predict the trajectories of the cells with respect to their dielectric properties, varying excitation voltage, and frequency.

AC electrokinetic immobilization of organic dye molecules.

The application of inhomogeneous AC electric fields for molecular immobilization is a very fast and simple method that does not require any adaptions to the molecule's functional groups or charges. Here, the method is applied to a completely new category of molecules: small organic fluorescence dyes, whose dimensions amount to only 1 nm or even less. The presented setup and the electric field parameters used allow immobilization of dye molecules on the whole electrode surface as opposed to pure dielectrophoretic applications, where molecules are attracted only to regions of high electric field gradients, i.e., to the electrode tips and edges. In addition to dielectrophoresis and AC electrokinetic flow, molecular scale interactions and electrophoresis at short time scales are discussed as further mechanisms leading to migration and immobilization of the molecules. Graphical Abstract.

Toward a Reliable Synaptic Simulation Using Al-Doped HfO2 RRAM.

The potential in a synaptic simulation for neuromorphic computation has revived the research interest of resistive random access memory (RRAM). However, novel applications require reliable multilevel resistive switching (RS), which still represents a challenge. We demonstrate in this work the achievement of reliable HfO2-based RRAM devices for synaptic simulation by performing the Al doping and the postdeposition annealing (PDA). Transmission electron microscopy and operando hard X-ray photoelectron spectroscopy results reveal the positive impact of Al doping on the formation of oxygen vacancies. Detailed I-V characterizations demonstrate that the 16.5% Al doping concentration leads to better RS properties of the device. In comparison with the other reported results based on HfO2 RRAM, our devices with 16.5% Al-doping and PDA at 450 °C show better reliable multilevel RS (∼20 levels) performance and an increased on/off ratio. The 16.5% Al:HfO2 sample with PDA at 450 °C shows good potentiation/depression characteristics with low pulse width (10 µs) along with a good On/Off ratio (>1000), good data retention at room temperature, and high temperature and good program/erase endurance characteristics with a pulse width of 50 ns. The synapse features including potentiation, depression, and spike time-dependent plasticity were successfully achieved using optimized Al-HfO2 RRAM devices. Our results demonstrate the beneficial effects of Al doping and PDA on the enhancement of the performances of RRAM devices for the synaptic simulation in neuromorphic computing applications.

Investigation of the Oxidation Behavior of Graphene/Ge(001) Versus Graphene/Ge(110) Systems.

The oxidation behavior of Ge(001) and Ge(110) surfaces underneath the chemical vapor deposition (CVD)-grown graphene films has been investigated experimentally and interpreted on the basis of ab initio calculations. Freshly grown samples were exposed to air for more than 7 months and periodically monitored by X-ray photoelectron spectroscopy, scanning electron microscopy, and Raman spectroscopy. The oxidation of Ge(110) started with incubation time of several days, during which the oxidation rate was supposedly exponential. After an ultrathin oxide grew, the oxidation continued with a slow but constant rate. No incubation was detected for Ge(001). The oxide thickness was initially proportional to the square root of time. After 2 weeks, the rate saturated at a value fivefold higher than that for Ge(110). We argue that after the initial phase, the oxidation is limited by the diffusion of oxidizing species through atomic-size openings at graphene domain boundaries and is influenced by the areal density and by the structural quality of the boundaries, whereby the latter determines the initial behavior. Prolonged exposure affected the surface topography and reduced the strain in graphene. In the last step, both the air-exposed samples were annealed in vacuum at 850 °C. This removed oxygen from the substrate and restored the samples to their initial state. These findings might constitute an important step toward further optimization of graphene grown on Ge.

Towards CMOS Integrated Microfluidics Using Dielectrophoretic Immobilization.

Dielectrophoresis (DEP) is a nondestructive and noninvasive method which is favorable for point-of-care medical diagnostic tests. This technique exhibits prominent relevance in a wide range of medical applications wherein the miniaturized platform for manipulation (immobilization, separation or rotation), and detection of biological particles (cells or molecules) can be conducted. DEP can be performed using advanced planar technologies, such as complementary metal-oxide-semiconductor (CMOS) through interdigitated capacitive biosensors. The dielectrophoretically immobilization of micron and submicron size particles using interdigitated electrode (IDE) arrays is studied by finite element simulations. The CMOS compatible IDEs have been placed into the silicon microfluidic channel. A rigorous study of the DEP force actuation, the IDE's geometrical structure, and the fluid dynamics are crucial for enabling the complete platform for CMOS integrated microfluidics and detection of micron and submicron-sized particle ranges. The design of the IDEs is performed by robust finite element analyses to avoid time-consuming and costly fabrication processes. To analyze the preliminary microfluidic test vehicle, simulations were first performed with non-biological particles. To produce DEP force, an AC field in the range of 1 to 5 V (peak-to-peak) is applied to the IDE. The impact of the effective external and internal properties, such as actuating DEP frequency and voltage, fluid flow velocity, and IDE's geometrical parameters are investigated. The IDE based system will be used to immobilize and sense particles simultaneously while flowing through the microfluidic channel. The sensed particles will be detected using the capacitive sensing feature of the biosensor. The sensing and detecting of the particles are not in the scope of this paper and will be described in details elsewhere. However, to provide a complete overview of this system, the working principles of the sensor, the readout detection circuit, and the integration process of the silicon microfluidic channel are briefly discussed.

Graphene Schottky Junction on Pillar Patterned Silicon Substrate.

A graphene/silicon junction with rectifying behaviour and remarkable photo-response was fabricated by transferring a graphene monolayer on a pillar-patterned Si substrate. The device forms a 0.11 eV Schottky barrier with 2.6 ideality factor at room temperature and exhibits strongly bias- and temperature-dependent reverse current. Below room temperature, the reverse current grows exponentially with the applied voltage because the pillar-enhanced electric field lowers the Schottky barrier. Conversely, at higher temperatures, the charge carrier thermal generation is dominant and the reverse current becomes weakly bias-dependent. A quasi-saturated reverse current is similarly observed at room temperature when the charge carriers are photogenerated under light exposure. The device shows photovoltaic effect with 0.7% power conversion efficiency and achieves 88 A/W photoresponsivity when used as photodetector.