114333 research outputs found
Sort by
Automated TEM Reveals Intercrystalline Correlations of Conjugated Polymers
No full textTransmission electron microscopy (TEM) continues to transform polymer science by revealing key aspects of chain packing, phase separation and nanoscale structure. The development of instrumentation and data analyses tools is driving the field forward and enabling new experiments. Here, we use automated high-resolution TEM (HRTEM) and image processing to identify the structure of a conjugated polymer used in organic electronics. Analysis of more than 600 HRTEM images reveals lattice parameters and orientation correlations between crystals, including the preferred alignment of neighboring crystals along the same crystallographic direction that is likely the result of liquid crystalline order.This article is published as Fair, Ryan A., Dhruv Gamdha, Joshua T. Del Mundo, Abigail M. Fenton, Agatha O’Connell, Karen C. Bustillo, Esther W. Gomez, Andrew M. Minor, Baskar Ganapathysubramanian, and Enrique D. Gomez. "Automated tem reveals intercrystalline correlations of conjugated polymers." Macromolecules (2025).
doi: https://doi.org/10.1021/acs.macromol.5c02888.Funding from the National Science Foundation under Award DMR-1905550 and DMR-2515754 and the Office of Naval Research under Award N00014-19-1-2453 is gratefully acknowledged. D.G. and B.G. acknowledge support from DMR-2323716. The X-ray work conducted at the Stanford Synchrotron Radiation Lightsource, SLAC National Accelerator Laboratory, was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under contract no. DE-AC02-76SF00515. Work at the Molecular Foundry was supported by the Office of Science, Office of Basic Energy Sciences, of the U.S. Department of Energy under Contract No. DE-AC02-05CH11231. MatLab and Python scripts used for 4D-STEM analysis were provided by Colin Ophus of the Molecular Foundry
Recycling waste claystone into value-added and eco-friendly cementitious materials for concrete industry decarbonisation
No full textThe increasing demand for sustainable cement production has prompted the exploration of alternative supplementary cementitious materials (SCMs). This study explores the influence of incorporating waste-derived Australian claystone (WAC) as SCM on the hydration kinetics, internal microstructure evolution, and macro-scale properties of blended cement systems. WAC was used in both raw and calcined forms, with calcination conducted at 400 °C, 600 °C, and 800 °C to enhance pozzolanic reactivity. The results demonstrate that calcined WAC, particularly at 800 °C, significantly promotes the generation of C–S–H and C–A–S–H gels while reducing calcium hydroxide (CH) content, thereby accelerating hydration and refining pore structure. The Ca/Si ratio of hydration gels decreases with curing age and is further reduced by WAC addition, especially in the samples containing thermally activated WAC. Among all cementitious mixtures incorporating WAC, the formulation containing 10 % WAC thermally activated at 800 °C achieves superior mechanical and transport performance. Specifically, this mix demonstrates the highest compressive strength and the lowest rates of water uptake and capillary absorption, indicating optimized matrix densification and hydration efficiency. The results highlight the potential of the thermally activated WAC as a high-performance, eco-friendly SCM for concrete decarbonisation.This article is published as Qu, Fulin, Juanjuan Ren, Like Tang, Wengao Liu, Kejin Wang, and Wengui Li. "Recycling waste claystone into value-added and eco-friendly cementitious materials for concrete industry decarbonisation." Journal of Cleaner Production 543 (2026): 147583. doi: https://doi.org/10.1016/j.jclepro.2026.147583.The authors would like to acknowledge the support from the National Natural Science Foundation of China (No. 52425213) and Australian Research Council (ARC), Australia (DP260100885, FT220100177, LP240200692, LP230100288, DP220101051, DP220100036)
Bridging Solid-State Chemistry and Electrocatalysis: Electronic Structure, Surface Dynamics, and Descriptor Discovery in Transition Metal Phosphides for Hydrogen Evolution Reaction
No full textThis dissertation investigates the structure–property–activity relationships of transition-metal
phosphides (TMPs) for the hydrogen evolution reaction (HER), emphasizing how controlled
compositional and electronic modulation governs catalytic behavior. A systematic exploration of
binary and ternary phosphides—spanning the NiSi₁₋ₓPₓ substitutional series, binary M₂P (M = Ni,
Co, Fe) single crystals, polymorphic Fe₂P structures, and ordered ternary M′₂M₁₂P₇ and BaM2P2
compounds—was undertaken to reveal the fundamental descriptors linking crystal structure and
electronic configuration to intrinsic HER activity.
The initial studies on the NiSi₁₋ₓPₓ system established how Si/P substitution tunes the electronic
structure and hydrogen adsorption energy, providing a benchmark for electron-count-driven
catalytic design. Investigation of flux-grown single crystals of M₂P phases revealed pronounced
facet-dependent HER activity, correlating with orientation-specific surface terminations and
metal–phosphorus covalency. Comparative analysis of orthorhombic and hexagonal Fe₂P
polymorphs further demonstrated the impact of structural anisotropy and d-band alignment on
catalytic behavior.
The discovery and expansion of the Zr₂M₁₂P₇ and M′₂M₁₂P₇ (M = Ni, Co, Fe) families introduced
ordered ternary phosphides as model systems for electronic structure engineering. Through
extensive synthesis, XANES spectroscopy, and electrochemical measurements of more than
twenty compositions, a reproducible volcano-type relationship between the d-band center and
intrinsic HER activity was identified, establishing the d-band energy as a universal electronic
descriptor across metallic phosphides. Complementary in situ XAS studies of Ni–P catalysts
elucidated dynamic electronic and structural changes under reaction conditions, confirming the
stability of low-valent phosphide motifs as active species. Finally, studies on BaNi₂P₂ extended
the design concept to layered pnictides, revealing structure-dependent electronic flexibility
relevant to both catalysis and superconductivity.
Collectively, this dissertation provides a coherent framework for understanding and predicting
catalytic activity in TMPs through quantitative electronic descriptors. The integrated
experimental dataset forms the foundation for future machine-learning analyses to uncover
hidden multivariate descriptors and guide the rational design of next-generation HER catalyst
Safety-driven learning for perception and decision making in autonomous systems
No full textAutonomous vehicles and robots are increasingly used in safety-critical settings, but their perception
modules are typically trained and evaluated in isolation from the planning and control systems that act on
those outputs. This dissertation argues that perception should be designed, evaluated, and trained with the
entire closed loop in mind—specifically, with respect to system-level safety.
This work makes three key practical contributions. (1) Uncertainty-aware model selection. An
evaluation procedure is introduced to measure how reliable a model’s probability estimates are for each
object class, instead of average accuracies. This matters because some classes carry much higher safety
stakes than others (e.g., pedestrians vs. traffic cones). The procedure helps stakeholders choose the model
that is most trustworthy on high-priority classes identified by a hazard analysis. (2) Decision-oriented
perception outputs. Lightweight “task heads” at the perception–control boundary are added to answer the
binary questions planners actually use (e.g., ‘Is there enough evidence to brake?’), instead of exposing only
scores that must later be thresholded. This reduces brittle, back-and-forth action flips near confidence
cutoffs and improves consistency in ambiguous scenes. (3) Safety-informed fine-tuning. We develop a
feedback loop that runs the system in simulation with a conventional (non-differentiable) controller and a
set of explicit safety rules (a “rulebook”). Rule violations observed during rollouts are converted into
learning signals that shift the perception model toward behaviors that lower real safety risk.
Collectively, these components constitute a modular and incrementally adoptable pipeline:
uncertainty-aware selection supports pre-deployment model choice; decision-oriented heads enhance
planning robustness; and rule-guided fine-tuning aligns perception with system-level objectives without
changing the controller. Experiments across classification and detection benchmarks and high-fidelity
driving simulation show (i) higher reliability on safety-critical classes, (ii) fewer spurious action reversals
near thresholds, and (iii) substantial reductions in rule violations under both normal and adverse conditions.
Optimizing perception for how its outputs are used yields behavior that is more robust, interpretable, and
aligned with the demands of real-world autonomy
Apparent annual survival of adult Vermivora chrysoptera (Golden-winged Warbler) does not differ by sex or region
No full textUnderstanding range-wide demographic, spatial, and temporal variation in annual survival is essential for managing species of conservation concern. Multi-population models are useful tools for integrating diverse datasets, reducing biases, and deriving survival estimates across differing spatial scales. We conducted a range-wide, multi-population apparent annual survival analysis for a declining songbird, Vermivora chrysoptera (Golden-winged Warbler), using data from 18 sites across its breeding and nonbreeding grounds. This Nearctic-Neotropical migrant breeds in 2 disjunct regional populations, the Great Lakes and Appalachian Mountains, which are experiencing different rates of decline. We aimed to quantify regional-, site-, and sex-specific apparent annual survival estimates to identify geographic patterns or demographic factors influencing population declines. We used simulations to assess the precision of our estimates. Our models did not reveal a substantial difference in apparent annual survival between the Great Lakes (0.41, 95% credible interval (CrI):0.31–0.50) and the Appalachian regions (0.49, 95% CrI: 0.36–0.60), as CrIs overlapped. Site-specific estimates also showed no clear differences in apparent annual survival among sites representing both regional populations. Male apparent annual survival tended to be greater than female apparent annual survival in both regions, though CrI’s overlapped. Our study suggests demographic factors other than adult annual survival likely play a larger role in recent regional and range-wide population declines, such as productivity, juvenile/immature survival, or recruitment. Simulations indicate that improving recapture probability and study duration of datasets could lead to more precise apparent annual survival estimates. However, our model produced CrI ranges comparable to the most ideal data collection scenario, suggesting the lack of trends we found was not due to variability in our estimates. We stress the importance of addressing inherent biases in survival datasets and the need for standardized collaborative efforts to inform species conservation on a range-wide scale.This article is published as Emily N Filiberti, Amber M Roth, Wayne E Thogmartin, Ethan J Royal, Kyle R Aldinger, Ruth E Bennett, David A Buehler, Lesley Bulluck, Ronald A Canterbury, Richard Chandler, Sarah J Clements, Cameron J Fiss, Keith A Hobson, John Anthony Jones, David King, Gunnar R Kramer, Jeffery L Larkin, Darin J McNeil, Jeffrey D Ritterson, Anna Buckardt Thomas, Rachel Vallender, Steven L Van Wilgenburg, Petra Wood, Apparent annual survival of adult Vermivora chrysoptera (Golden-winged Warbler) does not differ by sex or region, Ornithology, Volume 143, Issue 1, January 2026, Pages 1–13, https://doi.org/10.1093/ornithology/ukaf049We are grateful for the U.S. Fish and Wildlife Service and the University of Maine Agricultural and Forest Experiment Station for funding the conceptualization and analysis for this project. This work is supported by multiple contributors, including the McIntire-Stennis project award numbers MEO-41622 and KY009043 and from the U.S. Department of Agriculture’s National Institute of Food and Agriculture. Data collection was supported in part by funding from the Minnesota Cooperative Fish and Wildlife Research Unit and the National Science Foundation
Pressure-Induced Martensitic Phase Transformation and Microstructure Evolution in nanograined Fe-7%Mn Alloy
No full textThe Fe-Mn-based alloys are receiving immense attention due to their applications in the third generation of advanced high-strength steels, owing to their high strength and ductility. A detailed in situ high-pressure structural phase transformation and microstructural evolution in nanograined Fe-7%Mn alloy has been performed using the axial synchrotron X-ray diffraction technique. The ambient BCC phase of Fe-7%Mn undergoes pressure-driven structural PT to the HCP phase at 11.4 GPa. Both BCC and HCP phases coexist up to 15.9 GPa; thereafter, they transform into a pure HCP phase, which remains stable up to the maximum pressure of 30.3 GPa. The XRD study reveals that the (110)b dense crystallographic plane of the BCC lattice transforms into a densely packed (002)h peak of the HCP lattice following the orientational relationship (110)∥(0001)h via diffusionless Burger’s martensitic crystallographic PT pathway. The evolution of crystallite size and microstrain with pressure shows a distinct change during the structural PT. The microstrain exhibits a sharp anomaly at around 10 GPa, suggesting that the microstructural changes precede the structural PT.This is a preprint from Sahu, Mrinmay, Sorb Yesudhas, Valery I. Levitas, and Dean Smith. "Pressure-Induced Martensitic Phase Transformation and Microstructure Evolution in nanograined Fe-7%Mn Alloy." arXiv preprint arXiv:2601.08202 (2026). doi: https://doi.org/10.48550/arXiv.2601.08202.V.I.L. greatly acknowledges ARO (W911NF2420145), NSF (DMR-2246991 and CMMI-2519764), and Iowa State University (Murray Harpole Chair in Engineering)
Optical coherence photoacoustic microscopy for 3D cancer model imaging with AI-assisted organoid analysis
No full textCancer organoids and cancer spheroids are 3D cell culture models with distinct yet overlapping purposes in cancer research. Various commercially available optical imaging techniques have been employed to study these cell cultures, but these methods suffer from various limitations such as the requirement of fluorescence labeling, complicated sample handling, and limited image volume size. In this work, we demonstrate a multimodal optical coherence photoacoustic microscopy (OC-PAM) system for the study of these models, overcoming these limitations. We first performed a longitudinal study using optical coherence microscopy (OCM) for breast cancer organoids. Using the OCM imaging results, artificial intelligence (AI)-based algorithms were developed to automatically segment individual organoids and classify their viability over time using a radiomics texture feature approach, enabling robust, quantitative tracking and classification at the single-organoid level. To supplement OCM’s contrast, we then performed OC-PAM imaging of spheroid models with both melanin positive and melanin negative cells. In the second study, the OC-PAM images clearly mapped the distribution of melanin positive cells hidden amongst melanin negative cells. These results suggest that OC-PAM coupled with AI techniques can be a powerful tool to study cancer organoids and cancer spheroids.This article is published as Deloria, Abigail J., Agnes Csiszar, Shiyu Deng, Mohammad Ali Sabbaghi, Francesco Branciforti, Lukasz Bugyi, Giulia Rotunno et al. "Optical coherence photoacoustic microscopy for 3D cancer model imaging with AI-assisted organoid analysis." Light: Science & Applications 15, no. 1 (2026): 106. doi: https://doi.org/10.1038/s41377-025-02177-2.This work is funded by H2020-ICT-2018-20 project REAP with the grant agreement ID 101016964, and by H2020-FETOPEN-2018-2020 project SWIMMOT with the grant agreement ID 899612. A.J.D. and S.D. were funded by the Joint PhD Program Medical University of Vienna/NTU “Kooperation Singapur” with grant number SO10300010. M.L. is funded by H2020-MSCA-IF- 2019 project SkinOptima with the grant agreement ID 894325
Transformer model to determine spatio-temporal relationships of variables, and interpretability for soybean seed yield, oil, and protein prediction
No full textAccurate in-season prediction of seed yield and seed composition traits such as oil and protein are useful for gaining accuracy and efficiency in soybean breeding. These predictions can also inform farmers, enabling them to improve their field management practices, and guide their market decisions. We report a Transformer-based deep learning framework built on 30 years of multi-environment performance data from the Northern and Southern Uniform Soybean Tests (UST) across North America. Unlike earlier studies on seed yield, oil and protein prediction that focus on limited years, regions, single modalities, we utilized a comprehensive dataset that includes weather, genotype, and management factors, ensuring a more holistic approach to soybean yield, oil, and protein prediction. Our model integrates multivariate time-series weather data with genotypic relationship information, maturity group, and geographic location, to predict variety performance in diverse environments. Our model captures complex temporal patterns associated with trait variability; showing high predictive accuracy (R2) of 77.6 ± 0.2%, 63.9 ± 4.7%, and 79.3 ± 2.3% for seed yield, oil, and protein, respectively. Additionally, for seed yield, we also evaluated multiple interpretability methods to assess feature importance for predictor variables and critical growing timepoints, and solar radiation and temperature were noted as the key predictors. Overall, these results demonstrate the usefulness of a Transformer-based model in trait predictions, and the utility of large cooperative datasets from breeding programs.This article is published as Ayanlade TT, Van der Laan L, Liu Q, Gangopadhyay T, Shook J, Singh A, Ganapathysubramanian B, Sarkar S and Singh AK (2026) Transformer model to determine spatio-temporal relationships of variables, and interpretability for soybean seed yield, oil, and protein prediction. Front. Artif. Intell. 9:1750108. doi: 10.3389/frai.2026.1750108The author(s) declared that financial support was received for this work and/or its publication. This work was supported by: (a) AI Institute for Resilient Agriculture (USDA-NIFA #2021-67021-35329), (b) COALESCE: COntext Aware LEarning for Sustainable CybEr-Agricultural Systems (NSF CPS Frontier #1954556), (c) Smart Integrated Farm Network for Rural Agricultural Communities (SIRAC) (NSF S & CC #1952045), (d) USDA CRIS Project IOW04714, (e) Plant Sciences Institute, (f) Iowa Soybean Association, (g) North Central Soybean Research Program, (h) R. F. Baker Center for Plant Breeding
Turns and Downturns in Aging Drivers
No full textAs cognitive decline progresses, older adults may self-regulate their driving. Avoidance of left turns across traffic is observable in naturalistic driving data but rarely self-reported. We studied 106 older adults using baseline and one-year follow-up neuropsychological assessments. In-vehicle sensors passively recorded driving behavior over 12 weeks. We identified 295,112 turns from vehicle heading changes. We used mixed-effects logistic regression to model the odds of turning left, with cognitive status category change from baseline to one-year follow-up as the predictor. Greater cognitive impairment, represented by movement to a more severe cognitive status category at one-year follow-up, was associated with reduced odds of turning left (odds ratio = 0.984, 95% confidence interval = 0.969–0.999; P value = .037). Left-turn avoidance may be a behavioral marker of early cognitive decline. Passive driving data could help detect functional changes, enabling intervention to preserve mobility and independence. Further research is needed to establish a clinical threshold of concern for decreasing trends in left turn frequency in older drivers.This is a preprint from Hardt, Marie Elizabeth, Guillermo Basulto-Elias, Heike Hofmann, Shauna Hallmark, Anuj Sharma, Jeffrey D. Dawson, Matthew Rizzo, and Jun Ha Chang. "Turns and Downturns in Aging Drivers." medRxiv (2026): 2026-02. doi: https://doi.org/10.64898/2026.02.04.26345564.This work was supported by the National Institutes of Health (NIH), the National Institute on Aging (NIA 5R01AG17177-18), and the University of Nebraska Medical Center Mind & Brain Health Labs. The funder played no role in the collection, analysis, and interpretation of data; in the writing of this manuscript; and in the decision to submit this article for publication
LexiSafe: Offline Safe Reinforcement Learning with Lexicographic Safety-Reward Hierarchy
No full textOffline safe reinforcement learning (RL) is increasingly important for cyber-physical systems (CPS), where safety violations during training are unacceptable and only pre-collected data are available. Existing offline safe RL methods typically balance reward-safety tradeoffs through constraint relaxation or joint optimization, but they often lack structural mechanisms to prevent safety drift. We propose LexiSafe, a lexicographic offline RL framework designed to preserve safety-aligned behavior. We first develop LexiSafe-SC, a single-cost formulation for standard offline safe RL, and derive safety-violation and performance-suboptimality bounds that together yield sample-complexity guarantees. We then extend the framework to hierarchical safety requirements with LexiSafe-MC, which supports multiple safety costs and admits its own sample-complexity analysis. Empirically, LexiSafe demonstrates reduced safety violations and improved task performance compared to constrained offline baselines. By unifying lexicographic prioritization with structural bias, LexiSafe offers a practical and theoretically grounded approach for safety-critical CPS decision-making.This is a preprint from Yang, Hsin-Jung, Zhanhong Jiang, Prajwal Koirala, Qisai Liu, Cody Fleming, and Soumik Sarkar. "LexiSafe: Offline Safe Reinforcement Learning with Lexicographic Safety-Reward Hierarchy." arXiv preprint arXiv:2602.17312 (2026). doi: https://doi.org/10.48550/arXiv.2602.17312