Methacrylonitrile is an indispensable chemical intermediate in aerospace and wind energy material manufacturing. Currently, the one-step isobutene ammoxidation process is the industrial mainstream for its production, but its drawbacks of high reaction temperature, many byproducts and high purification energy consumption have prompted researchers to explore alternatives. Although the two-step ammoxidation process has theoretical advantages, it is still in the experimental stage, with its economic and environmental benefits not systematically evaluated. Under unified conditions (99.5 9.10 t_CO_2-eq·t_MAN ^-1 ). Sensitivity analysis indicates yield is key to the two-step process’s competitiveness: each 1 0.37 t_CO_2-eq·t_MAN ^-1 , and 86.9
The majority of energy chemical processes require heat supply to sustain reactions. The conventional heat supply relies mainly on fossil fuel combustion, plagued by high energy consumption and intensive carbon emissions. Advanced heating technologies are urgently demanded for low-carbon transition. Electromagnetic induction heating (EIH), featuring non-contact heating, ultra-fast temperature ramp rate, and high energy efficiency, has emerged as a promising technical pathway for the low-carbon transformation of energy chemical engineering. This paper reviews the research progress of EIH utilization in energy chemical processes, focusing on performance regulation strategies and typical reaction applications of three susceptor categories: carbon-based, macroscopic metal, and magnetic nanoparticle susceptors. Carbon-based susceptors enable reactant-heat source integration for high-temperature carbon-involved endothermic reactions such as calcium carbide synthesis; macroscopic metal susceptors combine excellent machinability with heat-catalysis synergy and support tunable temperature gradients for staged conversion; magnetic nanoparticle susceptors feature high specific heating power and excel in microscale local hot spot intensification for medium-low temperature catalysis. Overall, EIH achieves deep heat-reaction coupling and serves as an effective electrification solution for strongly endothermic processes, with excellent adaptability to intermittent green electricity. Future key research directions include long-life susceptor optimization, multiphysics coupling simulation, and standardized techno-economic evaluation for industrial scale-up.
Semiconductor synthetic biology (SemiSynBio) is an emerging field with the aim of exploring distinctive advantages of biological system to tackle computational hard problems. SemiSynBio-Biocomputing solves computational problems using cells or tissue level bio-inspired system or devices. This new pathway explores the potential of biological computing systems and overcomes limits of traditional semiconductor technology, driven by the exponentially growing demand for high-performance, energy-efficient computing resource. This review provides a comprehensive overview of biocomputing at DNA/RNA-level and cell/tissue-level. SemiSynBio-Biocomputing builds an integrated system consisting with physiological processes, internal network interfaces and cross-talks among cells or tissues. The primary objective is to develop design methodologies to utilize cellular-scale networks and their natural communication capabilities and bypass the constrains of traditional algorithmic computation. The biological performance at cell/tissue-level offers a robust pathway to achieve self-organizing, self-repairing, resilient, distributed and adaptive biological computing systems, making it possible to construct a bio-computer with a variety of applications. Future directions focus on exploring the architecture for information processing and storage within biocomputing applications, particularly those with non-Von Neumann configurations. SemiSynBio opens up a new pathway towards biocomputing and is expected to address increasingly complex problems and applications, serving as a foundational platform for next generation bio-computers.
Metal-organic frameworks-based mixed matrix membranes (MMMs) often face a trade-off between interfacial compatibility and intrinsic pore selectivity. Herein, we report a multi-step crosslinking strategy for constructing UiO-66-AC@PEG by covalently bridging MAH-prefunctionalized UiO-66-NH2 with polyethylene glycol (PEG), enabling simultaneous interfacial engineering and pore regulation. The resulting crosslinked architecture incorporates abundant amide and ether oxygen moieties, which collectively generate CO2-philic domains and enhance CO2 sorption through dipole-quadrupole interactions. Meanwhile, MAH-PEG crosslinking induces pore constriction within UiO-66-NH2, strengthening size-sieving effects toward CO2/N2 separation. Moreover, the flexible long-chain architecture of PEG ensures uniform dispersion of UiO-66-AC@PEG within the Pebax matrix. It also effectively suppresses particle agglomeration, thereby minimizing nonselective interfacial defects. Consequently, the optimized UiO-66-AC@PEG/Pebax membrane achieves a CO2 permeability of 125.8 Barrer and a CO2/N2 selectivity of 81.1, corresponding to increases of 56.6
Accurate prediction of the research octane number (RON) is essential for clean fuel design, engine optimization, and gasoline blending, yet experimental measurements remain expensive, and mixture blending behavior is difficult to describe using conventional linear rules. To address this challenge, we developed an interpretable multimodal molecular representation framework based on latent space mixing. Because the reliability of mixture prediction depends on the quality of the pure-component representations, a multimodal encoder was built for pure components by integrating graph neural network embeddings, molecular access system (MACCS) fingerprints, and molecular descriptors selected through a residual-guided strategy. The trained pure-component encoder was then transferred to mixture prediction. For the pure-component test set, the model achieved an R2 value of 0.9373 and a mean absolute error (MAE) of 4.04. For the mixtures, the latent space model achieved an R2 of 0.9736 and an MAE of 1.46. The ablation results showed that the graph topology provided the strongest contribution, whereas the MACCS fingerprints and descriptors supplied complementary information. Atom-level attention analysis identified a graph-branch emphasis on the local structural environments associated with the RON. Finally, the mixture model was applied to fuel formulation design, where inverse searching under the target RON constraints generated a feasible candidate formulation. These results demonstrated that composition-weighted molecular embedding is a promising strategy for mixture RON prediction and fuel formulation.
Multi-scale superquadric discrete element method-computational fluid dynamics (DEM-CFD) coupling provides a powerful tool to resolve cross-scale behaviors inside fluidized bed reactors, yet several critical obstacles restrict its industrial application, including anisotropic contact hydrodynamics, incomplete heat transfer frameworks, immature non-spherical reaction models, and high computational cost for large-scale particle systems. This view systematically dissects the above bottlenecks, summarizes unresolved research gaps, and proposes targeted remedies: reduced-space Newton iteration for contact singularity, generalized shape-factor Nusselt correlations, superquadric coordinate-based shrinking core reaction models, and the synergy of coarse-grained particle models with graphics processing unit acceleration. Furthermore, this work elaborates on the transformative potential of artificial intelligence: deep learning can construct universal drag/heat transfer correlations from direct numerical simulation data, generate interpretable chemical reaction kinetics, and accelerate stiff solvers to cut simulation overhead. This view delivers clear development roadmaps for the next-generation multi-scale SuperDEM-CFD platform, supporting predictive design and real-time digital twin construction of gas-solid reactors.
Although HZSM-5 zeolite is widely used in catalytic cracking reactions, most relevant studies have not thoroughly examined the initial reaction behavior of n-hexane catalytic cracking, leading to an insufficient understanding of molecular-scale reaction mechanisms. Herein, we conducted precise catalyst tests to establish a molecular-scale reaction kinetic model for n-hexane catalytic cracking. The results showed that isohexane is an important primary product of the reaction. Two reaction networks, with and without n-hexane isomerization to isohexane, were compared, revealing that isohexane directly influences ethylene and propylene formation pathways and indirectly affects substances like isobutane. Results from the kinetic model showed that consideration of the pathway of n-hexane isomerization to isohexane helps reduce model errors. The detailed investigation indicated that the pathway of n-hexane isomerization exerts opposite effects on the formation of ethylene and propylene, and the model conclusions were confirmed by co-feeding experiments, which demonstrated that adding isohexane boosts ethylene selectivity and suppresses propylene formation. The present study deepens the understanding of the n-hexane cracking mechanism and provides a basis for the directional regulation of reaction pathways and the optimization of industrial process conditions.
The growing demand for lithium-ion batteries to operate under extreme conditions, including fast charging and wide temperature ranges, has significantly narrowed the margin between performance and safety. Under such demanding scenarios, battery failure arises from the complex coupling of electrochemical, thermal, and mechanical degradation. Critically, the phenomena of lithium plating, interfacial growth, and gas generation drive thermal instability, and also produce corresponding physical signals for early warning. However, conventional safety strategies still rely heavily on delayed external measurements, which are often insufficient to capture early internal thermal and degradation evolution. This review critically examines the emerging paradigm of deeply integrating health monitoring with thermal management. By linking fundamental degradation mechanisms to the physical origins of diagnostic signals, we evaluate emerging internal perception technologies, with a focus on advanced thermal, mechanical and gas-based sensing. We further summarize the transition of thermal management from external regulation to localized internal intervention. We then discuss how multimodal sensing signals can be fused to interpret battery states, assign risk levels, and coordinate adaptive thermal responses. Ultimately, this review outlines future directions centered on intelligent, self-regulating battery architectures, enabled by the integration of multifunctional materials, embedded sensing, and digital-twin-driven data fusion.
The limited structural stability and intrinsic electronic conductivity of LiMn2O4 (LMO) remain key challenges hindering its application in electrochemically switched ion exchange (ESIX)-based lithium extraction. In this study, a zirconium-modified LiMn2O4 (LZMO) film electrode was developed for selective lithium extraction from brine. Combined experimental and density functional theory analyses revealed that Zr modification simultaneously enhanced electronic conductivity and structural stability by tuning the electronic structure and strengthening Mn-O hybridization. The modulation of the electronic structures accelerated charge transfer and elevated the electrochemical kinetics. In addition. the optimized local coordination of Mn-O lowered the proportion of Mn3+ and strengthened metal-oxygen bonding, which effectively suppressed Mn dissolution and improved structural integrity. For treating the real brine (Qarhan Salt Lake), the lithium intercalation capacity of the optimized LZMO-0.03 film electrode reached 22.16 mg·g−1, and the energy consumption for lithium extraction was only 3.61 Wh·mol−1 under constant current-constant voltage mode. The calculated separation factor for Li/Mg is as high as 685.43. After 20 successive intercalation-deintercalation cycles, Mn dissolution decreased to approximately 51.8
The selective hydrogenation of dimethyl oxalate (DMO) to methyl glycolate (MG) is critical for biodegradable polyglycolic acid production, but MG is prone to overhydrogenation to ethylene glycol. Ni3P/SiO2 catalysts show promise in selective hydrogenation due to their unique electronic structure and thermal stability. The conventional strategies to enhance the performance of these catalysts rely on maximizing the carrier specific surface area to expose more active sites. In this study, we synthesized three spherical SiO2 carriers with similar particle diameters but different internal structures to load Ni3P. Unexpectedly, among the three catalysts, the highest-surface-area Ni3P/SiO2-II catalyst exhibited the best Ni3P dispersion yet the poorest DMO conversion (only 55.4
To address the challenge of synergistic optimization of the multi-physical property parameters for the gas diffusion layer—a core component of anion exchange membrane water electrolyzers (AEMWEs)—this study establishes an integrated analytical framework combining multi-physics mechanism simulation and active learning-based intelligent optimization. Based on a three-dimensional steady-state multi-physics coupling numerical model of AEMWEs, the regulatory mechanisms of permeability, electrical conductivity, and porosity in the mass transport and heat transfer processes were elucidated. On this basis, a multi-model weighted ensemble active learning surrogate model was constructed to optimize the ele0ctrochemical performance of the electrolyzer. The model achieved a root mean square error (RMSE) of 16.5539 and a mean absolute error (MAE) of 12.9643 during the training process. On the test set, the model achieved a coefficient of determination (R2) of 0.9994, an RMSE of 5.0240, and an MAE of 4.0694. The optimal parameter combination for electrochemical performance identified by the proposed framework (a permeability of 1.18 × 10−10 m2, an electrical conductivity of 2208.86 S·m−1, and a porosity of 0.605) effectively suppresses local heat accumulation and eliminates mass transport bottlenecks. Detailed post-optimization analyses further demonstrated that the optimal parameter combination achieved a favorable balance between the optimal electrochemical response and acceptable multi-physics field distribution uniformity.
To address the limitations of existing methods in capturing long-term temporal dependencies, local capacity regeneration, and nonlinear degradation characteristics in lithium-ion battery remaining useful life prediction, this paper proposes a coordinate-aware Mamba2 framework based on state-space modeling. Mamba2 is adopted as the backbone to model long-range degradation evolution, while a coordinate feature attention network is introduced in the feature extraction stage to enhance the selection and representation of key degradation in00formation. Additionally, a swiGLU-gated residual network is constructed for feature transformation and prediction, improving the modeling of complex dynamic relationships and endpoint stability. Through the collaborative design of these modules, the proposed framework effectively captures both global degradation trends and local fluctuation patterns. Experiments were conducted on the National Aeronautics and Space Administration, Tongji University, and Xi’an Jiaotong University datasets under single-variable input, multivariable input, and cross-dataset generalization settings. The mean absolute error values are lower than 0.0098, 0.0016, and 0.0083, while the root mean square error values are lower than 0.0172, 0.0024, and 0.0111, respectively. Results demonstrate that coordinate-aware Mamba2 achieves superior accuracy, lower computational cost, faster training and inference, and stronger robustness to different prediction starting points.
γ-Al2O3 is one of the most widely used catalyst supports in heterogeneous catalysis, yet its catalytic role remains unclear due to intrinsic structural disorder and complex surface species. This paper describes the development of a high-accuracy Al–O–H machine learning interatomic potential combined with global optimization to explore the structural landscape of γ-Al2O3 and its influence on propane dehydrogenation over single-atom Pt catalysts. Two energetically favorable structures, γ-no aluminum vacancy (NAV) and γ-aluminum vacancy (AV), are identified with energies lower than the conventional model by up to 34.44 meV·(f.u.)−1. These optimized structures expose abundant penta-coordinated Al3+ sites on the (100) surface, serving as preferred anchoring sites for Pt atoms. Simulated X-ray diffraction patterns indicate that γ-AV shows better qualitative agreement with experimental data. Surface phase diagrams further reveal that defect-rich surfaces are thermodynamically stabilized under realistic reaction conditions. Catalytic calculations demonstrate that γ-NAV and γ-AV significantly reduce propane dehydrogenation activation barriers through enhanced metal-support interactions associated with penta-coordinated Al3+ sites and defect-modulated local environments. These results suggest that γ-Al2O3 is better described as an ensemble of defect-rich surface configurations rather than a single crystal structure. These findings establish a direct relationship between atomic structure, defect chemistry, and catalytic performance in γ-Al2O3.
To address the key bottlenecks in the state of health prediction of lithium-ion batteries under random variable load conditions, such as poor physical consistency, insufficient feature representation, and non-physical fluctuations in the prediction curves, this paper proposes a high-precision state of health prediction model that combines physical information constraints and multi-dimensional features, aiming to enhance the safety and long-term reliability of the battery management system. A neural network-transformer framework based on physical information was constructed, converting the consistency laws of irreversible battery aging into regularization loss terms in the training process to ensure physical rationality. Multidimensional coupled features including voltage, current, and cycle count were extracted to characterize the aging characteristics of the battery, and Bayesian optimization was applied to adaptively adjust key hyperparameters. Experimental results show that this model outperforms the comparison models in terms of MAE, RMSE, and R2 metrics, effectively suppressing non-physical fluctuations, balancing high prediction accuracy and physical rationality, and providing technical support for the full life cycle health management and safe operation of lithium-ion batteries.
Electrical capacitance tomography (ECT) is an indispensable non-intrusive diagnostic tool for fluidized beds. Although recent advancements have successfully yielded robust sensors capable of surviving high temperatures, temperature-induced permittivity variations pose a formidable challenge to accurate phase reconstruction in high-temperature applications. To systematically quantify and mitigate the unaddressed thermal interference, this study integrates a numerical framework, encompassing different sensor architectures with varying diameters to represent geometries from laboratory to industrial-scale reactors, with high-temperature experiments. The results reveal that although temperature fluctuations induce significant dielectric drifts, the normalized sensitivity distribution maintains exceptional stability. Conversely, raw capacitance measurements are highly susceptible to thermal perturbations within both packed bed and column wall. Crucially, increasing wall thickness drastically amplifies measurement errors for adjacent electrode pairs. To improve measurement robustness, excluding adjacent electrode measurements is shown to effectively suppress temperature-induced deviations and extend the permissible operating range. In exchange for sacrificing fine-scale boundary resolution, the resulting trade-off prioritizes core-region tomographic integrity, mitigating severe thermal interference. Furthermore, the study establishes a theoretical equivalence between thermal distortions and inherent noise, enabling direct application of noise-based criteria to define operational limits. These findings provide practical guidelines for the design and deployment of robust ECT systems in high-temperature environments.
Microwave ultra-high temperature (UHT) processing is promising for liquid food sterilization, yet its inactivation mechanisms against thermophilic bacteria remain inadequately understood. This study investigated the bactericidal efficacy and mechanisms of microwave UHT treatment against vegetative cells of Geobacillus stearothermophilus ATCC 7953. Using a single-mode microwave system at 2450 MHz, bacterial suspensions were subjected to various power-time combinations achieving 136 ± 1 °C followed by immediate cooling. A nonlinear relationship between power and inactivation efficacy was observed: optimal reductions of approximately 5 log CFU·mL−1 were achieved at 150 W·mL−1 for 40 s and 300 W·mL−1 for 20 s, while intermediate powers yielded inferior outcomes. Mechanistic analyses revealed that microwave treatment induced significant membrane damage, suppressed metabolic activity, and dramatically elevated intracellular reactive oxygen species and malondialdehyde levels, with the 300 W·mL−1 treatment generating the highest oxidative stress. Scanning electron microscopy confirmed distinct morphological alterations without electroporation. The similar trends observed between oxidative markers and bactericidal efficacy suggest that oxidative stress-mediated lipid peroxidation may constitute a primary inactivation mechanism, with low-power prolonged exposure promoting cumulative damage and high-power short-duration treatment triggering acute oxidative burst. These findings elucidate the power-time synergistic mechanism of microwave UHT inactivation and provide a theoretical foundation for process optimization.
Droplets exhibit distinct wetting characteristics and transport behavior on solid substrates at the macroscopic and microscopic scales. This behavior is critical for understanding liquid mass transfer on fibrous membranes. To better understand and control the mass transfer of liquids on fibrous membranes, we combined in situ visualization techniques with multiphase flow simulations. This approach enabled us to systematically explore the effects of various factors, including fiber wettability, fiber diameter, and fiber spacing, on liquid transfer behavior. Furthermore, we successfully elucidated the transfer mechanisms governing droplet transport on individual fibers and between adjacent fibers. Based on our findings, we constructed composite fiber membranes with varying fiber diameters and wettability structures. The validity of the proposed approach was verified by comparing fog collection, droplet wetting, and liquid permeation efficiency. Consequently, this study establishes a transferable cross-scale framework and proposes a general design strategy for constructing fibrous membranes tailored to diverse application requirements.
Converting CO2 into CO via reverse water gas shift (RWGS) reaction is a key step for carbon recycling. Molybdenum trioxide (MoO3) is a promising precatalyst due to its high activity and near-unity CO selectivity, yet the role of support properties remains unclear. To address this, a series of MoO3-based catalysts supported on MgO, γ-Al2O3, SiO2, TiO2, ZrO2, and CeO2 were prepared. Systematic characterizations show that MoO3 undergoes in situ carburization to Mo2C, and the extent of carburization correlates positively with catalytic activity. The formation of active Mo2C is governed by the metal oxide-support interaction (MOSI): strong MOSI between MoO3 and basic supports (MgO, CeO2) promotes stable solid solutions that suppress carburization, whereas acidic and amphoteric supports preserve MoO3 crystallites, enabling efficient carburization and high RWGS performance. Among all catalysts, MoOxCy/SiO2 exhibits the weakest MOSI, the highest surface Mo2C concentration, and thus superior mass-specific activity. The ionic potential of the support serves as a descriptor for MOSI strength, while the specific surface area introduces a certain deviation for amphoteric supports (γ-Al2O3, TiO2, ZrO2). This work provides clear support selection criteria and a theoretical foundation for rational design of high-performance Mo-based RWGS catalysts.
Aqueous zinc-ion hybrid capacitors (ZIHCs) are promising for sustainable energy storage, but their specific capacity is severely limited by the kinetic and capacity mismatch between capacitive carbon cathodes and zinc anodes. Herein, we present a synergistic space-confinement and activation strategy to synthesize hierarchical porous carbon nanosheets (CMK-x, x represents the pyrolysis temperature) derived from coal tar pitch. The optimized CMK-700 features a high specific surface area of 2223.9 m2·g−1 and a maximized ultramicropores volume of 0.4836 cm3·g−1. Combined ex-situ characterization and molecular dynamics simulations, a unique dual-ion relay storage mechanism enabled by spatial domain separation across different pore structures was unraveled. Specifically, the larger [Zn(H2O)6]2+ ions are stored in larger micropores (> 1 nm) at a voltage window of 0.3–1.9 V, while smaller hydrated protons ([H3O]+) act as ‘charge relays’ to penetrate the ultramicropores below 0.3 V. Such a proton accommodation through reversible chemical hydrogen adsorption/desorption contributes significant surface pseudocapacitance. Consequently, the assembled ZIHCs deliver an impressive high specific capacity of 368.1 mAh·g−1 at 0.5 A·g−1 and outstanding long-term stability with 86.39
Plasma-based gas conversion has emerged as increasingly prominent sustainable technology for chemical production, offering significant advantages such as mild operating conditions, instantaneous control, and flexibility in scales. However, the inherent complexity of its multidimensional parameter space makes traditional experimental optimization resource-intensive. Machine learning (ML) presents a transformative method to efficiently explore such intricate scientific phenomena, yet its application in the field remains in its infancy. Current efforts are constrained by fragmented, small-scale experimental datasets that lack standardization across different reactor configurations and measurement protocols. Data quality issues, inconsistent reporting of performance metrics, and the absence of critical plasma and catalyst descriptors further hinder model development. Consequently, most ML studies are limited to simple predictive models that interpolate within narrow operational domains, offering little generalizability or mechanistic insight. This critical review provides a comprehensive analysis of ML methodologies applied to plasma-based gas conversion, using CO2 conversion as a base case. We outline the general ML workflow and key algorithms, discuss their applications with state-of-the-art examples, and critically evaluate current limitations. Finally, we identify emerging challenges and future opportunities to guide the field toward more robust, generalizable, and physically as well as chemically meaningful ML applications.
