This article presents a 148-nW always-on wake-up system that drastically reduces the system power consumption of Internet of Things (IoT) sensor nodes while oftentimes operating in random-sparse-event (RSE) scenarios. To significantly reduce the long-term average (LTA) power consumption and realize multiapplication and intelligent event detection, three techniques are proposed: 1) In a three-stage pipelined event-driven architecture, a frame generator and a convolutional neural network intelligent inference engine (CNN IIE) in stage III are event-driven by the preliminary detectors in stage II, and the detectors are triggered by a level-crossing (LC) analog-to-digital converter (ADC), i.e., stage I, dramatically reducing the overall power consumption. 2) The clock-free pulse-based instant rate of change (IROC) feature extractor directly processes the asynchronous pulses of the LC-ADC outputs in the temporal domain instead of utilizing a conventional power-hungry frequency-domain method. 3) A reconfigurable IROC, the frame generator, and the CNN IIE provide adaptive intelligence for various IoT events, enhancing the accuracy of multipurpose detection with ultralow power. We demonstrate two artificial intelligence IoT (AIoT) applications at 0.6-V VDD. For electrocardiogram (ECG) recognition, one example works at a typical event rate (ER) of 4800/h, with an active power of 1.68 $μ \textW$ and a precision of up to 99%; the other is used for keyword spotting (KWS), where the chip achieves 378 nW at 720/h ER and 94% accuracy. The LTA power is bounded to 148 nW, while the event-driven chip is on call and waiting for events; this chip dominates the AIoT device battery life in RSE scenarios.
Generally, computer-science-oriented artificial neural networks (ANNs) and neuroscience-oriented spiking neural networks (SNNs) are two main approaches to develop brainspired non von Neumann computing systems. The goal of exploring complex artificial intelligence (AI) systems demandsgeneral neuromorphic hardware platforms compatible with both of them. However, as a result of the obvious differences in their fundamental mathematical expression and coding scheme, many neuromorphic platforms or deep neural network (DNN) accelerators accommodate only one of them. This brief presents a reconfigurable scalable neuromorphic chip based on digital leaky integrate-and-fire (LIF) neuron model targeting low-cost large-scale systems. By unifying ANN and SNN paradigms within a LIF neuron framework with point-to-point (P2P) communication, the chip can accommodate most popular neural networks. The chip adopts distributed on-chip memory architecture with a capacity of 64K neurons and 64M synapses. It achieves a peak throughput of 12.29 GSOP/s at 1.2 V, 192MHz and peak energy efficiency of 2.64 pJ/SOP at 890 mV, 24MHz.The results of implementations of a spike-based spatio-temporal memory model and ternary-weight event-based convolutional neural networks (CNNs) demonstrate outstanding compatibility of the chip.
This work introduces a second-order voltage-controlled oscillator (VCO)-based continuous-time delta-sigma modulator (CTDSM) that incorporates a distributed-input VCO as the second-stage integrator and quantizer. The distributed-input VCO topology virtually eliminates the VCO's voltage-to-frequency (V-F) parasitic pole. One of the key ideas of this article is to demonstrate the use of a capacitive-π network in the modulator's loop filter to break the constraint between the size of the modulator's inner capacitive digital-to-analog converter (DAC) and the factor by which the front-end Gm-C integrator is impedance scaled. This, in turn, helps to significantly reduce both analog and digital powers. The prototype chip has been fabricated in a 40-nm CMOS process. Despite not using any DAC calibration or explicit dynamic element matching (DEM) circuits, the worst case spurious-free dynamic range (SFDR) is -82 dBc across the signal bandwidth. The fabricated CTDSM achieves a 71.8-dB signal-to-noise-and-distortion ratio (SNDR) and a 74.5-dB dynamic range (DR) in a 10-MHz bandwidth at 655 MS/s, yielding an SNDR-based Walden figure of merit (FoM) of 45.6 fJ/step, an SNDR-based Schreier FoM of 167.2 dB, and a DR-based Schreier FoM of 169.9 dB.
V-Fe concentrate ore was applied to activate peroxydisulfate (PDS) for carbamazepine (CBZ) degradation. The excellent performance of V-Fe concentrate ore was mainly ascribed to the quick electron transfer from surface ≡V(III) and ≡V (IV) to ≡Fe(III) for ≡Fe(II) regeneration, which was confirmed by XPS and XAS analyses. This accelerated ≡Fe(II) regeneration could thus lead to quick formation of HO, SO4−, O2− and effective degradation of CBZ. The degradation rate of CBZ could be also expressed by a kinetic model, i.e., −d[CBZ]/dt = (0.83 mM-0.55 min-1(g/L)-0.65) [CBZ]0.29[PDS]1.26[V-Fe]0.65. Combined with the measured intermediates and the results of DFT calculation, CBZ degradation pathway was proposed systematically. Moreover, this catalyst displayed excellent recyclability and general applicability for a broad substrate scope. This study suggests low valent vanadium makes crucial contributions to the high activity of V-Fe-based catalysts, and improves the understanding of electron transfer mechanism between V and Fe in PDS activation process.
Herein, through supramolecular gel assisted pre-configuration, a novel bamboo-like porous graphitic carbon nitride (g-C3N4) deposited with single-atom Fe was successfully prepared and used for sulfite (S(IV)) activation. Unexpectedly, owing to the presence of single-atom Fe, hybrid material with only 2.5‰ of Fe exhibited 16 times higher S(IV) activation efficiency for diclofenac removal than pure g-C3N4 under visible light irradiation. Moreover, a synergetic effect of Fe and g-C3N4 was found to play the dominate role, and the synergetic factor was calculated to be 0.84. The synergetic mechanism mainly related to the generation of surface Fe-S(IV) complex, which could be affected by S(IV) species or the presence of H2PO4−. DCF removal was significantly enhanced at alkaline condition, but the enhancement was mainly attributed to photocatalysis but not synergetic effect. Decarboxylation, hydroxylation, chlorine abstraction and cleavages at bridging N atom were the main degradation pathways of DCF, which agreed well with Fukui index prediction. The toxicity of DCF was alleviated during the degradation process through successful mineralization of chlorine atoms.
Perfluorooctane sulfonate (PFOS) has drawn increasing attention due to its omnipresence and adverse health effects. We prepared a new adsorptive photocatalyst, Ga/TNTs@AC, based on activated carbon and TiO2, and tested the adsorption and subsequent solid-phase photodegradation of PFOS. Ga/TNTs@AC showed faster adsorption kinetics and higher affinity for PFOS than the parent AC, and could degrade 75.0% and mineralize 66.2% of pre-sorbed PFOS within 4-h UV irradiation. The efficient PFOS photodegradation also regenerates Ga/TNTs@AC, allowing for repeated uses without invoking chemical regenerants. The superior photoactivity is attributed to the oxygen vacancies, which not only suppressed recombination of the e−/h+ pairs, but also facilitated O2− generation. Both h+ and O2− played critical roles in the PFOS degradation, which starts with cleavage of the sulfonate group and converts it into PFOA that is then decarboxylated and defluorinated following the stepwise defluorination mechanism. Ga/TNTs@AC holds the potential for more cost-effective PFOS degradation.
Exploring the specific characteristics of pharmaceuticals and personal care products (PPCPs) via adsorption and degradation are scientific and practical significance to control their release to water matrix. In this work, a good adsorbent and ion-exchange material, i.e., titanate nanotubes (TNTs), was employed for adsorption of ciprofloxacin (CIP, a model PPCPs). The adsorption behaviors and mechanisms of CIP with different dissociated species by TNTs were studied through both experimental and theoretical calculations. The multilayered TNTs with high BET surface area (272.3 m2/g) and large pore volume (1.26 cm3/g) exhibited good adsorption property for CIP. The CIP species (i.e., CIP+, CIP±, CIP−) at various pH exhibited significantly different adsorption favorability. Adsorption kinetics and isotherms data revealed that TNTs offered the high uptake for CIP+ (Qmax = 464.47 μmol/g or 153.90 mg/g at pH 5) than CIP± and CIP−. Characterizations indicated the formation of Ti−O−N linkage between CIP molecules and TNTs after adsorption, suggesting the chemical interaction between CIP and TNTs. Density functional theory (DFT) calculations reveal variation on pH affects the protonation/deprotonation state of CIP, and then changes the distribution of molecular orbitals and the electrostatic potential (ESP) energy of CIP. ESP follows the trend as: CIP+ (180.57 kcal/mol) > CIP± (146.78 kcal/mol) > CIP− (12.30 kcal/mol), indicating the side of piperazine ring in CIP oriented to TNTs dominates the CIP adsorption. The integrated experimental and theoretical results, for the first time, suggest that ESP energy can serve as the indicator and predictor of adsorption ability for the PPCPs molecules with various speciation, and can help to deeply describe the adsorption mechanism of PPCPs. In addition, TNTs have great application for the removal of PPCPs through adsorption in practical wastewater treatment area.
Obtaining elastic properties of organic-rich shales through conventional geomechanical testing could be challenging due to availability of good quality core plugs and significant heterogeneous nature of the samples. In this regard, force spectroscopy methods, nanoindentation and atomic force microscopy (AFM) are two main powerful techniques to characterize elastic properties in nano/microscale. In this study, we investigate the applicability of these two methods on the same samples, by quantifying elastic modulus from the Bakken Shale. AFM provided us with modulus maps of higher resolution compared to the modulus maps from the nanoindentation that were created via geostatistical methods. Moreover, results from these methods were compared to demonstrate the advantages and shortcomings of each and discripancy in the outcome. To do so, multi-cluster deconvolution approach was adopted in the statistical analysis on the nanoindentation data, demonstrating 3 separate clusters and mechanical phases. AFM technique, similarly, distinguished three separate (mineral and organic) phases based on the corresponding modulus values, though with higher accuracy compared to nanoindentation and better distinction and less tolerance. It was found that nanoindentation, because it collects discrete datapoints that are farther apart from each other when thermal maturity is increased in the samples, would have difficulty to separate organic matter from intermediary phases. Overall, the range of modulus for each phase was larger in the data that was obtained by nanoindentation compared to the AFM which can be interpreted to the size of the tip and general higher resolution in the later one which is expected to probe a single particle rather than an aggregate of particles.