Peralta-Zamora P, Cornejo-Ponce L, Nagata N, Poppi R J
Universidade Estadual de Campinas, Instituto de Quimica, C.P. 6154, 13083-970, Campinas-SP, Brazil.
Talanta. 1997 Oct;44(10):1815-22. doi: 10.1016/S0039-9140(97)00055-6.
The simultaneous determination of lanthanide family elements is one of the greatest problems in analytical chemistry, due to the close similarity of their chemical properties. Spectrophotometric methods are generally of limited use, due to the various mutual spectral interferences involved. By using multivariate calibration methods (partial least-squares regression, PLSR), it was possible to obtain a model that adjusts itself perfectly to the values of the mixture concentrations used in the calibration. The model used absorption spectra in the 290-800 nm range for a set of 20 different mixtures of Ce, Pr, Nd and Sm, and made possible the determination of Ce, Pr and Nd concentrations of a commercial rare-earth product, with significantly greater precision than the conventional univariate calibration method. Determination of the Sm concentrations was not possible, since its concentration was below the concentrations used in the model definition.
由于镧系元素化学性质极为相似，同时测定镧系元素是分析化学中最棘手的问题之一。由于存在各种相互的光谱干扰，分光光度法的应用通常受到限制。通过使用多元校准方法（偏最小二乘回归，PLSR），可以获得一个能完美适配校准中所用混合浓度值的模型。该模型使用了一组包含铈（Ce）、镨（Pr）、钕（Nd）和钐（Sm）的20种不同混合物在290 - 800 nm范围内的吸收光谱，从而能够测定一种商业稀土产品中铈、镨和钕的浓度，其精度显著高于传统的单变量校准方法。由于钐的浓度低于模型定义中所用的浓度，所以无法测定其浓度。
