19 research outputs found
Predicting potentially hazardous chemical reactions using an explainable neural network
No full textPredicting potentially dangerous chemical reactions is a critical task for laboratory safety. However, a traditional experimental investigation of reaction conditions for possible hazardous or explosive byproducts entails substantial time and cost, for which machine learning prediction could accelerate the process and help detailed experimental investigations. Several machine learning models have been developed which allow the prediction of major chemical reaction products with reasonable accuracy. However, these methods may not present sufficiently high accuracy for the prediction of hazardous products which particularly requires a low false negative result for laboratory safety in order not to miss any dangerous reactions. In this work, we propose an explainable artificial intelligence model that can predict the formation of hazardous reaction products in a binary classification fashion. The reactant molecules are transformed into substructure-encoded fingerprints and then fed into a convolutional neural network to make the binary decision of the chemical reaction. The proposed model shows a false negative rate of 0.09, which can be compared with 0.47-0.66 using the existing main product prediction models. To provide explanations for what substructures of the given reactant molecules are important to make a decision for target hazardous product formation, we apply an input attribution method, layer-wise relevance propagation, which computes the contributions of individual inputs per input data. The computed attributions indeed match some of the existing chemical intuitions and mechanisms, and also offer a way to analyze possible data-imbalance issues of the current predictions based on relatively small positive datasets. We expect that the proposed hazardous product prediction model will be complementary to existing main product prediction models and experimental investigations.
In Situ Mapping and Local Negative Uptake Behavior of Adsorbates in Individual Pores of Metal-Organic Frameworks
No full textHerein, we report the adsorbate behavior in individual local pores of MIL-101, which is a metalorganic framework (MOF) with two heterogeneous mesopores and different metal sites, by combining adsorbate isotherms and in situ crystallography profiles. The in situ mapping shows that the substrate-adsorbate interaction affects the initial adsorption and pore condensation steps. The monolayer adsorption gradient changes greatly depending on the framework metal-adsorbate attraction force. Also, broad inflection points are found in adsorption isotherms, and the initial shape depends on the different metals. Besides, the capillary condensation at a pore draws adsorbates from other local pores. This leads to the local negative uptake behavior in individual pore isotherms. At higher pressure, they move to a larger space, whereas in a relatively low-pressure range the attraction force between the MOF framework and guest molecule influences the amount of rearranged guest molecules. Furthermore, the origin of the characteristic adsorption behavior based on the metals constituting the MOFs and the relative strength of substrateadsorbate and adsorbateadsorbate interactions are elucidated through the combined study of electron densities in pores, electron paramagnetic resonance spectroscopy spectra, and density functional theory and Monte Carlo simulations to uncover the previously veiled information on adsorption behavior.
고체-액체 계면에서의 물리화학적 특성 예측을 위한 멀티스케일 시뮬레이션
No full text학위논문(박사) - 한국과학기술원 : 화학과, 2019.8,[ix, 143 p. :]Solid-liquid interface is a place where various physicochemical reaction (such as heterogeneous catalyst reaction, wetting, DNA/protein folding etc.) take place and is a system that must be understood in order to regulate chemical reaction or properties. However, unlike other surfaces, the interface is not exposed and the direct observation is challenging. It triggered the active research of computational chemistry that can simulate the interface by controlling the environment at the atomic / molecular level. However, it is still challenging to quantitatively estimate even the wettability, which is the most fundamental concept of the interfacial phenomena. In this thesis, it is covered not only a single simulation method for catalytic reaction, which is one of the representative interfacial phenomena, but also a multiscale method for qualitative and quantitative prediction of the wettability, furthermore, the possibility of applying it to graphene and metal systems is also shown. In particular, in case of the interface where the solid and the liquid coexist, since the size of the system is large and high accuracy is required, the multiscale simulation method is applied. This will provide theoretical guidance to understand the wetting and the chemical reactions of current materials (that is not easy to measure by stability issues etc.) as well as future materials.한국과학기술원 :화학과
Multiscale Simulation Method for Quantitative Prediction of Surface Wettability at the Atomistic Level
No full textGeochronology, geochemistry, and its geological significance of the Damaoqi Permian volcanic sequences on the northern margin of the North China Block
No full textThe Suji volcanic sequences are situated on the boundary between the northern margin of the North China Block (NCB) and southern margin of the Inner Mongolia-Daxinganling Orogenic Belt (IMDOB). The sequences mainly consist of dacites and rhyolites. Zircons from the Suji rhyolite and dacite yielded weighted mean (206)pb/U-238 ages of 270.2 +/- 2.7 Ma and 268.0 +/- 3.8 Ma. These ages constrain the volcanic sequences erupted at the late Early Permian. The SiO2 content of the volcanic sequences ranges from 55.2 wt% to 78.8 wt%. These rocks exhibit high alkali contents, enrichment of large ion lithophile elements and light rare earth elements, and depletion in high field strength elements, with (Sr-87/Sr-86)(i) ratios of 0.70601-0.71427, epsilon(Nd)(t) of -9.5 to -11.8. The high (Sr-87/Sr-86)(i) and negative epsilon(Nd)(t) values are indicative of a significant contribution of crustal-derived rocks in the petrogenesis of the Suji volcanic rocks. The rhyolitic rocks have high ratios of Rb/Sr, K/Rb and 10000 x Ga/Al, and fractionated REE patterns with strong negative Eu anomalies. These geochemical features show a typical A(2)-type affinity, suggesting the late Early Permian Suji volcanic sequences formed in a post-collisional extensional setting. The distinct geochemistry features of the Suji volcanic sequences, combined with many previous studies from neighbouring regions, demonstrate that the final collision orogenesis in southern IMDOB took place as early as 270 Ma. (C) 2014 Elsevier Ltd. All rights reserved.Geosciences, MultidisciplinarySCI(E)[email protected],SI307-3199
Effect of NaBH4 on properties of nanoscale zero-valent iron and its catalytic activity for reduction of p-nitrophenol
No full textInternational audienceThe reduction of p-nitrophenol (p-NP) to p-aminophenol (p-AP) by nanoscale zero-valent iron (NZVI)/NaBH4 system in an oxygen environment was studied by means of liquid chromatography, spectroscopy (vibration and X-ray photoelectron), solid analyses (transmission electron microscopy and X-ray diffraction) and density functional theory (DFT) calculations. Addition of NaBH4 into NZVI suspension showed the disintegration of NZVI (60-100 nm), resulting in the formation of much smaller particles (15-40 nm) due to the chemical etching of outermost surfaces (i.e., magnetite). Interestingly, complete reduction of p-NP and high conversion efficiency of p-AP (> 98%) were observed in NZVI/NaBH4 system even after four recycling which is quite comparable with widely used noble metallic catalysts. Surface analysis confirmed that NaBH4 can prevent the oxidation of NZVI surface, leading to the continuous reduction of p-NP in oxygen environments. Experimental results and DFT calculations suggested that not only the formation of smaller NZVI but also thermodynamic preferences for reduction of p-NP on outermost surfaces of NZVI (i.e. magnetite) may significantly affect the reduction process of p-NP in NZVI/NaBH4 system. These novel findings can promote the development of new NZVI technologies which can be used for wastewater reductive treatment in oxygen environment
Multiscale Simulation Method for Quantitative Prediction of Surface Wettability at the Atomistic Level
No full textThe
solid–liquid interface is of great interest because
of its highly heterogeneous character and its ubiquity in various
applications. The most fundamental physical variable determining the
strength of the solid–liquid interface is the solid–liquid
interfacial tension, which is usually measured according to the contact
angle. However, an accurate experimental measurement and a reliable
theoretical prediction of the contact angle remain lacking because
of many practical issues. Here, we propose a first-principles-based
simulation approach to quantitatively predict the contact angle of
an ideally clean surface using our recently developed multiscale simulation
method of density functional theory in classical explicit solvents
(DFT-CES). Using this approach, we simulate the surface wettability
of a graphene and graphite surface, resulting in a reliable contact
angle value that is comparable to the experimental data. From our
simulation results, we find that the surface wettability is dominantly
affected by the strength of the solid–liquid van der Waal’s
interaction. However, we further elucidate that there exists a secondary
contribution from the change of water–water interaction, which
is manifested by the change of liquid structure and dynamics of interfacial
water layer. We expect that our proposed method can be used to quantitatively
predict and understand the intriguing wetting phenomena at an atomistic
level and can eventually be utilized to design a surface with a controlled
hydrophobic(philic)ity
도시활력: 대중교통중심 이동 및 활동에 대한 토지이용과 도시형태의 영향
No full text학위논문(박사) - 한국과학기술원 : 조천식녹색교통대학원, 2020.8,[vi, 150 p. :]This dissertation aimed to understand the environmental factors enhancing urban vitality. The city of Seoul in South Korea was tested to explore the relationships between transit-oriented urban mobility, urban activity, land use, and urban form through a data-driven approach. The analyses in this dissertation were conducted over three steps. The first analysis validated the interactive relationship between transit-oriented urban mobility and urban activity, from the geographical and temporal perspectives. The second analysis grouped the subway stations based on the similarity in temporal profiles of ridership using clustering analysis and evaluated the effects of land use and urban form on temporally different trajectories and volumes of ridership using trajectory modeling. In the last analysis, the author defined the reasonable areal unit to analyze urban activity from the urban vitality perspective, and examined the influential attributes of land use and urban form for vibrant activity using spatial regression modeling. The outcomes of these analyses offer a chance to understand the current urban structure and the requirements for creating vibrant environments in cities. The results show that the city has a severed structure by regional functions, but the dense, widespread transit network supports to sustain vibrant citizens’ mobilities and activities at the city level as it reduces the auto-dependency of citizens. At the neighborhood level, the results verify that transit- and walking-friendly environments encourage a vibrant community, which involves the well-connected streets, small block size, and small plot size. In addition, the analysis result of activity indicates that the residential use should be placed in balance with the neighborhood commercial, education, and welfare uses. The findings of this dissertation can provide the fields of urban and transportation planning with a comprehensive insight into creating a vibrant and sustainable city and also improving the quality of citizens’ daily lives.한국과학기술원 :조천식녹색교통대학원
Polymeric Carbon Nitride with Localized Aluminum Coordination Sites as a Durable and Efficient Photocatalyst for Visible Light Utilization
No full textMeasuring Haptic Experience: Elaborating the HX model with scale development
No full textIf an author previously posted their submitted version of the article in any of the following locations, he or she will need to replace the submitted version with the accepted version and add the IEEE copyright notice (© 2021 IEEE). When the article is published, the posted version should be updated with a full citation to the original IEEE publication, including DOI. No other changes may be made to the accepted article.Designers increasingly employ haptic feedback with the aim to improve user experience (UX). Designers and researchers currently use qualitative methods or demos for feedback, but neither approach scales to large studies or remote work. We build upon the recent Haptic Experience (HX) model and report on progress towards measuring the five constructs for designing haptic experiences: Harmony, Expressivity, Autotelics, Immersion, and Realism. We describe initial findings from scale development, specifically, from item generation (N=23) and exploratory factor analysis (N=261). Our results provide initial evidence that vibrotactile experiences are effectively modeled by five factors, enriched description of each factor, and guidelines for quantitatively measuring HX.NSERC Discovery Gran
