Detection of fruit odors using an electronic nose
The electronic nose is a device that uses an array of chemical sensors (usually 4 to 32) tied to a data-processing system that mimics the way a nose works. The sensors each produce independent responses to the different chemical elements within a given sample, and the sum of all the sensors' responses is a ‘pattern’ corresponding to that odor. To identify an unknown sample, the data-processing system compares its responses to a library of previously-measured patterns.
The detection capability required for a sensor to identify the presence of contraband is an extremely complex issue. Two primary factors governing the odor source are essential: it must emit chemical signals (odors); and it should have enough of the chemical signal to reach some level of equilibrium (steady state) within the package. Since such equilibriums are usually attained in reasonably short periods of time (hours), it can be assumed that the latter parameter is, in most instances, met. The emissions themselves, however, are the key to detection since the nature of the various emissions released from the source will determine what chemical signals are ultimately available for detection.
The typical electronic nose can detect most odors in the mid to low parts-per-million (volume for volume) range: low ppm (v/v). Unfortunately, however, many applications require detection in the high to mid ppb (v/v), so we would like to see an increase in sensitivity of 10 to 100. To this end, we have focused on evaluating the abilities of two commercially-available sensor array instruments—the Cyranose 320 and Airsense Portable electronic noses—to detect and characterize a diverse set of odors that included fruit, various solvents, and pure chemicals including volatile organic compounds (VOCs). In certain cases, the chemical analytes were diluted to very low concentrations in order to estimate sensor-array detecting limits. Then, a thermal desorption unit (the Airsense Enrichment Desorption Unit 2) was coupled to the arrays to determine degrees of detection-limit reduction and signal enhancement possible for the sensor arrays when challenged to detect and characterize odors.
Dilute concentrations of limonene, ethyl acetate, and diisopentyl sulfide (DIPS) were prepared in order to determine the approximate analyte concentration at which the electronic noses could no longer accurately detect or identify the chemical odor. Visual estimations based on instrument-generated smell prints of the low concentrations approximately define the detection limits for the instrument. The dilute samples were prepared by injecting specific quantities of pure chemical into a 2.4L round-bottomed flask equipped with a septum cap suitable for sampling via a tube. Approximate concentration detection limits are estimated for each analyte in Table 1.
Bananas were used as a ‘real world’ test of the utility of the pre-concentration to aid the electronic nose in identifying an odor. As shown in Figure 1, the pre-concentration greatly increases the odor response and, in this case, allowed unambiguous identification of the banana odor.
The data suggests that the unaided electronic noses are enhanced by odor pre-concentration to various degrees, depending on the particular analyte in question. Therefore, we can conclude that in order to achieve maximum sensor responses to the fruit and chemical odors included in this study, the electronic nose should be enhanced in this way. Odor pre-concentration expands the working concentration range for the electronic nose, thereby increasing its overall usefulness and viability in real-world sensing applications.
