Deviations from bilinearity in multivariate voltammetric calibration models.

Area of Science:

• Analytical Chemistry
• Chemometrics
• Electrochemistry

Background:

• Multivariate calibration models often suffer from a lack of bilinearity, particularly in voltammetry.
• Overlapping voltammetric signals distort signal shape, sensitivity, and potential axis, leading to linearity loss.
• This linearity loss significantly degrades the performance of chemometric models designed for linear data.

Purpose of the Study:

• To address the challenge of non-bilinearity in multivariate calibration for voltammetric data.
• To improve the predictive capabilities of chemometric models by enhancing data linearity.
• To introduce and validate Orthogonal Signal Correction (OSC) as a pre-processing technique.

Main Methods:

• Application of Orthogonal Signal Correction (OSC) as a data pre-processing technique.
• Utilized simulated voltammetric signals to demonstrate the OSC methodology.
• Applied OSC to real-world samples for the determination of scandium in the presence of nickel.

Main Results:

• Orthogonal Signal Correction successfully removed non-linear effects from voltammetric signals.
• The pre-processing enabled the development of more effective Partial Least Squares (PLS) models.
• Accurate determination of nanomolar scandium concentrations was achieved despite high and variable nickel excess.

Conclusions:

• Orthogonal Signal Correction is a powerful tool for mitigating non-bilinearity in voltammetric data.
• OSC significantly enhances the performance and predictive accuracy of multivariate calibration models.
• The method proves effective for complex analytical challenges, such as trace analysis in the presence of interferents.