arXiv Science⌕ Search

arXiv subjects

Kai Gong

Publications and source records attributed to Kai Gong.

At least 19 recordsLinked to original sources

Gel-Chemistry-Controlled Heavy-Metal Transport and Retention in Cementitious Gel Nanopores: A Molecular Dynamics Study

Cementitious materials are widely used for hazardous-waste encapsulation, yet molecular mechanisms governing heavy-metal ion retention across gel chemistries remain insufficiently resolved. Previous MD studies have focused on individual gel systems and pore-averaged metrics, limiting direct comparison across gel chemistries. Here, classical molecular dynamics simulations were used to investigate adsorption-controlled mobility of Pb2+, Ba2+, and Cs+ in nanopores of C-S-H, C-(N)-A-S-H, and N-A-S-H gels. Combining pore-averaged and spatially resolved diffusivity, residence-time analysis, ion-density profiles, 2D adsorption maps, radial distribution functions, coordination analysis, and interfacial binding-strength descriptors establishes a framework linking gel surface chemistry to mobility suppression while distinguishing local mobility from spatial ion partitioning effects on pore-averaged transport. Ion mobility is reduced in all gel nanopores relative to bulk, but the extent and mechanism vary strongly with gel chemistry. C-(N)-A-S-H with higher interfacial Al/Si substitution exhibits the strongest retention, driven by ion accumulation around Al-linked oxygen species via an ion-exchange-like mechanism with charge-balancing Na+. C-S-H immobilizes ions primarily through surface hydroxyl oxygens and Ca-mediated linkages, whereas N-A-S-H exhibits more distributed binding environments. Pb2+ and Ba2+ show broadly similar immobilization mechanisms, whereas Cs+ exhibits more gel-dependent interactions with silicate and aluminosilicate oxygen sites. A relative total binding strength (rTBS) descriptor shows a strong positive correlation with ion immobilization across gel types, ion species, and pore sizes. These results clarify gel- and ion-specific retention mechanisms and link interfacial binding strength to pore-averaged mobility suppression in idealized cementitious nanopores.

physics.chem-ph↗

OpenCollab: A Multi-Agent Coding Framework with Programmable Collaboration and Controllable Runtime

Multi-agent coding systems are designed to tackle complex software engineering tasks through collaboration. However, existing evaluations typically assume configured organizations are followed faithfully, whereas reality differs. This behavioral gap, combined with differences in underlying system components, prevents clear attribution of observed gains. To this end, we introduce OpenCollab, a multi-agent coding framework that provides a unified infrastructure for programmable collaboration and controllable runtime. Specifically, OpenCollab unifies organization design, enforces experimental control on a shared runtime, and tracks execution through fine-grained event streams. On this basis, we define Adherence to quantify whether the declared organization is actually realized. Our experiments reveal that agents collaborate very differently across configurations: changing any single dimension shifts Adherence, from 47.2% to as high as 97.2%. Furthermore, extensive agentic coding benchmarks show that OpenCollab (Duo) establishes new SOTA performance compared to the mainstream harnesses such as Mini-SWE-agent, OpenHands, Codex CLI, and Claude Code, showing that a well-designed organization can outperform strong existing harnesses, while OpenCollab's single-agent configuration uses the fewest tokens across all evaluated suites. OpenCollab establishes a unified multi-agent infrastructure for easy programmable collaboration and controlled causal evaluation.

cs.SE↗

Reaction mechanisms of sodium carbonate activation in ultra-high-volume slag-cement blends

Ultra-high-volume slag-cement (UHVS) blends can substantially reduce clinker ratio, but slow slag reaction limits strength development; Na2CO3 is a less caustic and lower-carbon activator for alkali-activated materials, yet its moderate alkalinity can similarly delay strength development. Here, dry-blended Na2CO3 (0-20 wt.% relative to water) was incorporated into a one-part binder with 90 wt.% ground-granulated blast-furnace slag and 10 wt.% Portland cement (PC) to determine whether this integration could mitigate slow strength development in both systems. Compressive strength testing, isothermal calorimetry, time-resolved quantitative X-ray diffraction, Fourier-transform infrared spectroscopy, aqueous solution analysis, and thermodynamic modeling were combined to determine how Na2CO3 controls phase assemblage, alkalinity, calculated porosity and strength development. At 5-10 wt.% Na2CO3, calcite-dominated carbonate reactions promoted a rapid alkalinity increase and accelerated early-age reaction and strength development; however, subsequent alkalinity decline limited later-age slag reaction and gel formation. In contrast, 20 wt.% Na2CO3 favored substantial early formation of gaylussite and carbonate-AFm phases, which moderated the initial pH rise and delayed early gel accumulation. The larger alkali inventory and evolving phase-solution partitioning helped sustain alkalinity at later ages, enabling continued slag dissolution and gel formation. This pathway produced higher reaction extent and lower calculated porosities, with compressive strength approaching that of neat PC. Screening-level cradle-to-gate life-cycle assessment indicates a 66-72% reduction in global warming potential relative to PC, depending on Na2CO3 production pathways. These results identify carbonate-mediated control of alkalinity persistence as key to sustaining slag reaction in UHVS systems.

physics.chem-ph↗

Large language model-enabled automated data extraction for concrete materials informatics

The promise of data-driven materials discovery remains constrained by the scarcity of large, high-quality, and accessible experimental datasets. Here, we introduce a generalizable large language model (LLM)-powered pipeline for automated extraction and structuring of materials data from unstructured scientific literature, using concrete materials as a representative and particularly challenging example. The pipeline exhibits robust performance across a broad range of LLMs and achieves an $F_1$ score of up to 0.98 for diverse composition--process--property attributes. Within one hour, it extracts nearly 9,000 high-quality records with over 100 attributes from a corpus screened from more than 27,000 publications, enabling the construction of the largest open laboratory database for blended cement concrete. Machine learning analyses underscore the importance of large, diverse, and information-rich datasets for enhancing both in-distribution accuracy and out-of-distribution generalization to unseen materials. The proposed pipeline is readily adaptable to other materials domains and accelerates the development of scalable data infrastructures for materials informatics.

cond-mat.mtrl-sci↗

Automating literature screening and knowledge synthesis with large language models

Scientific literature is expanding rapidly, yet much of its knowledge remains difficult to organize and reuse due to the substantial workload needed to navigate large, heterogeneous literature corpora. Here, we develop a generalizable large language model (LLM)-powered framework that automates literature screening, structured information extraction, and knowledge synthesis within a unified workflow. We benchmarked 11 state-of-the-art proprietary and open-weight LLMs across three materials science screening tasks differing in scope, corpus size, and class imbalance. Seven LLMs achieved balanced accuracies above 0.90 across all tasks, with the best-performing LLM reaching 0.96--0.98, demonstrating the potential of LLMs to accurately screen large, highly imbalanced scientific corpora. The framework also substantially reduces manual effort, screening 10,000 records within four minutes compared with approximately 83 person-hours for manual screening. Beyond literature screening, the framework extracts structured information from full-text articles, enabling content-level quantification of research landscapes that extends traditional bibliometric analysis based on publication metadata. This capability reveals methodological trends and underexplored directions, as shown for machine learning in concrete materials science. We further demonstrate its utility for information and data discovery by identifying publications containing silicate glass density data that are difficult to retrieve using conventional search strategies, which can support downstream hypothesis development, research design, and data-driven scientific discovery. Overall, these results demonstrate the broad applicability of LLM-powered literature analysis across scientific domains and its potential to transform large, unstructured literature corpora into structured, analyzable knowledge at scale.

cs.CL↗

Reactivity-Informed Machine Learning for Performance Prediction and Design Space Exploration of Alkali-Activated Slag

Establishing quantitative relationships among mix design, raw material properties, curing conditions, and performance remains a long-standing challenge in cementitious materials, particularly for alkali-activated materials with variable precursor and activator chemistry. Here, we curated the largest literature-derived alkali-activated slag (AAS) dataset to date, comprising over 3100 compressive strength records, 155 chemically distinct ground granulated blast-furnace slags (GGBSs), and 24 attributes incorporating precursor chemistry, fineness, and reactivity. Multiple machine learning (ML) algorithms were benchmarked across progressively enriched feature scenarios, demonstrating that integrating GGBS compositions, fineness, curing conditions, and specimen geometry improves predictive performance. The average metal oxide dissociation energy (AMODE), a physically interpretable representation of precursor reactivity, provides a compact alternative descriptor to explicit oxide compositions while enabling comparable predictive performance. Model interpretation revealed physically consistent trends from heterogeneous data, including non-monotonic effects of Na2O dosage and silicate modulus, reduced predicted strength at higher water content and larger specimen size, and coupled oxide-level effects more coherently represented by AMODE than by individual oxide contents. Statistically constrained design space exploration reveals reactivity-dependent trade-offs among strength, embodied CO2 emissions, and cost. The design maps identify high-strength regions with substantially lower CO2 emissions than OPC-based references at similar cost. Overall, this work demonstrates how reactivity-informed ML can extract physically meaningful trends from heterogeneous AAS data and guide source-dependent binder design. The curated dataset is publicly accessible to support advances in cement and concrete research.

cond-mat.mtrl-sci↗

Physics-Informed Glass-Structure Descriptors for Assessing the Intrinsic Reactivity of Mixed Amorphous-Crystalline Precursors in Alkali-Activated Materials

Rapid and reliable assessment of the intrinsic reactivity of amorphous aluminosilicates is critical for their application in alkali-activated materials (AAMs) and blended cements. Although physics-informed glass-structure descriptors have demonstrated strong structure-reactivity relationships for predominantly amorphous systems, their extension to heterogeneous precursors with mixed crystalline-amorphous phases has been limited. Here, quantitative X-ray diffraction combined with bulk compositional analysis was used to reconstruct the effective amorphous compositions of five fly ashes (FAs) and three ground granulated blast-furnace slags (GGBSs). These compositions served as inputs for molecular dynamics simulations employing a melt-and-quench approach to generate atomic-scale structural models of the glassy phases. Based on these structures, the previously introduced descriptors, i.e., average metal oxygen dissociation energy and average metal oxygen bond strength, were refined to cover a broader compositional space spanning SiO2-Al2O3-TiO2-Fe2O3-CaO-MgO-MnO-Na2O-K2O. The refined descriptors exhibit strong inverse correlations with multiple independent reactivity indicators, including cumulative heat release from isothermal calorimetry, bound water content from thermogravimetric analysis, and compressive strength, for both single precursors and binary FA-GGBS blends activated with NaOH. These results demonstrate that physics-informed glass-structure descriptors can be extended from ideal amorphous systems to heterogeneous mixed-phase precursors and capture relative intrinsic reactivity trends in alkaline solutions. The proposed framework provides a transferable, structure-informed basis for comparative assessment of precursor reactivity that complements experimental testing and may inform precursor screening and mix designs for AAM and blended cement systems.

cond-mat.mtrl-sci↗

Unveiling the Thermoelectric Properties of Group III-Nitride Biphenylene Networks

After the synthesis of the carbon biphenylene network (C-BPN), research has increasingly focused on adapting elements from other groups of the periodic table to this lattice structure. In this study, the direction-dependent electronic, thermal, and thermoelectric (TE) properties of semiconducting group-III (group-III = B, Al, Ga, In) nitride biphenylene networks are investigated using the non-equilibrium Green's function formalism in combination with first-principles calculations. Phonon spectra and force field molecular dynamics (MD) simulations were used to asses the dynamically and thermally stable structures. At room temperature, the lowest phonon thermal conductance values are obtained for InN-BPN, with $κ_{\mathrm{ph}}$ = 0.12 nW/K/nm and $κ_{\mathrm{ph}}$ = 0.21 nW/K/nm along the armchair and zigzag directions, respectively. The nearly dispersionless valence-band region between the $Γ$--$X$ symmetry points causes a sharp increase in the $p$-type electronic transmission, which significantly enhances the $p$-type thermoelectric figure of merit, $zT$. Among the investigated group-III nitride BPNs, InN-BPN exhibits the best performance, with a $p$-type $zT$ value of 2.33 in the zigzag direction at 800 K.

cond-mat.mtrl-sci↗

Insights into Ionic Diffusion in C-S-H Gel Pore from Molecular Dynamics Simulations: Spatial Distributions, Energy Barriers, and Structural Descriptor

Understanding transport behavior in nanoconfined environments is critical to many natural and engineering systems, including cementitious materials, yet its molecular-level mechanisms remain poorly understood. Here, molecular dynamics (MD) simulations were used to investigate Na, Cl, and water diffusion inside a 4 nm calcium-silicate-hydrate (C-S-H) pore channel over temperatures ranging from 300 K to 360 K. Spatially resolved analysis revealed strong suppression of diffusivity near the solid-liquid interface and gradual recovery toward the pore center. Arrhenius analysis further quantified the spatial variation of activation energy barriers and intrinsic mobilities across the pore channel, showing distinct confinement effects. The spatially resolved structural analysis uncovers a mechanistic transition from structure-controlled to hydrodynamics-controlled transport regimes with increasing distance from the pore surface. A structural descriptor, total coordination strength (TCS), was introduced, providing a predictive link between local liquid structure and molecular mobility within approximately 1 nm of the interface. Beyond approximately 1 nm, suppressed diffusivities were well captured by an empirical model inspired by the Darcy-Brinkman framework. To the best of our knowledge, this is the first MD study to comprehensively resolve the spatial heterogeneity of transport, thermal kinetics, and structure within cementitious nanopores. These findings deepen the fundamental understanding of nanoscale transport phenomena and suggest that tailoring the nanochannel structure and interfacial chemistry of cementitious gels, for example surface coordination environments, pore size distributions, and adsorption sites, may offer a promising strategy to suppress ionic ingress and enhance the durability of cement-based materials.

physics.chem-ph↗

Ultra-thin Carbon Biphenylene Network as an Anisotropic Thermoelectric Material with High Temperature Stability Under Mechanical Strain

Carbon biphenylene network (C-BPN), which is an ultra-thin material consisting of carbon atoms arranged in square-hexagonal-octagonal (4-6-8) periodic rings, has intriguing properties for nano-scale device design due to its unique crystal structure. Here, using the Landauer formalism in combination with first-principles calculations, we show that C-BPN is a highly stable thermoelectric material at elevated temperatures under mechanical strain, where its thermoelectric efficiency can be anisotropically engineered. Transport calculations reveal that C-BPN's transmission spectrum has significant degrees of directional anisotropy and it undergoes a metal-insulator transition under strain, which leads to an increase in its Seebeck coefficient. C-BPN's lattice thermal conductance can be selectively tuned up to 35% bidirectionally at room temperature by strain engineering. Enhancement in its power factor and the suppression of its lattice thermal conductance improves the p-type figure of merit up to 0.31 and 0.76 at 300 and 1000~K, respectively. Our findings reveal that C-BPN has high potency to be used in thermoelectric nano devices with selective anisotropic properties at elevated temperatures.

cond-mat.mtrl-sci↗

Landmark-guided Diffusion Model for High-fidelity and Temporally Coherent Talking Head Generation

Audio-driven talking head generation is a significant and challenging task applicable to various fields such as virtual avatars, film production, and online conferences. However, the existing GAN-based models emphasize generating well-synchronized lip shapes but overlook the visual quality of generated frames, while diffusion-based models prioritize generating high-quality frames but neglect lip shape matching, resulting in jittery mouth movements. To address the aforementioned problems, we introduce a two-stage diffusion-based model. The first stage involves generating synchronized facial landmarks based on the given speech. In the second stage, these generated landmarks serve as a condition in the denoising process, aiming to optimize mouth jitter issues and generate high-fidelity, well-synchronized, and temporally coherent talking head videos. Extensive experiments demonstrate that our model yields the best performance.

cs.CV↗

Decentralized Proximal Method of Multipliers for Convex Optimization with Coupled Constraints

In this paper, a decentralized proximal method of multipliers (DPMM) is proposed to solve constrained convex optimization problems over multi-agent networks, where the local objective of each agent is a general closed convex function, and the constraints are coupled equalities and inequalities. This algorithm strategically integrates the dual decomposition method and the proximal point algorithm. One advantage of DPMM is that subproblems can be solved inexactly and in parallel by agents at each iteration, which relaxes the restriction of requiring exact solutions to subproblems in many distributed constrained optimization algorithms. We show that the first-order optimality residual of the proposed algorithm decays to $0$ at a rate of $o(1/k)$ under general convexity. Furthermore, if a structural assumption for the considered optimization problem is satisfied, the sequence generated by DPMM converges linearly to an optimal solution. In numerical simulations, we compare DPMM with several existing algorithms using two examples to demonstrate its effectiveness.

math.OC↗

Distributed Proximal-Correction Algorithm for the Sum of Maximal Monotone Operators in Multi-Agent Network

This paper focuses on a class of inclusion problems of maximal monotone operators in a multi-agent network, where each agent is characterized by an operator that is not available to any other agents, but the agents can cooperate by exchanging information with their neighbors according to a given communication topology. All agents aim at finding a common decision vector that is the solution to the sum of agents' operators. This class of problems is motivated by distributed convex optimization with coupled constraints. In this paper, we propose a distributed proximal point method with a cumulative correction term (named Proximal-Correction Algorithm) for this class of inclusion problems of operators. It's proved that the Proximal-Correction Algorithm converges for any value of a constant penalty parameter. In order to make the Proximal-Correction ALgorithm computationally implementable for a wide variety of distributed optimization problems, we adopt two inexact criteria for calculating the proximal steps of the algorithm. Under each of these two criteria, the convergence of Proximal-Correction Algorithm can be guaranteed, and the linear convergence rate is established when the stronger one is satisfied. In numerical simulations, both exact and inexact versions of Proximal-Correction Algorithm are executed for a distributed convex optimization problem with coupled constraints. Compared with several alternative algorithms in the literature, the exact and inexact versions of Proximal-Correction both exhibit good numerical performance.

math.OC↗

Push-Pull Based Distributed Primal-Dual Algorithm for Coupled Constrained Convex Optimization in Multi-Agent Networks

This paper focuses on a distributed coupled constrained convex optimization problem over directed unbalanced and time-varying multi-agent networks, where the global objective function is the sum of all agents' private local objective functions, and decisions of all agents are subject to coupled equality and inequality constraints and a compact convex subset. In the multi-agent networks, each agent exchanges information with other neighboring agents. Finally, all agents reach a consensus on decisions, meanwhile achieving the goal of minimizing the global objective function under the given constraint conditions. For the purpose of protecting the information privacy of each agent, we first establish the saddle point problem of the constrained convex optimization problem considered in this article, then based on the push-pull method, develop a distributed primal-dual algorithm to solve the dual problem. Under Slater's condition, we will show that the sequence of points generated by the proposed algorithm converges to a saddle point of the Lagrange function. Moreover, we analyze the iteration complexity of the algorithm.

math.OC↗

Development of physics-based compositional parameters for predicting the reactivity of amorphous aluminosilicates in alkaline environments

The reactivity of amorphous aluminosilicates in alkaline environments is important to many applications, including blended Portland cements and alkali-activated materials (AAMs). Here, two physics-based compositional parameters are derived to describe the relative reactivity of CaO-Al2O3-SiO2 and CaO-MgO-Al2O3-SiO2 glasses in alkaline environments: (i) a modified average metal oxide dissociation energy (AMODE) parameter developed from molecular dynamics simulations; and (ii) a topology constraint parameter derived from topological constraint theory. Both parameters are seen to generally outperform existing compositional parameters from the literature for a wide range of aluminosilicate glasses. Given that both the modified AMODE and topology constraint parameters can be calculated directly from chemical compositions, this study represents a major step forward in connecting aluminosilicate glass compositions with their reactivity in alkaline environments. The limitations of these compositional parameters, including their inability to capture the impact of Ca versus Mg on glass reactivity, have also been discussed.

cond-mat.mtrl-sci↗

Density functional modeling of the binding energies between aluminosilicate oligomers and different metal cations

Interactions between negatively charged aluminosilicate species and positively charged metal cations are critical to many important engineering processes and applications, including sustainable cements and aluminosilicate glasses. In an effort to probe these interactions, here we have calculated the pair-wise interaction energies (i.e., binding energies) between aluminosilicate dimer/trimer and 17 different metal cations using a density functional theory (DFT) approach. Analysis of the DFT-optimized structural representations for the clusters (dimer/trimer + cation) shows that their structural attributes (e.g., interatomic distances) are generally consistent with literature observations on aluminosilicate glasses. The DFT-derived binding energies are seen to vary considerably depending on the type of cations (i.e., charge and ionic radii) and aluminosilicate species (i.e., dimer or trimer). A survey of the literature reveals that the difference in the calculated binding energies between different cations can be used to explain many literature observations on the impact of cations on materials properties (e.g., glass corrosion, mineral dissolution, and ionic transport). Analysis of all the DFT-derived binding energies reveals that the correlation between these energy values and the ionic potential and field strength of the cations are well captured by 2nd order polynomial functions (R2 values of 0.99-1.00 are achieved for regressions). Given that the ionic potential and field strength of a given metal cation can be readily estimated using well-tabulated ionic radii available in the literature, these simple polynomial functions would enable rapid estimation of the binding energies of a much wider range of cations with the aluminosilicate dimer/trimer, providing guidance on the design and optimization of sustainable cements and aluminosilicate glasses and their associated applications.

cond-mat.mtrl-sci↗

Data-driven prediction of room temperature density for multicomponent silicate-based glasses

Density is one of the most commonly measured or estimated materials properties, especially for glasses and melts that are of significant interest to many fields, including metallurgy, geology, materials science and sustainable cements. Here, two types of machine learning (ML) models (i.e., random forest (RF) and artificial neural network (ANN)) have been developed to predict the room-temperature density of glasses in the compositional space of CaO-MgO-Al2O3-SiO2-TiO2-FeO-Fe2O3-Na2O-K2O-MnO (CMASTFNKM), based on ~2100 data points mined from ~140 literature studies. The results show that the RF and ANN models give accurate predictions of glass density with R2 values, RMSE, and MAPE of ~0.96-0.98, ~0.02-0.03 g/cm3 and ~0.59-0.79%, respectively, for the 15% testing set, which are more accurate compared with empirical density models based on ionic packing ratio (with R2 values, RMSE, and MAPE of ~0.28-0.91, ~0.05-0.15 g/cm3, and ~1.40-4.61%, respectively). Furthermore, glass density is shown to be a reliable reactivity indicator for a range of CaO-Al2O3-SiO2 (CAS) and volcanic glasses due to its strong correlation (R2 values above ~0.90) with the average metal-oxygen dissociation energy (a structural descriptor) of these glasses. Analysis of the predicted density-composition relationships from these models (for selected compositional subspaces) suggests that the ANN model exhibits a certain level of transferability (i.e., ability to extrapolate to compositional space not (or less) covered in the database) and captures known features including the mixed alkaline earth effects for (CaO-MgO)0.5-(Al2O3-SiO2)0.5 glasses.

cond-mat.mtrl-sci↗

Time-dependent phase quantification and local structure analysis of hydroxide-activated slag via X-ray total scattering and molecular modeling

Here, an approach to quantify the amorphous-to-disordered/crystalline transformation occurring in NaOH-activated ground granulated blast-furnace slag (GGBS) is outlined that combines atomistic modeling with in situ pair distribution function (PDF) analysis. Firstly, by using force-field molecular dynamics (MD) simulations, a detailed structural representation is generated for the amorphous GGBS that is in agreement with experimental X-ray scattering data. Use of this structural representation along with literature-derived structures for the reaction products allows for real space X-ray PDF refinement of the alkaline activation of GGBS, resulting in the quantification of all phases and the degree of reaction (DOR) as a function of reaction time. All phases and the DOR are seen to approximately follow a logarithmic-type time-dependent behavior up to 5 months, while at early age (up to 11 hours) the DOR is accurately captured by a modified pseudo-single step first-order reaction model. Lastly, the evolution of DOR is found to agree with several other complementary in situ data containing quantitative reaction information, including isothermal conduction calorimetry, Fourier transform infrared spectroscopy, and quasi-elastic neutron scattering.

cond-mat.mtrl-sci↗