Hong Kong University of Science and Technology
Hong Kong University of Science and Technology Institutional RepositoryNot a member yet
162821 research outputs found
Sort by
Eco-friendly candle soot based composite PCMs for thermal regulation in 4680 tabless lithium-ion cells
No full textAs demand for high-performance electric vehicles grows, effective and sustainable battery thermal management is critical. We upcycle industrial by-products: candle soot and waste wood carbon into composite phase-change materials (CPCMs) by blending with paraffin and a small fraction of expanded graphite, then integrate the CPCMs with 4680 tabless lithium-ion battery (LIB). The waste-derived fillers increase thermal conductivity and volumetric heat capacity relative to neat paraffin and improve anti-leakage stability at elevated temperature. At the cell level, the optimal composition (5 wt% candle soot, 3 wt% expanded graphite, balance paraffin) lowers the peak surface temperature by ≈ 20 °C during a 1 C discharge compared with a no-PCM control, mitigates hot spots, and maintains operation near ∼ 40 °C. The materials are inexpensive, scalable, and compatible with simple processing, enabling lightweight thermal buffers without active cooling penalties. By validating these CPCMs directly on large-format cells rather than only at material scale, this work establishes a circular-economy pathway to safer, more uniform thermal regulation in next-generation EV batteries. Performance was consistent across 0.5–1 C, underscoring robustness.</p
Persuasive Calibration
No full textWe introduce and study the persuasive calibration problem, where a principal aims to provide trustworthy predictions about underlying events to a downstream agent to make desired decisions. We adopt the standard calibration framework that regulates predictions to be unbiased conditional on their own value, and thus, they can reliably be interpreted at face value by the agent. Allowing a small calibration error budget, we aim to answer the following question: what the optimal predictor is and how to compute it under this calibration error budget, especially when there exists incentive misalignment between the principal and the agent? We focus on the standard ℓ𝑡-norm Expected Calibration Error (ECE) metric.We develop a general framework by viewing predictors as post-processed versions of perfectly calibrated predictors. Using this framework, we first characterize the structure of the optimal predictor. Specifically, when the principal’s utility is outcome-independent and for ℓ1-norm ECE, we show: (1) the optimal predictor is over-(resp. under-) confident for high (resp. low) true expected outcomes, while remaining perfectly calibrated in the middle; (2) the miscalibrated predictions exhibit a collinearity structure with the principal’s utility function. On the algorithmic side, we provide an FPTAS for computing approximately optimal predictor for general principal utility and general ℓ𝑡-norm ECE. Moreover, for the ℓ1- and ℓ∞-norm ECE, we provide polynomial-time algorithms that compute the exact optimal predictor
Multifunctional polymeric composites with self-healing and self-lubricating functionalities
No full textRepeated sliding or rolling contact under long-term tribological service can easily induce surface damages in tribological materials in the form of surface wear or surface/in-depth microcracks. Therefore epoxy matrix composites (EMCs) with hexamethylene diisocyanate (HDI)- or wax-filled microcapsules were developed for their self-healing or self-lubricating functionality to autonomously repair their surface damages or lubricate their surfaces, respectively, and their tribological properties were systematically investigated. The increased contents of HDI- or wax-filled microcapsules decreased the friction and wear of the EMCs through their self-healing or self-lubricating process with released core liquids. In addition, the EMCs exhibited a more significant improvement in their fracture toughness with a higher content of HDI-filled microcapsules. The EMCs coincorporated with HDI- and wax-filled microcapsules had both self-healing and self-lubricating functionalities, with better tribological performance for the higher fraction of wax-filled microcapsules. It could be concluded that the incorporation of microencapsulated chemicals was an effective way to achieve the multifunctional properties of the EMCs.</p
Intergenerational pathway to nature connectedness
No full textNature connectedness has gained recognition for its profound benefits to individuals' well-being and the planet's health. Despite existing evidence on factors associated with it, an integrated understanding of how nature connectedness develops within family contexts remains underexplored. To address this gap, we propose the Intergenerational Pathway to Nature Connectedness, a model that comprehensively elucidates the intergenerational processes through which parents' nature experiences during their own childhood influence their children's nature connectedness. Utilizing survey data with a sample of over 2357 parent-child dyads, we observed significant relationships among four key variables: parental childhood experience with nature, parental nature connectedness at present, child engagement in nature experience, and child nature connectedness. Results suggest that parents who have more experience with nature during their own childhood are more likely to possess strong connectedness with nature presently, which, in turn, motivate them to arrange more nature experience for their children, cultivating similar connectedness in them. The proposed intergenerational pathway contributes to the literature by providing an integrated framework for understanding the familial processes underlying connections to nature and presenting practical implications for intervention strategies.</p
Learning interpretable environment-dependent stochastic discrepancy equations for bias correction and epistemic uncertainty quantification of tropical cyclone models
No full textThis paper presents a generalizable stochastic discrepancy discovery framework to integrate observation into calibrating environment-dependent tropical cyclone (TC) models. Existing environment-dependent TC models often suffer from biases due to oversimplified physics, while their deterministic nature prohibits proper quantification of epistemic uncertainties arising from modeling inadequacies. To simultaneously address the task of bias correction and uncertainty quantification, this paper treats the existing model as prior knowledge and discovers parameter-efficient interpretable stochastic governing equations for their modeling discrepancies against historical TC observations, leveraging symbolic regression and stochastic processes. As a proof-of-concept, we demonstrate our approach in improving the TC track and intensity simulations in the Western Northern Pacific basin, through individual historical TC hindcasts and statistical validation. Influence from the track and intensity model uncertainties is measured. We also focus on the practical task of typhoon wind hazard assessment. Our estimated wind speed generally agrees with the code recommendations, and the confidence intervals are well calibrated to include results from alternative models. Overall, the proposed framework provides a principled approach to enhance the environment-dependent TC models, paving the way for more informed TC simulation under changing climates.</p
Protonation-engineered MOF-801 thin films for efficient Congo red removal
No full textCongo red (CR) contamination poses severe threats to aquatic ecosystems, demanding advanced treatment solutions. This work presents a high-performance filtration platform based on a protonated metal-organic framework (MOF) thin film for efficient CR removal. Zirconium-based MOF-801 is synthesized and characterized, demonstrating a high intrinsic adsorption capacity for CR, which is further enhanced (>2000mg/g at pH 2.2) by a facile HCl protonation treatment. To simplify the solid-liquid separation of powders, the protonated MOF-801 is engineered into a thin film on a porous nylon support via vacuum filtration. This composite film acts as an adsorbent layer. The protonated film achieved >95% CR removal at a low MOF:CR mass ratio of 1.5, which is only 20% of the adsorbent required using powder MOF-801. Such a low adsorbent required is attributed to a coupled adsorption-filtration mechanism, where convective flow sustains the delivery of CR to protonated surface sites, facilitating the formation of a cohesive, multilayered dye overlayer. This work establishes a scalable strategy for dye wastewater remediation by combining interfacial chemistry and membrane technology.</p
Acceleration Practices for Gifted Learners in Hong Kong SAR China
No full textThis chapter discusses the evolution and current status of acceleration practices for gifted learners in Hong Kong. The authors also address some of the challenges faced in implementing these approaches and suggest future directions to overcome some of the barriers. Case studies are presented for the innovative Dual Programme offered at the Hong Kong University of Science and Technology (HKUST) and the Mentorship Scheme provided by the Hong Kong Academy for Gifted Education (HKAGE). The concluding sections recommend future directions for the advancement of acceleration practices in Asian settings
AI regulatory strategies for digital sovereignty: The role of geopolitics and technological disparities
No full textThis article examines the European Union’s (EU) strategy to assert digital sovereignty through stringent artificial intelligence (AI) regulation, situated within evolving geopolitical dynamics and widening technological disparities with the United States (US). A game-theoretic model is developed to analyze the strategic dilemma confronting EU regulators, who must balance the protection of domestic market share for local firms against maintaining access to leading AI technologies. The analysis shows that stringent regulation constitutes a rational strategy only when it disproportionately constrains foreign firms and when foreign technological advantages in the domestic market are significant but not overwhelming. Applying this framework to the EU AI Act’s provisions on general-purpose AI, the study illustrates that the effectiveness of the EU’s approach is contingent on its redistributive effect on domestic market power, the level of US technological competitiveness, and the state of transatlantic cooperation.</p
Divergent mechanisms in the addition of (NHC)Au(I)–H and (NHC)Cu(I)–H across alkynes
No full textCopper(I) hydride complexes are widely recognized as reactive intermediates in numerous transformations and typically undergo syn-addition across alkynes. Although copper and gold both belong to Group 11, (NHC)Au(I)–H displays a contrasting anti-addition behavior. In this work, we systematically investigate the reactivity differences between (NHC)Cu(I)–H and (NHC)Au(I)–H using density functional theory. Our results reveal that the observed stereoselectivity originates from the relative stability of the three-coordinated metal-alkyne intermediates. Specifically, the distortion required for (NHC)Au(I)–H to engage in syn-addition is energetically very unfavorable compared to its copper(I) counterpart, destabilizing the corresponding gold-alkyne intermediate and thus favoring anti-addition.</p
A bispecific antibody designed to act as a NRP2/PLXNA1 agonist mimics anticancer activity of SEMA3F
No full textNeuropilin-2 (NRP2) is a pleiotropic receptor with diverse roles across biological systems. Recent work detailed its role as an immunomodulatory receptor target that is currently being explored in clinical development for interstitial lung diseases, establishing it as a viable therapeutic target. To mediate its diverse effects, NRP2 interacts with endogenous ligands, including semaphorins (SEMAs) and vascular endothelial growth factors, signaling via ligand-induced heterodimerization with various receptor families. One of these ligands, SEMA3F exhibits well-documented tumor-suppressive activities mediated through NRP2 and plexinA1 (PLXNA1). Despite its observed benefits, SEMA3F is not therapeutically viable due to the multifaceted nature of its functions through non-NRP2–mediated interactions, leading to concerns around potential toxicity. Here, we describe development of bispecific antibodies (bsAbs) that dimerize PLXNA1 and NRP2, selectively mimicking the beneficial aspects of SEMA3F signaling as a basis for a novel anticancer therapy. Using a single B cell–based mAb discovery platform, anti-PLXNA1 mAbs with diverse lineages were generated and combined with anti-NRP2 mAbs to produce over 200 PLXNA1-NRP2 bsAbs. Antibodies were screened in cell-based assays (receptor dimerization, phospho-AKT, oncogene expression, and cell proliferation), yielding one bsAb capable of mimicking NRP2-mediated SEMA3F activities in all assays. Structural studies revealed that this bsAb binds to PLXNA1/NRP2 at sites distinct from the SEMA3F-binding site, but in a manner that allows proper spacing for receptor complex formation and flexibility of conformational changes for signaling. This study demonstrates the potential of these receptors as targets for agonistic bsAbs development and provides the groundwork for further exploration in tumor models.</p