Dichromatic model based illuminant estimation under time-varying light sources

유준상 Greudate School, Korea University 2021 국내박사

Bilayer oxide semiconductor gas sensors with nanoscale overlayer for highly selective, sensitive, and humidity-independent gas sensing

정성용 Greudate School, Korea University 2021 국내박사

Metal oxide chemiresistors have been widely used to detect diverse gases owing to their advantages such as simple structure, cost-effectiveness, miniaturization, and facile integration. However, they often show similar selectivity to a range of different gases and exhibit insufficient gas response for use in actual applications. Moreover, the removal of moisture dependence of gas sensing characteristics remains unsolved. Accordingly, in this thesis, three bilayer sensors composed of gas sensing film and nanoscale oxide overlayer have been suggested as the promising platforms for achieving highly selective, sensitive, and humidity-independent gas sensing characteristics, which include 1) the SnO2 sensors with a Cr2O3 catalytic overlayer, 2) Pd-SnO2 sensors with a Co3O4 catalytic overlayer, and 3) In2O3 sensors with a moisture-blocking Tb4O7 overlayer. The bilayer sensor provides a high tunability for gas selectivity and response via the separate control of catalytic activation and gas sensing reaction, enabling on-demand gas sensors. Furthermore, the moisture dependent gas sensing characteristics can be mitigated by coating the moisture-blocking overlayer. For demonstrating the promising potential of this bilayer concept, the sensing mechanisms underlying the selective, sensitive, and humidity-independent gas sensing characteristics are investigated in relation to the sensing film, catalytic overlayer materials, and configuration of the bilayer films. First, a bilayer sensor with a SnO2 sensing layer and a nanoscale catalytic Cr2O3 overlayer was used for the exclusive detection of sub-ppm-level ethylene, which can be hardly achieved by simple loading of catalysts on oxide nanostructures. The response (Ra/Rg – 1, where Ra represents the resistance in air and Rg represents the resistance in gas) of the proposed bilayer sensor to 2.5 ppm ethylene was 16.8 at 350 C, which was 2.6–15.3 times higher than those to food-related interfering gases (TMA, DMA, and NH3), ubiquitous ethanol, and indoor pollutants (HCHO and CO). Furthermore, the SnO2 sensor coated with the nano-thin catalytic Cr2O3 overlayer exhibited rapid sensing kinetics and good stability as well as measurable sensor resistance (200–300 kΩ) by using a conventional electric circuit. The sensing mechanism underlying the ultraselective and highly sensitive ethylene detection involved the effective catalytic oxidation of interfering gases into less- or non-reactive species, without limiting the analyte gas transport. The sensor can be used to assess fruit freshness by monitoring the ripening of various fruits. Second, the nanoscale catalytic Co3O4 overlayer was coated on Pd-loaded SnO2 sensing film for selective and sensitive detection of benzene. An ultra-selective and sensitive detection of carcinogenic benzene vapor with negligible interferences from other major indoor pollutants is not only critical but also challenging because the BTX gases (benzene, toluene, and xylene) are generally less reactive to a majority of oxide semiconductor gas sensors and the similar chemical structures of BTX gases hamper their discrimination by chemiresistive variation. In this study, a Pd-loaded SnO2 sensing film coated with a nano-thin catalytic Co3O4 overlayer showed an ultrahigh response (resistance ratio = 88) to 5 ppm benzene with negligibly low cross-responses to the other representative and ubiquitous indoor pollutants such as toluene, xylene, HCHO, CO, and ethanol. The sensor response toward benzene was enhanced by reforming highly stable benzene into more active and smaller species, while the cross-responses to the other indoor pollutants became low through the catalytic oxidation of the gases into less- or non-reactive forms. Such a high gas selectivity to chemically stable benzene is also explained with respect to the synergistic combination of catalytic Co3O4 layer and Pd-SnO2 yolk-shell micro-reactor. Lastly, a novel sensor design with a moisture-blocking Tb4O7 overlayer was suggested as a facile and universal strategy to mitigate the water poisoning without scarifying intrinsic sensing properties. A submicrometer-thick coating of Tb4O7 overlayer on In2O3 sensors effectively eliminated the humidity dependence of the gas-sensing characteristics without significantly altering the gas response, selectivity, and sensor resistance, while pure In2O3 sensors exhibited humidity-dependent gas sensing behaviors. Furthermore, the general validity of the water-blocking effect using the Tb4O7 overlayer was confirmed in diverse gas sensors using SnO2, ZnO, and Pd-loaded SnO2. The negligible moisture interference of the bilayer sensor is explained in terms of the hydrophobic nature of the Tb4O7 overlayer and the prevention of –OH radical formation by the interaction between Tb4O7 and In2O3. Three novel functionalities of bilayer sensors with nanoscale oxide overlayer in this thesis are 1) the suppression of cross-responses to interfering gases through oxidative filtering, 2) the enhancement of responses to less reactive gases through gas reforming, and 3) the reliable and precise gas sensing through moisture-blocking. These bilayer sensors with high selectivity, sensitivity, and moisture endurance can open new pathways toward high-performance gas sensors and electronic noses.

Feedback networks and curriculum learning using deep learning

민정기 Greudate School, Korea University 2021 국내석사

Compared to feed-forward networks that are generally studied in deep learning, feedback network shows superior results in terms of model cost and performance. Also, the network performance can be improved more effectively by using curriculum learning to increase the learning effect. This dissertation presents feedback network and curriculum learning for seismic event classification. In order to compare the proposed model and the base-line model, two types of experiments were conducted using the Korean Peninsula earthquake dataset and STanford EArthquake Dataset(STEAD). In the first experiment, an experiment was conducted to classify natural earthquakes among seismic events to confirm the classification performance of the proposed model. The model proposed in this experiment improved classification accuracy by an average 0.5% compared to the base-line model, and True Positive Rate(TPR) and False positive Rate(FPR) improved by an average 1.5% and 0.3% respectively. In the second experiment, an experiment was conducted to check the classification performance in more general cases by classifying natural and artificial earthquakes among seismic events. The model proposed in this experiment improved the classification accuracy by an average 0.3% compared to the base-line model, and TPR and FPR improved by 1.4% and 0.4%, respectively.

Synthesis and photophysical properties of flavanthrone-based fluorophores with narrow emission spectrum

허선형 Greudate School, Korea University 2021 국내석사

Fluorescent organic light-emitting diodes (OLEDs) are cost-effective and stable with high color purity. However, they suffer from low internal quantum efficiency (IQE) of 25% because generated triplet excitons can’t be directly utilized by radiative process. To avoid exciton loss via non-radiative pathways, thermally activated delayed fluorescence (TADF) materials have been developed. TADF molecules are consisted of donor-acceptor (D-A) moieties with separated highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO), yielding a small energy gap between the first singlet and triplet excited states (ΔEST). TADF materials can achieve 100% IQE but large dipole moment caused by the D-A structure can induce charge transfer (CT) character, resulting in large Stokes shift and broad emission. A broad emission with poor color purity of TADF materials is detrimental to realization of high-resolution OLED display. New flavanthrone-based fluorophores, 8,16-bis(3,5-di-tert-butylphenyl)benzo[h]benzo[5,6]acridino[2,1,9,8-klmna]acridine (FC-DTBP) and 8,16-bis(triisopropylsilyl)benzo[h]benzo[5,6]acridino[2,1,9,8-klmna]acridine (FC-OTIPS) are designed and synthesized for OLED applications. The flavanthrone core has a rigid and symmetric structure. To prevent the intermolecular aggregation and enhance the solubility, tert-butyl phenyl and triisopropylsilyl ether groups are introduced to FC-DTBP and FC-OTIPS, respectively. Both compounds show narrow green emissions and high photoluminescence quantum yields (PLQYs). The solvatochromism test is performed to study CT characteristics, confirming negligible CT properties of both compounds. By density functional theory (DFT) calculation, HOMO, LUMO and the lowest excited singlet state (S1) are calculated. The calculated HOMO and LUMO levels are shallower than experimental energy levels. Based on thermogravimetric analysis (TGA) measurement, FC-OTIPS shows poorer thermal stability compared to FC-DTBP. This phenomenon is predicted to be caused by instability of triisopropylsilyl ether group. The hyperfluorescence OLED device with a fluorescent dopant of FC-DTBP shows a bright electroluminescence emission at 536 nm, 17.8% of maximum external quantum efficiency (EQE), a full width at half maximum (FWHM) of 30 nm with Commission Internationale de l'Eclairage (CIE) coordinates of (0.36, 0.62).

Tungstate-zirconia-supported Ru as a multifunctional catalyst for three step production of deoxygenated hydrocarbons from 2-methylfuran

조현민 Greudate School, Korea University 2021 국내석사

The multifunctionality of tungstate-zirconia-supported Ru catalyst was observed for three step conversion of 2-methylfuran to deoxygenated C10-15 hydrocarbon fuels, where condensation, hydrogenation, and hydrodeoxygenation were performed using the single catalyst of tungstate-zirconia-supported Ru catalyst. By manipulating the reaction conditions, particularly reaction temperature and feeding reactant for each reaction step, tungstate-zirconia-supported Ru exhibited the different catalytic roles for different reaction stages. Compared to other catalysts including Pd and Ni supported on tungstate-zirconia, tungstate-zirconia-supported Ru exhibited good condensation and hydrogenation activity along with the good hydrodeoxygenation activity, achieving selective production of C15 hydrocarbons, rather than C5 and C10 compounds. The improved condensation and hydrogenation on tungstate-zirconia-supported Ru can be attributed to the moderate acidity and the good hydrogenation activity. Multifunctionality of tungstate-zirconia-supported Ru was further characterized using pyridine-FT-IR, TEM, and XRD, which confirmed its adjusted activity in the different reactions.

DNA-barcoding of forensically important carrion beetles (Coleoptera: Silphidae) in Korea

김혜영 Greudate School, Korea University 2021 국내석사

한국의 송장벌레는 2아과 11속 26 종이 알려져 있으며, 북미지역의 8속 30종과 비교해 볼 때 매우 높은 종 다양성을 지니고 있다. 종 분류에 관한 전통적인 방법은 외부 형태적인 특징에 의존하고 있다. 하지만 표현형의 변이가 심하고 그 형태가 비슷하기 때문에 종 구분이 쉽지 않다. 특히 사건 현장에서 자주 발견되는 유충의 경우 성충보다 더욱 어렵다. 또한 이런 분류학적 특징은 곤충의 전문가가 아니면 알 수 없고, 분류하는 방법을 익히는 데도 오랜 시간이 걸린다. 이전부터 유럽 등 외국에서는 송장벌레의 DNA 데이터베이스 구축이 활발하게 연구되어 왔으나, 국내에서는 상대적으로 연구가 적은 편이다. 따라서 본 연구는 송장벌레과의 Mitochondrial cytochrome oxidase subunit I (COI) 유전자를 이용하여 한국 시식성 송장벌레과의 DNA barcode 라이브러리를 구축하였다. 연구는 수집한 송장벌레 표본의 DNA를 추출하여 새로 개발된 프라이머를 이용해 sequence를 확보하였으며, 결과는 NCBI에 등재되어있는 Sequence와 비교분석하였다. 그 결과, 형태적 종동정이 명확하고 출현 빈도가 높은 7종(Necrodes littoralis, Necrodes nigricornis, Necrophila (Calosilpha) brunneicollis brunneicollis, Thanatophilus rugosus, Thanatophilus sinuatus, Nicrophorus quadripunctatus, Nicrophorus concolor)에 대해 Sequence를 얻었으며, 차후에 송장벌레과에 속한 전체 종에 대하여 DNA 바코드 라이브러리를 구축하고 이를 토대로 법의 곤충으로서 중요한 송장벌레과 유충의 종 분류에 활용하고자 한다.