Data-driven prediction and interpretation of paracetamol and amoxicillin adsorption onto<i> Cordia</i><i> myxa</i>-derived activated carbon
DIAMOND AND RELATED MATERIALS, cilt.168, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 168
- Basım Tarihi: 2026
- Doi Numarası: 10.1016/j.diamond.2026.113899
- Dergi Adı: DIAMOND AND RELATED MATERIALS
- Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Chemical Abstracts Core, Chimica, Compendex, INSPEC, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO)
- Ondokuz Mayıs Üniversitesi Adresli: Evet
Özet
Pharmaceutical residues such as paracetamol (PCM) and amoxicillin (AMX) persist in aquatic environments due to their incomplete removal in conventional wastewater treatment plants. In this study, a sustainable Cordia myxa-derived activated carbon (CMAC) was produced via NaOH activation and comparatively evaluated against commercial activated carbon (CAC). Batch experiments were conducted under varying pH (3-11), adsorbent dosage (0.25-2.0 g L-1), contact time (5-240 min), initial concentration, stirring speed, and temperature. The synthesized CMAC exhibited a higher surface area and more developed mesoporous structure than CAC. Kinetic modelling showed that pseudo-second-order and Elovich models adequately described the adsorption behaviour, highlighting the importance of surface-related adsorption processes. Isotherm results revealed a significantly higher Langmuir capacity for PCM on CMAC (qmax = 334.25 mg g-1), whereas AMX showed higher capacity on CAC (qmax = 202.82 mg g-1), demonstrating adsorbent-adsorbate selectivity. Thermodynamic analysis showed PCM adsorption was spontaneous and exothermic, while AMX adsorption on CMAC was endothermic and favored at higher temperatures. Machine learning models demonstrated strong predictive capability for adsorption capacity prediction, achieving R2OOF values of 0.941 for PCM and 0.891 for AMX. SHAP analysis identified initial concentration, carbon type, and dosage as key predictors. The integrated experimental-ML framework enables accurate prediction and mechanistic insight for adsorption-based water treatment.