Most gas sensors can only detect a single gas species (e.g., a smoke detector). Some more advanced gas sensors called "electronic noses" are able to distinguish among a few gas species in a gas mixture. However, there is still no "universal" gas sensor, or a single device that could detect a very broad spectrum of gases under a wide range of operating conditions. This project seeks develop a preliminary design of a such a device using computational modeling and simulations. A portable universal gas sensor would have enormous socioeconomic impact, similar in magnitude to the digital camera, and would revolutionize health diagnostics. The investigator plans to create a roadmap for the first universal gas sensing device whose capabilities could match, and potentially surpass, those of a dog's nose. The approach will use molecular simulations to accurately model the behavior of volatile organic compounds (VOCs). Also, large-scale computational optimization will be used to search the vast combinatorial space of possible sensing arrays. The primary outcome of the proposed research will be a computationally-derived roadmap for future "electronic nose" research. The proposed research is expected to have a significant educational impact via the production of high-quality scientific movies for both undergraduates and the broader public. The investigator will produce a movie that will cover the complex mixtures of VOCs emitted from the human body, touching on the many engineering applications that can benefit from large-scale computational/simulation approaches.
The overall objective of the proposed research is to develop a "computational road map" guiding experimental design of a universal gas sensor. It has been shown that trained dogs can identify various illnesses through smell along, indicating that detection of diseases via gas sensing is possible. However, prior "electronic noses" have been unable to compete with canine olfactory abilities because researchers have been narrowly focused on gas sensing devices with arrays of 20-30 elements (or less). A dog's nose, on the other hand, has 300 million olfactory receptors, with corresponding receptor types numbering in the thousands. It is unlikely that an electronic nose with less than 30 sensing elements can match a dog's nose any more than a digital camera with less than 30 pixels can match a human eye. Moreover, prior work has been entirely experimentally driven - computational modeling has only been used to analyze the data generated after the fact. Designing large arrays (100-1000+ elements) presents a combinatorial challenge and hence is a classic "big data" problem, which can only be solved using computational data science and engineering methods. The investigator will use grand canonical Monte Carlo simulations to model complex gas mixture adsorption in arrays of metal-organic frameworks (MOFs), which are crystalline nanoporous materials. By computationally exploring different combinations of MOFs, the investigator will then determine, in principle, what kind of array could match the sensing performance benchmarks of a biological nose. The project will examine fundamental questing including: how does performance scale with array size; at what sizes does one cross benchmark olfactory performance of animals; and what material characteristics have the greatest influence on sensing performance?
This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.
