Search

Core-shell nanostructured poly(3,4-ethylenedioxythiophene) (PEDOT)/carbon nanotube (CNT) composites with high thermoelectric power factor have been successfully synthesized via a facile in-situ chemical solution polymerization method. In this approach, the addition of sodium dodecyl sulfate (SDS) not only improves the dispersion of CNTs but also acts as soft templates to promote the ordered structure of PEDOT layer growing along CNTs. Raman spectra analysis confirms the existence of strong interfacial interaction between PEDOT and CNTs in such a core-shell nanostructure that facilitates charge transfer in the composites. As a result, an optimal power factor of 157 μW m−1 K−2 is achieved in PEDOT/CNT composite film, which is much higher than that of most reported PEDOT based composites fabricated by chemical solution polymerization method. Furthermore, the composites allow the fabrication of thermoelectric yarns via a simple dip-coating method. Textile-based thermoelectric generator based on segmented thermoelectric yarn that comprises 100 pairs of P-N junctions (p-type PEDOT/CNT composite yarns and n-type polyethyleneimine (PEI) doped PEDOT/CNT composite yarns) is assembled by embroidery of thermoelectric yarns into spacer fabric. The generator with high voltage output, good mechanical flexibility and air stability displays good potential as wearable power source.

Additional file 1: Table S1. Descriptive statistics by season. Table S2. Descriptive statistics by quartiles of rural population. Table S3. Descriptive statistics for complete-case analysis. Table S4. Percent increase in daily CVD admissions and 95% confidence intervals (CI) per 10 µg/m3 for all PM2.5 products. Table S5. Quasi-Akaike’s Information Criterion (qAIC) for select seasonal and long-term trends. Figure S1. Percent increase in daily CVD admissions rates per interquartile range for all PM2.5 products. Figure S2. Percent increase in daily CVD admissions rates for subset excluding AQS. Figure S3. Percent increase in lag 0–1 CVD admissions rates per 10 µg/m3 for all PM2.5 products. Figure S4. Sensitivity analysis adding bank holiday as a potential confounder. Figure S5. Time series of average PM2.5 exposure estimates by PM2.5 product.

Ambient exposure to fine particulate matter (PM2.5) and ground-level ozone (O3) is identified as a leading risk factor for global disease burden. A major limitation to advancing our understanding of the cause and impacts of air pollution is the lack of observations with the spatial and temporal resolution needed to observe variability in emission, chemistry and population exposure. Satellite remote sensing, which fills a spatial gap in ground-based networks, is playing an increasingly important role in atmospheric chemistry. This thesis exploits satellite remote sensing observations to: (1) estimate human exposure to PM2.5 from remotely sensed aerosol optical properties; (2) identify the chemical regimes of surface O3 formation using satellite observations of O3 precursors. In the first part, we use a forward geophysical approach to derive PM2.5 distributions from satellite AOD at 1 km2 resolution over the northeastern US by applying relationships between PM2.5 and AOD simulated from a regional air quality model (CMAQ). We use multi-platform ground, airborne and radiosonde measurements to quantify multiple sources of uncertainties in the satellite-derived PM2.5. We find that uncertainties in satellite-derived PM2.5 are largely attributed to the varying relationship between PM2.5 and AOD that depends on the aerosol vertical distribution, speciation, aerosol optical properties and ambient relative humidity. To assess the value of remote sensing to improve PM2.5 exposure estimate, we compile multiple PM2.5 products that include information from remote sensing, ground-based observations and models. Evaluating these products using independent observations, we find the inclusion of satellite remote sensing improves the representativeness of surface PM2.5 mostly in the remote areas with sparse monitors. Due to the success of emission control, PM2.5-related mortality burden over NYS decreased by 67% from 8410 (95% confidence interval (CI): 4, 570 – 12, 400) deaths in 2002 to 2750 (95% CI: 700 – 5790) deaths in 2012. We estimate a 28% uncertainty in the state-level PM2.5 mortality burden due to the choice of PM2.5 products, but such uncertainty is much smaller than the uncertainty (130%) associated with the exposure-response function. The second part of the thesis focuses on ground-level O3. O3 production over urban areas is non-linearly dependent on the availability of its precursors: nitrogen oxides (NOx) and volatile organic compounds (VOCs). A major challenge in lowering ground-level O3 in urban areas is to determine the limiting species for O3 production (NOx-limited or VOC-limited). We use satellite observations of NO2 and HCHO to infer the relative abundance of NOx versus VOCs, thus to identify the O3 chemical regime. We first use a global chemical transport model (GEOS-Chem) to evaluate the uncertainties of using satellite-based HCHO/NO2 to infer O3 sensitivity to precursor emissions. Next, we directly connect this space-based indicator, retrieved consistently from three satellite instruments, to spatiotemporal variations in O3 recorded by on-the-ground monitors from 1996 to 2016. The nationwide emission reduction has led the O3 formation over U.S. urban areas to shift from VOC-limited to NOx-limited regime. Urban O3 monitors reveal trends consistent with this regime transition. Nonetheless, it is a major challenge for these retrievals to accurately depict day-to-day variability within urban cores. TROPOspheric Monitoring Instrument (TROPOMI) which launched in 2017, offers an unprecedented view to infer O3 chemistry at fine spatial and temporal scales. As an example, we use TROPOMI HCHO/NO2 to identify short-term changes in O3 sensitivity during the California Camp Fire. We find that the emissions from wildfires lead to NOx-saturated ozone formation near the fire source but NOx -limited conditions downwind. This thesis bridges basic research in atmospheric chemistry, which advances the state-of-science related to O3 and PM2.5 pollution from urban to global scales, and applied research in air quality management and public health, by quantifying the health benefits of emission control, and informs policymakers on which emission reductions to focus so as to maximize the cost-effectiveness of pollution controls. We show how space-based measurements can complement in situ networks and model simulations by providing information on the spatial heterogeneity and temporal evolution of PM2.5 exposure and O3 chemical regimes, which will lay the scientific foundation for interpreting future products retrieved from upcoming geostationary platforms.

Neural networks are vulnerable to small adversarial perturbations. Existing literature largely focused on understanding and mitigating the vulnerability of learned models. In this paper, we demonstrate an intriguing phenomenon about the most popular robust training method in the literature, adversarial training: Adversarial robustness, unlike clean accuracy, is sensitive to the input data distribution. Even a semantics-preserving transformations on the input data distribution can cause a significantly different robustness for the adversarial trained model that is both trained and evaluated on the new distribution. Our discovery of such sensitivity on data distribution is based on a study which disentangles the behaviors of clean accuracy and robust accuracy of the Bayes classifier. Empirical investigations further confirm our finding. We construct semantically-identical variants for MNIST and CIFAR10 respectively, and show that standardly trained models achieve comparable clean accuracies on them, but adversarially trained models achieve significantly different robustness accuracies. This counter-intuitive phenomenon indicates that input data distribution alone can affect the adversarial robustness of trained neural networks, not necessarily the tasks themselves. Lastly, we discuss the practical implications on evaluating adversarial robustness, and make initial attempts to understand this complex phenomenon., Comment: ICLR 2019, Seventh International Conference on Learning Representations

advertorch is a toolbox for adversarial robustness research. It contains various implementations for attacks, defenses and robust training methods. advertorch is built on PyTorch (Paszke et al., 2017), and leverages the advantages of the dynamic computational graph to provide concise and efficient reference implementations. The code is licensed under the LGPL license and is open sourced at https://github.com/BorealisAI/advertorch .

Although State Implementation Plans (SIPs) typically rely on observations from ground-based networks and regulatory models, satellite data is increasingly available to state agencies and can also inform and supplement state implementation plans to improve air quality. An advantage of satellite data is that it provides information for a broader area than sampled by ground-based networks. This document provides examples and guidance for using satellite products of formaldehyde (HCHO) and nitrogen dioxide (NO2) to inform ground-level ozone sensitivity to emissions of nitrogen oxides (NOx) versus volatile organic compounds (VOC) in state implementation plans. Analysis of changes in ozone sensitivity over periods where emission controls have been implemented can provide insights into the efficacy of those past strategies and the likely efficacy of proposed future emission control programs.