Microfluidic synthesis of functional nanomaterials for artificial olfactory sensing
Biological sensory perception happens as a result of processing various stimuli that are detected at a specific sensory body depending on a type of stimuli. Such processing through the sensory bodies is categorized into five “senses” which are well-known as the sense of sight, hearing, touch, taste, and smell. As technology advances, most of the senses have been successfully mimicked, and even commercialized in the form of camera, microphone, pressure sensor, and electronic tongue. However, it is still a grand challenge to realize practically useful artificial olfaction mainly because of extreme diversity and transient feature of odors, while there are a variety of potential applications for artificial olfaction.
One of the promising ideas for artificial olfaction would be obtained by looking at how the actual olfaction system works to discriminate odors. Extensive studies have revealed that there are several hundred types of olfactory receptors in our nose, and they interact differently with the constituents of odors so that a specific pattern is formed. We learn such patterns, and discriminate what an odor is by referring to our own library. To mimic this mechanism, multiple sensors where each of them has different surface property are aligned. Such a system is so-called a sensor array and utilized for odor discrimination.
We have been working on the development of sensor arrays and their applications for odor analysis. To realize an artificial olfaction system, there are various elements to be optimized from microscopic to macroscopic point of view―receptor materials, sensor geometry, analyte introduction/release, and data acquisition/analysis. Among them, receptor materials are crucial to discriminate a wide variety of odors which frequently have slight difference. Roughly, following conditions need to be fulfilled: high sensitivity, tunable selectivity, chemical stability, and humidity resistance. We have reported that inorganic oxide-based functional nanoparticles were useful for this purpose. Sequential flow reaction using microchannels was employed to synthesize silica-coated titania nanoparticles with various surface functionalities (Figure 1).1-3
Figure 1 Overview of the sequential flow reaction for functional nanoparticle synthesis (Chem. Commun. (2015))
Since these nanoparticles are mainly composed of silica, they are chemically robust enough at least in gas sensing, meaning that reuse and long-term use are possible. Furthermore, these nanoparticles showed ppm or ppb level sensitivity to a variety of volatile organic compounds. Taking advantage of these nanoparticles, we have demonstrated some applications which are not limited to discrimination but extended to even quantification of odors. Specifically, the composition of ternary mixture composed of water, methanol, and ethanol was determined successfully just based on its odor (Figure 2).2
Figure 2 Odors containing water, methanol, and ethanol were quantitatively analyzed by a data-driven sensing approach combined with systematic material design (ACS Sens. (2018)).
In this study, a machine learning model was trained with dataset beforehand, and then the composition of a few sample odors was predicted. Interestingly, alcohol content (%) of liquors was also predicted simply with their odors which usually contain many volatile compounds other than water and ethanol (Figure 3).3 These examples seem to have following important implications: successful odor sensing in humidified condition (we are always in a humidified environment), and specific information extraction (e.g. concentration) from odors. Considering real-world applications for us now and in the future, such remarkable accomplishments are appreciated.
Figure 3. Schematic of the alcohol content quantification in combination with various nanoparticles-functionalized sensor array and machine learning (Sci. Rep. (2017)).
More and more materials, which are basically hydrophobic but have a different affinity for analytes, are required to achieve a versatile olfactory sensor. We are now trying to prepare various materials with controlled size, shape, and structure by means of microfluidic techniques.
References
- K. Shiba, T. Sugiyama, T. Takei, G. Yoshikawa, Chem. Commun. 51, 15854-15857 (2015).
- K. Shiba, R. Tamura, T. Sugiyama, Y. Kameyama, K. Koda, E. Sakon, K. Minami, H. T. Ngo, G. Imamura, K. Tsuda, G. Yoshikawa, ACS Sens. 3, 1592-1600 (2018).
- K. Shiba, R. Tamura, G. Imamura, G. Yoshikawa, Sci. Rep. 7, 3661 (2017)
Contact: kshiba@seas.harvard.edu