- Santos-Prieto, F. de los
- Eric, Deborah
- Portí, Marc
- Martín-Martínez, Javier
- Roca, Elisenda
- Ramon, Eloi
- Ogier, Simon
- Nafría, Montserrat
- Fernández, Francisco V.
- Castro-López, Rafael
Organic thin-film transistors (OTFTs) are attractive for low-cost printed electronics, and they can also enable lightweight hardware-security primitives such as Physical Unclonable Functions (PUFs). In differential OTFT-PUFs, time-zero mismatches between pairs of nominally identical transistors provide the entropy. However, reliability detractors like off-time aging, thermal annealing, electrical stress, or circuit non-idealities, induce bit flips, bias, and non-random outputs. This paper proposes a reliability-aware transistor-pairing strategy, where a genetic algorithm (GA) searches the combinatorial transistor pairing space and selects those maximizing reliability under multiple detractors while maintaining near-ideal PUF behavior. A fabricated OTFT array is first characterized at several bias conditions before and after controlled stresses, establishing baseline behavior and calibrating stochastic drift models. These models can generate large synthetic populations enabling systematic exploration of biasing, circuit non-idealities and GA hyperparameters, while also exposing the limits of fresh-only stability rules. The resulting guidelines are then applied to the fabricated array, where GA-selected pairings yield high reliability while maintaining near-ideal unpredictability and uniqueness. For practical deployment, this work shows that stressed measurements are only needed once to calibrate the model, after which chip-specific pairing can be successfully optimized pre-deployment from fresh measurements alone by running the GA on model-predicted degradation., This work has been supported by grants PID2022–136949OB-C21 and PID2022–136949OB-C22 funded by MICIU/AEI/10.13039/501100011033 and by “ERDF/EU”. This work has been also funded by grant ProyExcel_00536 funded by Consejería de Universidad, Investigación e Innovación of Junta de Andalucía and by grant USECHIP (TSI–069100–2023–001), project funded by the Secretary of State for Telecommunications and Digital Infrastructure, Ministry for Digital Transformation and Civil Service and by the European Union-NextGenerationEU/PRTR . A preliminary version of this work was presented at the 2025 edition of the International Conference on Synthesis, Modeling, Analysis and Simulation Methods, and Applications to Circuit Design (SMACD)., Peer reviewed