For neural network (NN) applications at the edge ofAI, computing-in-memory (CIM) demonstrates promising energyefficiency. However, when the network size grows while fulfillingthe accuracy requirements of increasingly complicated applicationscenarios, significant memory consumption becomes an issue.Model pruning is a typical compression approach for solvingthis problem, but it does not fully exploit the energy efficiencyadvantage of conventional CIMs, because of the dynamic distributionof sparse weights and the increased data movement energyconsumption of reading sparsity indexes from outside the chip.Therefore, we propose a vector-wise dynamic-sparsity controllingand computing in-memory structure (DS-CIM) that accomplishesboth sparsity control and computation of weights in SRAM, toimprove the energy efficiency of the vector-wise sparse pruningmodel. Implemented in a 65 nm CMOS process, the measurementresults show that the proposed DS-CIM macro can save upto 50.4% of computational energy consumption, while ensuringthe accuracy of vector-wise pruning models. The test chip canalso achieve 87.88% accuracy on the CIFAR-10 dataset at 4-bitprecision in inputs and weights, and it achieves 530.2TOPS/W(normalized to 1 bit) energy efficiency.
High-precision large dynamic-range (DR) current-sensing front-ends are widely used in biomedical applications, such as patch-clamp, molecular concentration detection, and gene sequencing. The new gene sequencers require low-noise analog front-ends capable of sensing large DR current (>100 dB) down to sub-pA-level. At this level of precision, oversampled data converters are usually used. However, given the limited oversampling ratio in high throughput applications, it is very challenging to achieve a sub-pA-level sensitivity and >100dB DR within the limited area and energy budgets [1]. In [2], a 140dB DR is achieved using a multi-bit delta-sigma modulator (DSM), but the power consumption is over 1mW and the current sensitivity is limited to 6.3pA. An hourglass ADC achieving a 100fA sensitivity and 140dB DR is presented in [3], but is limited by conversion rate and relatively high power consumption (295μW). For a 100Hz bandwidth, its noise floor increases to 18pA.
Peracetic acid (PAA), an alternative disinfectant of chlorine, has drawn increasing attention in the application of wastewater treatment. However, little is known about the influence of water matrices on PAA-induced organic micropollutant (OMP) degradation. Here, we found that the coexisting bromide ions (Br–) in water can trigger the oxidation of OMP during PAA treatment but probably result in higher ecotoxicity. Br– can efficiently decompose PAA with a species-specific rate constant (kPAAH/Br–) of 0.198 ± 0.003 M–1·s–1. The thus generated HOBr led to a significant abatement (31.8–81.3%) of OMPs (17α-ethinylestradiol, sulfamethoxazole, naproxen, and phenol) after a 1 h reaction at pH 7.1. The coexisting H2O2 component in the PAA solution can competitively consume HOBr and inhibit OMP transformation. The OMP degradation in the PAA/Br– process was highly pH-dependent and preferred acidic conditions. Furthermore, a comprehensive model was established to simulate the reaction kinetics of the OMP degradation by the PAA/Br– process with good accuracy. High-performance/electrospray ionization-triple quadrupole mass spectrometry results indicated the generation of various brominated products, with higher model-predicted toxicity than their parent compounds. This work significantly advances the understanding of the role of Br– in OMP oxidation by PAA and alerts the possible environmental health risks.
Accurate identification of radicals in advanced oxidation processes (AOPs) is important to study the mechanisms on radical production and subsequent oxidation-reduction reaction. The commonly applied radical quenching experiments cannot provide direct evidences on generation and evolution of radicals in AOPs, while electron paramagnetic resonance (EPR) is a cutting-edge technology to identify radicals based on spectral characteristics. However, the complexity of EPR spectrum brings uncertainty and inconsistency to radical identification and mechanism clarification. This work presented a comprehensive study on identification of radicals by in-situ EPR analysis in four typical UV-based homogenous AOPs, including UV/H2O2, UV/peroxodisulfate (and peroxymonosulfate), UV/peracetic acid and UV/IO4− systems. Radical formation mechanism was also clarified based on EPR results. A reliable EPR method using organic solvents was proposed to identify alkoxy and alkyl radicals (CH3C(=O)OO·, CH3C(=O)O· and ·CH3) in UV/PAA system. Two activation pathways for radical production were proposed in UV/IO4− system, in which the produced IO3·, IO4·, ·OH and hydrated electron were precisely detected. It is interesting that addition of specific organic solvents can effectively identify oxygen-center and carbon-center radicals. A key parameter in EPR spectrum for 5,5-dimethyl-1-pyrroline N-oxide (DMPO) spin adduct, AH, is ranked as: ·CH3 (23 G) >·OH (15 G) >IO3· (12.9 G) >O2·− (11 G) ≥·OOH (9–11 G) ≥IO4· (9–10 G) ≥SO4·− (9–10 G) >CH3C(=O)OO· (8.5 G) > CH3C(=O)O· (7.5 G). This study will give a systematic method on identification of radicals in AOPs, and shed light on the insightful understanding of radical production mechanism.
Peracetic acid (PAA)-based system is becoming an emerging advanced oxidation process (AOP) for effective removal of organic contaminants from water. Various approaches have been tested to activate PAA, while no previous researches reported the application of metal-organic frameworks (MOFs) materials for PAA activation. In this study, zeolitic imidazole framework (ZIF)-67, a representative MOFs, was facile synthesized via direct-mixing method at room temperature, and tested for PAA activation and sulfachloropyridazine (SCP) degradation. The as-synthesized ZIF-67 exhibited excellent performance for PAA activation and SCP degradation with 100% of SCP degraded within 3 min, owing to the specific MOFs structure and abundant Co2+ sites. The pseudo-first-order kinetic model was applied to fit the kinetic data, with rate constant k1 of ZIF-67 activated PAA system 34.2 and 156.5 times higher than those of conventional Co3O4 activated PAA and direct oxidation by PAA. Radical quenching experiments and electron paramagnetic resonance (EPR) analysis indicated that CH3C(O)OO• played a major role in this PAA activation system. Then, the Fukui index based on density functional theory (DFT) calculation was used to predict the possible reaction sites of SCP for electrophilic attack by CH3C(O)OO•. In addition, the degradation pathway of SCP was proposed based on Fukui index values and intermediates detection, which mainly included the S-N bond cleavage and SO2 extrusion and followed by further oxidation, dechlorination, and hydroxylation. Therefore, ZIF-67 activated PAA is a novel strategy and holds strong potential for the removal of emerging organic contaminants (EOCs) from water.
The release of microplastics (MPs) especially those with sizes less than 10 μm from effluent of wastewater treatment plants (WWTPs) is one of the major sources of plastics into aquatic environment. To reduce the discharge of MPs into environment, it is essential to further enhance their removal efficiencies in WWTPs. In present study, to boost the removal performance of MPs in sand filtration systems (units that commonly employed in WWTPs to remove colloidal pollutants), six types of biochar fabricated from three raw biomass materials (i.e. lignin, cellulose, and woodchips) at two pyrolysis temperatures (400 °C and 700 °C) was respectively amended into sand columns as thin permeable layer. We found that adding all six types of biochar into sand columns as thin permeable layer could greatly improve the retention of MPs with the diameter of 1 μm under either slow (4 m/d) or fast flow rates (160 m/d) due to the high adsorption capability of biochar. Woodchip-derived biochar exhibited the highest MPs retention performance, which was followed by cellulose-derived biochar and then lignin-derived biochar. Moreover, for biochar derived from three raw biomasses, increasing pyrolysis temperature could improve MPs retention performance. The direct observation of real-time plastics retention processes on different types of biochar via a visible flow chamber showed that woodchip-derived biochar especially that fabricated at 700 °C exhibited more MPs trapping processes relative to lignin and cellulose-derived biochar due to their more complex surface morphology. Thus, the highest MPs retention performance was achieved in sand columns with amendment by 1 wt% woodchip-derived biochar fabricated at 700 °C. More importantly, we found that for these modified sand filtration column systems, complete MPs removal could be achieved in real river water and actual sewage water, in multiple filtration cycles, longtime filtration process (100 pore volumes injection) as well as with interval flow conditions. Moreover, biochar could be regenerated and reused as thin permeable layer to effectively remove MPs. The results of this study clearly showed that biochar especially woodchip-derived biochar fabricated at 700 °C had the potential to immobilize MPs especially those with small sizes in WWTPs.
Magnetite nanoparticles (MNPs) with varied organic coatings (OCs) which improved their stability have broad environmental applications. However, the adsorbed amounts and layer thickness of varied OCs onto MNPs during the synthesis were generally not or poorly characterized, and their interactions with natural organic matter (NOM) were still in progress. In this study, acetic (HAc), citric (CA), and polyacrylic acid (PAA) were selected as model OCs, the adsorption behaviors of OCs on MNPs were characterized under varied aqueous C/Fe ratios, and the aggregation behaviors of MNPs with varied OCs (OC-MNPs) at neutral pH (7.0 ± 0.2) with NaCl (5–800 mM) in the presence/absence of NOM were systematically investigated. Under low aqueous C/Fe ratio, the adsorbed amounts of model OCs as –COOH/Fe ratio followed the order: CA ≈ PAA >> HAc. With high aqueous C/Fe ratio, the maximum adsorbed masses of OC-MNPs were similar. The adsorbed layer thicknesses of OC-MNPs were thoroughly characterized using three different methods, all showing that the adsorbed layer of PAA was thicker than that of CA and HAc. Derjaguin–Landau–Verwey–Overbeek (DLVO) and extended DLVO (EDLVO) calculations showed that electrostatic and van der Waals forces were dominant for CA-MNPs and HAc-MNPs stabilization; while steric repulsion played major roles in stabilizing PAA-MNPs, probably due to a thicker PAA layer. In the presence of NOM, stability behaviors of all OC-MNPs were similar, ascribing to the much greater amounts of NOM adsorbed than the OCs, causing greater steric repulsion. This study provides new mechanistic insights which could help better understand the effects of varied OCs on MNPs' colloidal stability.
