Carbon-based flexible thermoelectric generators enhanced by dual-elastomer design, AI prediction, and life cycle optimization

The rapid expansion of wearable and portable electronics has intensified the search for compact, sustainable energy sources capable of continuous operation without frequent recharging. Flexible thermoelectric generators (FTEGs), which convert low-grade heat into electricity via the Seebeck effect, offer a promising solution but are often constrained by the limited thermal conductivity of surface materials and mechanical robustness. This work develops an FTEG incorporating graphene, carbon nanotubes, and carbon black into dual-elastomer encapsulation, Ecoflex as the inner layer for compliance, and PDMS as the outer shell for structural stability. Graphene integration delivers the highest thermal conductivity (4.24–5.63 W m−1 K−1) and markedly enhanced tensile strength. Under a 75 °C temperature gradient, the optimized device generates 0.021 W, surpassing the unfilled PDMS control by over three orders of magnitude. Finite element simulations, with a 2.38% deviation from experimental results, confirm uniform heat flow and current density. Substitution of Bi2Te3 with a high-performance MgAgSb/Mg3.2Bi1.5Sb0.5 pair in simulations increased output power by approximately 60%, demonstrating strong potential for next-generation devices. An artificial neural network (ANN) model trained on 189 datasets achieved excellent predictive performance of R2 = 0.9978 and mean error = 1.68%, enabling rapid design evaluation. Life cycle assessment identifies Bi2Te3 and energy-intensive processing as the main contributors to a carbon footprint of 5.57 kg CO2-eq, approximately 89% lower than conventional rigid TEGs. The results demonstrate a combined materials-modeling-sustainability framework that advances FTEGs for wearable electronics, IoT devices, and low-grade heat recovery applications.