文献类型：期刊文献

中文题名：近红外光谱结合膜富集技术测定饮料中微量邻苯二甲酸二异辛酯的含量

英文题名：Determination of Trace Diisooctylphthalate in Drink by NIR Spectroscopy Coupled with Membrane Enrichment Technique

作者：陈贵平[1];童佩瑾[1];耿金培[2];李晓玉[2];李秀勇[2];杜一平[1]

机构：[1]华东理工大学上海市功能材料重点实验室化学与分子工程学院,上海200237;[2]烟台市出入境检验检疫局,山东烟台264000

年份：2012

卷号：31

期号：5

起止页码：605

中文期刊名：分析测试学报

外文期刊名：Journal of Instrumental Analysis

收录：CSTPCD;;北大核心:【北大核心2011】；CSCD:【CSCD2011_2012】；

基金：国家自然科学基金项目(20975039);烟台市科技局项目(2010145)

语种：中文

中文关键词：近红外光谱;聚醚砜膜;漫反射;邻苯二甲酸二异辛酯;饮料;膜富集技术

外文关键词：near infared spectroscopy; polyethersulfone membrane; diffuse reflection; diisooctylphthalate; drink; membrane enrichment technique

摘要：提出了一种基于在线膜富集的近红外漫反射光谱技术,对饮料中的微量塑化剂邻苯二甲酸二异辛酯(DEHP)进行快速检测。采用聚醚砜膜对饮料中的DEHP进行富集,将富集DEHP的膜直接进行近红外漫反射检测。参考DEHP的透射近红外光谱,对波数进行选择,以4 420～4 060、4 700～4 540、6 040～5 600cm-1作为建模的波数区间。通过比较原始光谱、多元散射校正、一阶求导、二阶求导及其组合,考察了光谱预处理方法对模型的影响,用去一交互验证法建立了偏最小二乘(PLS)模型,并用所建立的校正模型对校正集样品进行了预测。结果表明,在选定的波数区间,当用一阶求导对校正集光谱进行预处理时,所建立的模型对校正集的预测效果最佳,在隐变量数为7时,对校正集所有样品的校正均方根误差(RMSEC)为0.188 7mg/L。用此模型对预测集样品进行预测时,DEHP的质量浓度在0.5～5.0 mg/L范围内,预测均方根误差(RMSEP)为0.232 4 mg/L,平均相对预测误差为6.29%。
A method based on on-line enrichment technique and diffuse reflection near infrared spectroscopy（DRNIRS） was developed to detect rapidly trace plasticizer diisooctylphthalate（DEHP） in drink.DEHP solutions were passed through and concentrated with the polyethersulfone（PES） membrane and then the surface of the membrane was detected directly to collect diffuse reflection NIR（DRNIR） spectra.The wavenumber regions of 4 420-4 060,4 700-4 540,6 040-5 600 cm-1 were selected to build calibration models according to the transmission NIR spectrum of DEHP.The original spectral data were pretreated by multiplicative scattering correction（MSC）,first derivative,second derivative,multiplicative scattering correction combined with first derivative and multiplicative scattering correction combined with second derivative,respectively.Meanwhile,partial least squares（PLS） calibration models with leave-one-out cross validation were established,and samples in the prediction dataset were predicted by the calibration models.The results demonstrated that the predictive results of the PLS calibration models were the best when the spectral data were pretreated by first derivative,and the latent variable number was 7.The root mean squared error of calibration（RMSEC） for the calibration samples was 0.188 7 mg/L,and the root mean squared error of prediction（RMSEP） for the prediction samples was 0.232 4 mg/L with mean relative error（MRE） of 6.29% in the DEHP mass concentration range of 0.5-5.0 mg/L.