Nonlinear regression in adsorption studies: which software to use?

Resumo

This study compares the performance of Excel 365, RStudio, Jupyter Notebook, Origin, and Matlab for nonlinear regression in adsorption studies. Nonlinear regression is vital for modeling adsorption processes or any other process involving adsorption, such as the removal of methylene blue by activated carbon from monkfish seeds (CAFT). The experimental setup involved kinetic and isotherm models, specifically pseudo-second-order kinetics and the Langmuir isotherm, using a Gauss-Newton algorithm based on the least squares method applied to experimental data from methylene blue adsorption (pH 5.5) on CAFT, an activated carbon from monkfish seeds. Data input varied in complexity and time from one program to another. Excel 365, for example, was easy to use and required minimal programming knowledge. Nonlinear regressions were performed using the Solver add-in or Gauss-Newton algorithm. RStudio offered advanced statistical tools for nonlinear regression via the nls function but required knowledge of R programming. Jupyter Notebook, similar to RStudio, allowed for multiple programming languages like Python, nonlinear regressions were performed using the curve_fit function. Origin is specially designed for scientific data analysis and offered intuitive tools for nonlinear regression without programming knowledge but lacks a free version. Finally, Matlab provided extensive mathematical modeling capabilities, with nonlinear regression available through the Curve Fitter tool. It requires programming knowledge and is not free, though a mobile version is available. The results of the kinetic and isotherm modeling are presented in Tables 1 and 2. As expected, all software produced comparable statistical results with minimal differences but differed greatly in ease of use, availability, and graphical illustration capabilities [1,2].