浙江农业学报 ›› 2016, Vol. 28 ›› Issue (12): 2123-2129.DOI: 10.3969/j.issn.1004-1524.2016.12.23

• 生物系统工程 • 上一篇    

基于均匀设计的土壤重金属PXRF检测方法优化研究

杨桂兰1, 商照聪1,2, 李良君1, 倪晓芳1,2,*, 章明洪1,2   

  1. 1.上海化工研究院,上海 200062;
    2.上海天科化工检测有限公司,上海 200062
  • 收稿日期:2016-02-16 出版日期:2016-12-15 发布日期:2017-01-05
  • 通讯作者: 倪晓芳, E-mail: nxf_sds@163.com
  • 作者简介:杨桂兰(1990—),女,山东莒县人,硕士研究生,从事X射线荧光技术分析检测研究。E-mail: ygl090206@163.com
  • 基金资助:
    上海市青年科技启明星计划(16QB1401500)

Application of uniform design method in optimizing PXRF determination methods of heavy metals in soil

YANG Gui-lan1, SHANG Zhao-cong1,2, LI Liang-jun1, NI Xiao-fang1,2,*, ZHANG Ming-hong1,2   

  1. 1. Shanghai Research Institute of Chemical Industry, Shanghai 200062, China;
    2. Shanghai TECH. Chemical Industry Testing Co., Ltd, Shanghai 200062, China
  • Received:2016-02-16 Online:2016-12-15 Published:2017-01-05

摘要: 应用便携式X射线荧光光谱法(PXRF,portable X-ray fluorescence)对长三角地区某农田土壤中Cr、Cu、Zn、As和Pb等5种土壤中常见重金属元素进行测定。应用均匀设计实验方法,并结合二次多项式逐步回归法建立整体平均相对标准偏差(5种元素的相对标准偏差的均值)与检测条件之间的回归模型。结果显示,最优检测条件为风干样品、0.125 mm粒径、制样压力3 MPa、检测时间115 s。3组验证实验的预测相对误差小于5%,表明该模型预测能力良好。在该条件下,PXRF对Cr、Cu、Zn、As、Pb的最低检出限分别为15.0、4.12、3.22、2.22、3.14 mg·kg-1。33个土壤样品的PXRF检测值与ICP-OES检测值之间的一元线性回归模型的决定系数R2均大于0.97,7个未知样的预测相对误差全部低于10%,表明在均匀设计法优化得到的最佳检测条件下,PXRF检测结果通过简单的一元线性回归模型校正后,可满足定量分析的要求。

关键词: PXRF, 土壤, 重金属, 检测条件, 均匀设计法

Abstract: Total concentrations of Cr, Cu, Zn, As and Pb in farm land soil from Yangtze River Delta region were determined by portable X-ray fluorescence spectroscopy (PXRF). A regression model between the overall average of relative standard deviation (RSD) and relevant factors was established using uniform design and quadratic polynomial stepwise regression method. It was shown that the optimal detection conditions were obtained as air dried sample, 0.125 mm particle size, 3 MPa sampling pressure and 115 s detection time. The relative error between the overall average of RSD (the average of 5 elements' relative standard deviation) for experimental values and the overall average of RSD for predicted values in regression model verification test were less than 5%, indicating a good predictive ability of the regression model. Under such optimal detection conditions, the detection limits of PXRF for Cr, Cu, Zn, As, and Pb were 15.0, 4.12, 3.22, 2.22 and 3.14 mg·kg-1, respectively. For total concentrations of Cr, Cu, Zn, As and Pb in 33 soil samples determined by PXRF and ICP-OES, the linear regression models' coefficients R2 were all greater than 0.97, and the relative error between model predicted values and ICP-OES measured value of 7 unknown samples were less than 10%. The results showed that under the optimal detection conditions, test data by PXRF reached a quantitative analysis level through linear regression model calibration.

Key words: PXRF, soil, heavy metals, detection conditions, uniform design experimentation

中图分类号: