Ultra-low Energy Research Articles

Triboelectric potential tuned oxide artificial tactile sensory platform with ultra-low power consumption

Sophistiated cognitive platform calls for multifunctional intelligent sensory devices with ultra-low power consumption. Triboelectric nanogenerator (TENG) provides an energy-efficient strategy to gate neuromorphic transistor with multiple sensing applications. To this end, a self-powered artificial tactile sensory neuron is proposed. Fish gelatin (GEL) and carboxylated-chitosan (CCs) are cross-linked to gate indium tin oxide (ITO) neuromorphic transistor. Thanks to the extremely strong protonic/electronic interfacial coupling at GEL-CCs based hydrogel/channel interface, good electrical performances are obtained at low voltage of only ∼0.8 V. The device demonstrates fundamental synaptic behaviors. Moreover, it exhibits ultra-low energy consumption of only ∼31.1 aJ for a single synaptic response with a good sensitivity of ∼4.65 dB. TENG is fabricated using polydimethylsiloxane elastomer and polytetrafluoroethylene, demonstrating good sensitivity of ∼8.0 V/kPa and good stabilities under pressure of 400 Pa. Thus, a tactile sensory platform is built with co-substrate configuration, integrating TENG and ITO neuromorphic transistor. Image edge detection is conceptually demonstrated by adopting Sobel algorithm. Furthermore, a wearable arterial pulse monitor is proposed based on the haptic sensory platform. The present work provides a candidate solution for realizing environment-friendly, energy-saving neuromorphic system and tactile perception system, demonstrating potentials in portable intelligent platform, prosthetics and human-machine interface.

Hierarchical processing enabled by 2D ferroelectric semiconductor transistor for low-power and high-efficiency AI vision system

The growth of data and Internet of Things challenges traditional hardware, which encounters efficiency and power issues owing to separate functional units for sensors, memory, and computation. In this study, we designed an α-phase indium selenide (α-In2Se3) transistor, which is a two-dimensional ferroelectric semiconductor as the channel material, to create artificial optic-neural and electro-neural synapses, enabling cutting-edge processing-in-sensor (PIS) and computing-in-memory (CIM) functionalities. As an optic-neural synapse for low-level sensory processing, the α-In2Se3 transistor exhibits a high photoresponsivity (2855 A/W) and detectivity (2.91 × 1014 Jones), facilitating efficient feature extraction. For high-level processing tasks as an electro-neural synapse, it offers a fast program/erase speed of 40 ns/50 µs and ultralow energy consumption of 0.37 aJ/spike. An AI vision system using α-In2Se3 transistors has been demonstrated. It achieved an impressive recognition accuracy of 92.63% within 12 epochs owing to the synergistic combination of the PIS and CIM functionalities. This study demonstrates the potential of the α-In2Se3 transistor in future vision hardware, enhancing processing, power efficiency, and AI applications.

Forecasting the carbon footprint of civil buildings under different floor area growth trends and varying energy supply methods

The energy consumption and carbon footprint of buildings are significantly impacted by variations in building area and the number of households. Therefore, it is crucial to forecast the growth trend of building area and number of households. A validated time series model is used to predict the new building area in Jilin Province from 2023 to 2030. The new building area in Jilin Province is expected to exhibit two trends of growth in the future: rapid growth (S1) and slow growth (S2). By 2030, under the S1 growth trend, the residential construction area and public building construction area in Jilin Province are expected to be 30.26 Mm2 (million square meters) and 7.23 Mm2, respectively. If the future floor area grows slowly under the S2 trend, the new floor area of different types will be 8.26 Mm2 and 1.33 Mm2 by 2030, respectively. The population growth shows a downward trend. Therefore, the energy consumption and carbon footprint of new buildings with different growth trends of floor areas and the number of households can be predicted. The energy consumption of new buildings shows an increasing trend from 0.32 Mtce in 2023 to 0.55 Mtce in 2030 under the S1 trend and a slight downward trend under the S2 trend. The carbon footprint is expected to be reduced by 0.017–0.311 million tons of CO2 when using heat pumps to supply 10–50% of the heat and wind and solar to supply 10–50% of the electricity. For every 10% increase in the use of ultra-low energy buildings, the energy consumption of civil buildings decreases in the range of 0.0063–0.028 Mtce. If the use of heat pumps and renewable energy increases by 10%, the energy consumption of civil buildings decreases in the range of 0.0054–0.0249 Mtce.

Modulating the ferromagnetism of Fe3GeTe2 with 3d transition metal adsorption and strain-engineering

Two-dimensional ferromagnetic materials hold great promise to develop energy-efficient magnetoelectric memory devices and next-generation spintronics. However, one of the crucial challenges for these materials is the realization of tunable magnetocrystalline anisotropy (MCA) to balance thermal stability and energy efficiency. Here, we systematically study the adsorption effects of 3d transition metals (3d-TMs) on the electronic structure and magnetic property of the Fe3GeTe2 (FGT) monolayer. The adsorption systems exhibit different ground state configurations depending on the adatoms, while the controlled perpendicular magnetic anisotropy has also been achieved. Notably, the Mn/FGT system can maintain the out-of-plane magnetic orientation with a changing amplitude of MCA energy up to 3.057 erg/cm2 as the external strain varies from −4% to 1%. In contrast, the Fe/FGT structure undergoes spin reorientation from in-plane to out-of-plan magnetization with a distinct modification behavior of MCA. We elucidate that the underlying atomistic mechanism mainly arises from the alteration of Fe-derived 3d-orbital states in response to the strain effect, leading to competitive changes in the different coupling states. These findings can not only provide useful guidance to optimize two-dimensional magnets for fundamental research but also reveal the promising potential of TMs/FGT materials for the development of ultra-low energy spintronic devices.

Water hyacinth derived hierarchical porous biochar absorbent: Ideal peroxydisulfate activator for efficient phenol degradation via an electron-transfer pathway

In this paper, a facile hydrothermal pretreatment and molten salt activation route was presented for preparing a self-doped porous biochar (HMBC) from a nitrogenous biomass precursor of water hyacinth. With an ultrahigh specific surface area (2240 m2 g−1), well-developed hierarchical porous structure, created internal structural defects and doped surface functionalities, HMBC exhibited an excellent adsorption performance and catalytic activity for phenol removal via peroxydisulfate (PDS) activation. Specifically, the porous structure promoted the adsorption of PDS on HMBC, forming a highly active HMBC/PDS* complex and thereby increasing the oxidation potential of the system. Meanwhile, the carbon defective structure, graphitic N and CO groups enhanced the electron transfer process, favoring the HMBC/PDS system to catalyze phenol oxidation via an electron transfer dominated pathway. Thus, the system degraded phenol effectively with an ultralow activation energy of 4.9 kJ mol−1 and a remarkable oxidant utilization efficiency of 8.2 mol mol-oxidant−1 h−1 g−1. More importantly, the system exhibited excellent resistance to water quality and good adaptability for decontaminating different organic pollutants with satisfactory mineralization efficiency. This study offers valuable insights into the rational designing of a low-cost biochar catalyst for efficient PDS activation towards organic wastewater remediation.

Emulation of neuro-mimetic dynamics via GaSb/Si heterojunction V-DGTFET leaky-integrate-fire silicon neuron

A leaky integrate and fire (LIF) neuron property is emulated using vertically developed double gate GaSb/Si tunnel field effect transistor (V-DGTFET). The suggested device can precisely replicate the neuronal dynamics after considering and proper validating the numerical simulations. Silvaco-TCAD tool is exploited for emulation of neuronal attributes and found in good agreement with earlier published research data. The article reports the membrane threshold voltage of 0.47 V which is 3.2x, 2.13x, 2.13x, 1.28x times less as compared to silicon nanowire, MOSFET, phase change device and Bulk FinFET neuron. The energy required for mimicking the neurons is computed as 0.748 aJ which is 1.2 × 109, 4 × 107, 4.68 × 107, 1.34 × 107, 7.59 × 106, 8.42 × 103, 4.30 × 103 3.34 × 105, 2.41 × 105, 1.52 × 106, 26.7,2, 3.88 × 103 times less compared to CMOS, phase change device, SOI CMOS, PCMO RRAM, SiNPN, Bulk FinFET, BTBT based PDSOI MOS, FBFET, LBIMOS, DGJLFET, Si NIPIN, DL-TFET, silicon nano wire respectively. The biasing at second gate is used to induce potential well at channel region to accumulate hole carriers produced due to impact ionization. The threshold current is achieved at 0.20 V drain voltage which is 15x, 14x, 10x, and 1.5x less as compared to FinFET, PD-SOI MOSFET, LBIMOS, and DLTFET respectively. Furthermore, calculated spiking frequency of V-DGTFET LIF neuron is 534 GHz. Thus, remarkable enhancement in frequency of operation and an ultra-low energy per spike makes this neuron device a potential solution for implementing the neuro-mimetic behaviour at large scale.

Federated reinforcement learning based task offloading approach for MEC-assisted WBAN-enabled IoMT

The exponential proliferation of wearable medical apparatus and healthcare information within the framework of the Internet of Medical Things (IoMT) introduces supplementary complexities pertaining to the elevated Quality of Service (QoS) of intelligent healthcare in the forthcoming 6G era. Healthcare services and applications need ultra-reliable data transfer and processing with ultra-low latency and energy usage. Wireless Body Area Network (WBAN) and Mobile Edge Computing (MEC) technologies enabled IoMT to handle large amounts of data sensing, transmission, and processing while maintaining good QoS. Traditional frame aggregation (FA) systems in WBAN, on the other hand, create an excessive number of control frames during data transmission, resulting in significant latency and energy consumption, as well as a lack of flexibility. A Federated Reinforcement Learning (FRL) based TO Approach is recommended in this research. In the beginning, different types of service-related information were separated into queues with equal QoS needs. The duration of the FA was then automatically determined by the aggregation vertex based on energy consumption, latency, and throughput using FRL. Finally, based on the existing status, the amount of tasks offloaded was determined. The simulation results demonstrate that, as compared to the baseline schemes, the suggested FRLTO efficiently reduces energy consumption and latency while enhancing throughput and total WBAN utilization. Numerical results show that the proposed scheme improves the throughput by 37.06% and reduced the energy consumption by around 69.84% and time delay by about 6.23%, as compared to the state-of-the-art existing baseline schemes.

Fully reconfigurable MEMS-based second-order coupled-resonator optical waveguide (CROW) with ultra-low tuning energy.

Integrated microring resonators are well suited for wavelength-filtering applications in optical signal processing, and cascaded microring resonators allow flexible filter design in coupled-resonator optical waveguide (CROW) configurations. However, the implementation of high-order cascaded microring resonators with high extinction ratios (ERs) remains challenging owing to stringent fabrication requirements and the need for precise resonator tunability. We present a fully integrated on-chip second-order CROW filter using silicon photonic microelectromechanical systems (MEMS) to adjust tunable directional couplers and a phase shifter using nanoscale mechanical out-of-plane waveguide displacement. The filter can be fully reconfigured with regard to both the ER and center wavelength. We experimentally demonstrated an ER exceeding 25 dB and continuous wavelength tuning across the full free spectral range of 0.123 nm for single microring resonator, and showed reconfigurability in second-order CROW by tuning the ER and resonant wavelength. The tuning energy for an individual silicon photonic MEMS phase shifter or tunable coupler is less than 22 pJ with sub-microwatt static power consumption, which is far better than conventional integrated phase shifters based on other physical modulation mechanisms.

Carbon-saving benefits of various end-of-life strategies for different types of building structures

Against the backdrop of advocating for ultra-low energy consumption buildings both domestically and internationally, much attention has been attracted to energy consumption and carbon emissions at the end-of-life (EoL) stage from the whole life cycle of buildings. Recycling of construction and demolition wastes (C&DW) is an inevitable step in achieving the goals of ultra-low energy consumption and “carbon neutrality.” In this paper, the life cycle assessment (LCA) theory was employed to study the carbon-saving benefits of four different EoL strategies (i.e., recycling, remanufacturing, reuse, and the integrated strategy) for various building structures (including frame, frame shear wall, shear wall, and light steel structures), in which the carbon-saving potential (CSP) was served as the evaluation index. The results show that compared with the traditional demolition and landfill disposal methods, the implementation of integrated management strategies for structural buildings in this study can reduce carbon emissions by 150.7–246.8 kgCO2-e/m2, with the carbon-savings of 11.9–34.8 times the carbon emissions generated by the abandoned landfill. Among the four structures, the light steel structure has the greatest CSP, reaching (73.75–77.24%), followed by the frame shear wall structure (42.67–46.93%), the frame structure (41.19–45.11%), and the shear wall structure (39.00–43.79%). When the building life span is 50 years, all CSPs of the four structural buildings significantly increased, at this point, deconstruction of the building is the most reasonable approach.

Abstract 15009: Extra-Low Energy Defibrillation With Epicardial Implantable Cardioverter Defibrillator Networks With Electrically Conductive Fibers in Disease Models

Introduction: We previously reported that extra-low energy (<0.6J) defibrillation is possible with the epicardial ICD net lead system (Ep-ICD) in normal canine hearts. Hypothesis: The aim of this study was to test whether the Ep-ICD can terminate ventricular fibrillation (Vf) with the same energy in two disease models (heart failure and long QT syndrome (LQT)). Methods: 1. Ep-ICD was constructed by fine tungsten fibers at the top and bottom of the cardiac mesh (Figure 1), connected to the RV and SVC coils of the DF-1 lead, respectively. 2. Heart failure was induced by rapid right ventricular pacing (170 bpm, 8 weeks). 3. Vf was induced by placing a 9V battery on the surface of the heart. DC shock was delivered with a minimum energy of 0.4J (Medtronic Viva XT). 4. Defibrillation threshold (DFT) was defined as 75% of defibrillation success. 5. LQT was simulated by reducing the conductance (Gk) of the IKS current by (x 1.0(CTL) or 0.8(LQT)) by computer simulation. Results: 1. DFT(J) of Ep-ICD in the canine HF model: 0.8, 1.2, 1.2. 3, Minimum DF energy(J): 0.6, 0.6, 0.8, 2. Maximum DF energy(J): 1.2, 1.6, 5, 52. Immediate return to sinus rhythm (SR) was achieved in most cases after DC delivery. In a small number of cases, brief periods of VT occurred, with rapid return to SR. No tachycardic arrhythmias occurred after return to SR. 3. The DFT was 0.39 for LQT and 0.52 for CTL. Conclusions: 1. DFT of Ep-ICD increased from 0.6 J in normal canine heart to 1.2 J in canine HF model, but still much lower energy used in current ICD system (36 J). 2. The Ep-ICD lead system is the least traumatic and still within the acceptable range while conscious. 3. DFT of LQT with Ep-ICD is 25% lower than normal heart. 4. Ultra-low energy DC can avoid conduction abnormalities or malignant arrhythmias after shock. Hearts return to SR immediately. 5. Ep-ICD can deliver the DC shock before loss of consciousness, improving quality of life and safety of ICD therapy.

PZT-Enabled MoS2 Floating Gate Transistors: Overcoming Boltzmann Tyranny and Achieving Ultralow Energy Consumption for High-Accuracy Neuromorphic Computing.

Low-power electronic devices play a pivotal role in the burgeoning artificial intelligence era. The study of such devices encompasses low-subthreshold swing (SS) transistors and neuromorphic devices. However, conventional field-effect transistors (FETs) face the inherent limitation of the "Boltzmann tyranny", which restricts SS to 60 mV decade-1 at room temperature. Additionally, FET-based neuromorphic devices lack sufficient conductance states for highly accurate neuromorphic computing due to a narrow memory window. In this study, we propose a pioneering PZT-enabled MoS2 floating gate transistor (PFGT) configuration, demonstrating a low SS of 46 mV decade-1 and a wide memory window of 7.2 V in the dual-sweeping gate voltage range from -7 to 7 V. The wide memory window provides 112 distinct conductance states for PFGT. Moreover, the PFGT-based artificial neural network achieves an outstanding facial-recognition accuracy of 97.3%. This study lays the groundwork for the development of low-SS transistors and highly energy efficient artificial synapses utilizing two-dimensional materials.

Physical Layer-Aware Digital-Analog Co-Design for Photonic Convolution Neural Network

Photonic convolution neural network (CNN) has shown tremendous capability to provide Tera-level operations/sec (OPS) for streaming object detection with ultra-low energy consumption. However, due to the nature of analog photonic computing which is constrained by accumulated noise from various sources, its physical operating precision is significantly lower than that of electronic artificial intelligence (AI) accelerators powered by tensor processing unit (TPU) and graphic processing unit (GPU). In this paper, we propose a robust digital-analog co-designed photonic CNN using error learning and Kirsch edge enhancement. Experimental results show that, nearly 96.1% recognition accuracy, compared with 34.2% in baseline photonic CNN, can be achieved at global 2-bit precision through physical error learning in dense layer and edge enhancement by photonic Kirsch operator. The proposed photonic CNN architecture is promising to provide 10 Tera-OPS scale computing speed deployed by the state of art PAM-4 optical modules with high energy efficiency.

Research on the current compensation mechanism of FJTL in ERSFQ

Energy-efficient rapid single flux quantum (ERSFQ) devices have unique energy advantages with a power consumption of less than 0.001fJ per single-bit switch, even considering the energy required for the refrigeration system, achieving ultra-low energy consumption of sub-fJ/bit. Combined with its typical operating frequency in the tens of GHz range, ERSFQ circuits have the potential to realize computing chips with extremely high computational power and energy efficiency, making them highly promising for applications in the post-Moore era. However, the performance improvement of ERSFQ circuits faces two challenges. One is how to reduce the size of their feeding Josephson transmission line (FJTL), which is required to stabilize the voltage bias of the ERSFQ circuits. The second challenge is how to improve their bias margin. The measured bias margin of ERSFQ circuits is relatively smaller than for Rapid Single Flux Quantum (RSFQ) circuits. This study investigates the current compensation mechanism of the FJTL and derives a formula for the compensating current ΔI in the main logic circuit by each pulse input to the FJTL. Based on the formula, a method is proposed to enhance the bias margin of ERSFQ circuits while reducing the size of FJTL by applying feeding pulses into the FJTL. We applied this method to an 8-bit ERSFQ shift register (SR), where the size of its FJTL is 4 JTLs (each JTL consists of 2 Josephson junctions). The measurement demonstrated an improvement of the bias margin from [84 %, 104 %] to [38 %, 104 %], confirming the feasibility of the proposed method. Furthermore, by comparing the measurement results of an 8-bit ERSFQ SR, where the size of its FJTL is 6 JTLs, it is demonstrated that applying this method can also reduce the size of the FJTL while maintaining a large margin.

Stretchable photosynaptic transistor with an ultralow energy consumption conferred using conjugated block copolymers/perovskite quantum dots nanocomposites

Emulation of photonic synapses offers a promising avenue for developing low-energy consumption of soft electronics, neurologically inspired robotics, and neuromorphic network computation. In this paper, a fully stretchable photosynaptic device with ultralow energy consumption using intrinsically stretchable poly(δ-decanolactone) (PDL)-based conjugated block copolymers (BCPs) with perovskite quantum dots (PeQDs) is first reported. The findings reveal that selectively choosing solvents for the PDL-based BCPs effectively regulates the assembly of P3HT and the accommodation of PeQDs, leading to improved self-aggregation of PeQDs, increased grain size, and optimized interfaces between P3HT and PeQDs The BCPs/PeQDs composite effectively emulates significant features of photonic synapses, such as paired-pulse facilitation (PPF), spike-dependent and short/long-term neuroplasticity, demonstrating excellent performance, including the fastest response time (1 ms), the highest current contrast (4.9 × 105), PPF (1.93) and ultra-low energy consumption (0.3 aJ) at an operating voltage of –0.1 mV. Furthermore, the BCP/PeQDs exhibit remarkable neuromuscular synapse characteristics, including high strain and bending tolerance and spike-dependent plasticity, enabling the devices to achieve high classification accuracy in artificial neural network simulations during tensile strain. The accommodation solvent selectivity of BCP/PeQDs suggests a promising strategy for advancing neurologically soft electronics, human-like pattern recognition, and neuromorphic computation.

Bright Transparent Scintillators with High Fraction BaCl2 : Eu2+ Nanocrystals Precipitation: An Ionic-Covalent Hybrid Network Strategy toward Superior X-Ray Imaging Glass-Ceramics.

Metal halide crystals are bright but hygroscopic scintillator materials that are widely used in X-ray imaging and detectors. Precipitating them in situ in glass to form glass ceramics (GCs) scintillator offers an efficient avenue for large-scale preparation, high spatial resolution, and excellent stability. However, precipitating a high fraction of metal halide nanocrystals in glass to maintain high light yield remains a challenge. Herein, an ionic-covalent hybrid network strategy for constructing GCs scintillator with high crystallinity (up to ≈37%) of BaCl2 : Eu2+ nanocrystals is presented. Experimental data and simulations of glass structure reveal that the Ba2+ -Cl- clustering promotes the high crystallization of BaCl2 nanocrystals. The ultralow phonon energy (≈200 cm-1 ) of BaCl2 nanocrystals and good Eu reduction effect enable high photoluminescence inter quantum efficiency (≈80.41%) in GC. GCs with varied crystallinity of BaCl2 : Eu2+ nanocrystals demonstrate efficient radioluminescence and tunable scintillator performance. They either outperform Bi4 Ge3 O14 single crystal by over 132% steady-state light yield or provide impressive X-ray imaging resolutions of 20 lp mm-1 . These findings provide a new design strategy for developing bright transparent GCs scintillators with a high fraction of metal halide nanocrystals for X-ray high-resolution imaging applications.

Dynamic Simulation Analysis of Influencing Factors of Energy Consumption of Public Building Envelope in Central China

It difficult for buildings in cold regions to reasonably meet ultra-low energy consumption requirements. To explore the influencing factors of building energy consumption in this type of climatic region, this paper takes a typical public building in Zhengzhou, China as a case study to dynamically simulate an entire building. The influence of the window glass, window-to-wall ratio (WWR), thermal insulation material thickness and other factors on building energy consumption is considered. The sensitivity of envelope design factors to energy consumption is analyzed by the orthogonal test. The results indicate that the effect of changing the technical parameters of the external envelope structure on building energy efficiency is significant. Moreover, range analysis is used to determine the primary and secondary factors with regard to their influence on the energy consumption of the building. The WWR, the type of external window glass and the thickness of the external wall insulation layer should be considered first in energy-saving design; the energy-saving potential of the roof insulation thickness is not as significant as these factors. The results of the study can provide a reference for the energy-saving design of public buildings in Central China.

Superamphiphobic surface with high aperture ratio interconnected pore structures for anti−condensation and repelling hot fluids

Pipeline transportation suffers from high-viscosity crude oil, wax-deposition, and hydrates blockage. Superamphiphobic surfaces with excellent repellency to various liquids have been proposed to solve those problems. However, they may lose superamphiphobicity once their micro-nanostructures were wetted by condensate droplets or hot fluids that require heat to improve liquidity. Here, we report a superamphiphobic P(HFIO)@SiO2@SS with a high aperture ratio pore structures (denoted as the PSSH) to resist water/oil-droplet-condensation and repel hot fluids. Its interconnected high aperture ratio pore structures and ultralow surface energy fluorinated side chains enhance the penetration resistance of droplets to micro-nanostructures and the driving force for condensate droplets to spontaneously roll away from its surface. Besides, it prevents crude oil (25 °C−72 °C) from sticking to its surface (25 °C), as well as molten paraffin droplets (104 °C) from adhering to the surface (104 °C), increasing the flow rate of 4–35 times. In addition, the PSSH reduces adhesive force to hydrates (3.2 °C) and wax (25 °C) by 92.1% and 89.4%, respectively. The PSSH demonstrates the great potential in handling pipeline blockage and improving the transport efficiency of fluids, offering alternative strategy to prepare superamphiphobic surfaces for oil and gas transportation, heat transfer, and other applications.

Minimizing energy consumption of IoT devices for O-RAN based IoT systems

The implementation of Open Radio Access Network (O-RAN) architecture in Internet of Things (IoT) systems has garnered significant attention as a means to fulfill the stringent requirements of ultra-low latency and ultra-low energy consumption in future IoT systems. Although the traditional edge network architecture has been extensively employed, it continues to pose challenges in terms of synchronizing the integration of global and local information for the design of an optimal offloading strategy in edge servers, thus hindering the reduction of latency and energy consumption in IoT devices. In an effort to decrease the latency and energy consumption of IoT devices (IoTDs), we propose a computation offloading problem and employs the Successive Convex Approximation (SCA) algorithm to convert the non-convex problem into a convex problem. The proposed strategy aims to minimize the energy consumption of IoTDs while ensuring Quality of Service (QoS) requirements. The results of the experiment demonstrate that the average energy consumption of terminal devices can be reduced by 20%. Additionally, this strategy is found to be more effective in reducing IoTDs' latency and energy consumption as compared to the traditional edge network architecture.

Electrostatic aid-free photo-floating gate two dimensional MoS2 synaptic transistors

Synaptic transistors are important component of neuromorphic computing systems, which is promising to reduce data traffic, time delay, and energy cost. However, most of synaptic behaviors normally need the aid of electrostatic gate voltage, which induces considerable consumption. Here, we propose a structure of the electrostatic aid-free photo-floating gate transistor based on the MoS2/MoOx/WSe2 heterojunctions, in which the MoOx acts as a unipolarity barrier layer and WSe2 functions as a photo-floating gate layer. Compared with the traditional floating-gate transistors, the designed device exhibits the distinct characteristics of visual synapse behaviors without the electrostatic aid gate, including the multilevel storage property of 13 stages, paired-pulse facilitation, the transition of short time plasticity to long time plasticity, and learning-forgetting-learning. This work provides a feasible strategy to construct neuromorphic computing device with ultra-low energy consumption.

Energy‐efficient organic photoelectric synaptic transistors with environment‐friendly CuInSe2 quantum dots for broadband neuromorphic computing

AbstractPhotoelectric synaptic device is a promising candidate component in brain‐inspired high‐efficiency neuromorphic computing systems. Implementing neuromorphic computing with broad bandwidth is, however, challenging owing to the difficulty in realizing broadband characteristics with available photoelectric synaptic devices. Herein, taking advantage of the type‐II heterostructure formed between environmentally friendly CuInSe2 quantum dots and organic semiconductor, broadband photoelectric synaptic transistors (BPSTs) that can convert light signals ranging from ultraviolet (UV) to near‐infrared (NIR) into post‐synaptic currents are demonstrated. Essential synaptic functions, such as pair‐pulse facilitation, the modulation of memory level, long‐term potentiation/depression transition, dynamic filtering, and learning‐experience behavior, are well emulated. More significantly, benefitting from broadband responses, information processing functions, including arithmetic computing and pattern recognition can also be simulated in a broadband spectral range from UV to NIR. Furthermore, the BPSTs exhibit obvious synaptic responses even at an ultralow operating voltage of −0.1 mV with an ultralow energy consumption of 75 aJ per event, and show their potential in flexible electronics. This study presents a pathway toward the future construction of brain‐inspired neural networks for high‐bandwidth neuromorphic computing utilizing energy‐efficient broadband photoelectric devices.