Statistical analysis of adsorption isotherm models and its appropriate selection

Area of Science:

• Environmental Science
• Materials Science
• Physical Chemistry

Background:

• Adsorption research extensively studies various adsorbent-adsorbate systems and isotherm models.
• However, a comprehensive evaluation of model applicability across diverse systems is lacking.

Purpose of the Study:

• To assess the performance of twelve common isotherm models in correlating adsorption data for activated carbon with various solutes.
• To identify the most effective models for diverse adsorbent-adsorbate systems.

Main Methods:

• Literature data for activated carbon adsorption of solutes with molecular masses ranging from 78.118 to 932 g/mol were compiled.
• Twelve isotherm models were applied using Polymath® software, with nonlinear optimization to determine model parameters.
• Model performance was evaluated using Arithmetic Average Relative Deviation (AARD%) and Corrected Akaike's Information Criterion (AICcorrected).

Main Results:

• The Baudu Isotherm and Langmuir-Freundlich isotherm exhibited the lowest overall AARD% values (2.6% and 2.8%, respectively).
• The Langmuir-Freundlich isotherm yielded the lowest AICcorrected value (15.859), indicating superior model efficiency.
• The Marczewski-Jaroniec isotherm showed the highest AARD% (23.5%) and AICcorrected (59.283).

Conclusions:

• The Langmuir-Freundlich and Baudu isotherms are highly effective for correlating adsorption data across a range of activated carbon and solute systems.
• Model selection based on statistical metrics like AARD% and AICcorrected is crucial for accurate adsorption data interpretation.