Laboratoire Nanotechnologies et Nanosystèmes

The low communication bandwidth between memory and processing units in conventional von Neumann machines does not support the requirements of emerging applications that rely extensively on large sets of data. More recent computing paradigms, such as high parallelization and near‐memory computing, help alleviate the data communication bottleneck to some extent, but paradigm‐shifting concepts are required. In‐memory computing has emerged as a prime candidate to eliminate this bottleneck by colocating memory and processing. In this context, resistive switching (RS) memory devices is a key promising choice, due to their unique intrinsic device‐level properties, enabling both storing and computing with a small, massively‐parallel footprint at low power. Theoretically, this directly translates to a major boost in energy efficiency and computational throughput, but various practical challenges remain. A qualitative and quantitative analysis of several key existing challenges in implementing high‐capacity, high‐volume RS memories for accelerating the most computationally demanding computation in machine learning (ML) inference, that of vector‐matrix multiplication (VMM), is presented. The monolithic integration of RS memories with complementary metal–oxide–semiconductor (CMOS) integrated circuits is presented as the core underlying technology. The key existing design choices in terms of device‐level physical implementation, circuit‐level design, and system‐level considerations is reviewed and an outlook for future directions is provided.

We present a droplet-based surface acoustic wave (SAW) system designed to viably detach biological cells from a surface and sort cell types based on differences in adhesion strength (adhesion contrast) without the need to label cells with molecular markers. The system uses modulated SAW to generate pulsatile flows in the droplets and efficiently detach the cells, thereby minimizing the SAW excitation power and exposure time. As a proof of principle, the system shows efficient sorting of HEK 293 from A7r5 cells based on adhesion contrast. Results are obtained in minutes with sorting purity and efficiency reaching 97% and 95%, respectively.

. Our approach offers great potential for the development of fast specific detection systems based on affinity monitoring.

The present study presents a three-dimensional numerical analysis using the finite element method of nanofluid enhanced heat transfer in micro heat exchanger equipped with triangular fins. The new configuration based on an existent system where a jet impingement supplies a microchannel structure. A modification of the heat exchanger geometry in the z-direction is added allowing a uniform wall temperature profile. The micro heat exchanger is assumed to be well insulated. The hot fluid (water) flows in the lower channel with a fixed velocity (uw_in = 20 mm/s) and cold fluid (CNT-water nanofluid) flows in the upper channel which is equipped with triangular fins with a velocity (unf_in) ranged from 5 to 45 mm/s. The nanofluid is considered homogeneous with temperature-dependent thermophysical properties and the CNT nanoparticles volume fraction is varied from 0 to 5%. The results are presented in term of thermal and flow fields, heatlines, overall heat transfer coefficient, thermal effectiveness, and thermal performance factor (TPF). It was found that the performances of the heat exchanger are significatively improved using the CNT nanofluid and the triangular fins. But the TPF increase with the CNT volume fractions and decreases with the fin’s height.

Thermal management complexity increases in high-performance chips, where the heat loads vary spatially and temporally, while liquid cooling systems are usually designed for most stringent stationary conditions. Several works developed heat transfer enhancement techniques to increase the cooling capacity of liquid cooled heat sinks, but pumping power is increased in a permanent way due to the addition of elements within the channels. Here, a liquid cooling self-adaptive heat sink that can efficiently adapt the distribution of its heat extraction capacity to time dependent and non-uniform heat load scenarios is proposed. Numerical design of the mesoscale cooling device with bimorph metal/SMA fins, definition of the fabrication and training procedure of the SMA fins to reach the desired behavior and experimental assessment is presented. The capacity of the self-adaptive fins to locally boost the heat transfer is numerically and experimentally demonstrated. Results obtained show that the self-adaptive fins can improve the temperature uniformity by 63% with respect to plain channel. The reduction in thermal resistance using bimorph metal/SMA fins sample allows the surface maximum temperature gradient to remain almost constant although heat flux increases. Energy savings are maximized in applications where partial load intervals contributes significantly to the overall operating period.

Resistive switching and transport mechanisms of Al2O3/TiO2−x memristor crosspoint devices have been investigated at cryogenic temperatures down to 1.5 K, for the future development of memristor-based cryogenic electronics. We report successful resistive switching of our devices in the temperature range of 300–1.5 K. The current–voltage curves exhibit negative differential resistance effects between 130 K and 1.5 K, attributed to a metal–insulator transition of the Ti4O7 conductive filament. The resulting highly nonlinear behavior is associated with an ION/IOFF diode ratio of 84 at 1.5 K, paving the way for selector-free cryogenic passive crossbars. Temperature-dependent thermal activation energies related to the conductance at low bias (20 mV) are extracted for memristors in a low resistance state, suggesting hopping-type conduction mechanisms. Finally, the transport mechanism analysis at 1.5 K indicates that for all resistance states, the conduction follows the space-charge limited current model in low fields, whereas trap-assisted tunneling dominates in higher fields.

Organic Electrochemical Transistors are considered today as a key technology\nto interact with biological medium through their intrinsic ionic-electronic\ncoupling. In this paper, we show how this coupling can be finely tuned (in\noperando) post-microfabrication via electropolymerization technique. This\nstrategy exploits the concept of adaptive sensing where both transconductance\nand impedance are tunable and can be modified on-demand to match different\nsensing requirements. Material investigation through Raman spectroscopy, atomic\nforce microscopy and scanning electron microscopy reveals that\nelectropolymerization can lead to a fine control of PEDOT microdomains\norganization, which directly affect the iono-electronic properties of OECTs. We\nfurther highlight how volumetric capacitance and effective mobility of\nPEDOT:PSS influence distinctively the transconductance and impedance of OECTs.\nThis approach shows to improve the transconductance by 150% while reducing\ntheir variability by 60% in comparison with standard spin-coated OECTs.\nFinally, we show how to the technique can influence voltage spike rate hardware\nclassificationwith direct interest in bio-signals sorting applications.\n

Abstract Resistive switching (RS) devices are promising forms of non-volatile memory. However, one of the biggest challenges for RS memory applications is the device-to-device (D2D) variability, which is related to the intrinsic stochastic formation and configuration of oxygen vacancy (V O ) conductive filaments (CFs). In order to reduce the D2D variability, control over the formation and configuration of oxygen vacancies is paramount. In this study, we report on the Zr doping of TaO x -based RS devices prepared by pulsed-laser deposition as an efficient means of reducing the V O formation energy and increasing the confinement of CFs, thus reducing D2D variability. Our findings were supported by XPS, spectroscopic ellipsometry and electronic transport analysis. Zr-doped films showed increased V O concentration and more localized V O s, due to the interaction with Zr. DC and pulse mode electrical characterization showed that the D2D variability was decreased by a factor of seven, the resistance window was doubled, and a more gradual and monotonic long-term potentiation/depression in pulse switching was achieved in forming-free Zr:TaO x devices, thus displaying promising performance for artificial synapse applications.

In this paper, hexagonal germanium dioxide (GeO2) nanostructures with different morphologies and sizes were successfully synthesized by a simple and fast electrodeposition method. We investigated the electrochemical growth mechanism and the structural and optical properties of the products through Fourier transform infrared spectroscopy (FTIR), dispersive X-ray spectroscopy (EDX), X-ray diffraction (XRD) and cathodoluminescence (CL). The results reveal that the electrodeposited GeO2 nanostructures are pure, dense, and highly crystalline. The XRD analyses indicate that the resulting GeO2 crystals only show peaks related to the α-quartz structure. CL measurements exhibit strong blue and green light emissions related to oxygen vacancies in the core of the GeO2 crystals. The resulting nanostructures may have potential application in integrated optical devices in the future.

Bringing bodies close together at sub-micron distances can drastically enhance radiative heat transfer, leading to heat fluxes greater than the blackbody limit set by Stefan-Boltzmann law. This effect, known as near-field radiative heat transfer (NFRHT), has wide implications for thermal management in microsystems, as well as technological applications such as direct heat to electricity conversion in thermophotovoltaic cells. Here, we demonstrate NFRHT from microfabricated hotplates made by surface micromachining of [Formula: see text]/[Formula: see text] thin films deposited on a sacrificial amorphous Si layer. The sacrificial layer is dry etched to form wide membranes ([Formula: see text]) separated from the substrate by nanometric distances. Nickel traces allow both resistive heating and temperature measurement on the micro-hotplates. We report on two samples with measured gaps of [Formula: see text] and [Formula: see text]. The membranes can be heated up to [Formula: see text] under vacuum with no mechanical damage. At [Formula: see text] we observed a 6.4-fold enhancement of radiative heat transfer compared to far-field emission for the smallest gap and a 3.5-fold enhancement for the larger gap. Furthermore, the measured transmitted power exhibits an exponential dependence with respect to gap size, a clear signature of NFRHT. Calculations of photon transmission probabilities indicate that the observed increase in heat transfer can be attributed to near-field coupling by surface phonon-polaritons supported by the [Formula: see text] films. The fabrication process presented here, relying solely on well-established surface micromachining technology, is a key step toward integration of NFRHT in industrial applications.

This paper compares plasmonic substrates manufactured by three lithography methods: E-beam, soft and hard UV NanoImprint Lithography. The different plasmonic modes existing in samples made of an array of gold nanostructures on gold film are investigated for biochemical detections taking advantage of Surface Plasmon Resonance Imaging (SPRI) and Surface-Enhanced Raman Scattering (SERS). Recently, it has been shown that this geometry of substrate is of great interest for both SPRI and SERS measurements. A comparison of their performances obtained by the different lithographic methods is provided. In particular, due to limitations in NanoImprint Lithographic techniques, the impact of sidewall geometry of nanostructures is investigated in regard to plasmonic properties. Thus, experimental optical characterization analyses have been carried out on samples and compared with the numerical simulations.

Novel computing architectures based on resistive switching memories (also known as memristors or RRAMs) have been shown to be promising approaches for tackling the energy inefficiency of deep learning and spiking neural networks. However, resistive switch technology is immature and suffers from numerous imperfections, which are often considered limitations on implementations of artificial neural networks. Nevertheless, a reasonable amount of variability can be harnessed to implement efficient probabilistic or approximate computing. This approach turns out to improve robustness, decrease overfitting and reduce energy consumption for specific applications, such as Bayesian and spiking neural networks. Thus, certain non-idealities could become opportunities if we adapt machine learning methods to the intrinsic characteristics of resistive switching memories. In this short review, we introduce some key considerations for circuit design and the most common non-idealities. We illustrate the possible benefits of stochasticity and compression with examples of well-established software methods. We then present an overview of recent neural network implementations that exploit the imperfections of resistive switching memory, and discuss the potential and limitations of these approaches.

surface grown at 420 °C become larger at higher growth temperatures, and 3D islands appear on the surface at 565 °C. Atomic force microscopy phase measurements reveal a composition difference between the islands and the pits, whereas the sample grown at 420 °C appears to be homogeneous. Confocal Raman spectra show that all the N atoms are bonded to Al instead of Ga. Using apertureless scanning near-field optical microscopy, the luminescence of a gold tip is mapped over the surface of the sample grown at 565 °C. We extract the shift of the tip's surface plasmon resonance and determine the variation in the refractive index between the islands and the pits to be close to 0.2. Numerical simulations of the tip luminescence while in contact with the sample predict a similar variation of ∼0.3 in the refractive indices between AlGaAs islands and AlN pits, a substantially smaller value than the difference in the bulk refractive indices of the two media (∼1.8), which we attribute to a convolution of material distribution in an uneven topography. The excellent agreement between simulation and experiments supports the hypothesis of nitrogen-clustering in the pits.

An adaptive inference method for crossbar (AIDX) is presented based on an optimization scheme for adjusting the duration and amplitude of input voltage pulses. AIDX minimizes the long-term effects of memristance drift on artificial neural network accuracy. The sub-threshold behavior of memristor has been modeled and verified by comparing with fabricated device data. The proposed method has been evaluated by testing on different network structures and applications, e.g., image reconstruction and classification tasks. The results showed an average of 60% improvement in convolutional neural network (CNN) performance on CIFAR10 dataset after 10000 inference operations as well as a 78.6% error reduction in image reconstruction.

Exploration of memristors' behavior at cryogenic temperatures has become crucial due to the growing interest in quantum computing and cryogenic electronics. In this context, our study focuses on the characterization at cryogenic temperatures (4.2 K) of TiO2−x-based memristors fabricated with a CMOS-compatible etch-back process. We demonstrate a so-called cryogenic reforming (CR) technique performed at 4.2 K to overcome the well-known metal-insulator transition (MIT), which limits the analog behavior of memristors at low temperatures. This cryogenic reforming process was found to be reproducible and led to a durable suppression of the MIT. This process allowed to reduce by ∼20% the voltages required to perform DC resistive switching at 4.2 K. Additionally, conduction mechanism studies of memristors before and after cryogenic reforming from 4.2 to 300 K revealed different behaviors above 100 K, indicating a potential change in the conductive filament stoichiometry. The reformed devices exhibit a conductance level that is 50 times higher than ambient-formed memristor, and the conduction drop between 300 and 4.2 K is 100 times smaller, indicating the effectiveness of the reforming process. More importantly, CR enables analog programming at 4.2 K with typical read voltages allowing to store up to 4 bits of information on a single CR memristor. Suppressing the MIT improved the analog switching dynamics of the memristor leading to ∼250% larger on/off ratios during long-term depression (LTD)/long-term potentiation (LTP) resistance tuning. This enhancement opens up the possibility of using TiO2−x-based memristors to be used as synapses in neuromorphic computing at cryogenic temperatures.

International audience

This article proposes a dimensionless thermal analysis of micro-fabricated membranes as Pirani gauges for pressure measurements. Use of dimensionless numbers simplifies the mode l and facilitates understanding. Our model's predictions are consistent with experimental results obtained from heated suspended SiO2/SiN membranes with a sub-micrometer separation distance with the substrate (500 nm). Other systems reported in the literature also confirmed the modelling. This framework is a powerful prediction tool as it allows a study of the effects of key parameters on the sensing pressure range, including geometry, material properties and radiative heat fluxes. It also addresses the effect of the operating mode, either constant temperature or constant power. Furthermore, we propose a methodology for rapid design of Pirani gauge arrays to reach a broad range of pressure measurements from atmospheric pressure to high vacuum (10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-3</sup> Pa) without microcontroller use. We prove that this new formalism offers a way to optimize the geometry to reach the target application, which essentially depends on the power consumption, material choice and the pressure measurement range.

Herein, the effect of a 2 nm thin aluminum layer inserted between the ferroelectric layer and the top electrode in a TiN//TiN stack deposited by reactive magnetron sputtering is investigated. The oxidation of the interfacial layer during annealing due to scavenging of the impacts both the ferroelectric properties and the electrical conductivity of the junction. It is shown that the overall conductivity of the junction is boosted 20 folds while the resistance ratio between the positive and negative polarization states is increased from 1.3 up to 3.7. Through a systematic analysis of programming conditions, pulse duration, and height, we show that both the remanent polarization and On/Off current ratio can be enhanced at the expanse of the endurance leading to a trade‐off.

Love wave (L-SAW) sensors have been used to probe cell monolayers, but their application to detect changes beyond the focal adhesion points on cell monolayers, as viscosity changes on the cytoskeleton, has not been explored. In this work we present for the first time a Love wave sensor with tuned penetration depth and sensitivity to potentially detect mechanical changes beyond focal adhesion points of cell monolayers. We designed and fabricated a Love wave sensor operating at 30 MHz with sensitivity to detect viscous changes between 0.89 and 3.3 cP. The Love wave sensor was modeled using an acoustic transmission line model, whereas the response of interdigital transducers (IDTs) was modeled with the Campbell’s cross-field circuit model. Our design uses a substrate with a high electromechanical coupling coefficient (LiNbO3 36Y-X), and an 8-µm polymeric guiding layer (SU-8). The design aims to overcome the high insertion losses of viscous liquid environments, and the loss of sensitivity due to the low frequency. The fabricated sensor was tested in a fluidic chamber glued directly to the SU-8 guiding layer. Our experiments with liquids of viscosity similar to those expected in cell monolayers showed a measurable sensor response. In addition, experimentation with SaOs-2 cells within a culture medium showed measurable responses. These results can be of interest for the development of novel cell-based biosensors, and novel characterization tools for cell monolayers.

This paper presents the design and the fabrication of an ultra-thin vapor chamber exclusively composed of silicon, aimed to be integrated in microelectronic chips to spread high-density hot spots. A process flow fully compatible with the presence of a circuit on the front side has been developed and a 1 x 1 cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> prototype with an internal vapor cavity and a wick thickness of 210 μm has been designed and fabricated. The wick is composed of a matrix of micropillars with 5 μm diameter and 30 μm high in a square arrangement. The cavity is obtained by plasma activated direct bonding of two wafers with complementary cavities. The spreading performances have been estimated by a finite element method (FEM) modeling and presents higher performances compared to copper heat spreader with 4°C less temperature difference for a 4 W and 1 × 1 mm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> hotspot.