• Membrane Journal
• /
• v.22 no.6
• /
• pp.415-423
• /
• 2012

For hybrid water treatment of carbon fiber ultrafiltration and photocatalyst, we used the hybrid module that was composed of photocatalyst packing between tubular membrane outside and module inside. Photocatalyst was PP (polypropylene) bead coated with $TiO_2$ powder by CVD (chemical vapor deposition) process. Water back-flushing of 10 sec was performed per every period of 10 min to minimize membrane fouling for modified solution, which was prepared with humic acid and kaolin. Resistance of membrane fouling ($R_f$) decreased as humic acid concentration changed from 10 mg/L to 2 mg/L, and finally the highest total permeate volume ($V_T$) could be obtained at 2 mg/L, which was the same with the previous results. Then, treatment efficiencies of turbidity and humic acid were above 98.9% and 88.7%, respectively, those did not depend on the humic acid concentration. However, the treatment efficiency of humic acid increased a little as the humic acid concentration increased in the previous results.

• Proceedings of the Korean Institute of Electrical and Electronic Material Engineers Conference
• /
• 2008.06a
• /
• pp.158-158
• /
• 2008

모바일 정보통신기기를 중심으로 실장모듈의 초소형화, 고집적화로 인해 접속단자의 피치가 점점 미세화 됨에 따라 플립칩 본딩용 접착제에 함유되는 무기충전제인 실리카 필러의 크기도 미세화되고 있다. 본 연구에서는 NCP (non-conductive paste)의 실리카 필러의 크기가 COB(chip-on-board) 플립칩 패키지의 신뢰성에 미치는 영향을 조사하였다. 실험에 사용된 실리카 필러는 Fused silica 3 종과 Fumed silica 3종이며 response surface 실험계획법에 따라 혼합하여 최적의 혼합비를 정하였다. 테스트베드로 사용된 실리콘 다이는 투께 $700{\mu}m$, 면적 5.2$\times$7.2mm로 $50\times50{\mu}m$ 크기의 Au 도금범프를 $100{\mu}m$ 피치, peripheral 방식으로 형성시켰으며, 기판은 패드를 Sn으로 finish 하였다. 기판을 플라즈마 전처리 후 Panasonic FCB-3 플립칩 본더를 이용하여 플립칩 본딩을 수행하였다. 패키지의 신뢰성 평가를 위해 $-40^{\circ}C{\sim}80^{\circ}C$의 열충격시험과 $85^{\circ}C$/85%R.H.의 고온고습시험을 수행하였으며 Die shear를 통한 접합 강도와 4-point probe를 통한 접속저항을 측정하였다.

• Journal of Convergence for Information Technology
• /
• v.11 no.9
• /
• pp.28-34
• /
• 2021

In the case of object detection using deep learning, both accuracy and real-time are required. However, it is difficult to use a deep learning model that processes a large amount of data in a limited resource environment. To solve this problem, this paper proposes an object detection model for small embedded devices. Unlike the general detection model, the model size was minimized by using a structure in which the pre-trained feature extractor was removed. The structure of the model was designed by repeatedly stacking lightweight convolution blocks. In addition, the number of region proposals is greatly reduced to reduce detection overhead. The proposed model was trained and evaluated using the public dataset PASCAL VOC. For quantitative evaluation of the model, detection performance was measured with average precision used in the detection field. And the detection speed was measured in a Raspberry Pi similar to an actual embedded device. Through the experiment, we achieved improved accuracy and faster reasoning speed compared to the existing detection method.

• Proceedings of the Korean Vacuum Society Conference
• /
• 2012.02a
• /
• pp.289-289
• /
• 2012

터치패널은 키보드나 마우스와 같은 입력장치를 사용하지 않고, 스크린에 손가락, 펜 등을 접촉하여 입력하는 방식이다. 누구나 쉽게 입력할 수 있는 장점으로 인해 기존에는 현금인출기, 키오스크 등 공공분야에 주로 많이 사용되어 왔으나, 최근의 터치스크린은 휴대폰, 게임기, 네비게이션, 노트북 모니터 등 개인정보기기의 입력장치로 활용분야가 넓어져가고 있다. 최근의 정전용량 방식의 터치패널은 디스플레이 패널 위에 올여지는 형태의 Add on type이며, 테블렛의 출현으로 터치패널의 사이즈가 커지면서 인듐산화물 투명성 전도막의 두께가 두꺼워지고, 이로 인하여 광학적 특성인 투과율이 저하되는 문제가 발생하여 투과율을 높여주기 위한 새로운 전도박막 제조방법이 요구되는 상황이다. 현재의 고글절 산화물(TiO2)과 저굴절 산화물(SiO2)의 적층형태의 저반사 특성의 다층막은 주로 플라즈마 보조의 전자빔 증착기를 이용하여 제조되기 때문에, 저반사 특성이 우수하지만 대면적 크기의 대량생산에는 적합하지가 않다. 그리고 태양전지의 에너지 변환효율도 태양전지로 흡수되는 태양광의 량에 크게 의존하기 때문에, 태양전지로 흡수되는 태양광 량을 높이기 위하여 태양전지의 가장 위층에 혹은 모듈 제작시 커버유리의 내부에 저반사 특성을 갖는 박막을 코팅한다. 특히 박막태양전지의 경우는 대면적의 유리위에 저반사 코팅을 해야 한다. 본 연구에서는 In-line magnetron sputtering system을 사용하여 소다라임 유리 기판 위에 고글절 산화물(Nb2O5)과 저굴절 산화물(SiO2)의 2층 적층형태의 "SiO2/Nb2O5/SiO2/Nb2O5/SLG" 다층 박막을 증착하고, 저반사의 광학적 특성을 하였고, 이를 논하고자한다. 일반적으로 빛이 투과되는 투명한 기판이 공기층에 노출되어있을 경우에 기판의 양면에서 공기층과의 계면에서 각각 4%의 반사율 즉, 총 8%의 반사율을 갖는데, 본 연구의 다층 박막에서는 530에서 540nm 파장 영역에서 투과율은 95% 이상, 반사율은 4.8% 이하이었다. 이 결과는 터치패널과 박막태양전지 시장의 Needs에 대응할 수 있기 때문에 산업의 응용측면에서 매우 중요한 연구 성과를 얻었다고 말할 수 있다. (본 연구는 지식경제부 사업화연게기술개발 연구지원금으로 일부 이루어졌음).

• Membrane Journal
• /
• v.6 no.3
• /
• pp.173-181
• /
• 1996

Common volatile organic compounds(VOCs) such as toluene and methanol were removed successfully from N$_{2}$ using a novel silicone-coated hollow fiber membrane module. This novel membrane is a thin film composite(TFC) and was highly efficient in removing VOCs selectively from a N$_{2}$ stream. This membrane had some innate advantages over other silicone-based membrane in that the selective barrier was ultrathin(~1 $\mu$m) and the porosity of the polypropylene substrate was high which leads to a low permeation resistance. The substram was very strongly bonded to the coating layer by plasma polymerization and can withstand a very high pressure. A small hollow fiber module having a length of 25cm and 50 fibers could remove 96~99% of toluene as well as methanol vapors when the feed flow rate was up to 60cc/min. The percent removal of VOCs were even higher when the feed inlet concentration was higher. This process is especially suitable for treating streams having a low flow rate and high VOCs concentration. The permeances of VOCs through this membrane was in the range of $4~30 \times 10^{-9}gmol/sec \cdot cm^{2}\cdot cmHg$ for both toluene and methanol, and nitrogen permeance was between $3~9 \times 10^{-10}gmol/sec \cdot cm^{2} \cdot cmHg$. High separation factor between 10~55 for toluene/N$_{2}$ and 15~125 for methanol/N$_{2}$ were obtained depending on the feed flow rate ranges and feed VOCs concentration levels.

• Membrane Journal
• /
• v.22 no.5
• /
• pp.359-368
• /
• 2012

For hybrid water treatment of high turbidity water, we used the hybrid module that was composed of photocatalyst packing between tubular membrane outside and module inside. Photocatalyst was PP (polypropylene) bead coated with $TiO_2$ powder by CVD (chemical vapor deposition) process. Water back-flushing of 10 sec was performed per every period of 10 min to minimize membrane fouling for modified solution was prepared with humic acid and kaolin. Resistance of membrane fouling ($R_f$) decreased as humic acid concentration changed from 10 mg/L to 2 mg/L, and finally the highest total permeate volume ($V_T$) could be obtained at 2 mg/L, which was the same with the previous results. Then, treatment efficiencies of turbidity and humic acid were above 98.9% and 88.7%, respectively. As results of treatment portions of UF, UF + $TiO_2$, and UF + $TiO_2$ + UV processes, turbidity was treated little by photocatalyst adsorption, and photo-oxidation. However, treatment portions of humic acid by adsorption and photo-oxidation were 2.5% and 12.3%, respectively. Compared with the previous results, treatment portions of humic acid by adsorption and photo-oxidation were different depending on membrane material and pore size. As simplified the process, the membrane fouling resistance after 180 minutes' operation ($R_{f,180}$) increased and the final permeate flux decreased a little.

• Journal of the Korea Convergence Society
• /
• v.12 no.11
• /
• pp.45-51
• /
• 2021

With the development of deep learning, semantic segmentation methods are being studied in various fields. There is a problem that segmenation accuracy drops in fields that require accuracy such as medical image analysis. In this paper, we improved PSPNet, which is a deep learning based segmentation method to minimized the loss of features during semantic segmentation. Conventional deep learning based segmentation methods result in lower resolution and loss of object features during feature extraction and compression. Due to these losses, the edge and the internal information of the object are lost, and there is a problem that the accuracy at the time of object segmentation is lowered. To solve these problems, we improved PSPNet, which is a semantic segmentation model. The multi-scale attention proposed to the conventional PSPNet was added to prevent feature loss of objects. The feature purification process was performed by applying the attention method to the conventional PPM module. By suppressing unnecessary feature information, eadg and texture information was improved. The proposed method trained on the Cityscapes dataset and use the segmentation index MIoU for quantitative evaluation. As a result of the experiment, the segmentation accuracy was improved by about 1.5% compared to the conventional PSPNet.