Near-infrared( NIR) spectrum was applied for the rapid quantitative determination of benzene homologues( toluene,ethyl-benzene,p-xylene, m-xylene and o-xylene) in paint thinner of solvent based coatings for woodenware. After samples of paint thinner were collected,firstly gas chromatography( GC) was used to detect content of benzene homologues;secondly their NIR spectra were collected;finally partial least square( PLS) was used to develop a linear correlation between their NIR spectra and content of ben-zene homologues. Root mean square error of calibration(RMSEC)of benzene homologues were between(0. 47~1. 40)%,and their square of the correlation coefficient( R2 ) were between 0. 956~0. 988. Root mean square error of prediction( RMSEP) were between (0. 73~2. 32)%,and their square of the correlation coefficient(R2)were between 0. 951~0. 986. NIR model prediction effect was good. The NIR quantitative method was rapid,simple,accurate,and it could be applied in determination of toxic and harmful sub-stances in coatings.%建立使用近红外光谱法( NIR)快速测定溶剂型木器涂料稀释剂中甲苯、乙苯、对二甲苯、间二甲苯和邻二甲苯等苯系物含量方法。收集涂料稀释剂样品，使用气相色谱法( GC)测定苯系物含量，并采集其近红外光谱信息，采用偏最小二乘法( PLS )建立NIR光谱与苯系物含量的线性关系模型。苯系物校正均方差( RM-SEC)在(0．47~1．40)%之间、相关系数(R2)在0．956~0．988之间；预测均方差(RMSEP)在(0．73~2．32)%之间、相关系数( R2)在0．951~0．986之间。 NIR模型预测效果良好，定量方法快速、简单、准确，可在检测涂料的有毒有害物质中推广应用。
展开▼
