無機氣體表面聲波感測器研製與應用

Preparation and Application of Surface Acoustic Wave Sensor for Inorganic Gases

蔡顯仁(Hsien-Jen Tsai)

指導教授 ： 施正雄

國立臺灣師範大學/理學院/化學系/碩士(2004年)

摘要

本研究中研製一多頻道氣體表面聲波感測器(surface acoustic wave sensor, SAW)以偵測空氣中二氧化氮(nitrogen dioxide, NO2)及一氧化碳(carbon monoxide, CO)，在表面聲波晶片上塗佈Ru3+/ cryptand[2,2]與Zn2+/ cryptand[2,2]辨識膜分別對NO2及CO作感測，本研究發現待測氣體與塗佈物間吸附現象為物理吸附，可反覆使用具良好再現性，並且使用壽命超過一個月；由氣體濃度效應研究中發現，本研究研製的表面聲波感測器頻率訊號與CO及NO2濃度皆有良好的線性關係，其中Ru3+/ crypatnd[2,2]及Zn2+/ cryptand[2,2]偵測NO2及CO的偵測下限分別為0.176與0.699 ppm，皆低於恕限量及排放標準。本研究亦探討環境溫度和濕度對表面聲波感測器的影響，各種空氣中有機揮發污染物對表面聲波感測器偵測NO2與CO可能造成的干擾亦被探討。
本研究亦藉由數學統計方法中的主成分分析法(principal component analysis, PCA)來確定本研究所選擇的塗佈物足以分辨NO2及CO，由研究結果顯示在二維的X-Y主成份分數散佈圖(PCA scores plot)，其中NO2與CO各自成群且相互分離即可證明確實足以分辨NO2及CO。另外亦使用類神經網路中監督型倒傳遞神經網路(back propagation network, BPN)來作確認及辨別NO2和CO，倒傳遞神經網路系統亦顯示所選的表面聲波塗佈物Ru3+/ crypatnd[2,2]和Zn2+/ cryptand[2,2]確可分辨NO2及CO作為定性分析，多變項複迴歸分析技術(multivariate multiple regression analysis)亦用來作定量分析。

關鍵字

表面聲波感測器 ； 主成份分析 ； 倒傳遞類神經網路 ； 迴歸分析

並列摘要

A multichannel surface acoustic wave (SAW) gas sensor system was prepared to detect NO2 and CO in the air. The coated Ru3+/ cryptand[2,2] and Zn2+/ cryptand[2,2] SAW crystals were applied to recognize NO2 and CO, respectively. The physical adsorption was found for the adsorption of these inorganic gases onto respective coating materials. The SAW sensor also showed good reproducibility and good enough lifetime of ≧ 30 days for detection of NO2 and CO. The detection limits of this SAW sensor with Ru3+/ cryptand[2,2] and Zn2+/ cryptand[2,2] coatings for NO2 and CO were 0.172 and 0.699 ppm respectively, which were lower than occapational exposure limits for both gases and implied that the developed SAW sensor in this study could be employed for environmental analysis for both gases. The concentration effect of NO2 and CO on the frequency responses of the SAW sensor was studied and showed good linear responses with the concentrations of NO2 and CO, respectively. Effects of temperature and humidity on the SAW sensor were also investigated and discussed. Furthermore, the interference of some organic vapors to the detection of NO2 and CO with the SAW sensor was also studied and discussed.
The principal component analysis (PCA) was also applied in this study to confirm that appropriate coating materials for NO2 and CO were selected. Two dimension PCA scores plot showed good separation between NO2 and CO which implied that NO2 and CO can be distinguished clearly by the two-channel SAW sensor. In addition, an artificial neural network, using back propagation network (BPN), was also used to recognize NO2 and CO gases and it shows the distinction of these inorganic gases qualitatively by the two-channel SAW sensor with Ru3+/ crypand[2,2] and Zn2+/ crypatnd[2,2] coatings. The quantitative analysis for NO2 and CO were also studied by the multivariate multiple regression analysis.

並列關鍵字

surface acoustic wave sensor ； principal component analysis ； back propagation network ； regression analysis