Abstract
Nanofiltration (NF) offers a promising low-energy approach for producing decaffeinated instant coffee, but requires membranes that combine high caffeine selectivity, strong permeance, and retention of key coffee matrix compounds. Here, graphene oxide membranes (GOMs) were modified with sodium alginate (SA) and Ca2+ crosslinking to control nanochannel structure and separation performance. The results show that membrane modification does not follow a simple monotonic trend; instead, small structural and chemical changes generate non-linear and competing effects on permeance and selectivity through coupled transport, adsorption, and confinement mechanisms. To identify the most promising membrane, a multi-criteria evaluation framework (MCEF) combining entropy weighting and TOPSIS was established using five key performance indicators. This approach identified CGSM-1 as the best-performing flat membrane, which also showed promising performance in hollow-fibre form. In coffee filtration tests, CGSM-1 reduced caffeine concentration from 24.64 to 12.95 μg/mg while retaining measurable levels of HMF, trigonelline, NMP, and choline. These findings demonstrate the potential of engineered GOMs for selective caffeine removal and scalable decaffeination applications.

This publication has 40 references indexed in Scilit: