Jingai Shao

Establishing quantitative relationships between the multidimensional structural characteristics of hard carbon and its sodium-storage performance remains challenging because different electrochemical metrics are associated with distinct and coupled structural factors. In this study, a literature-derived database containing 498 polymer-derived hard-carbon samples and 27 structural and process descriptors was established, covering elemental composition, carbon chemical states, pore structure, crystal structure, and carbonization parameters. Nine machine-learning algorithms were evaluated, and a target-specific Top-5 weighted ensemble strategy was employed to integrate complementary predictive models. Under the current random-split evaluation protocol, the ensemble model achieved test-set R² values of 0.927, 0.929, 0.882, 0.934, and 0.964 for initial Coulombic efficiency (ICE), initial charge capacity (Qcha), initial discharge capacity (Qdis), sloping capacity (Qslop), and plateau capacity (Qplat), respectively. The cycling-retention model (retN) showed a comparatively lower R² of 0.759 and is therefore treated as an exploratory analysis rather than as a primary predictive benchmark. SHAP, partial dependence analysis, and feature-interaction analysis revealed distinct model-attributed relationships across the different performance metrics. Pore-related descriptors showed the highest importance for ICE and Qdis, elemental-composition descriptors were strongly associated with Qcha and Qslop, carbonization parameters showed the strongest association with Qplat, and carbon chemical states were prominently associated with cycling retention. The resulting response trends and interaction patterns identify data-supported favorable structural regions within the observed literature domain rather than experimentally validated universal optima. This work provides an interpretable multi-target framework for comparing structure–performance relationships in polymer-derived hard carbon and for generating experimentally testable hypotheses for sodium-ion battery anode design.

Converting polyethylene (PE) waste into liquid fuels through catalytic hydrocracking provides an attractive route for plastic upcycling, yet achieving efficient hydrogen activation and selective hydrocracking over catalysts remains challenging. In this work, a starch-templated hierarchical HZSM-5 was employed as the support to tailor the dispersion and catalytic behavior of Ru nanoparticles. By systematically tuning the Ru loading (1-6 wt%) and calcination temperature (400-600oC), the structure-performance relationship governing LDPE hydrocracking toward liquid fuels was established. The hierarchical pore architecture promoted the dispersion of Ru nanoparticles and facilitated the diffusion of bulky PE-derived intermediates, while oxygen-deficient defect sites strengthened the metal-support interaction (MSI) and accelerated hydrogen activation. Among the catalysts investigated, the sample containing 3 wt% Ru calcined at 550 oC exhibited the highest liquid yield (77.56%) with a diesel-range (C10-C22) selectivity of 68.54%. This superior performance is attributed to the synergistic effect of highly dispersed Ru nanoparticles and abundant defect-associated oxygen sites, which creates balanced hydrogenation and hydrogenolysis functionalities. These findings demonstrate that the catalytic performance of hierarchical Ru/HZSM-5 is governed by the cooperative regulation of pore architecture, Ru nanostructure, surface oxygen defect structures, and MSI rather than Ru loading alone. To quantitatively describe the MSI-performance relationship, a MSI coefficient KMSI was established by integrating multiple structural and electronic parameters. Quantitative analysis verifies that the liquid yield displays a positive correlation with KMSI, providing a rational strategy for designing efficient hierarchical zeolite catalysts for plastic waste valorization.

Developing biomass‐based solid acid catalysts and converting biomass‐derived molecules into high‐value chemicals has a dual promoting effect on the high‐value utilization of waste biomass. This study introduces a groundbreaking one‐pot hydrothermal strategy integrating biomass with MIL‐101(Cr) precursor to fabricate hierarchical carbon‐based catalysts for the highly efficient and selective dehydration of xylose to furfural, achieving 67.2%–72.0% furfural yield (a 33% enhancement over conventional biomass‐derived catalysts) under optimized conditions (180°C, 3 h) in a toluene/water biphasic system. The modified hydrochar exhibits an ultra‐high surface area of 2119.3 m ² /g, abundant dual Lewis/Brønsted acid sites, and dispersed Cr ³⁺ active centers, enabled by synergistic MOF‐templated modification engineering and biomass hydrothermal carbonization. Moreover, the correlation between the physicochemical properties and catalytic performance of modified hydrochar was analyzed. Notably, biomass feedstocks with low ash content and high nitrogen content can yield modified hydrochar with more acidic active sites, which exhibits superior catalytic activity. This study provides new ideas for the high‐value utilization of waste biomass.

A MOF-mediated Structural Modulation of biomass-derived carbon into flower-like architectures, achieving ORR activity (0.83V vs. RHE) comparable to Pt/C and enabling uniform Ni doping for CO2 electroreduction. This design delivers >91% FECO at-1.0 V vs RHE with 84.8% stability over 48-hour operation, outperforming irregular morphologies.

The N-doped biochar is recognized as a promising, cost-effective, and efficient material for CO₂ adsorption. However, achieving efficient enrichment of N-containing adsorption sites and improving their accessibility remains a bottleneck problem that restricts the adsorption performance of N-doped biochar. Herein, a synthesis strategy for nitrogen-doped biochar by one-pot ionothermal treatment of biomass and zeolitic imidazolate framework (ZIF) precursors accompanied by pyrolysis is demonstrated. Through ion thermal ZIF modification, biochar exhibits a controllable flower-like morphology with effective enrichment of nitrogen elements (nitrogen retention rates ranging from 62% to 88%). After pyrolysis, this regular morphology is retained, and a developed hierarchical pore structure is formed. Compared with pristine biochar and ZIF-derived carbon, ZIF-modified biochar has superior CO₂ adsorption capacity (up to 3.5 mmol/g) and excellent CO₂/N₂ adsorption selectivity (up to 38.6). The CO₂ adsorption capacities of ZIF-modified biochars have a good linear relationship with both bulk and surface N content, with correlative coefficients of around 0.998 and 0.950, respectively. This positional indifference reflects the effective accessibility of N-containing sites, which can be attributed to the ordered flower-like morphology and hierarchical pore structure of ZIF-modified biochar. The DFT results confirmed the importance of the number and accessibility of such N-containing sites for CO₂ adsorption.