The synergistic solvent extraction behavior and kinetic mechanism of light rare earth elements using a mixed Cyanex 572/Cyanex 302 system were investigated. The binary extractant system showed substantially higher extraction efficiency than either extractant used alone. Slope analysis confirmed the participation of both Cyanex 572 and Cyanex 302 in the formation of a mixed-ligand complex. The extraction experiments were performed using a Cyanex 572/Cyanex 302 synergistic system diluted in kerosene as the organic phase. A synthetic nitrate solution containing Nd(III), Pr(III), and La(III) at initial concentrations of 0.015, 0.004, and 0.014 M, respectively, was contacted with the organic phase at an aqueous-to-organic phase ratio of 1:1. The extraction behavior was investigated over a pH range of 1–6, and the optimum condition was obtained at approximately pH 5. The separation performance of the system was evaluated, and separation factors of βNd/Pr = 3.67 and βNd/La = 14.29 were obtained under the optimized conditions. Kinetic analysis revealed positive reaction orders for Cyanex 572 and Cyanex 302 and an inverse dependence on hydrogen ion concentration. The apparent reaction orders were 0.75 for Cyanex 572, 1.14 for Cyanex 302, and −1.05 for hydrogen ions, indicating the important role of proton exchange. A diffusion-controlled mechanism was proposed, involving dissociation of the dimeric extractant, rapid formation of a neutral interfacial metal complex, and subsequent diffusion of this complex into the organic phase. Similar extraction rate constants were obtained for La(III), Pr(III), and Nd(III), ranging from 2.24 × 10⁻² to 2.57 × 10⁻² s⁻¹. The obtained extraction and separation results demonstrate the potential of the Cyanex 572/Cyanex 302 system for Nd(III) recovery; however, further counter-current studies are required for process-scale validation.
The performance ceiling of high-capacity mesoporous adsorbents in liquid-phase packed beds, where maximizing internal surface area severely restricts transport kinetics, necessitates hierarchical architectures. This study presents hierarchical supraparticles (SPs), which are spherical assemblies consisting of close-packed mesoporous silica particles, designed to improve both capacity and mass transport within micro-packed beds compared to packing materials with unimodal pore size. Here, SPs featuring 2.4 nm mesopores for capacity and ∼80 nm macropores for transport were synthesized and compared to a unimodal control (Wakosil®60) via Rhodamine B breakthrough analysis in micro-packed beds. By decoupling capacity from diffusion kinetics, hierarchical SPs overcome the severe mass-transfer limitations typical of unimodal porous materials. SP-packed beds achieved a simultaneous increase in saturation capacity and a three-fold higher apparent internal diffusivity, which proves that the architecture bypasses internal transport resistance. Consequently, under constant operating pressure (30 kPa), SP-packed beds showed a three-fold increase in volumetric breakthrough productivity compared to the unimodal porous control of similar size. Furthermore, SPs that were 3.7 times smaller matched the productivity of the larger unimodal standard; the high flow resistance inherent to smaller packed beds was entirely offset by accelerated intraparticle kinetics. Dimensionless analysis using the Mass Fourier Number confirmed that both architectures follow a universal scaling of bed utilization, but the kinetic bottleneck shifted to higher throughputs for hierarchical SPs. These findings demonstrate that hierarchical supraparticles drive process intensification in liquid-phase packed beds by overcoming the coupled constraints of pressure drop, diffusion, and capacity.
Enhancing the performance of electrochemical reactors has received substantial attention, but studies on mass transport enhancement by pulsatile flow and its associated power consumption remain unknown. The effects of net electrolyte flow rate, pulsation frequency, pulse amplitude, and geometry of 3D printed electrodes on mass transport and pumping power were studied and compared against conventional steady state flow. The limiting current was measured to calculate the volumetric mass transport coefficient based on the copper deposition reaction on 3D-printed stainless steel electrodes. The pumping power density of different electrode geometries, including Body-centred cubic, Octahedron, and Baffle designs, was determined from the oscillatory pressure drop and oscillatory flow rate. Body-centred cubic with a unit cell size of 1.25 mm and Baffle electrode geometries showed a clear enhanced mass transport relative to steady state flow at higher pumping power density. The highest measured mass transport was achieved by the Body-centred cubic electrode, with a maximum improvement of approximately 30% over steady state flow. Pulsatile flow was shown to increase mass transport at the same pumping power, indicating that it is a viable strategy for improving the performance of electrochemical reactors.
Benzene is one of the most hazardous air pollutants. In this study, the performance of a biofilter packed with granular corn cob, which was inoculated with the white rot fungus Irpex lacteus, was investigated for the treatment of a benzene-contaminated air stream. The biofilter was operated for 90 days at benzene concentrations of 0.19–1.91 g/m3 and empty bed residence times of 30–120 s. Additionally, 10 h of aeration per day to operate as intermittent daily and weekend shutdown periods were applied to simulate realistic industrial emission scenarios. The performance of the biofilter was examined regarding removal efficiency, elimination capacity, biomass growth, CO₂ production, pressure drop, and biofilm development. The stability of the biofilter was evaluated under shutdown periods. It was observed that there was an acclimatization period of 27 days within the biofilter. Maximum benzene removal efficiency was 95.11 % at an inlet concentration of 1.23 g/m³. The maximum elimination capacity reached 129.23 g/m³.h at an inlet loading rate of 235.24 g/m³.h. Concentration of fungal biomass increased from 0.31 × 10⁸ CFU/g to a maximum value of 17.8 × 10⁸ CFU/g in phase II. The CO₂ production and elimination capacity showed a strong linear correlation (R² > 0.98). Scanning electron microscopy (SEM) observation also verified substantial fungal biofilm formation on the corn cob surface. The results demonstrated that the combination of Irpex lacteus and corn cob granules represents a beneficial and sustainable biofiltration system for benzene-contaminated air streams under dynamic operating conditions, which offers a promising alternative for industrial emission control.
This work explores the functional role of extracellular polymeric substances (EPS) in modulating microplastic (MP) toxicity during seed germination, with a focus on tomato (Solanum lycopersicum) as a model crop. EPS was used at three concentrations (10, 50, and 100 µg/mL) under both In-situ and Ex-situ conditions in the presence of polyvinyl chloride (PVC), polyethylene (PE), and polystyrene (PS), each introduced at 3% of EPS weight. To simulate environmental exposure, MPs were incubated for 45 days in agricultural soil (AS) and sludge-amended agricultural soil (SAS), after which EPS was extracted and subsequently tested for its interaction with MPs. Comparative analyses of AS and SAS derived EPS were conducted to evaluate their influence on seed germination and early seedling development. Seed germination percentage, root and shoot elongation, biomass accumulation (wet and dry weight), seedling vigour indices (SVI I and II), and chlorophyll content were quantified to assess MP-induced phytotoxicity. Ex-situ EPS-MP mixtures consistently suppressed seedling growth as compared to In-situ treatments, with marked reductions in elongation, biomass, and chlorophyll synthesis. Among the three polymers, PE shows the strongest inhibitory effects across both soil systems, highlighting its pronounced persistence and toxicity potential. Notably, EPS extracted from SAS mitigated these adverse outcomes, enhancing germination efficiency and growth indices relative to AS derived EPS. This suggests that sludge-origin EPS, enriched in organic matter and nutrient fractions, may act as a compensatory nutrient that offsets MP induced stress. These findings underscore the dual role of EPS as both a mediator of MP toxicity and a potential nutritional buffer, with implications for soil resilience and sustainable crop establishment in MPs contaminated agroecosystems.
Polyolefins, particularly polyethylene and polypropylene, dominate plastic waste streams yet remain difficult to separate and recycle because of their chemical inertness and low solubility in most of the solvents. Therefore, developing solvent systems that enable selective polyolefin processing while minimizing trial-and-error experimentation is a key challenge for sustainable materials management. Here, we present an atomistic simulation workflow to guide the design and screening of ionic liquids (ILs) as potential media for polyolefin dispersion. The workflow combines molecular dynamics simulations with radial distribution functions, mean-squared displacement, and Voronoi analysis to quantify polymer aggregation, ion coordination, and interfacial interactions, respectively. The simulation protocol was validated against experimentally benchmarked reference systems. To identify optimal solvents for polyolefin processing, we apply this workflow to a family of phosphonium ILs with systematically varied cation alkyl chain length and sulfonate anion structure. The analysis reveals that maximizing nonpolar surface area of cations and anions suppresses polymer aggregation and promotes polyolefin dispersion. Beyond identifying promising ILs, this study establishes a computationally guided strategy to accelerate the development of alternative solvent systems for polyolefin separation and recycling, reducing experimental screening effort and supporting more sustainable plastics processing.
In this manuscript, we propose a hydrometallurgical flowchart for recycling spent NiMH batteries by evaluating successive stages of REEs precipitation, ozonation, Al and Fe removal, selective separation of Zn and Y, and separation of Ni and Co. NiMH batteries are important for hybrid vehicles, complementing the use of Li-ion batteries in electric vehicles to contribute to the energy transition in transport. REEs were precipitated as double sulfates (above 99% purity) using Na2CO3 at pH 1.0, 25°C, and 60min. Ozonation enabled the selective oxidation and removal of Mn at pH 1.0 and 25°C after 240 min. Al and Fe were removed by Na2CO3 precipitation at pH 3.5, 80°C, and 120min. The selective extraction of Y and Zn was carried out using 20% Cyanex 272 at 25°C, pH 3.5, an aqueous/organic (A/O) ratio of 1/1, and 15 min of contact time. The separation of Co and Ni was achieved through two solvent extraction contacts with Cyanex 272, followed by selective precipitation with Na2CO3 and C2H2O4, resulting in basic Ni carbonate and Co oxalate with purities of 97% and 99%, respectively. The proposed process demonstrates how different hydrometallurgical techniques can be integrated into a single purification flowsheet, enabling the sequential removal of impurities and the recovery of high-purity REE, Ni, and Co products. Although developed for spent NiMH batteries, the proposed downstream purification strategy may also provide a basis for adapting similar hydrometallurgical flowsheets to other battery recycling systems.
Vapor pressure is a fundamental thermophysical property governing phase equilibrium, separation, and process design, yet reliable experimental data remain limited for most synthesized chemicals. In this work, we develop a physics-constrained hybrid machine learning framework to systematically improve vapor pressure predictions from the PR+COSMOSAC equation of state (EOS). The model operates in reduced temperature (Tr) and pressure (Pr) space and combines molecular fingerprints with discretized vapor-pressure curves generated by the PR+COSMOSAC EOS to predict reduced vapor pressure over a broad temperature range. Thermodynamic consistency is enforced through a hard constraint ensuring lnPr=0 at the critical point. Using a dataset of 18,979 chemically diverse compounds for model development and evaluation, we demonstrate that the proposed framework effectively corrects systematic deficiencies in PR+COSMOSAC EOS predictions with limited training data. The model reduces the average absolute relative deviation from 213% for the standalone PR+COSMOSAC EOS to 64%, while requiring only 40% of the compounds for training. The results further show that incorporating physically informed EOS features significantly enhances predictive performance and generalizability compared with purely data-driven approaches. In addition, the proposed framework provides practical estimates of prediction reliability by learning the expected prediction error from ensemble predictions and their associated uncertainties. These results demonstrate that physics-constrained machine learning is an effective strategy for integrating thermodynamic theory with data-driven modeling to achieve reliable thermophysical property prediction for chemically diverse compounds.
Solubility measurements of Abiraterone acetate in SC-CO₂ were carried out over temperatures ranging from 308 to 338 K and pressures from 120 to 270 bar. The experimental results demonstrate that solubility increases significantly with pressure at all temperatures, while temperature exhibits a dual effect: at low pressures, rising temperature reduces solubility due to the dominant density loss, whereas at high pressures, enhanced sublimation pressure improves solubility. A crossover point is clearly observed near 120 bar. Three distinct modeling approaches were applied and rigorously compared. Among cubic equations of state, the SRK EoS combined with the van der Waals type I and II mixing rules provided the most accurate predictions, yielding an AARD of 4.83%, outperforming the PR EoS/vdW type I and II, the PR EoS/KM, and the sPC-SAFT model (AARD = 18.19%), which proved overly complex for this system. The modified Wilson’s expanded liquid theory model delivered superior correlation with an AARD of 2.26%, reflecting strong solute-solvent attractive interactions. Furthermore, 31 semi-empirical correlations were evaluated; the Sodeifian et al. model I exhibited the lowest error (AARD = 1.52%), while the modified Bartle et al. model offered the best trade-off between simplicity and accuracy. Enthalpy analysis confirmed that the overall dissolution process is endothermic with negative solvation enthalpy, which indicates solvent-solute intermolecular interactions partially compensate for the high vaporization energy of the solid drug.
The focus of this article is on the analysis of transient breakthrough experiments measured in laboratory scale units. The scale up of such breakthrough data to industrial scale pressure swing adsorption (PSA) units requires simulations of the adsorption/desorption cycles in a sequence of multiple beds. This article provides a description of the simulation methodology that is essential for interpretation of laboratory scale data, and scaling up.A number of important conclusions are drawn from detailed analysis of large number of experimental breakthrough experiments, along with corresponding simulations.In the event that experimental breakthroughs are “sharp”, this is a strong indication of the absence of intra-crystalline diffusion influences. A consequence of intra-crystalline diffusion influences is that breakthrough characteristics attain distended characteristics. Commonly, but not always, distended breakthroughs result in diminished productivities of the desired product in mixture separation. In exceptional cases, intra-crystalline diffusion serves to complement and enhance the separations dictated by mixture equilibrium thermodynamics; such synergy manifests, for example, for 2MP/22DMB separations in MFI zeolite. For kinetically driven separations of C3H6/C3H8 mixtures using ZIF-8, and KAUST-7, both particle sizes and gas-particle contact times influence separations to a significant extent. The modelling of kinetically driven separations, requires thermodynamic coupling effects to be duly accounted for by use of the Maxwell-Stefan formulation for intra-crystalline diffusion.
Efficient recycling of waste electrical and electronic equipment (WEEE) is a major priority for the circular economy and sustainable resource management. This study proposes and investigates an integrated mechanical, pyrometallurgical, and electrochemical process for recovering copper from waste printed circuit boards (WPCBs), focusing on smelting metal concentrates into anodes (∼69 wt.% Cu) and subsequent electrorefining. Electrorefining of the anodes was carried out in an acidic sulfate medium (50 g⋅L⁻¹ Cu²⁺, 200 g⋅L⁻¹ H2SO4) at a constant temperature of 60°C, at three applied cell voltages: 0.25 V, 0.30 V and 0.35 V. Experimental results indicated an optimal working window between 0.25 V and 0.30 V. In this range, cathodic deposits containing approximately ∼99% wt.elemental Cu and a maximum copper recovery yield of ∼ 89% were obtained. Increasing the cell voltage to 0.35 V caused a kinetic destabilization of the process. The XRD results are consistent with the proposed formation of Cu2O under conditions favoring local Cu⁺ accumulation and secondary precipitation; however, the present experimental data do not allow direct confirmation of the proposed intermediate Cu⁺ species. Also, X-ray fluorescence spectroscopy (ED-XRF) revealed a undesired incorporation of silver at the cathode (up to 0.97 wt.% at 0.35 V), may be attributed to secondary chemical dissolution induced by dissolved oxygen and by mechanical entrainment of fine particles from the anode slime (insoluble anodic residue) under the effect of agitation. The integrated method proved superior in selectivity, limiting the co-deposition of impurities such as Ni and Zn and offering an efficient solution for e-waste recycling.
Lignin is a renewable resource co-generated at large scale in biorefineries and has the potential to replace a significant fraction of fossil-derived aromatic polymer building blocks. However, technical lignins typically exhibit high polydispersity, low reactivity, and poor solubility, limiting direct valorisation. Continuous reductive catalytic depolymerization (RCD) of lignin enables the production of large quantities of well-defined lignin-derived building blocks suitable for material applications. To enable industrial implementation and to facilitate the transition from batch processing to continuous flow operation, maximizing catalyst lifetime and performance through optimized catalyst bed design is essential. This study investigates the impact of catalyst geometry on process stability and depolymerization efficiency during continuous RCD of hydrolysis lignin. Palladium on γ-Al2O3 catalyst was shaped as spheres, wash-coated honeycombs, and 3D-printed monoliths. The catalyst geometry strongly affected results of screening tests conducted at 215°C. Spherical catalysts enabled effective depolymerization but failed after 25 h of continuous operation due to rapid pressure buildup associated with coke formation. Honeycomb structures enabled stable operation for 49 h but showed limited depolymerization efficiency. Optimized 3D-printed monoliths with high open frontal area provided the best compromise, combining improved process stability with efficient depolymerization. In a full-reactor configuration, 3D-printed catalysts permitted stable RCD for 84 h, yielding sufficient depolymerized product for the synthesis of lignin-based acrylic resins exhibiting competitive mechanical and thermal properties. The results of this study clearly demonstrate the potential of 3D-printing for shaping RCD catalysts and highlight the relevance of this lignin valorisation pathway for biobased polymer applications.
Rising atmospheric CO₂ concentrations require the development of efficient carbon capture technologies. This study establishes a three-bed nine-step vacuum pressure swing adsorption (VPSA) process integrated with a replacement (RP) step for CO₂ capture from dry flue gas. Both experimental and simulation results confirm the RP step effectively enhances CO₂ purity. To clarify the internal mechanism whereby RP step improves CO₂ purity, a condition-driven autoregressive CNN-LSTM (AR-CNN-LSTM) model to accelerate process simulation and optimization is proposed. The model captures the spatiotemporal distributions of state variables at cyclic steady state, with Simpson’s integration applied to calculate process performance indicators. The model exhibits better interpretability and generalization than Multi-Layer Perceptron. It maintains stable prediction even under operating conditions 15% outside the training operating parameter range. Combined with NSGA-II algorithm, two dual-objective optimization problems are solved. SHAP analysis shows that higher feed flow rate, RP step flow rate, vacuum flow rate, and longer RP step time effectively increase CO₂ purity. The results demonstrate high post-RP CO₂ solid concentration contributes to high CO2 purity. Optimal parameters are selected via TOPSIS method, maintaining purity within acceptable levels (75% - 85%), achieving above 95% CO₂ recovery, a productivity of 0.079 kgCO₂/kgads/h and a specific energy consumption of 321 kWh/tCO₂. Compared with previous study, this VPSA process improves both CO₂ purity and recovery, with productivity increased by 3.43 times and specific energy consumption only rising by 15.05%. The AR-CNN-LSTM improves interpretability of data-driven surrogate models and offers a new way to explore optimal operating conditions beyond original parameter boundaries.
Soil contamination by petroleum hydrocarbons is one of the global environmental challenges. This study presents an integrated bio-electrokinetic framework that enhances petroleum remediation in contaminated soils by combining bioremediation with an optimized electrokinetic system for improved TPH degradation. Therefore, a phytoremediation approach was developed using bermudagrass for soils contaminated with 1.5%, 3%, and 4.5% crude oil under different moisture conditions (field capacity, FC, 0.8FC, and 0.6FC), electric current of 0, 1.5, and 3 V cm-1, and either 3%wt Pseudomonas fluorescens or a 3%wt combined application of dried Chlorella biomass and Pseudomonas fluorescens. The high R² values of the models indicated the satisfactory performance of Response Surface Methodology (RSM) in optimizing the effectiveness of the remediation system. The incorporation of desiccated Chlorella biomass alongside Pseudomonas fluorescens constitutes the core innovation of the proposed integrated bioremediation strategy, overcoming the logistical challenges associated with live algal cultures while enhancing microbial activity, facilitating passive biosorption and bioavailability of petroleum hydrocarbons, and improving soil water retention. The integrated bioelectric remediation approach showed superior TPH removal from soil (up to 65.60%, 75.63%, and 80.58% in soils with 4.5, 3, and 1.5%wt oil contamination, respectively) along with a significant increase in TPH accumulation in bermudagrass biomass, plant dry weight, and its resilience compared with those of their standalone approaches. To achieve the highest possible removal efficiency, the following optimal values were predicted: an electrical current of 3 V cm-1, a biological treatment consisting of a 3%wt combined application of dried Chlorella biomass and Pseudomonas fluorescens, and moisture conditions of 80%, 100%, and 100% field capacity (FC) for 1.5%, 3%, and 4.5% contamination levels, respectively. Notably, continuous post-harvest plant regrowth and vitality highlighted the long-term restorative potential of the applied remediation approach and its capability to facilitate the uptake and processing of contaminants. In other words, this study proposes a practical, data-based, environmentally friendly, and scalable approach for the efficient remediation of petroleum hydrocarbon-contaminated soils.
The pervasive contamination of aquatic environments by tetracycline (TC) antibiotics poses severe threats to ecosystem integrity and public health, primarily due to its chemical stability, resistance to conventional treatment, and role in propagating antibiotic resistance. Herein, we report the rational design and synthesis of cobalt/zirconium-implanted boron and nitrogen co-doped hierarchical carbon nanotubes (CoZr@BCN) via controlled pyrolysis of a porphyrinic Z-MOF (PCN-222) at 900 °C in the presence of polyethylene glycol, boric acid, and urea as auxiliary sources. This innovative strategy yields hollow carbon nanotubes (approximately 800 nm in length, with an average diameter of 100 nm and an average wall thickness of 15 nm) featuring a hierarchically mesoporous structure. Comprehensive characterization confirms the coexistence of B/N-doped graphitic carbon nitride within a moderately disordered graphitic matrix, as indicated by an ID/IG ratio of 1.41. Notably, under visible light irradiation, CoZr@BCN showed 98.1% degradation of TC (20 mg L-1) within 30 min at pH 6 (catalyst dosage: 0.75 g L-1), accompanied by ∼89% total organic carbon removal, demonstrating near-complete mineralization rather than mere chromophore destruction. The degradation follows pseudo-first-order kinetics (k1 = 0.122-0.068 min-1 for 10-50 mg L-1 TC), with radical scavenging studies establishing the reactive species order: photogenerated holes (h+) > superoxide radicals (O2•⁻) > hydroxyl radicals (•OH). Remarkably, CoZr@BCN retains over 94% of its degradation efficiency after five consecutive cycles with less than 1% loss in activity, underscoring exceptional structural stability and recyclability. By elucidating the underlying mechanistic pathways, this study provides critical insights for the advancement of next-generation water treatment technologies.
Temperature transition and organic loading rate (OLR) are major operational factors governing anaerobic digestion performance; however, the influence of temperature-transition strategy on subsequent reactor resilience during increasing substrate loading remains insufficiently understood. Six continuous stirred tank reactors were operated under distinct strategies: constant mesophilic operation (37°C); gradual transitions to 42, 46, or 55°C; rapid stepwise transition to 46°C; and abrupt transition to 46°C. After thermal adaptation, OLR was progressively increased from 3 to 10 kg VS m−3 d−1 while hydraulic retention time (HRT) simultaneously decreased. Reactor performance and methanogenic community responses were assessed using methane yield, biogas composition, volatile fatty acids, total ammoniacal nitrogen, calculated free ammonia, and quantitative PCR (qPCR). Moderate temperature increases to 42–46°C transiently improved methane yield but produced no sustained long-term gains. Transition to 55°C caused pronounced short-term inhibition, followed by recovery after extended acclimation and stable methane production at the highest loading conditions. Segmented regression identified a methane-yield transition region centered at 6.75 kg VS m−3 d−1 (95% confidence interval: 4.92–8.57), while generalized least-squares modelling and block-bootstrap analyses supported the robustness of the reactor-response patterns. Greater loading tolerance was associated with increased abundances of Methanosarcinaceae and hydrogenotrophic methanogens, indicating restructuring of the quantified methanogenic community. Overall, reactor resilience was related not only to final operating temperature but also to the transition pathway and acclimation time. These findings provide an engineering basis for adaptive temperature management and loading optimization in continuously operated agricultural anaerobic digesters.
This review provides a comprehensive analysis of machine learning (ML)-assisted and data-driven design strategies for carbon dots (CDs), covering the transition from property prediction and synthesis-parameter optimization toward emerging inverse-design and autonomous-discovery frameworks. CDs exhibit complex structural heterogeneity and nonlinear synthesis–structure–property relationships, which limit conventional rational design approaches and create challenges for reliable data-driven modeling. This review systematically discusses the applications of machine learning methods, including ensemble learning, active learning (AL), Bayesian optimization (BO), uncertainty quantification, and explainable artificial intelligence, for predicting optical properties, quantum yield, emission behavior, and application-related performance of CDs. The distinctions among forward property prediction, parameter optimization, AL, inverse design, closed-loop experimentation, and autonomous discovery are critically evaluated. Current studies mainly focus on prediction and optimization within predefined chemical spaces, whereas complete inverse design requires target-property definition, candidate generation, physical and synthesizability constraints, ranking strategies, and experimental validation. Remaining challenges include data standardization, descriptor representation, model validation, uncertainty calibration, interpretability, and cross-system transferability. Future progress will rely on integrating machine learning with physically informed models, standardized multimodal datasets, and automated experimentation to establish more reliable and transferable principles for data-driven carbon-dot design.
Catalytic reforming requires feed information that can be transferred to kinetic-model lumps because naphtha composition influences aromatics formation, hydrogen production and coke-related catalyst deactivation. An integrated Molecular-Type Homologous Series (MTHS) characterisation and lumped kinetic-reactor modelling method is developed for continuous catalytic reforming (CCR). A data-adapted flexible-lump formulation is used to transfer carbon-number-based (CN-based) or pseudo-component-based (PC-based) MTHS feed matrices to the CCR kinetic model according to available industrial feed data. A 40-lump reaction network is formulated for reactor modelling. Coke formation is represented in the catalytic reforming reaction network and used to describe coke-related deactivation and carbon/hydrogen elemental conservation. The kinetic-reactor model is calibrated for a four-reactor industrial CCR unit using available base-case industrial feed assays and plant operating data. Pressure behaviour, coke deposition, reactor profiles and operating-variable responses are then assessed as process and mechanistic consistency checks. Feed density and hydrocarbon-family contents are predicted with errors below 0.50 %. After calibration, reactor outlet temperature errors are below 0.66 K, and final-product lump flowrate errors are below 1.6 kmol/h. The operating-variable responses follow expected catalytic reforming trends under changes in feed and operating conditions.
Efficient hydrogen production via ammonia decomposition is considered a promising direction for future carbon-free energy systems. In this study, a catalyst system based on Ruthenium (Ru) as the main active metal, promoted by Potassium (K) and Ceria (CeO2) on γ-Al2O3 support was developed, which was chosen for its high surface area, low cost, and excellent thermal stability. Potassium acts as an electronic promoter that can enrich the Ru surface and assist the removal of nitrogen-containing intermediates, while CeO2 provides a redox-active oxide environment that can participate in Ru–Ce interfacial modification. As a result of these cooperative promotional effects, the K-Ru/Ce/γ-Al2O3 catalyst achieved 100% NH3 conversion at 475 °C, with an H2 formation rate of 30.67 mmol H2·gcat-1 min-1, even when operated at a high WHSV of 30,000 mL·gcat-1 h-1. The optimized catalyst also maintained stable NH3 conversion during a 60 h time-on-stream test at 450 °C, giving an average conversion of 90.94%. These results indicate competitive performance among recently reported Ru-based catalysts. The structural and surface characteristics of the catalysts were further clarified by TEM-EDS, XPS, XRD, and BET analyses, which supported that the enhanced catalytic performance was mainly associated with electronic modulation of Ru and modification of the Ce–O/Ru–Ce–O interfacial environment, rather than with major geometric changes in Ru particle size.
Solar-driven interfacial evaporation (SDIE) holds great promise for sustainable water purification and desalination. However, the complicated fabrication procedures of evaporators remain a challenge. More critically, many existing evaporators rely on polymer networks as mechanical skeletons, which often compromise porosity and water transport, leading to limited evaporation rates. Here, we report a polymer-free self-supporting graphene aerogel prepared via partial chemical reduction of graphene oxide (GO). During reduction, the GO sheets crosslink through strong π-π stacking interactions, forming a three-dimensional porous network in which the reduced graphene itself serves as the structural backbone. The resulting aerogel features high hydrophilicity, broadband light absorption of approximately 97% under wet conditions, and a porous architecture that facilitates rapid water transport and steam escape. By systematically varying the GO sheet size and concentration, we reveal how these parameters govern the aerogel’s intermediate water content, which collectively determine the evaporation performance. The optimized aerogel achieves an evaporation rate of 3.98 kg m−2 h−1 and an efficiency of 96.03% Under 1 sun illumination, with >99.9% ion rejection and stable cycling performance. This work establishes a structure-property relationship in polymer-free graphene aerogels, offering insights for designing high-performance solar evaporators beyond conventional polymer-based systems.
