There is a need in the Homeland Security Enterprise for small, lightweight, and low-cost chemical sensors for many real-world applications. One important aspect is ensuring food protection against both accidental and intentional chemical contamination. Factors such as production batch size, shelf life, and quality control procedures already in place could affect the number of people impacted. Emerging sensing technologies and data analytic tools from the fields of nanomaterials and machine learning provide an opportunity for low-cost microsensors that could be widely distributed to provide onsite and real-time awareness of contamination events. To survey the challenges around this objective, a functionalized carbon nanotube-based set of sensors were evaluated for the ability to identify chemical vapors and detect contamination in complex mixtures in common food matrices such as apple juice. It was found that the detector could identify pure chemicals in dry lab air as well as contamination that was present in the headspace over food samples. Technical challenges were identified, with the most significant being variable signature responses between the three different identically configured detectors. Strategies for mitigation of sensor variability were evaluated, including machine learning techniques as well as sensor calibration procedures.
|