SafeDrop kicks off — a digital dog nose for on-site diagnostics

SafeDrop sets out to emulate mammalian olfaction: a “digital dog nose” that fuses silicon-photonic interferometers with thin-film transistors in co-localised K-cells, giving synchronous optical and electrical readouts from the same sensing site. Thirty-two K-cells carry cross-responsive organic semiconductor coatings, and machine learning decodes the resulting high-dimensional fingerprints.

The point of the dual readout is orthogonality. A photonic interferometer and a transistor channel respond to the same coating in different ways, so two independent signatures are available where most sensors offer one — which is what makes the target of better than 99% classification accuracy and better than 95% false-positive reduction plausible.

The intended applications are all cases where results are currently measured in days rather than minutes: heavy metals and pathogens in water at parts-per-billion levels, point-of-care urinary tract infection diagnostics, and pesticides, toxins and spoilage in food. Over four years the project moves from proof of concept at TRL 3 to validated demonstrators at TRL 7, targeting under one watt of power and a cartridge cost of ten euros or less, with an open design kit and standards contributions alongside.

Flowphys contributes the simulation and optimisation layer: two-phase microfluidics modelling for the pump-free cartridge, design of experiments and surrogate models for sensor design, multi-objective optimisation with uncertainty quantification, and techno-economic analysis. The microfluidic work draws on solver capability recently extended down to sub-millimetre scales.

SafeDrop is coordinated by the Fraunhofer Institute for Applied Polymer Research and brings together ten partners across seven European countries.

Links: safedrop-project.eu, Expertise: electromagnetics and photonics, Expertise: optimisation and UQ.