Functionalized ZnO TNs will be integrated into a well-defined in-sensor preprocessor that performs time-domain feature extraction and on-device adaptation within the sensing substrate, showing network-level neuromorphic behaviour (nonlinear response, dynamic charge transport, fading-memory effects) rather than isolated synaptic events. The prototype will process spatiotemporal signals for tasks such as anomaly detection, signal transformation, and trend estimation, thereby reducing downstream classifier load and enabling real-time operation at the edge.
To ensure reproducibility and stability, rigorous electrical characterization will be conducted to quantify latency (end-to-end detection delay), energy per update, accuracy parity with a software baseline (e.g., AUROC), dynamic range, and 24-h drift under repeated operation. The tunability window will be mapped via controlled UV and chemical perturbation sweeps where relevant. The potential for scalable fabrication will be explored through additive, low-temperature processes (e.g., spray coating) to deposit functionalized TNs on diverse substrates, including flexible and lightweight materials—consistent with the project’s green electronics principles. The prototype will undergo iterative test–optimize cycles to verify stable operation under defined conditions at TRL 3. Collaboration with industrial partners will inform enclosure, interfacing, and line-side integration, providing early evidence of future scalability and structured transition pathways for ZnO TN–based neuromorphic computing.
