近红外光谱法快速测定溶剂型木器涂料中甲苯、乙苯和二甲苯

Rapid determination of contents of toluene,ethylbenzene and xylene in solvent based coatings for woodenware by NIRS

建立了使用近红外光谱法（NIR）快速测定溶剂型木器涂料中甲苯、乙苯和二甲苯的方法。收集涂料样品,使用气相色谱法（GC）测定苯系物含量。采用聚乙烯密实袋封装聚氨酯类、硝基类或醇酸类涂料,应用积分球透漫射采样方式采集清漆和漫射采样方式采集色漆的近红外光谱。采用偏最小二乘法,分别建立清漆和色漆的近红外光谱与苯系物线性关系模型。校正集均方差在0.43%-1.32%之间、相关系数R在0.9046-0.9766之间。验证集均方差在0.591%-1.73%之间。对未知样品预测,清漆样品预测值相对偏差〈15%;色漆样品预测值相对偏差〈20%。两个定量模型预测效果良好。该2个NIR定量方法适用于对溶剂型木器涂料中甲苯、乙苯和二甲苯含量进行快速测定。

A method was built for the rapid quantitative determination of contents of toluene, ethylbenzene, p- xylene, m-xylene and o-xylene in solvent based coatings for woodenware by Near-infrared spectrometry （NIR）. After coating samples were collected, gas chromatography （ GC ） was used to detect the contents of benzene homologues. Polyethylene plastic bags were used to package polyurethane, nitrocellulose and alkyd coatings, then their NIR spectra were collected by integrating sphere transmittance-diffuse-reflectance for varnishes and diffuse-reflectance for paints. Partial least square （PLS） was used to develop two linear correlation models for varnish and paint respectively between their NIR spectra and the contents of benzene homologues. Room mean square errors of calibration （ RMSEC ） were 0. 430% - 1. 320%, and the correlation coefficients （R） were 0. 9046 -0. 9766. Room mean square errors of prediction （RMSEP） were 0. 591% - 1. 730%. For prediction results of unknown samples, relative deviations of forecast for varnish samples values were less than 15%, and relative deviations of forecast values for paint samples were less than 20%. Prediction effect from two quantitative models is good. The two NIR quantitative methods could be applied to the rapid determination of contents of toluene, ethylbenzene and xylene in solvent based coatings.