Optimal Bio-Inspired Sniffing for Improved E-Nose Detection

Abstract

It is common knowledge that dogs have a keen sense of smell. Many people have sought to capitalize on such outstanding olfactory sense by using canines with the special forces, as drug detecting tools, and for medical diagnosis [1]. Small mammals such as mice have been shown to quickly learn how to almost perfectly discriminate between two presented odors [2]. On the other size extreme, elephants are extremely adept with their olfactory system and are being used to detect substances at low concentration such as TNT [3]. A key component to the mammalian sense of smell at all size scales is the dynamic sniff cycle. While there has been work done to discover the airflow patterns inside mammalian nasal cavities [4], little is known of the fluid mechanics of gathering and concentrating odors onto sensors. Elucidating the odor collection response of an oscillatory airflow could be applied to greatly enhance the abilities of electronic noses. Blowing fluid past a sensor can enhance signal more than passive diffusion [5]. While constant speed flow might be the easiest to implement, there are inherent advantages in oscillator flow like a sniff. For instance, noise fluctuations can be used as a feature for better identification [6]. Such methods are commonly referred to as fluctuation enhanced sensing (FES) [7]. There have also been other attempts at using bio-inspired sniffing electronic noses which use each sniff cycle as a unique trial [8]. Our central aim is to enhance the amount of information gleaned from inexpensive metal oxide gas sensors by applying a sniffing-inspired flow pattern across the sensor. When metal oxide sensors are exposed to a reducing gas such as ethanol, the resistance of the tin dioxide layer decreases. Accordingly, higher concentrations of ethanol elicit increased resistance changes. The goal becomes bringing analyte to the gas sensors by manipulating the flow so that the sensor experiences the highest concentration and duration of flow as quickly as possible. We hypothesize that using sniffing type will best pre-concentrate air flows. The sniffing airflow successfully gives the sensor information on a shorter time scale than simply diffusion alone. We find the amplitude of the resistance response during oscillation is synchronized to the sniffing frequency, and increases with ethanol concentration. Thus, oscillating the flow profile around a metal oxide sensor can provide an additional level of temporal information which could enhance sensor performance. There is a limit to increasing frequency however. At higher frequencies, the chemical-laden air flows faster near the walls of the chamber where the sensors are arranged. Without sufficient duration near the sensors, decreased sensor resistance response is experienced. Such trade-off leads to an ideal case where a system sniffs quickly enough to utilize the novel temporal information but slowly enough so the signal does not disappear into noise. The novel temporal information provides information on the order of a few seconds whereas the standard steady state feature takes over a minute to retrieve information about the analyte in the sensor’s environment. A device utilizing this new set of information has the potential to recognize an analyte in 20% of the time it takes a standard device without this new information. [1] T. Jezierski et al.,“Efficacy of drug detection by fully-trained police dogs varies by breed, training level, type of drug and search environment,” Forensic Sci. Int., vol. 237, pp.112–118, 2014.[2] N. Bodyak and B. Slotnick, “Performance of Mice in an Automated Olfactometer: Odor Detection, Discrimination and Odor Memory,” Chem. Senses, vol. 24, no. 6, pp.637–645, 1999.[3] A. K. Miller et al., “African elephants (Loxodonta africana) can detect TNT using olfaction: Implications for biosensor application,” Appl. Anim. Behav. Sci., vol. 171, pp.177–183, 2015.[4] M. E. Staymates et al., “Biomimetic Sniffing Improves the Detection Performance of a 3D Printed Nose of a Dog and a Commercial Trace Vapor Detector,” Sci. Rep., vol. 6, p.36876, 2016.[5] Y. Wang, J. Xing, and S. Qian, “Selectivity Enhancement in Electronic Nose Based on an Optimized DQN,” Sensors , vol.17, no.10, Oct. 2017.[7] P. Sedlák, P. Kuberský, and F. Mívalt, “Effect of various flow rate on current fluctuations of amperometric gas sensors.” [8] J. Smulko and M. Trawka, “Gas selectivity enhancement by sampling-and-hold method in resistive gas sensors,” Sens. Actuators B Chem., vol. 219, pp.17–21, Nov. 2015.[9] A. Ziyatdinov, J. Fonollosa, L. Fernández, A. Gutierrez-Gálvez, S. Marco, and A. Perera, “Bioinspired early detection through gas flow modulation in chemo-sensory systems,” Sens. Actuators B Chem., vol. 206, pp. 38–547, Jan. 2015.