scispace - formally typeset
Search or ask a question

Showing papers by "STMicroelectronics published in 2020"


Journal ArticleDOI
TL;DR: In this article, a tensile strain was applied to a 300nm-thick GeSn layer with 5.4 at% Sn, which is an indirect-bandgap semiconductor as-grown, to transform it into a direct-band gap semiconductor that supports lasing.
Abstract: Strained GeSn alloys are promising for realizing light emitters based entirely on group IV elements. Here, we report GeSn microdisk lasers encapsulated with a SiNx stressor layer to produce tensile strain. A 300 nm-thick GeSn layer with 5.4 at% Sn, which is an indirect-bandgap semiconductor as-grown, is transformed via tensile strain engineering into a direct-bandgap semiconductor that supports lasing. In this approach, the low Sn concentration enables improved defect engineering and the tensile strain delivers a low density of states at the valence band edge, which is the light hole band. We observe ultra-low-threshold continuous-wave and pulsed lasing at temperatures up to 70 K and 100 K, respectively. Lasers operating at a wavelength of 2.5 μm have thresholds of 0.8 kW cm−2 for nanosecond pulsed optical excitation and 1.1 kW cm−2 under continuous-wave optical excitation. The results offer a path towards monolithically integrated group IV laser sources on a Si photonics platform. Continuous-wave lasing in strained GeSn alloys is reported at temperatures of up to 100 K. The approach offers a route towards a group-IV-on-silicon laser.

138 citations


Journal ArticleDOI
TL;DR: Biomedicine, catalysis and sensing are the application areas mainly discussed in this review, highlighting advantages of laser-synthesized nanoparticles for these types of applications and, once partially resolved, the limitations to the technique for large-scale applications.
Abstract: Laser synthesis emerges as a suitable technique to produce ligand-free nanoparticles, alloys and functionalized nanomaterials for catalysis, imaging, biomedicine, energy and environmental applications. In the last decade, laser ablation and nanoparticle generation in liquids has proven to be a unique and efficient technique to generate, excite, fragment and conjugate a large variety of nanostructures in a scalable and clean way. In this work, we give an overview on the fundamentals of pulsed laser synthesis of nanocolloids and new information about its scalability towards selected applications. Biomedicine, catalysis and sensing are the application areas mainly discussed in this review, highlighting advantages of laser-synthesized nanoparticles for these types of applications and, once partially resolved, the limitations to the technique for large-scale applications.

104 citations


Journal ArticleDOI
29 Nov 2020-Sensors
TL;DR: An extensive review of the low-cost particulate matter sensors currently available on the market, their electronic characteristics, and their applications in published literature and from specific tests shows that most of the reviewed LCPMS can accurately monitor PM changes in the environment and exhibit good performances with accuracy that, in some conditions, can reach R2 values up to 0.99.
Abstract: The concerns related to particulate matter’s health effects alongside the increasing demands from citizens for more participatory, timely, and diffused air quality monitoring actions have resulted in increasing scientific and industrial interest in low-cost particulate matter sensors (LCPMS). In the present paper, we discuss 50 LCPMS models, a number that is particularly meaningful when compared to the much smaller number of models described in other recent reviews on the same topic. After illustrating the basic definitions related to particulate matter (PM) and its measurements according to international regulations, the device’s operating principle is presented, focusing on a discussion of the several characterization methodologies proposed by various research groups, both in the lab and in the field, along with their possible limitations. We present an extensive review of the LCPMS currently available on the market, their electronic characteristics, and their applications in published literature and from specific tests. Most of the reviewed LCPMS can accurately monitor PM changes in the environment and exhibit good performances with accuracy that, in some conditions, can reach R2 values up to 0.99. However, such results strongly depend on whether the device is calibrated or not (using a reference method) in the operative environment; if not, R2 values lower than 0.5 are observed.

73 citations


Journal ArticleDOI
TL;DR: The use of HD computing to classify electroencephalography (EEG) error-related potentials for noninvasive brain–computer interfaces is described and achieves on average 5% higher single-trial classification accuracy compared to a conventional machine learning method on this task.
Abstract: The mathematical properties of high-dimensional (HD) spaces show remarkable agreement with behaviors controlled by the brain. Computing with HD vectors, referred to as “hypervectors,” is a brain-inspired alternative to computing with numbers. HD computing is characterized by generality, scalability, robustness, and fast learning, making it a prime candidate for utilization in application domains such as brain–computer interfaces. We describe the use of HD computing to classify electroencephalography (EEG) error-related potentials for noninvasive brain–computer interfaces. Our algorithm naturally encodes neural activity recorded from 64 EEG electrodes to a single temporal–spatial hypervector without requiring any electrode selection process. This hypervector represents the event of interest, can be analyzed to identify the most discriminative electrodes, and is used for recognition of the subject’s intentions. Using the full set of training trials, HD computing achieves on average 5% higher single-trial classification accuracy compared to a conventional machine learning method on this task (74.5% vs. 69.5%) and offers further advantages: (1) Our algorithm learns fast: using only 34% of training trials it achieves an average accuracy of 70.5%, surpassing the conventional method. (2) Conventional method requires prior domain expert knowledge, or a separate process, to carefully select a subset of electrodes for a subsequent preprocessor and classifier, whereas our algorithm blindly uses all 64 electrodes, tolerates noises in data, and the resulting hypervector is intrinsically clustered into HD space; in addition, most preprocessing of the electrode signal can be eliminated while maintaining an average accuracy of 71.7%.

63 citations


Proceedings ArticleDOI
16 Feb 2020
TL;DR: An active interposer integrating a Switched Capacitor Voltage Regulator (SCVR) for on-chip power management, flexible system interconnect topologies between all chiplets for scalable cache coherency support, and energy-efficient 3D-plugs for dense inter-layer communication is presented.
Abstract: In the context of high-performance computing and big-data applications, the quest for performance requires modular, scalable, energy-efficient, low-cost manycore systems. Partitioning the system into multiple chiplets 3D-stacked onto large-scale interposers - organic substrate [1], 2.5D passive interposer [2] or silicon bridge [3] -leads to large modular architectures and cost reductions in advanced technologies by the Known Good Die (KGD) strategy and yield management. However, these approaches lack flexible efficient long-distance communications, smooth integration of heterogeneous chiplets, and easy integration of less-scalable analog functions, such as power management [4] and system IOs. To tackle these issues, this paper presents an active interposer integrating: i) a Switched Capacitor Voltage Regulator (SCVR) for on-chip power management; ii) flexible system interconnect topologies between all chiplets for scalable cache coherency support; iii) energy-efficient 3D-plugs for dense inter-layer communication; iv) a memory-IO controller and PHY for socket communication. The chip (Fig. 2.3.7) integrates 96 cores in 6 chiplets in 28nm FDSOI CMOS, 30-stacked in a face-to-face configuration using 20µm-pitch micro-bumps (µ-bumps) onto a 200 mm2 active interposer with 40µm-pitch Through Silicon Via (TSV) middle in a 65nm technology node. Even though complex functions are integrated, active-interposer yield is high thanks to the mature 65nm node and a reduced complexity (0.08transistors/µm2), with 30% of interposer area devoted to a SCVR variability-tolerant capacitors scheme.

40 citations


Journal ArticleDOI
TL;DR: A 128 single-photon avalanche diode (SPAD) motion detection-triggered time-of-flight (ToF) sensor is implemented in STMicroelectronics 40 nm CMOS SPAD foundry process to acquire depth frames only when inter-frame intensity changes are detected.
Abstract: A 128 $\times $ 128 single-photon avalanche diode (SPAD) motion detection-triggered time-of-flight (ToF) sensor is implemented in STMicroelectronics 40 nm CMOS SPAD foundry process. The sensor combines vision and ToF ranging functions to acquire depth frames only when inter-frame intensity changes are detected. The 40 $\mu \text{m}\,\,\times $ 20 $\mu \text{m}$ pixels integrate two 16-bit time-gated counters to acquire ToF histograms and repurpose them to compare two vision frames without the requirement for additional out-of-pixel frame memory resources. An embedded column-parallel ToF and vision processor performs on-chip vision frame comparison and binary frame output compression as well as controlling the time-resolved histogram sampling. The sensor achieves a maximum 32.5 kframes/s in vision modality and 500 frames/s in motion detection-triggered ToF over a measured 3.5 m distance with 1.5 cm accuracy. The vision function reduces the sensor power consumption by 70% over continuous ToF operation and allows the sensor to gate the ToF laser emitter to reduce the system power when no motion activity is observed.

36 citations


Journal ArticleDOI
TL;DR: A comprehensive review of the different signal conditioning front-end topologies targeted in particular at sub-femtofarad resolution reveals that resolution, measuring time, area occupation and energy/conversion lower than 100 aF, 1 ms 0.1 mm2, and 100 pJ/conv.
Abstract: Capacitance detection is a universal transduction mechanism used in a wide variety of sensors and applications. It requires an electronic front-end converting the capacitance variation into another more convenient physical variable, ultimately determining the performance of the whole sensor. In this paper we present a comprehensive review of the different signal conditioning front-end topologies targeted in particular at sub-femtofarad resolution. Main design equations and analysis of the limits due to noise are reported in order to provide the designer with guidelines for choosing the most suitable topology according to the main design specifications, namely energy consumption, area occupation, measuring time and resolution. A data-driven comparison of the different solutions in literature is also carried out revealing that resolution, measuring time, area occupation and energy/conversion lower than 100 aF, 1 ms 0.1 mm 2 , and 100 pJ/conv. can be obtained by capacitance to digital topologies, which therefore allow to get the best compromise among all design specifications.

34 citations


Journal ArticleDOI
TL;DR: In this paper, a low-cost chemical bath deposition (CBD) and thermal annealing was used to construct a nanostructured and porous NiO thin film (nanoporous NiO film) composed of NiO nanoparticles (30−50 nm).
Abstract: NiO-based NO2 sensors operating at room temperature are attracting great attentions due to their promising energy and cost saving performances. However, only a few reports showed high sensitivity and selectivity in the sub-ppm concentration range and low limit of detection (LoD). In this work, we designed and fabricated by a low-cost chemical bath deposition (CBD) and thermal annealing a nanostructured and porous NiO thin film (nanoporous NiO film) composed of NiO nanoparticles (30−50 nm). The nanoporous NiO film was then applied as sensing element for the NO2 detection at room temperature, demonstrating a high response to sub-ppm level NO2 (140 ppb), excellent selectivity and stability, and a very low LoD of 20 ppb. The NO2 sensing mechanism was investigated and satisfactorily modelled by two energetically different and independent adsorption sites at room temperature. Both sites contribute to the NO2 detection at room temperature while only one site contributes at higher temperatures. The described low-cost fabrication method and the discussed superior NO2 sensing performances at room temperature make the nanoporous NiO film a promising NO2 sensor for environmental monitoring.

34 citations


Journal ArticleDOI
TL;DR: GeSn alloys are nowadays considered as the most promising materials to build Group IV laser sources on silicon (Si) in a full complementary metal oxide semiconductor compatible approach.
Abstract: GeSn alloys are nowadays considered as the most promising materials to build Group IV laser sources on silicon (Si) in a full complementary metal oxide semiconductor-compatible approach. Recent GeS...

33 citations


Journal ArticleDOI
08 Dec 2020-Sensors
TL;DR: An unobtrusive, wearable, and wireless system for the pre-screening and follow-up in the domestic environment of specific sleep-related breathing disorders and results are encouraging: sensitivity and precision around 90% were achieved in detecting more than 500 apnea episodes.
Abstract: We propose an unobtrusive, wearable, and wireless system for the pre-screening and follow-up in the domestic environment of specific sleep-related breathing disorders. This group of diseases manifests with episodes of apnea and hypopnea of central or obstructive origin, and it can be disabling, with several drawbacks that interfere in the daily patient life. The gold standard for their diagnosis and grading is polysomnography, which is a time-consuming, scarcely available test with many wired electrodes disseminated on the body, requiring hospitalization and long waiting times. It is limited by the night-by-night variability of sleep disorders, while inevitably causing sleep alteration and fragmentation itself. For these reasons, only a small percentage of patients achieve a definitive diagnosis and are followed-up. Our device integrates photoplethysmography, an accelerometer, a microcontroller, and a bluetooth transmission unit. It acquires data during the whole night and transmits to a PC for off-line processing. It is positioned on the nasal septum and detects apnea episodes using the modulation of the photoplethysmography signal during the breath. In those time intervals where the photoplethysmography is detecting an apnea, the accelerometer discriminates obstructive from central type thanks to its excellent sensitivity to thoraco-abdominal movements. Tests were performed on a hospitalized patient wearing our integrated system and the type III home sleep apnea testing recommended by The American Academy of Sleep Medicine. Results are encouraging: sensitivity and precision around 90% were achieved in detecting more than 500 apnea episodes. Least thoraco-abdominal movements and body position were successfully classified in lying down control subjects, paving the way toward apnea type classification.

32 citations


Journal ArticleDOI
TL;DR: A versatile single complementary metal–oxide–semiconductor chip forming a platform to address personalized needs through on-chip multimodal optical and electrochemical detection that will reduce the number of tests that patients must take is presented.
Abstract: Precision metabolomics and quantification for cost-effective rapid diagnosis of disease are the key goals in personalized medicine and point-of-care testing. At present, patients are subjected to multiple test procedures requiring large laboratory equipment. Microelectronics has already made modern computing and communications possible by integration of complex functions within a single chip. As More than Moore technology increases in importance, integrated circuits for densely patterned sensor chips have grown in significance. Here, we present a versatile single complementary metal–oxide–semiconductor chip forming a platform to address personalized needs through on-chip multimodal optical and electrochemical detection that will reduce the number of tests that patients must take. The chip integrates interleaved sensing subsystems for quadruple-mode colorimetric, chemiluminescent, surface plasmon resonance, and hydrogen ion measurements. These subsystems include a photodiode array and a single photon avalanche diode array with some elements functionalized to introduce a surface plasmon resonance mode. The chip also includes an array of ion sensitive field-effect transistors. The sensor arrays are distributed uniformly over an active area on the chip surface in a scalable and modular design. Bio-functionalization of the physical sensors yields a highly selective simultaneous multiple-assay platform in a disposable format. We demonstrate its versatile capabilities through quantified bio-assays performed on-chip for glucose, cholesterol, urea, and urate, each within their naturally occurring physiological range.

Journal ArticleDOI
TL;DR: An efficient method to estimate jitter in a chain of CMOS inverters in the presence of multiple noise sources, including the power supply noise, input data noise, and the ground bounce noise is presented.
Abstract: This paper presents an efficient method to estimate jitter in a chain of CMOS inverters in the presence of multiple noise sources, including the power supply noise, input data noise, and the ground bounce noise. For this purpose, necessary noise transfer functions are derived and the recently developed EMPSIJ method is advanced to handle cascaded CMOS inverter stages. Results from the proposed method are compared with the results from a conventional EDA simulator, which demonstrate a significant speed-up using the proposed method for a comparable accuracy.

Journal ArticleDOI
20 Jul 2020
TL;DR: In this paper, a 40 Gbps direct detection of chip-integrated silicon-germanium avalanche p-i-n photo receiver driven with low-bias supplies at 1.55 µm wavelength is reported.
Abstract: Photodetectors are cornerstone components in integrated optical circuits and are essential for applications underlying modern science and engineering. Structures harnessing conventional crystalline materials are typically at the heart of such devices. In particular, group-IV semiconductors such as silicon and germanium open up more possibilities for high-performing on-chip photodetection thanks to their favorable electrical and optical properties at near-infrared wavelengths and processing compatibility with modern chip manufacturing. However, scaling the performance of silicon-germanium photodetectors to technologically relevant levels and benefiting from improved speed, reduced driving bias, enhanced sensitivity, and lowered power consumption arguably remains key for densely integrated photonic links in mainstream shortwave infrared optical communications. Here we report on a reliable 40 Gbps direct detection of chip-integrated silicon-germanium avalanche p-i-n photo receiver driven with low-bias supplies at 1.55 µm wavelength. The avalanche photodetection scheme calls upon fabrication steps commonly used in complementary metal-oxide-semiconductor foundries, alleviating the need for complex epitaxial wafer structures and/or multiple ion implantation schemes. The photo receiver exhibits an internal multiplication gain of 120, a high gain-bandwidth product up to 210 GHz, and a low effective ionization coefficient of ∼0.25. Robust and stable photodetection at 40 Gbps of on–off keying modulation is achieved at low optical input powers, without any need for receiver electronic stages. Simultaneously, compact avalanche p-i-n photodetectors with submicrometric heterostructures promote error-free operation at transmission bit rates of 32 Gbps and 40 Gbps, with power sensitivities of −12.8dBm and −11.2dBm, respectively (for 10−9 error rate and without error correction coding during use). Such a performance in an on-chip avalanche photodetector is a significant step toward large-scale integrated optoelectronic systems. These achievements are promising for use in data center networks, optical interconnects, or quantum information technologies.

Journal ArticleDOI
TL;DR: In this paper, supercontinuum generation in nitrogen-rich (N-rich) silicon nitride waveguides fabricated through back-end complementary-metal-oxide-semiconductor (CMOS)-compatible processes on a 300mm platform was reported.
Abstract: We report supercontinuum generation in nitrogen-rich (N-rich) silicon nitride waveguides fabricated through back-end complementary-metal-oxide-semiconductor (CMOS)-compatible processes on a 300 mm platform. By pumping in the anomalous dispersion regime at a wavelength of 1200 nm, two-octave spanning spectra covering the visible and near-infrared ranges, including the O band, were obtained. Numerical calculations showed that the nonlinear index of N-rich silicon nitride is within the same order of magnitude as that of stoichiometric silicon nitride, despite the lower silicon content. N-rich silicon nitride then appears to be a promising candidate for nonlinear devices compatible with back-end CMOS processes.

Journal ArticleDOI
TL;DR: It is shown that nonlinear devices such as electronic transistors exhibit major advantages enabling realization of low-cost and portable circuits for the emerging applications in these frequency ranges.
Abstract: We review the recent advances on the implementation of electronic circuits that operate in the millimeter-wave (30–300 GHz) and terahertz (300–3000 GHz) frequency ranges. The focus of this article is on nonlinear phenomena in electronics. The different implementations of nonlinear circuits for the sake of millimeter-wave and terahertz signal generation are studied in this paper. The challenges of signal generation are examined and the benefits and limitations of different schemes of signal generation are discussed. It is shown that nonlinear devices such as electronic transistors exhibit major advantages enabling realization of low-cost and portable circuits for the emerging applications in these frequency ranges. We also review linear and nonlinear design methodologies employing the properties of electromagnetic waves. The electronic systems designed based on the presented ideas are shown to push the previously unbeatable limits of operation in millimeter-wave and terahertz frequency ranges. A discussion on remaining challenges and future directions concludes the paper.

Book ChapterDOI
10 May 2020
TL;DR: A duplex-based authenticated encryption scheme based on a new permutation called Friet-P, designed with a novel approach for cryptographic permutations and block ciphers that takes fault-attack resistance into account, is presented.
Abstract: In this work we present a duplex-based authenticated encryption scheme \(\textsc {Friet}\) based on a new permutation called \(\textsc {Friet-P}\). We designed \(\textsc {Friet-P}\) with a novel approach for cryptographic permutations and block ciphers that takes fault-attack resistance into account and that we introduce in this paper.

Journal ArticleDOI
TL;DR: In this article, the authors describe the realization process of a microfluidic tool, from computer-aided design (CAD) to the proof-of-concept application as a capture device for circulating tumor cells (CTCs).

Journal ArticleDOI
TL;DR: NLR-5 deserves prospective validation to identify mUC patients with poor prognosis following ICIs to identify patients with metastatic urothelial cancer unlikely to benefit from immune-checkpoint inhibitors.
Abstract: To identify patients with metastatic urothelial cancer (mUC) unlikely to benefit from immune-checkpoint inhibitors (ICIs). We explored the predictive and prognostic values of baseline neutrophil-to-lymphocyte ratio (NLR), with cut-offs ≥ 3 and ≥ 5, and of a urothelial immune prognostic index (UIPI, based on increased NLR and LDH), on 146 patients. NLR and UIPI significantly predicted progressive disease and progression-free survival with both cut-offs (p = 0.0069, p = 0.0034, p = 0.0160, p = 0.0063; p < 0.001, p = 0.021, p = 0.014, p = 0.026; for NLR-3, NLR-5, UIPI-3, UIPI-5, respectively) and overall survival when NLR cut-off was ≥ 5 (p = 0.03 and p = 0.024, for NLR-5 and UIPI-5, respectively). NLR-5 deserves prospective validation to identify mUC patients with poor prognosis following ICIs.

Journal ArticleDOI
TL;DR: In this paper, the authors focus on various sources of nonlinearities induced by the interaction with the surrounding fluid and by occurrence of geometric large transformations in piezo-micromirrors with large opening angles.
Abstract: In this work, we address the modelling of piezo-micromirrors with large opening angles. We focus on various sources of nonlinearites induced by the interaction with the surrounding fluid and by occurrence of geometric large transformations. Specific attention is also devoted to the piezoelectric coupling coefficient. Piezoelectric thin films have gained attention as key materials for the actuation of micro devices since they provide high drive forces, enabling devices with higher performances compared to electrostatic actuation. However they also induce material nonlinearities that cannot be neglected. The model is validated using experimental data obtained for various excitation levels. [2020-0229]

Journal ArticleDOI
TL;DR: In this paper, the opto-electrical properties of hetero-structured pin photodetectors were investigated for use in power efficient optical links operating at 40 Gbps, with a device energy dissipation of only few fJ per bit.
Abstract: Optical interconnects are promising alternatives to copper-based wirings in on-chip communications. Recent advances in integrated group-IV nanophotonics should address a range of challenges related with speed, energy consumption, and cost. Monolithically integrated germanium pin photodetectors on silicon-on-insulator (SOI) waveguides are indispensable devices in this buoyant research field. Here, we comprehensively investigate the opto-electrical properties of hetero-structured pin photodetectors. All photodetectors were fabricated on top of 200-mm SOI substrates using industrial-scale semiconductor manufacturing processes. Under a low-bias voltage supply of 1 V, pin photodetectors exhibit dark-currents from 5 nA to 100 nA, dark current densities from 0.404 A/cm 2 to 0.808 A/cm 2 , responsivities in a range of 0.17 A/W to 1.16 A/W, and cut-off frequencies from 7 GHz to 35 GHz, respectively. Such achievements make them promising for use in power-efficient optical links operating at 40 Gbps, with a device energy dissipation of only few fJ per bit.

Journal ArticleDOI
TL;DR: In this article, the lasing characteristics of GeSn alloys with Sn contents ranging from 7% to 10.5% were examined and the results offer new perspectives for future designs of geSn-based laser sources.
Abstract: GeSn alloys are nowadays considered as the most promising materials to build Group IV laser sources on silicon (Si) in a full complementary metal oxide semiconductor-compatible approach. Recent GeSn laser developments rely on increasing the band structure directness, by increasing the Sn content in thick GeSn layers grown on germanium (Ge) virtual substrates (VS) on Si. These lasers nonetheless suffer from a lack of defect management and from high threshold densities. In this work we examine the lasing characteristics of GeSn alloys with Sn contents ranging from 7 \% to 10.5 \%. The GeSn layers were patterned into suspended microdisk cavities with different diameters in the 4-\SI{8 }{\micro\meter} range. We evidence direct band gap in GeSn with 7 \% of Sn and lasing at 2-\SI{2.3 }{\micro\meter} wavelength under optical injection with reproducible lasing thresholds around \SI{10 }{\kilo\watt\per\square\centi\meter}, lower by one order of magnitude as compared to the literature. These results were obtained after the removal of the dense array of misfit dislocations in the active region of the GeSn microdisk cavities. The results offer new perspectives for future designs of GeSn-based laser sources.

Proceedings ArticleDOI
01 Jun 2020
TL;DR: In this article, the performance of 28nm FD-SOI transistors was evaluated for the first time down to ultra low temperatures (UL T), at T = 1 00mK.
Abstract: Variability of28nm FD-SOI transistors is evaluated for the first time down to ultra low temperatures (UL T), at T= 1 00mK. High performance is achieved at UL T for short channel transistors, with $\mathrm{I}_{\mathrm{ON}} > 1\mathrm{mA}\mu \mathrm{m}$ and $\mathrm{I}_{\mathrm{OFF}}$ below the equipment accuracy $(\mathrm{V}_{\mathrm{TH}})$ and current gain factor $(\beta)$ variabilities. Besides that, we demonstrated that the increase of $\mathrm{V}_{\mathrm{TH}}$ and $\beta$ variabilities at low temperature remains reasonably low in comparison to RT values and other CMOS technologies, so that it should not be detrimental to circuit operation in this range of temperatures.

Proceedings ArticleDOI
18 Nov 2020
Abstract: REACTION is an EU Innovation Action Funded Project which aims to develop the first worldwide 200mm Silicon Carbide (SiC) Pilot Line for SiC based Power technology. The project, which mainly addresses the Smart Mobility societal challenge (car electrification), will allow to match the ever-increasing demand of requirements in terms of quality and cost constraint for next decade generation’s power electronics. A major strength of the project is the complete value chain implementation of the Pilot Line, which integrates and optimizes the cooperation and partnership among producers of 8" SiC substrates and equipment developers, as well as SiC process technologists, RTOs and end-users partners, in order to deliver the final applications. Hence, to achieve innovative SiC power devices with improved performances, along with cost and size reduction, are the most relevant requirements addressed in the project that are expected to lead to a new stronger European supply chain for very compact SiC converters, ideal for the addressed Automotive application.

Journal ArticleDOI
TL;DR: A video object segmentation framework that leverages the combined advantages of user feedback for segmentation and gamification strategy by simulating multiple game players through a reinforcement learning (RL) model that reproduces human ability to pinpoint moving objects and using the simulated feedback to drive the decisions of a fully convolutional deep segmentation network.
Abstract: Integrating human-provided location priors into video object segmentation has been shown to be an effective strategy to enhance performance, but their application at large scale is unfeasible. Gamification can help reduce the annotation burden, but it still requires user involvement. We propose a video object segmentation framework that leverages the combined advantages of user feedback for segmentation and gamification strategy by simulating multiple game players through a reinforcement learning (RL) model that reproduces human ability to pinpoint moving objects and using the simulated feedback to drive the decisions of a fully convolutional deep segmentation network. Experimental results on the DAVIS-17 benchmark show that: 1) including user-provided prior, even if not precise, yields high performance; 2) our RL agent replicates satisfactorily the same variability of humans in identifying spatiotemporal salient objects; and 3) employing artificially generated priors in an unsupervised video object segmentation model reaches state-of-the-art performance.

Proceedings ArticleDOI
01 Feb 2020
TL;DR: The ever-increasing demand for high network capacities and escalating data centers have pushed the boundary from discrete transceivers toward the integration of monolithic electro-optical ICs, and the cost advantage of a fully integrated silicon solution will eventually push classical discrete implementations toward obsolescence.
Abstract: The ever-increasing demand for high network capacities and escalating data centers have pushed the boundary from discrete transceivers toward the integration of monolithic electro-optical ICs [1]. Furthermore, complex modulation schemes such as PAM-4 have been introduced to improve the trade-off between circuit bandwidth, power consumption, data-rate and optical-link range compared to NRZ signaling. Therefore, the cost advantage of a fully integrated silicon solution will eventually push classical discrete implementations toward obsolescence.

Journal ArticleDOI
TL;DR: It is demonstrated that conjugation of a trehalose moiety to silybin significantly increases both water solubility and stability in blood serum without significantly compromising its anti-aggregation properties.
Abstract: Alzheimer’s disease (AD) is linked to the abnormal accumulation of amyloid β peptide (Aβ) aggregates in the brain. Silybin B, a natural compound extracted from milk thistle (Silybum marianum), has ...

Journal ArticleDOI
TL;DR: In this article, near interface oxide traps (NIOTs) in lateral 4H-SiC MOSFETs were investigated combining transient gate capacitance measurements (C-t) and state of the art scanning transmission electron microscopy in electron energy loss spectroscopy (STEM-EELS) with sub-nm resolution.
Abstract: Studying the electrical and structural properties of the interface of the gate oxide (SiO2) with silicon carbide (4H-SiC) is a fundamental topic, with important implications for understanding and optimizing the performances of metal-oxide-semiconductor field effect transistor (MOSFETs). In this paper, near interface oxide traps (NIOTs) in lateral 4H-SiC MOSFETs were investigated combining transient gate capacitance measurements (C-t) and state of the art scanning transmission electron microscopy in electron energy loss spectroscopy (STEM-EELS) with sub-nm resolution. The C-t measurements as a function of temperature indicated that the effective NIOTs discharge time is temperature independent and electrons from NIOTs are emitted toward the semiconductor via-tunnelling. The NIOTs discharge time was modelled taking into account also the interface state density in a tunnelling relaxation model and it allowed to locate traps within a tunnelling distance up to 1.3nm from the SiO2/4H-SiC interface. On the other hand, sub-nm resolution STEM-EELS revealed the presence of a Non-Abrupt (NA) SiO2/4H-SiC interface. The NA interface shows the re-arrangement of the carbon atoms in a sub-stoichiometric SiOx matrix. A mixed sp2/sp3 carbon hybridization in the NA interface region suggests that the interfacial carbon atoms have lost their tetrahedral SiC coordination.

Journal ArticleDOI
TL;DR: A custom Human Activity Recognition system is presented based on the resource-constrained Hardware (HW) implementation of a new partially binarized Hybrid Neural Network, which achieves much higher accuracy than Binarized Neural Network.
Abstract: A custom Human Activity Recognition system is presented based on the resource-constrained Hardware (HW) implementation of a new partially binarized Hybrid Neural Network. The system processes data in real-time from a single tri-axial accelerometer, and is able to classify between 5 different human activities with an accuracy of 97.5% when the Output Data Rate of the sensor is set to 25 Hz. The new Hybrid Neural Network (HNN) has binary weights (i.e. constrained to +1 or −1) but uses non-binarized activations for some layers. This, in conjunction with a custom pre-processing module, achieves much higher accuracy than Binarized Neural Network. During pre-processing, the measurements are made independent from the spatial orientation of the sensor by exploiting a reference frame transformation. A prototype has been realized in a Xilinx Artix 7 FPGA, and synthesis results have been obtained with TSMC CMOS 65 nm LP HVT and 90 nm standard cells. Best result shows a power consumption of $6.3~\mu \text{W}$ and an area occupation of 0.2 mm $^{\mathbf {2}}$ when real-time operations are set, enabling in this way, the possibility to integrate the entire HW accelerator in the auxiliary circuitry that normally equips inertial Micro Electro-Mechanical Systems (MEMS).

Journal ArticleDOI
TL;DR: In this article, the effect of surface oxidation on the crystallization of Ge-rich Ge-Sb-Te materials was studied, promising for Phase Change Memories working at high temperatures (>350°C).

Proceedings ArticleDOI
01 Sep 2020
TL;DR: This paper describes how Deep but Tiny Neural Networks (DTNN) can be designed to be parsimonious and can be automatically converted into a STM32 microcontroller-optimized C-library through X-CUBE-AI toolchain, and proposes the integration of the obtained library with Miosix, a Real Time Operating System tailored for resource constrained and tiny processors, which is an enabling factor for system scalability and multi tasking.
Abstract: Collecting vast amount of data and performing complex calculations to feed modern Numerical Weather Prediction (NWP) algorithms require to centralize intelligence into some of the most powerful energy and resource hungry supercomputers in the world. This is due to the chaotic complex nature of the atmosphere which interpretation require virtually unlimited computing and storage resources. With Machine Learning (ML) techniques, a statistical approach can be designed in order to perform weather forecasting activity. Moreover, the recently growing interest in Edge Computing Tiny Intelligent architectures is proposing a shift towards the deployment of ML algorithms on Tiny Embedded Systems (ES). This paper describes how Deep but Tiny Neural Networks (DTNN) can be designed to be parsimonious and can be automatically converted into a STM32 microcontroller-optimized C-library through X-CUBE-AI toolchain; we propose the integration of the obtained library with Miosix, a Real Time Operating System (RTOS) tailored for resource constrained and tiny processors, which is an enabling factor for system scalability and multi tasking. With our experiments we demonstrate that it is possible to deploy a DTNN, with a FLASH and RAM occupation of 45,5 KByte and 480 Byte respectively, for atmospheric pressure forecasting in an affordable cost effective system. We deployed the system in a real context, obtaining the same prediction quality as the same DNN model deployed on the cloud but with the advantage of processing all the necessary data to perform the prediction close to environmental sensors, avoiding raw data traffic to the cloud.