Explosives Detection with Nanofabricated Chemical Sensor Arrays and Machine Learning
Abstract
Approved for Public ReleaseWe propose an experimental research program to investigate the fabrication and testing of nanoscale chemi,cal sensor arrays for explosives detection. Detection and identification of hidden/ buried explosives as well as chemical precursors, for explosives manufacture is imperative for protecting armed forces, peacekeepers, and populations living in conflict zones. Mine,detection dogs are often used for these purposes, but the training, resting, and maintenance of canines limits their application in,the field. Moreover, handler ? dog interactions are complex and limit operation scale. There is a need for low cost portable detecti,on systems that can be deployed in the field with minimal training. Nanofabricated devices combined with integrated circuits offer a, means for producing low cost, low power, portable electronic sensor devices that can be widely used to identify explosive materials, through discrimination of chemical properties. The critical components to such systems are the sensor elements where chemical prope,rties are transduced into electrical signals. Our previous work on this topic has shown that large arrays of chemiresistor elements,combined with machine learning are promising to detect and identify explosive materials. Machine learning benefits from large data s,treams produced by large arrays of detector elements fabricated as part of integrated sensor chips. Each sensor element of an array,interacts at the molecular level with volatiles (odors) produced by explosive materials and residues. Chemical diversity engineered,into the arrays and the combined action of many sensor elements produces signal patterns that are analyzed by machine learning to id,entify chemicals. Our technical approach is to fabricate sensor chips using nanofabrication methods and integrate them with gas-snif,fer flow cells as prototypes for explosive detection devices. We will expand the number of detector elements to > 100 and the divers,ity of sensor materials to ? 10 on each sensor chip, which will enhance the ability of the arrays to accurately identify explosive m,aterials based on their volatile emissions. The objective is to create prototypes that can be tested using vapor generators calibrat,ed and certified by researchers at the Naval Research Laboratories. Our prior work has shown promising results using commercially av,ailable NESTT scent kits for explosives detection training of canines. The outcome of this research project will be to determine if,our sensor approach can function under conditions relevant to field operation and using calibrated sources operated by NRL staff. Th,e potential impact for DOD is to have low cost, scalable sensor technologies for explosives detection that could augment and/or repl,ace current operations using detection dogs.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jul 13, 2022
- Source ID
- N000142212567
Entities
People
- Brian G. Willis
Organizations
- Office of Naval Research
- United States Navy
- University of Connecticut