液体阴极辉光放电-原子发射光谱是近些年兴起的一种水体金属元素检测技术。 该技术具有开放大气环境工作, 进样简便, 体积小, 运行费用低, 可同时检测多种金属元素等显著特征。 根据之前的研究工作可知, 金属元素的浓度不仅与自身的某一条谱线强度有关, 而且还与自身其他的谱线或者基体中其他元素的谱线强度有关。 为提高该技术的检测能力和精度, 降低实验过程中基体效应的影响, 以及更加充分地利用光谱信息, 采用多元线性回归法对光谱信息进行定量分析。 选取Pb Ⅰ 368. 35 nm和Pb Ⅰ 405.78 nm两条特征谱线, 建立Pb元素浓度与这两条光谱线强度的二元线性回归方程; 相比于标准曲线法, Pb元素的拟合度R2从0.986 5提高到0.998 7, 两组Pb测试液的相对误差从34.00%和29.00%降低到14.20%和1.51%。 为降低复杂成分中基体效应的影响, 建立Na的浓度与Na Ⅰ 589.38 nm, Zn Ⅰ 213.8 nm, PbⅠ405.78 nm和Hβ四条特征谱线强度的四元线性回归方程; 拟合度R2从标准曲线法的0.955 8提高到0.995 6, 两组Na测试液的相对误差从11.67%和14.71%降低到2.33%和3.57%。 以上结果表明: 相比于标准曲线法, 多元线性回归法可以降低实验过程中基体效应的影响, 并且能更加充分地利用光谱信息, 能提高拟合度R2, 以及降低测量的误差, 从而提高液体阴极辉光放电-原子发射光谱定量分析金属元素的精度。

In recent years, solution cathode glow discharge atomic emission spectroscopy (SCGD-AES), a booming technique for metal element determination in aqueous solution, has a large quantity of notable characteristics, such as working under atmospheric pressure, easy simple injection, little size, small volume, low operating costs, and simultaneous detection of multi-metal-elements. According to our previous researches, the concentration of metal elements was relative to not only one of its characteristic spectral lines, but also the other characteristic spectral lines of itself or the characteristic spectral lines of other elements. In order to improve the detection capability and accuracy of solution cathode glow discharge atomic emission spectroscopy, reduce the matrix effect in the complex solution and make full use of spectral information, the multivariate linear regression method was employed to make a quantitative analysis of spectral information. Two characteristic spectral lines of Pb Ⅰ 368.35 nm and Pb Ⅰ 405.78 nm were employed to establish the binary linear regression equation; compared with standard curve method, the degree of fitting R2 increased to 0.998 7 from 0.986 5, and the relative errors of two test solutions decreased to 14.20% and 1.51% from 34.00% and 29.00%, respectively. Quaternary linear regression equation using Na Ⅰ 589.38 nm, Zn Ⅰ 213.80 nm, Pb Ⅰ 405.78 nm and Hβ were also built to reduce the matrix effect in the complex solution; compared with standard curve method, the degree of fitting R2 increased to 0.995 6 from 0.955 8, and the relative errors of two test solutions decreased to 2.33% and 3.57% from 11.67% and 14.71%, respectively. The above results indicate that: compared with standard curve method, multivariate linear regression method has the ability to reduce the influence of matrix effect in the complex aqueous solution, take full advantage of spectral information, improve degree of fitting (R2)and reduce the errors, thus improve the accuracy of metal elements determination with solution cathode glow discharge atomic emission spectroscopy.