Hasil untuk "Materials Science"

Ting Zhu, Ju Li

Delia McGrath, Curtis Chong, Rohil Kulkarni et al.

Scientific reasoning in materials science requires integrating multimodal experimental evidence with underlying physical theory. Existing benchmarks make it difficult to assess whether incorporating visual experimental data during post-training improves mechanism-grounded explanation reasoning beyond text-only supervision. We introduce MATRIX, a multimodal benchmark for materials science reasoning that evaluates foundational theory, research-level reasoning, and the interpretation of real experimental artifacts across multiple characterization modalities. Using MATRIX as a controlled diagnostic, we isolate the effect of visual grounding by comparing post-training on structured materials science text alone with post-training that incorporates paired experimental images. Despite using relatively small amounts of multimodal data, visual supervision improves experimental interpretation by 10-25% and yields 5-16% gains on text-only scientific reasoning tasks. Our results demonstrate that these improvements rely on correct image-text alignment during post-training, highlighting cross-modal representational transfer. We also observe consistent improvements on ScienceQA and PubMedQA, demonstrating that the benefits of structured multimodal post-training extend beyond materials science. The MATRIX dataset is available at https://huggingface.co/datasets/radical-ai/MATRIX and the model at https://huggingface.co/radical-ai/MATRIX-PT.

Michael Schäfer, Ulrike Faltings, Björn Glaser

Abstract In the blast furnace and basic oxygen furnace route, pig iron and steel scrap are used as resources for steel production. The scrap content can consist of many different types of scrap varying in origin and composition. This makes it difficult to compile the scrap mix and predict the future chemical analysis in the converter. When compiling the scrap mix, steel manufacturers often rely on experience and trials. In this paper, we present a machine learning approach based on XGBoost to predict the chemical element content in the converter. Data from around 115000 heats were analyzed and a model was developed to better predict the content of the tramp elements copper, chromium, molybdenum, phosphorus, nickel, tin and sulphur at the end of the basic oxygen furnace process. The study shows that it is possible to predict the chemical element content for tramp elements in the converter based solely on data available in advance and routinely collected without the necessity of additional sensors or analysis of input material. Given the nature of scrap classifications for (external) scrap types, this is non-trivial. Furthermore, an online model was implemented, accessible via a defined synchronous interface, which allows to optimize the use of different scrap types by predicting the chemical content at the end of the basic oxygen furnace process and simulating with new combinations of input material. Not all types of steel scrap are always available. With the model developed, new scrap input constellations can now be created to ensure that the quality of the melt is maintained. However, for very accurate predictions, the data from the upstream processes must be of high quality and quantity. Efficient scrap management, monitoring of the scrap input and confusion checks.

Deyu Jiang, Miao Luo, Changxi Liu et al.

In this study, a Ti1.5Nb1Ta0.5Zr1Mo0.5 (TNTZM) high-entropy alloy was fabricated using laser powder bed fusion (LPBF). By integrating 63 sets of parameter trials with machine learning (ML) models, an optimised process window was identified, achieving a density of up to 99.9%. The combination of relatively high laser power and low scanning speed resulted in the formation of a stable cellular structure. Subsequent heat treatments at 700, 850, and 1000°C showed that while small-angle misorientations developed at cell-wall interfaces and medium-entropy (Ti–Zr–Mo) second-phase particles precipitated preferentially in the cell walls, the overall cellular architecture remained intact. Mechanical testing showed that these heat-treated samples exhibited yield strengths over 150 MPa higher than the as-built samples, while still retaining nearly 50% ductility under short-term heat treatment. In particular, small-angle grain boundaries and nanoscale second-phase particles together reinforce the cell walls and promote intracellular dislocation accumulation, thereby improving the overall mechanical properties of the alloy. These results demonstrate that combining ML-guided process design with targeted heat treatment is an effective method for additive manufacturing of refractory HEAs with high density and mechanical properties.

Lisley Madeira Coelho, Belayne Zanini Marchi, Pedro Henrique Poubel Mendonça da Silveira et al.

Abstract The use of Reclaimed Asphalt Pavement (RAP) in road base layers represents a solution to reduce the consumption of natural aggregates. However, the variability of RAP properties poses challenges to its application, particularly regarding mechanical behavior. This study investigates thermal compaction as a strategy to stabilize mixtures composed exclusively of RAP, introducing the concept of a warm base. Repeated load triaxial tests were conducted to evaluate the effects of compaction temperature on permanent deformation (PD) and resilient modulus (RM). The results indicate that increasing the compaction temperature significantly improves the mechanical behavior of RAP, reducing PD by up to 52% at the highest stress level. Additionally, the RM of RAP-M samples increased by approximately 187.13% compared to the maximum value of RAP-F samples and 389.05% compared to the minimum value. This approach enables the application of larger quantities of RAP in pavements, ensuring good structural quality while minimizing the effects of the material’s initial variability.

Jialu Suo, Li Guan, Peng Chen et al.

Electromagnetic shielding materials are pivotal for suppressing electromagnetic radiation and mitigating potential health risks that electronic devices may pose to humans. Beyond health protection, they also hold significant strategic value in safeguarding national information security and maintaining stability. In the research of electromagnetic shielding materials, continuous technological advancements and growing application demands have driven the emergence of various novel materials. Among these, liquid metal (LM) exhibits outstanding properties—including exceptional electrical conductivity, excellent fluidity, and superior deformability—which endow it with substantial potential for application in electromagnetic shielding. Looking ahead, with the continuous advancement in related technologies, liquid metal-based electromagnetic shielding materials are expected to provide effective solutions to key challenges such as electromagnetic pollution and interference. This contribution synthesizes the latest literature. First, it clarifies the nomenclature and classification of liquid metals, as well as the fundamental framework for electromagnetic shielding. Then, it systematically distills recent research advances based on four key design motifs. These motifs include monolithic liquid metal (LM) scaffolds, LM/conductive-filler blends, LM/magnetic particle composites, and architectured multifunctional architectures. Finally, this review identifies current bottlenecks in the field and outlines directions for future development, which aim to achieve ultra-lightweight, broadband, and intelligent LM-based electromagnetic shields.

Yuantiao Xie, Dajian Huang, Wenjie Tang et al.

High drying shrinkage remains a major challenge in alkali-activated materials (AAM). In this study, porous palm fiber (PF) was modified with NaOH solution, saturated with water via vacuum impregnation, and incorporated as an internal curing agent into metakaolin-based AAM. The influence of PF on hydration behavior and microstructure was examined using scanning electron microscopy (SEM) coupled with energy-dispersive X-ray spectroscopy (EDS), Fourier transform infrared spectroscopy (FT-IR), X-ray diffraction (XRD), Vicat setting tests, and low-field nuclear magnetic resonance (LF NMR). The results showed that PF incorporation promoted the formation of additional hydration products and enhanced microstructural densification. Flexural strength was significantly improved, with the sample containing 1.5 % of 6 mm PF achieving the highest enhancement—an increase of 66.7 % compared with the control. After 28 days, a small amount of PF slightly increased the compressive strength. Vertical mortar expansion tests confirmed that PF effectively reduced drying shrinkage, with the most pronounced reductions observed in mixtures containing 1 % of 3 mm PF and 1.5 % of 6 mm PF, which decreased shrinkage by 42.9 % and 41.1 %, respectively, compared with the MK-AAM control sample (P0, 0 % PF). Overall, PF markedly mitigates drying shrinkage, improves flexural performance, and shows strong potential as a sustainable and environmentally friendly internal curing agent for AAM.

Lei He, Xiaofeng Ma, Yongze Liu et al.

ABSTRACT Bacterial infection‐induced acute sinusitis is prevalent and can easily progress into chronic sinusitis, which is often difficult to treat due to the challenging nature of the site, increased environmental pollution, and bacterial drug resistance prevalent nowadays. To address these challenges, a flexible hydrogel (LM@P/S@CP@Hemin) that involves flexible wood‐modified logs, photoactive conjugated polymers, an immunomodulator, and an immobilization hydrogel was prepared for nasal cavity treatment. The flexible wood‐modified logs provide mechanical strength support. In vitro, experiments verified that the hydrogel could efficiently induce the photothermal effect under near‐infrared‐II laser irradiation after deeply penetrating bone and produce reactive oxygen species (ROS) to initiate the photodynamic effect for synergetically eliminating bacteria. The introduction of hemin endows LM@P/S@CP@Hemin hydrogel with a strong immunomodulatory effect on macrophages to achieve anti‐inflammation and cellular ROS clearance abilities, which avoids the excessive oxidative stress in the nasal cavity. The results showed that the hydrogel induced an anti‐bacterial effect with a 98.5% inhibition rate against methicillin‐resistant Staphylococcus aureus, hadexcellent clearance ability of excessive ROS, and promoted anti‐inflammatory M2 macrophage generation to relieve inflammation. Meanwhile, transcriptome sequencing and mRNA level measurements revealed that the hydrogel could regulate inflammatory‐related genes. In vivo, bacterial infection‐induced acute sinusitis rabbit model experiments and histological analysis further confirmed the great therapeutic effect of LM@P/S@CP@Hemin for acute sinusitis based on photothermal and photodynamic therapy. Therefore, LM@P/S@CP@Hemin is an excellent therapeutic material that can adapt to the nasal environment and treat acute sinusitis.

Jerry Junyang Cheung, Shiyao Shen, Yuchen Zhuang et al.

Despite recent advances in large language models (LLMs) for materials science, there is a lack of benchmarks for evaluating their domain-specific knowledge and complex reasoning abilities. To bridge this gap, we introduce MSQA, a comprehensive evaluation benchmark of 1,757 graduate-level materials science questions in two formats: detailed explanatory responses and binary True/False assessments. MSQA distinctively challenges LLMs by requiring both precise factual knowledge and multi-step reasoning across seven materials science sub-fields, such as structure-property relationships, synthesis processes, and computational modeling. Through experiments with 10 state-of-the-art LLMs, we identify significant gaps in current LLM performance. While API-based proprietary LLMs achieve up to 84.5% accuracy, open-source (OSS) LLMs peak around 60.5%, and domain-specific LLMs often underperform significantly due to overfitting and distributional shifts. MSQA represents the first benchmark to jointly evaluate the factual and reasoning capabilities of LLMs crucial for LLMs in advanced materials science.

Subham Ghosh, Abhishek Tewari

The rapid discovery of materials is constrained by the lack of large, machine-readable datasets that couple performance metrics with structural context. Existing databases are either small, manually curated, or biased toward first principles results, leaving experimental literature underexploited. We present an agentic, large language model (LLM)-driven workflow that autonomously extracts thermoelectric and structural-properties from about 10,000 full-text scientific articles. The pipeline integrates dynamic token allocation, zeroshot multi-agent extraction, and conditional table parsing to balance accuracy against computational cost. Benchmarking on 50 curated papers shows that GPT-4.1 achieves the highest accuracy (F1 = 0.91 for thermoelectric properties and 0.82 for structural fields), while GPT-4.1 Mini delivers nearly comparable performance (F1 = 0.89 and 0.81) at a fraction of the cost, enabling practical large scale deployment. Applying this workflow, we curated 27,822 temperature resolved property records with normalized units, spanning figure of merit (ZT), Seebeck coefficient, conductivity, resistivity, power factor, and thermal conductivity, together with structural attributes such as crystal class, space group, and doping strategy. Dataset analysis reproduces known thermoelectric trends, such as the superior performance of alloys over oxides and the advantage of p-type doping, while also surfacing broader structure-property correlations. To facilitate community access, we release an interactive web explorer with semantic filters, numeric queries, and CSV export. This study delivers the largest LLM-curated thermoelectric dataset to date, provides a reproducible and cost-profiled extraction pipeline, and establishes a foundation for scalable, data-driven materials discovery beyond thermoelectrics.

Onur Boyar, Indra Priyadarsini, Seiji Takeda et al.

Discovering materials with desirable properties in an efficient way remains a significant problem in materials science. Many studies have tackled this problem by using different sets of information available about the materials. Among them, multimodal approaches have been found to be promising because of their ability to combine different sources of information. However, fusion algorithms to date remain simple, lacking a mechanism to provide a rich representation of multiple modalities. This paper presents LLM-Fusion, a novel multimodal fusion model that leverages large language models (LLMs) to integrate diverse representations, such as SMILES, SELFIES, text descriptions, and molecular fingerprints, for accurate property prediction. Our approach introduces a flexible LLM-based architecture that supports multimodal input processing and enables material property prediction with higher accuracy than traditional methods. We validate our model on two datasets across five prediction tasks and demonstrate its effectiveness compared to unimodal and naive concatenation baselines.

Xiawei Yang, Mingxuan Yao, Yu Su et al.

In this paper, T-joints of Ti-6Al-3Nb-2Zr-1Mo titanium alloy were joined with friction stir welding, and microstructure evolution and forming mechanism were studied. The effect of using tungsten inert gas welding to heat additionally the FSW was investigated. Results show a strong effection microstructure of stir zone (SZ) due to the temperature gradient and fast cooling rate. The top and middle sections of SZ have a basketweave microstructure, while there is duplex microstructure at the bottom. When welding at 750 rpm-50 mm/min, the maximum tensile strength of the joint is similar to that of the base metal (BM). As the heat input increases, grain coarsening occurs, which reduces the joint tensile strength and the ability to plastically deform. The fracture mode changes from mixed fracture to ductile one. When TIG-assisted heat source is 20 mm in front of the tool and the power input is in 600 W, the temperature field produced is relatively uniform, which has a positive effect on the weld.

Byoung-Nam Park

In thin LiNi_1/3Mn_1/3Co_1/3O₂ (NMC111) electrodes, pseudocapacitive behavior is notably enhanced due to their increased surface-to-volume ratio, which intensifies the role of the electrode–electrolyte interface. This behavior is driven by fast, reversible redox reactions and ion intercalation occurring near the surface, where the shorter diffusion path allows for more efficient ionic transport. The reduced thickness of the electrodes shortens the Li-ion diffusion distance, improving the diffusion coefficient by a factor of 40 compared to thicker electrodes, where ion transport is hindered by longer diffusion paths. The increased surface area and shorter diffusion paths promote faster electrochemical kinetics, allowing for quicker ion intercalation and deintercalation processes. The thin-film configuration enhances pseudocapacitive charge storage, which is essential for applications requiring rapid charge and discharge cycles. As a result, the combination of improved Li-ion diffusion and enhanced surface activity contributes to superior electrochemical performance, offering higher power densities, faster energy delivery, and better rate capability. This improvement in performance makes thin NMC111 electrodes particularly advantageous for applications such as high-power energy storage systems, where fast kinetics and high power densities are critical. These findings highlight the importance of interface engineering and material morphology in optimizing the performance of Li-ion batteries and similar electrochemical energy storage devices.

Vahid Attari, Raymundo Arroyave

Materials data, especially those related to high-temperature properties, pose significant challenges for machine learning models due to extreme skewness, wide feature ranges, modality, and complex relationships. While traditional models like tree-based ensembles (e.g., XGBoost, LightGBM) are commonly used for tabular data, they often struggle to fully capture the subtle interactions inherent in materials science data. In this study, we leverage deep learning techniques based on encoder-decoder architectures and attention-based models to handle these complexities. Our results demonstrate that XGBoost achieves the best loss value and the fastest trial duration, but deep encoder-decoder learning like Disjunctive Normal Form architecture (DNF-nets) offer competitive performance in capturing non-linear relationships, especially for highly skewed data distributions. However, convergence rates and trial durations for deep model such as CNN is slower, indicating areas for further optimization. The models introduced in this study offer robust and hybrid solutions for enhancing predictive accuracy in complex materials datasets.

Yoel Zimmermann, Adib Bazgir, Zartashia Afzal et al.

Here, we present the outcomes from the second Large Language Model (LLM) Hackathon for Applications in Materials Science and Chemistry, which engaged participants across global hybrid locations, resulting in 34 team submissions. The submissions spanned seven key application areas and demonstrated the diverse utility of LLMs for applications in (1) molecular and material property prediction; (2) molecular and material design; (3) automation and novel interfaces; (4) scientific communication and education; (5) research data management and automation; (6) hypothesis generation and evaluation; and (7) knowledge extraction and reasoning from scientific literature. Each team submission is presented in a summary table with links to the code and as brief papers in the appendix. Beyond team results, we discuss the hackathon event and its hybrid format, which included physical hubs in Toronto, Montreal, San Francisco, Berlin, Lausanne, and Tokyo, alongside a global online hub to enable local and virtual collaboration. Overall, the event highlighted significant improvements in LLM capabilities since the previous year's hackathon, suggesting continued expansion of LLMs for applications in materials science and chemistry research. These outcomes demonstrate the dual utility of LLMs as both multipurpose models for diverse machine learning tasks and platforms for rapid prototyping custom applications in scientific research.

Radeen Mostafa, Mirza Nihal Baig, Mashaekh Tausif Ehsan et al.

In the field of Material Science, effective information retrieval systems are essential for facilitating research. Traditional Retrieval-Augmented Generation (RAG) approaches in Large Language Models (LLMs) often encounter challenges such as outdated information, hallucinations, limited interpretability due to context constraints, and inaccurate retrieval. To address these issues, Graph RAG integrates graph databases to enhance the retrieval process. Our proposed method processes Material Science documents by extracting key entities (referred to as MatIDs) from sentences, which are then utilized to query external Wikipedia knowledge bases (KBs) for additional relevant information. We implement an agent-based parsing technique to achieve a more detailed representation of the documents. Our improved version of Graph RAG called G-RAG further leverages a graph database to capture relationships between these entities, improving both retrieval accuracy and contextual understanding. This enhanced approach demonstrates significant improvements in performance for domains that require precise information retrieval, such as Material Science.

Bramantyo Bayu Aji, Yu-Hsiuan Huang, Masatsugu Oishi et al.

Selective emission of green light phosphor powder Y₄SiAlO₈N as the host material and Tb³⁺ as the activator was successfully achieved using spray pyrolysis (SP). Samples synthesized with various calcination temperatures and precursor concentrations indicated that the most suitable parameter for the synthesized powder is the calcination of 0.05 M Y_3.92SiAlO₈N:0.08Tb³⁺ at a temperature of 1600 °C. The effect of the selected parameters was substantiated by the high purity of the Y_3.92SiAlO₈N:0.08Tb³⁺ phase, as confirmed by X-ray diffraction (XRD) analysis. The Scherrer equation was used to calculate grain size. In addition, scanning electron microscopy (SEM) and energy-dispersive X-ray spectrometry (EDS) confirmed the presence of micron-sized particles, which matched well with the theoretical chemical composition. The specific surface area of the phosphor powder was determined using the Brunauer–Emmett–Teller method. Finally, fluorescence spectrometry was used to determine the luminescence properties. The correlation between the crystallinity of the phosphor powder and narrowing emission is also discussed.

Yan Ren, Jin-Rong Gao, Shou-Meng Cai et al.

In vitro regeneration was studied to protect the rare Chinese medicinal orchid Liparis nervosa (Thunb.) Lindl. The mixtures of protocorm and seeding and the stem tip were used as explants. The results revealed that the best essential medium for L. nervosa growth was 1/3 MS medium with 25 g · L–1 sucrose, 50 g · L–1 banana puree, 40 g · L–1 mashed potato, and 1.0 g · L–1 AC (MS1); MS1 medium with 0.5 mg · L–1 BA, 0.05 mg · L–1 2,4-D, and 1.5 mg · L–1 NAA was optimal for proliferation. When stem tips were cultured in a proliferation medium, four types of proliferation occurred: basal stem cluster bud (occurring at the basal node), tiller bud (occurring at the root), protocorm-like body (occurring at the plant’s base incision), and high-position bud (occurring on plant stem nodes other than the basal nodes). Four methods produced 10.12 proliferation coefficients. In the MS1 medium with 0.5 mg · L−1 NAA, the plantlets rooted 100%, and the rooted plantlets survived 100% after domestication and transplantation.

Muhammad Arshad, Lucia Sorba, Petra Rudolf et al.

The integration of carbon nanostructures with semiconductor nanowires holds significant potential for energy-efficient integrated circuits. However, achieving precise control over the positioning and stability of these interconnections poses a major challenge. This study presents a method for the controlled growth of carbon nanofibers (CNFs) on vertically aligned indium arsenide (InAs) nanowires. The CNF/InAs hybrid structures, synthesized using chemical vapor deposition (CVD), were successfully produced without compromising the morphology of the pristine nanowires. Under optimized conditions, preferential growth of the carbon nanofibers in the direction perpendicular to the InAs nanowires was observed. Moreover, when the CVD process employed iron as a catalyst, an increased growth rate was achieved. With and without the presence of iron, carbon nanofibers nucleate preferentially on the top of the InAs nanowires, indicating a tip growth mechanism presumably catalysed by a gold-indium alloy that selectively forms in that region. These results represent a compelling example of controlled interconnections between adjacent InAs nanowires formed by carbon fibers.