Home Featured Articles List

Self-supervised 3D AI data training alleviates model learning bottleneck

Artificial intelligence (AI) is vastly superior to human analysis in many ways—particularly in processing, analyzing and synthesizing massive quantities of information at incredibly fast speeds—but AI model performance is only as good as the data used to train it. Much like humans, AI models need to be taught to perform specific tasks, and oftentimes, the more quality data used to train a model, the more proficient the model will be at recognizing patterns or processing data, for example. Currently, one of the biggest bottlenecks for AI model implementation is the training phase, where models are taught to perform specific tasks using data. Prior to training, large amounts of data often require humans to annotate or structure data into specific formats that algorithms require. As models get larger, the amount of labor and time required to transform or appropriately label data can grow exponentially, delaying model training and deployment. To address this issue, a group of researchers from Fudan University and Nanyang Technological University wrote a comprehensive literature review outlining how AI models can supervise their own learning using point cloud data, or raw, three-dimensional (3D) spatial coordinates that correlate to the surface of an object, greatly reducing or eliminating the amount of human labor required to annotate large datasets for model training. The team published their review on April 22 in the journal Computational Visual Media, published by Tsinghua University Press. “Just as self-supervised learning has already revolutionized how AI understands natural language and 2D images, it is now becoming the key to unlocking 3D data. 3D point clouds are essential for technologies like self-driving cars and robotics, but training AI to understand them traditionally required humans to manually label millions of 3D coordinates, creating a great bottleneck. This technology is attracting immense attention because it allows the AI to learn generic, powerful 3D representations directly from raw, unlabeled data, paving the way for the emergence of true 3D foundation models,” said Ben Fei, research fellow at the Chinese University of Hong Kong and first author of the review paper. One popular self-supervised AI learning approach uses deep neural networks (DNNs) and assigns various pretext tasks for the network to solve. The pretext task serves as a temporary, artificially created problem to train an AI model on unlabeled data. This task forces the model to understand the fundamental structure and patterns of the data, which can later be applied to downstream real-world applications. Several different pretext tasks have been proposed for self-supervised learning, including point cloud reconstruction, or the conversion of unstructured, 3D data into a usable digital 3D model; contrastive learning, which teaches models to understand data through comparison; and multi-modal learning, which integrates many different types of data into a single unified model.  Pretext tasks share two common properties: 1) The visual features of point clouds must be captured by DNNs to solve the pretext task, and 2) the supervisory signal is generated from the data itself, which results in self-supervision, by exploiting its structure. “The main takeaway is that effective 3D representation learning is closely tied to choosing the right pre-training tasks that leverage the data itself rather than human labels. For instance, these self-supervised approaches allow a single model to learn rich, generic geometric features from the data's own structure. This comprehensive understanding serves as a powerful prior that can be easily transferred to various applications, from indoor robotics to outdoor autonomous driving. Our paper provides a comprehensive, unified roadmap that categorizes these learning schemes, showing how to systematically build and adapt these frameworks toward capable 3D foundation models,” said Fei. Despite these advances, hurdles still exist for self-supervised 3D model training. “Our next step is to overcome the challenges unique to 3D representation learning, including optimizing the massive computing power required and establishing higher-quality, standardized 3D pre-training datasets. Our ultimate goal is to break down the barriers between different modalities. By successfully scaling up 3D foundation models and aligning them with large language models and 2D vision, we hope to provide the crucial spatial intelligence needed to achieve true Artificial General Intelligence,” said Fei. Jingyi Xu, Yixuan Li, Weidong Yang, Qingyuan Zhou, Liwen Liu and Tianyue Luo from the School of Computer Science at Fudan University in Shanghai, China; and Ying He from the College of Computing and Data Science at Nanyang Technological University in Singapore, Singapore also contributed to this research. This research was supported by the JC STEM Lab of AI for Science and Engineering, funded by The Hong Kong Jockey Club Charities Trust, the MTR Research Funding (MRF) Scheme (CHU-24003), the Research Grants Council of Hong Kong (CUHK14213224) and the Ministry of Education, Singapore, Academic Research Fund Grant (RT19/22).  
Computer Science

The multifaceted science of where rubber meets road

Increasing pavement skid resistance on a London highway by 50% reduced the total number of accidents over four years by 45%, according to the United Kingdom Transport and Road Research Laboratory. The United States National Transportation Safety Board and Federal Highway Administration stated that poor pavement skid resistance is a major cause of driving accidents. These are just two sources cited by an international team of researchers who recently reviewed more than 3,000 studies on tire-road friction and found that while tire-road friction theory and simulations have been well developed across the research scales — rubber-pavement scale, the tire-road scale and the vehicle scale — the scales themselves remain largely unconnected. According to the researchers, who published their work on Feb. 10 in Friction, that disconnect is a major hindrance in advancing the field, and ultimately, road safety for all.  “The concept of pavement skid resistance pertains to the capacity of road surfaces to provide adequate friction during diverse vehicular operations, including braking, accelerating and cornering,” said co-corresponding author Yuchuan Du, professor, Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, China. “It is critical for preventing skidding incidents and ensuring driving safety, serving as a significant indicator for road performance evaluation and maintenance decisions.” Tire-road friction has been studied for more than half a century, Du said, explaining that early studies focused on improving the materials and structure of the tire. However, researchers later learned that simply improving tires did not reduce the proportion of skidding accidents, so more researchers in the field focused on the skid resistance of the road itself. “From the perspective of road researchers, the primary concern is the role that the pavement itself plays within the tire-rubber friction system, which is referred to as the pavement skid resistance,” Du said. “This is an inherent characteristic of the pavement within the tire-road friction system, and understanding the tire-road friction system is fundamental for evaluating the skid resistance of asphalt pavements.” From their analysis of the field’s scientific literature, the researchers found that studies typically fell into one of four research paradigms: experimental science, theoretical science, computational science and data science. They also found that the research can be categorized into the rubber-pavement scale, the tire-road scale and the vehicle scale. Across the paradigms and scales, the team found that experimental studies were limited by specific variables that have made reproducibility difficult, and that computational studies were limited by their ability to integrate friction mechanisms, making them difficult to interpret properly. “The synergistic development of the four research paradigms can promote and advance the understanding and application of tire-road friction mechanisms,” Du said. The researchers developed a multiscale architecture for tire-road friction research to clarify the different concepts and boundaries as they feed into two coefficients: rubber-pavement friction and tire-pavement adhesion. Both feed into determining the skidding risk of the vehicle, along with other variables. The two coefficients are determined by pavement factors, such as texture and aggregate; rubber-related material properties, like density; environmental factors, including temperature and humidity; tire-related physical conditions, such as pattern and pressure; and vehicle-related operation conditions, like load and velocity. “With the accumulation of data, data-based research has gradually become the primary research paradigm for evaluating pavement skid resistance,” Du said. “However, the integration of data with theoretical research remains a question that requires further exploration. Leveraging theory to enhance the predictive capabilities of data models and utilizing data mining to reveal the underlying mechanisms and patterns of pavement skid resistance are areas that warrant deeper investigation.” Collaborators include Zihang Weng and co-corresponding author Zhen Leng, Department of Civil and Environmental Engineering, Hong Kong Polytechnical University, China; Chenglong Liu and Difei Wu, Key Laboratory of Road and Traffic Engineering, Ministry of Education, Tongji University, China; and Bryan T. Adey, Institute of Construction and Infrastructure Management, Swiss Federal Institute of Technology, Switzerland. Weng is also affiliated with Tongji University and Hong Kong Polytechnic University Shenzhen Research Institute. The National Natural Science Foundation of China, the Shanghai Science and Technology Innovation Action Plan and the Carbon Neutrality Funding Scheme of PolyU supported this research.
Engineering

How we age: 3D organization of DNA offers clues to the underlying mechanisms of aging Reveals connections between aging and disease

A research team has examined how aging and diseases are connected by exploring the aging-related remodeling of chromatin architecture. This study of chromatin architecture, the three-dimensional organization of DNA, gives scientists a better understanding of aging and its underlying mechanisms. Their review is published in the journal Aging Research on May 8, 2026. With senescence, or the biological process of aging, an irreversible and systemic process occurs, characterized by declining cellular functions, loss of tissue homeostasis, and deterioration of overall physiological capacity. Scientists know that aging is a primary risk factor for a variety of chronic non-communicable diseases, such as Alzheimer’s disease, Parkinson’s disease, cardiovascular disease, metabolic syndrome, and cancer. In their review, the team synthesizes the recent studies that explore how multiscale chromatin reconfiguration influences gene regulation and cellular identity in senescence. They summarize the representative disease settings and implicated structural layers. The team further discusses the major technical challenges. These include the challenge of accurately capturing the biological and genetic differences between individual cells, the limitations of fixed-cell assays for capturing chromatin dynamics, and the difficulties in robust multi-omics integration.  The team proposes future directions for leveraging single-cell and spatiotemporal three-dimensional genomics to dissect the mechanisms linking senescence to aging and to inform the development of new therapeutic treatments. Because the global population is rapidly aging, scientists are looking to unravel the mechanisms of aging. “Understanding the mechanisms of aging and its intrinsic connections with disease holds profound social and biomedical implications,” said Professor Zhenyu Ju from Jinan University.  The research team notes that there are two related concepts that are the driving forces behind the aging process: cellular senescence and organismal aging. Organismal aging describes the process by which multicellular organisms decline in function over time. This decline involves multiple organ systems and includes the disruption of tissue homeostasis, along with the cumulative risk of age-associated diseases.  By contrast, cellular senescence describes the stable, irreversible cellular state where damaged or old cells permanently stop dividing but do not die. This cellular state is induced by specific stresses. Current studies show that cellular senescence represents one of the key cellular foundations of organismal aging and many age-related diseases. Cells undergo many multifaceted changes as they become senescent. Scientists have learned that progressive reorganization of higher-order chromatin structure is a pivotal regulatory layer. Chromatin is a mixture of DNA and proteins inside cells. They provide an organized storage system that folds the long strands of DNA so they fit inside the cells. This reorganization refers to the changes that occur over times as the DNA is folded into the nucleus.  The chromatin’s three-dimensional architecture goes through profound changes and dynamic remodeling as cellular senescence occurs and the cells age. Studies are showing that these architectural changes are closely related to key senescence-related changes that occur when the cells permanently stop dividing but remain active. These changes include genomic instability, transcriptional dysregulation, stem cell functional decline, and chronic inflammatory signaling. Scientists have recently begun to map patterns of three-dimensional genome remodeling and connect them to diseases and medical conditions that occur with aging. These range from Alzheimer’s disease to blood diseases to cancer.  “In recent years, advances in three-dimensional genome technologies have provided valuable tools for understanding the dynamic chromatin changes underlying aging and disease,” said Ju. Looking ahead, the team notes that the integration of single-cell technologies with dynamic three-dimensional genome mapping may enable the construction of systematic spatiotemporal models of chromatin evolution during aging. “This approach could not only help delineate causal relationships between chromatin remodeling and aging or disease phenotypes but also advance three-dimensional genomics from theoretical studies to clinical applications, offering novel strategies for aging intervention, cancer therapy, and the prevention of complex diseases,” said Ju. The research team includes Weicong Chen, Qiang Zhan, Feng Xiao, Zhenyu Ju, and Zhiyang Chen from the Key Laboratory of Regenerative Medicine of Ministry of Education, Institute of Aging and Regenerative Medicine, Department of Developmental & Regenerative Medicine, College of Life Science and Technology, Jinan University, Guangzhou, China. The research is funded by the National Key R&D Program of China, the National Natural Science Foundation of China, and the innovation team project of universities in Guangdong province. DOI Link:https://doi.org/10.26599/AGR.2025.9340072
Life Sciences and Medicine

Seismic performance of high-rise segmented rocking truss steel frame with multiple tuned mass dampers

High-rise buildings in seismically active regions face a critical engineering challenge: traditional seismic design approaches often struggle to control complex vibration patterns that can lead to catastrophic damage during major earthquakes. But the traditional rocking truss mainly controls the damage distribution of the structure according to the first-order vibration mode, which has limited applicability to high-rise structures. A research team led by Professor Zhi-Qian Dong at Dalian University of Technology has proposed a solution: a steel frame-segmented rocking truss-MTMD damping system. The system divides rocking structures into multi-segment trusses along the vertical direction, allowing deformation along multiple mode shapes while arranging MTMDs based on system vibration characteristics without increasing frame lateral stiffness. The research focused on a 35-story steel frame structure with a total height of 140 meters, analyzing single-segment, double-segment, and triple-segment rocking truss configurations under severe earthquake conditions. The team investigated critical parameters including segmentation number and position, lateral stiffness ratios between the frame and rocking truss, MTMD mass ratios and placement locations, and the effects of velocity-dependent dampers and self-centering energy-dissipative braces. The findings reveal that the segmented rocking truss system offers several distinct advantages over conventional designs. The segmentation effectively reduces the maximum bending moment in the middle of the rocking truss. By introducing hinged joints between segments, the system allows each rocking truss segment to deform along multiple mode shapes, significantly reducing internal forces while maintaining structural integrity. The integration of multiple tuned mass dampers arranged according to the system's vibration mode represents another innovation which can reduce the structure's maximum inter-story drift ratio by 7% to 29%. The researchers found that placing MTMDs at the top of the structure—where deformation of the first three vibration modes is largest—proves more effective than locating them at segment joints. With a mass ratio of just 1%, the MTMDs achieved an 8.5% reduction in maximum inter-story drift under certain ground motions, while a 4% mass ratio delivered average vibration reduction rates of 19.5% at the top. Additional energy dissipation is provided by velocity-dependent dampers, which the study found particularly effective. "Compared with self-centering energy dissipation braces, velocity-dependent dampers provide no additional stiffness and can dissipate seismic energy effectively," the researchers reported. Installation of these dampers reduced maximum inter-story drift ratios by 53% to 59% and decreased roof displacements by 24% to 37% under various earthquake scenarios. The research also explored variable stiffness designs for segmented rocking trusses, demonstrating that upper segments can use smaller cross-sections than lower segments without compromising seismic performance. This approach significantly reduces overall structural weight and steel consumption while maintaining effective lateral deformation control. The system's self-centering energy-dissipative braces, installed at each rocking truss segment base, further reduce residual structural deformation after earthquakes, facilitating faster post-earthquake recovery. This system is particularly well-suited for high-rise structures where higher-order modes play a significant role. Through incremental dynamic analysis (IDA) and vulnerability analysis, the results show the seismic performance of controlled structures is significantly better than that of uncontrolled structures. The segmented structural design effectively minimizes the impact of higher-order modes on internal forces and stress concentrations in the sway wall structure. Each sway wall segment can be designed and constructed independently based on its specific location and structural characteristics, making maintenance more convenient and flexible compared to a monolithic sway wall. The research team expects this innovative system to provide a practical and efficient solution for high-rise building seismic design, particularly in regions with high seismic risk. Future work will focus on optimizing the integration of these various damping mechanisms and further refining the vulnerability analysis of relevant influencing factors. The ultimate goal is to develop standardized design methodologies that can be readily implemented in engineering practice, making tall buildings more resilient against earthquakes while maintaining economic feasibility. Other contributors to this research include Hao Wu, Yi-Zhang Chai, Qing-Tao Meng from Dalian University of Technology, Hui-Dong Liu from China Construction Third Engineering Bureau Group, and Yi Zhao from Guangxi University and Rice University. This research was supported by the Open Project Program of Guangdong Provincial Key Laboratory of Intelligent Disaster Prevention and Emergency Technologies for Urban Lifeline Engineering, and the Opening Funds of State Key Laboratory of Building Safety and Built Environment & National Engineering Research Center of Building Technology. See the article: Seismic performance of high-rise segmented rocking truss steel frame with multiple tuned mass dampers
Physical Sciences and Engineering

Vortex-induced triboelectric nanogenerator for multidirectional wind energy harvesting enables efficient wind energy harvesting under low wind speed and high humidity

“We propose a vortex-induced vibration-based triboelectric nanogenerator (VIV-TENG) to enable efficient wind energy harvesting under low wind speed and complex environmental conditions,” says Prof. Chuyan Zhang from China University of Geosciences (Beijing). They published their study on May 13, 2026, in iEnergy. A new strategy for distributed wind energy harvesting With the growing demand for clean energy, harvesting distributed wind energy in urban environments has attracted increasing attention. However, conventional wind turbines are often limited by their large size, high cost, and strict installation requirements, making them unsuitable for small-scale and decentralized applications. Triboelectric nanogenerators (TENGs), pioneered by Wang Zhonglin, provide a promising alternative due to their advantages in low-frequency energy harvesting. Nevertheless, existing wind-driven TENG designs still face challenges such as low efficiency at low wind speeds, poor adaptability to multidirectional airflow, mechanical wear, and performance degradation in humid environments.  Vortex-induced vibration enables stable energy conversion To overcome these limitations, the research team developed a VIV-TENG that utilizes vortex-induced vibration instead of conventional rotational or fluttering mechanisms. When airflow passes through the device, periodic vortex shedding induces oscillations of the central axis, which drives multiple TENG units to generate electricity. “The vortex-induced vibration mechanism allows the device to operate without rotational components, reducing mechanical wear while maintaining stable output under multidirectional airflow,” the research team explains. The device features a symmetric structure with multiple TENG units arranged around a central axis, enabling efficient wind energy collection from various directions. In addition, an encapsulated design improves its environmental robustness.  Efficient output under low wind speed and high humidity The experimental results demonstrate that the VIV-TENG achieves stable output performance over a wide range of wind speeds. At a wind speed of 3.5 m/s, the device reaches a peak open-circuit voltage of 82.9 V and a short-circuit current of 13 μA. Notably, the device can operate at a low start-up wind speed of 0.9 m/s. It can also charge a 47 μF capacitor to 2 V within one minute. In addition, the device maintains stable performance in humid environments. The maximum average output power reaches 49.5 μW at 45% relative humidity and remains at 45.5 μW even at 85% humidity, demonstrating environmental adaptability. Toward self-powered systems in urban environments The device shows potential for practical applications. It can power multiple LEDs, drive small electronic devices such as clocks, and support self-powered sensing systems. The results suggest that the proposed device could serve as a feasible approach for distributed wind energy harvesting and self-powered systems. This study provides a possible pathway for utilizing low-speed wind energy in complex environments and indicates the potential of TENG-based technologies in urban energy systems. See the article: Vortex-driven triboelectric nanogenerator for multidirectional wind energy collection with humidity resistance
Information Sciences

Variants within the LOXHD1 may provide insight into the genetics of hearing loss

Hearing loss is a common cause of disability worldwide, with anywhere between 30-60% of cases being caused by genetic factors. LOXHD1 is a gene integral to essential protein interactions responsible for maintaining normal hair cell function. Certain variants in this gene can cause progressive or non-progressive congenital hearing impairments, called “pathogenic variants” in individuals. Researchers went deeper into this gene to learn more about the genetic causes of hearing loss in the Chinese population, and they uncovered novel variants within this gene to further explore these causes, in addition to what early intervention and potential medicinal therapies might look like for those who are predisposed to hearing loss. Researchers published their results in the Journal of Otology (https://doi.org/10.26599/JOTO.2026.9540055) in April 2026. 157 individuals with hearing loss and their families were part of this study. 60.51% of affected individuals (95 patients) had potential causal variants, most frequently seen in the SLC26A4 and GJB2 genes. Further analysis revealed 10 different variants within the LOXHD1 gene, with five novel ones being linked to hearing loss. This information unveils even more of the known mutation spectrum of the LOXHD1 gene and points to its significance in auditory function, making understanding the mechanisms of mutation an important part of getting to the root cause of genetic hearing loss. These variants have been found to significantly contribute to non-syndromic hearing loss, a type of hearing impairment that is hereditary, with hearing loss being the sole symptom. Variants in the gene can result from different types of mutations affecting the protein, such as misfolding or frameshift mutations, which can lead to substitutions in the amino acid sequence. These substitutions can lead to early termination of protein translation, an essential step in the sequence of events that create functional proteins. The stability and performance of these proteins can also be drastically affected when incorrect amino acids are present.  With this information, genetic screening becomes paramount to early intervention and establishing therapeutic options. Two patients received cochlear implants, one at the age of 10 months and the other at the age of 10 years, and were re-evaluated after 12 months. Results from the speech-language evaluation demonstrated expressive language development and auditory reception consistent with the patient’s age, showing the benefit of genomic sequencing and early intervention for non-syndromic hearing loss. Genomic sequencing especially can be particularly insightful in cases where cochlear implants are considered for guiding operative decisions and post-operative effects. “The next step is to functionally characterize these novel LOXHD1 variants to understand how they disrupt hearing at the molecular level,” said Kun Zhang, researcher at Qilu Hospital of Shangdong University and author of the study. With favorable outcomes in the two young patients and momentum gained in understanding LOXHD1 variants, researchers hope to use the knowledge found in this study to continue their work on establishing genetic variants in the LOXHD1 gene. Developing precise genetic tests can lead to personalized interventions for patients with hearing loss, and expanding upon this research can improve the quality of life for many people across the globe. Kun Zhang, Xijian Xin, Shiqi Huang, Bo Hou, Xinbo Xu, Xiao Han and Hanbing Zhang of the Department of Otorhinolaryngology at the Qilu Hospital of Shangdong University and the NHC Key Laboratory of Otorhinolaryngology at Shangdong University, and Peng Qu of the Dermatology Hospital of Shangdong First Medical University and Shangdong Provincial Institute of Dermatology and Venerology at the Shangdong Academy of Medical Sciences contributed to this research. The Natural Science Foundation of Shangdong Province supported this research. See the article: Identification of Novel LOXHD1 Variants in Chinese Patients with Non-Syndromic Hearing Loss
Life Sciences and Medicine

Atomic Tuning of Titanium-Chromium Nitride Catalysts Unlocks High-Performance Lithium-Sulfur Batteries

As fossil fuels continue to deplete and environmental pollution intensifies, the development and application of clean energy have garnered widespread attention. Lithium-sulfur (Li-S) batteries, with their remarkable theoretical specific capacity of 1675 mAh g-1 and energy density of 2600 Wh kg-1, which is about six times that of conventional lithium-ion batteries at 387 Wh kg-1. Additionally, they offer advantages including environmental sustainability, high safety, and low cost, making them a highly promising solution for future electrochemical energy storage. A research team led by Professor Jie Sun at Shaanxi Normal University has announced a major breakthrough in lithium-sulfur (Li-S) battery technology. Through atomic-level precision engineering, the team has developed a novel titanium-chromium nitride (TixCr1-xN) solid-solution catalyst that efficiently traps and rapidly converts polysulfides, which is the key culprit behind the short lifespan and poor efficiency of Li-S batteries. This innovation effectively addresses a fundamental barrier to the practical application of this high-energy-density technology. The team published their article in Nano Research on April 22, 2026. “The core innovation of this battery material research lies in our achievement of 'precision tuning' of the material’s electronic structure through continuous adjustment of the composition in TixCr1-xN solid-solution at atomic scale. This is not merely a simple material mixture, but a true form of solid-solution phase with atomic-level interface engineering,” stated Professor Jie Sun, the corresponding author of the paper from the School of Materials Science and Engineering at Shaanxi Normal University. Transition metal compounds leverage the “Lewis’s acid-base” interaction between metal ions and polysulfide anions to achieve strong chemical adsorption. Simultaneously, the d-orbitals of transition metals can couple with the frontier orbitals of polysulfide anions, facilitating electron transfer and enhancing polysulfide conversion efficiency. Among them, transition metal nitrides exhibit exceptional physicochemical stability and high electrical conductivity due to their stable metal lattice structure and nitrogen interstitial alloying effects, demonstrating potential as ideal sulfur host materials. These combined properties establish metal nitrides as promising candidate materials for lithium-sulfur battery cathodes. The research team synthesized flexible CNFs@TCN membranes using electrospinning and high-temperature nitridation methods. By varying the ratio of titanium and chromium precursors, they were able to continuously modulate the electronic structure of the resulting solid-solution material. This atomic-level electronic fine-tuning directly determines the material’s key properties. Both theoretical calculations and experimental results confirm that when the Ti/Cr atomic ratio reaches 1:2, the d-band center of the catalyst is optimally positioned. This configuration significantly enhances the adsorption energy for polysulfides compared to pure TiN or CrN, while providing an optimal pathway for charge transfer. As a result, the dual functions of “trapping” and “conversion” are synergistically strengthened, fundamentally addressing both the shuttle effect and slow reaction kinetics in Li-S batteries. The Li-S battery assembled with this material demonstrates a significant electrochemical breakthrough. The CNFs@TCN-1/2 electrode delivers a high specific capacity of 801 mAh g-1 and maintains 93% of its capacity after 600 cycles at 2 C, with an ultralow decay rate of only 0.012% per cycle. These results strongly validate the exceptional stability and effectiveness of the catalyst in suppressing the shuttle effect and enhancing conversion efficiency. Looking ahead, Professor Sun concluded, “This work demonstrates that atomic-level doping via solid-solution construction is a powerful strategy for regulating the catalytic performance of transition metal nitrides. It opens new avenues for designing highly efficient catalysts for complex multi-step reactions, extending beyond Li-S batteries to other energy conversion and storage fields.” Other contributors include Jiyuan Zhang, Weiye Zhang, Jiarui Xue, Nan Zhu, Yunping Ge, Zhibin Lei, Qi Li, Xuexia He, and Zonghuai Liu from the Key Laboratory of Applied Surface and Colloid Chemistry at Shaanxi Normal University. This work was funded by the Natural Science Basic Research Plan of Shaanxi Province (2025JC-YBMS-351, 2019JLP-12), the funds of Shaanxi Sanqin Scholars Innovation Team, and the Central University Foundation of Shaanxi Normal University (GK202302005). About Nano Research Nano Research is a peer-reviewed, open access, international and interdisciplinary research journal, sponsored by Tsinghua University and the Chinese Chemical Society, published by Tsinghua University Press on the platform SciOpen. It publishes original high-quality research and significant review articles on all aspects of nanoscience and nanotechnology, ranging from basic aspects of the science of nanoscale materials to practical applications of such materials. After 18 years of development, it has become one of the most influential academic journals in the nano field. Nano Research has published more than 1,000 papers every year from 2022, with its cumulative count surpassing 7,000 articles. In 2024 InCites Journal Citation Reports, its 2024 IF is 9.0 (8.7, 5 years), and it continues to be the Q1 area among the four subject classifications. Nano Research Award, established by Nano Research together with TUP and Springer Nature in 2013, and Nano Research Young Innovators (NR45) Awards, established by Nano Research in 2018, have become international academic awards with global influence.
Nanoscience and Nanotechnology

Silicon Nitride Ceramics Strengthened by Novel Intergrown Distorted Columnar-Cluster Microstructures

Silicon nitride (Si3N4), a widely adopted engineering ceramic, demonstrates significant potential in emerging applications such as Insulated Gate Bipolar Transistor (IGBT) packaging, semiconductor substrates, and bioceramics due to its high theoretical thermal conductivity and excellent biocompatibility. However, liquid-phase sintering, the dominant densification mechanism for Si3N4, restricts further performance enhancements. The inherent phase transformation of this high-temperature process typically yields a β-phase-dominated microstructure. The precipitated β-phase grains exhibit anisotropic growth with high growth rates along certain crystallographic directions, leading to the formation of an interlocking microstructure that enhances strength and toughness. Nevertheless, this high growth rate also tends to cause grain coarsening, making it difficult to precisely control microstructure evolution and thus limiting further improvements in properties. High-pressure lowers the required densification temperature below phase transition point, thereby enabling the fabrication of dense, equiaxed α-phase Si3N4 ceramics characterized by high hardness but intrinsically low toughness. Due to their distinct lattice stacking sequences, the α and β phases essentially represent a high-hardness phase and a high-toughness phase, respectively. The adjustments of processing parameters and powder composition to content allow the tuning of toughness and hardness by manipulating phase content, but they cannot overcome the performance bottleneck caused by the phase composition and grain morphology inherent to conventional sintering.  Although novel silicon nitride architectures with a certain degree of compromise between toughness and hardness can be synthesized through complex processing or extreme high-pressure techniques, these high-cost approaches have yet to achieve a definitive breakthrough in the synergistic optimization of both properties。 Notably, the previous study on high-pressure-assisted liquid-phase sintering revealed that high pressure promoted phase transformation through a mechanism involving stress-induced interfacial migration. This mechanism altering the microstructure by transcending the thermodynamic constraints of phase transformation avoided the performance limitations caused by microstructural and phase composition in conventional sintering. Recently, a team of materials scientists led by Zhengyi Fu from Wuhan University of Technology conducted a deeper investigation into the microstructural evolution and grain growth kinetics during metastable phase transitions under high-pressure stress. This work can provide insights that break the limitations of traditional liquid-phase sintering, offering a theoretical basis and guidance for the development of a new generation of high-performance Si₃N₄ ceramics. The team published their work in Journal of Advanced Ceramics on April 21, 2026. In this study, samples were prepared using 2%, 4%, and 6% sintering aids and sintered at 1550 °C and 1600 °C. which depends on previous experience with additive content and sintering temperature, and to prevent grain coarsening and performance degradation induced by excessive liquid phase formation. The effects of high pressure applied above the phase transformation temperature on phase transformation and grain growth during the final stage of densification are shown to define the microstructure evolution through interfacial stress, which varies with the liquid phase content. A low additive content (2 wt%) yielded high-hardness α-Si₃N₄, while increasing the liquid phase to 4–6 wt% facilitated complete phase transformation, thus resulting in high strength and toughness. Notably, the sample with 6 wt% additive sintered at 1600 °C achieved a simultaneous enhancement of strength (982 ± 63 MPa), toughness (10.2 ± 0.3 MPa·m¹/²), and hardness (20.1 ± 0.3 GPa). This improvement in performance is attributed to a unique twisted intergrowth mechanism, which involves stress-induced compression and shear during the ordered coalescence of precipitated particles during phase transformation.  About author: Xiao-Wei Qin is a Ph.D. candidate at Wuhan University of Technology, affiliated with the State Key Laboratory of Advanced Technology for Materials Synthesis and Processing. Under the supervision of Research Professor Wei Ji, his academic specialization focuses on high-performance structural ceramic materials. His current research primarily investigates the fundamental principles of novel sintering technologies for advanced ceramics, particularly nitrides and oxides.  Funding: This work was financially supported by the National Natural Science Foundation of China (52322207, 92163208, and 52494933), Fundamental and Interdisciplinary Disciplines Breakthrough Plan of the Ministry of Education of China (JYB2025XDXM408), the Natural Science Foundation of Hubei Province (2025AFA043, 2025CSA004), and the Independent Innovation Projects of the Hubei Longzhong Laboratory (2022ZZ-11). About Journal of Advanced Ceramics Journal of Advanced Ceramics (JAC) is an international academic journal that presents the state-of-the-art results of theoretical and experimental studies on the processing, structure, and properties of advanced ceramics and ceramic-based composites. JAC is Fully Open Access, monthly published by Tsinghua University Press, and exclusively available via SciOpen. JAC’s 2024 IF is 16.6, ranking in Top 1 (1/34, Q1) among all journals in “Materials Science, Ceramics” category, and its 2024 CiteScore is 25.9 (5/130) in Scopus database. ResearchGate homepage: https://www.researchgate.net/journal/Journal-of-Advanced-Ceramics-2227-8508
Ceramics

TrafficPerceiver enables instruction-driven understanding and segmentation in challenging traffic scenes

Understanding traffic scenes is a fundamental capability for intelligent transportation systems and autonomous driving. However, real-world traffic environments are often far from ideal, featuring adverse weather, low visibility, motion blur, and occlusions that significantly degrade perception performance. Existing vision-based methods are typically designed for standard scenarios and lack the ability to follow human instructions or perform fine-grained, target-level reasoning. To address these challenges, researchers from the School of Vehicle and Mobility at Tsinghua University propose TrafficPerceiver, a unified multimodal framework built upon a multimodal large language model (MLLM). TrafficPerceiver is designed to jointly support coarse-grained traffic scene understanding tasks, such as scene description and question answering, and fine-grained target-oriented segmentation tasks, such as isolating a specific vehicle, pedestrian, or road element according to a natural language instruction. The team published their study in Communications in Transportation Research (https://doi.org/10.26599/COMMTR.2026.9640008). Unlike conventional perception pipelines that rely on task-specific decoders, TrafficPerceiver aligns language and visual representations through a shared multimodal Transformer. A special segmentation token is introduced to directly associate textual instructions with relevant image regions, enabling efficient and interpretable segmentation without additional task-specific heads. To further enhance robustness under visually degraded conditions, the researchers introduce a reinforcement learning strategy based on Group Relative Policy Optimization (GRPO). Instead of optimizing absolute prediction scores, GRPO evaluates model outputs relative to other sampled responses within the same group. This encourages consistent instruction-following behavior and improves reasoning stability in challenging scenarios such as rain, fog, blur, and nighttime scenes. In addition, the team constructs a new dataset named Challenging Traffic Scene Understanding (CTSU), which focuses specifically on difficult real-world traffic environments. The dataset includes diverse conditions such as adverse weather, low illumination, occlusion, and regional variations in traffic infrastructure, and provides paired language instructions, textual responses, and pixel-level segmentation annotations.   Extensive experiments on the CTSU dataset and existing benchmarks demonstrate that TrafficPerceiver achieves superior performance in both traffic scene understanding and segmentation tasks, particularly under complex visual conditions. The results suggest that combining instruction-driven multimodal perception with reinforcement learning offers a promising direction for building more robust and interactive traffic perception systems. About Communications in Transportation Research Communications in Transportation Research was launched in 2021, with academic support provided by Tsinghua University and China Intelligent Transportation Systems Association. The Editors-in-Chief are Professor Xiaobo Qu, a member of the Academia Europaea from Tsinghua University and Professor Xiaopeng (Shaw) Li from University of Wisconsin–Madison. The journal mainly publishes high-quality, original research and review articles that are of significant importance to emerging transportation systems, aiming to serve as an international platform for showcasing and exchanging innovative achievements in transportation and related fields, fostering academic exchange and development between China and the global community. It has been indexed in SCIE, SSCI, Ei Compendex, Scopus, CSTPCD, CSCD, OAJ, DOAJ, TRID and other databases. It was selected as Q1 Top Journal in the Engineering and Technology category of the Chinese Academy of Sciences (CAS) Journal Ranking List. In 2022, it was selected as a High-Starting-Point new journal project of the “China Science and Technology Journal Excellence Action Plan”. In 2024, it was selected as the Support the Development Project of “High-Level International Scientific and Technological Journals”. The same year, it was also chosen as an English Journal Tier Project of the “China Science and Technology Journal Excellence Action Plan PhaseⅡ”. In 2024, it received the first impact factor (2023 IF) of 12.5, ranking Top1 (1/58, Q1) among all journals in "TRANSPORTATION" category. In 2025, its 2024 IF was announced as 14.5, maintaining the Top1 position (1/62, Q1) in the same category. From Volume 6 (2026), Communications in Transportation Research will be published by Tsinghua University Press on the SciOpen platform with the official journal website at https://www.sciopen.com/journal/2097-5023. We kindly request that all new manuscript submissions be made through the journal’s submission system at https://mc03.manuscriptcentral.com/commtr. For any submission-related inquiries, please contact the Editorial Office at commtr_e@mail.tsinghua.edu.cn.  
Physical Sciences and Engineering

Assessing the impact of Saitama Emissions Trading Scheme on energy consumption and economic performance at the facility level

Concerns over adverse effects on business development and employment have long been cited as a key reason for Japan's delay in implementing a national carbon market. As Japan plans to launch a nationwide mandatory ETS in 2026, a central question demands an answer: does carbon trading necessarily come at the cost of economic performance? A research team from Waseda University recently investigated the Saitama ETS, Japan's second regional emissions trading scheme launched in 2011. By examining energy use and economic performance at the facility level, the study provides empirical evidence on whether a voluntary ETS can achieve emissions reductions without compromising economic activity. The team published their findings in Energy and Climate Management on February 09, 2026. This study is examining the impact of the Saitama ETS on economic activity. By simultaneously considering both energy consumption and employment, the research assesses whether the scheme effectively reduces energy use and whether it has any adverse effects on production activities. Unlike the Tokyo ETS, the Saitama ETS focuses on the industrial sector and adopts a voluntary approach, encouraging regulated facilities to reduce CO₂ emissions without strict penalties. Using facility-level panel data from 2007 to 2018 and a difference-in-differences (DID) method, the team found that heavy oil consumption among regulated facilities declined by 5.69% in the first phase and 34.82% in the second phase, city gas decreased by 25.58%, while electricity consumption increased slightly by 3%–4%. As the authors note in the paper, "The observed decline in heavy oil use alongside a tendency for increased electricity consumption following the introduction of the ETS may partly indicate a shift in the energy composition of facilities. Rather than cutting overall energy use and potentially compromising production activity, regulated facilities in Saitama may have sought to achieve emissions reductions by shifting their energy mix toward lower-emission sources". More importantly, contrary to the widespread concern that carbon trading might harm economic performance, the study found no evidence of negative impacts on employment. In fact, a statistically significant positive effect emerged in the second compliance period, the number of employees in regulated facilities increased by 5.67%. This suggests that firms may have achieved emissions reductions not through blunt measures such as layoffs or production cuts, but through flexible strategies of transitioning toward lower-emission energy sources. These findings carry important implications for Japan's upcoming nationwide mandatory ETS. The voluntary nature of the Saitama ETS has granted firms flexibility that facilitated early compliance, but this design may face challenges under more stringent reduction targets. The authors concluded: "The Saitama ETS has proven effective, partly due to its relatively modest standards and the flexibility it offers to firms. However, when faced with more stringent reduction requirements, a voluntary mechanism carries the risk of becoming ineffective. Therefore, the 'voluntary nature' of the scheme must be carefully evaluated in the design of future ETSs in Japan and beyond." This research was supported by the Environment Research and Technology Development Fund (JPMEERF20202008) of Japan, the JSPS KAKENHI grant (JP21H04945), and the Waseda University Grant for Special Research Project (2025Q-002). See the Article: Assessing the impact of Saitama Emissions Trading Scheme on energy consumption and economic performance at the facility level
Humanities and Social Sciences
Total 180