An electronic nose is a type of vapor analyzer using an array of dissimilar sensors simulating the human olfactory response [Ref. 1]. A recognizable visual image of specific vapor mixtures (fragrances) is provided by the electronic nose, where vapor mixtures could possibly contain hundreds of different chemical species.
The electronic nose will quantify and characterize the various types of smells, universally. Sensors known for their chemical affinities are selected, and chemisorbing polymer films are typically used for this purpose. Multiple sensors can be employed; Figure 1 shows a histogram of sensor outputs produced by a serial polling of each sensor reading. However, the responses are not correlated, and sometimes several sensors respond to the same vapor e.g. overlap.
Due to this, it is nearly impossible to calibrate this type of electronic nose with test vapors consisting of multiple compounds. Another issue is the sensitivity as the vapor sample being tested by the array should be shared among all sensors in the array.
Figure 1. Histogram produced by serially polling sensor arrays
This article talks about a new type of electronic nose based upon fast chromatography. Sensor space is mathematically defined based on retention time slots. There is a significant improvement in the separation of different compounds. In the ideal or optimal response, there is no occurrence of overlap of sensor outputs with each sensor output corresponding to only one chemical compound or analyte.
Different chemical species have different retention times, and therefore, these vectors can be named as 'carbon tetrachloride’ or ‘cocaine.’ With an electronic nose based upon chromatography, orthogonality of retention time vectors is a function of the temperature profile and column chemical phase.
Fast Gas Chromatography as an Electronic Nose
Figure 2 shows a new fast gas chromatography system employing a surface acoustic wave (SAW) detector [Ref. 2].
The system includes a vapor preconcentrator, heated inlet, temperature-ramped and direct-heated GC column and a SAW detector. Excellent sensitivity is obtained since the SAW detector has pictogram sensitivity, and no dilution of vapor sample takes place. The system inlet can sample desorbed vapors, ambient air, or headspace vapors from liquid samples.
Figure 2. GC/SAW system diagram
When the analytes condense and evaporate on the surface of a quartz crystal, the SAW detector produces a variable frequency.
In Figure 3, the upper trace shows the derivative of frequency (column flux) and produces the familiar peaks of chromatography, and the lower trace shows the frequency histogram. The SAW detector is called an integrating detector since it measures the integral of the chromatogram peaks.
Figure 3. Gas Chromatograms used to create VaporPrint® images
In fast chromatography, peak widths are in milliseconds, the chromatogram duration is less than one minute, and retention time is resolved to within 20 milliseconds. Thus, potentially 500 sensors can be serially polled in less than one minute. The ‘sensor’ responses are almost orthogonal with only a minimal overlap effect. Thus, using a standard chemical mixture, the minimum detection level can be easily determined.
Figure 3 also displays VaporPrints® images [Ref. 3] of the detector frequency and derivative of frequency. These images are formed when the time variable is transformed to a radial angle with the beginning and end of the analysis occurring at 0° or vertical.
VaporPrint® Images and Pattern Recognition
A polar plot of chromatogram time with the radial direction being the sensor signal or the derivative of sensor signal offers a crucial graphical feature well-suited to electronic nose pattern recognition algorithms [Reg. 4, 5].
The SAW frequency corresponds to the total (integral) amount of analyte condensed while the SAW sensor determines the amount of analyte condensing (and evaporating) on a quartz surface. While the SAW crystal is the only integrating GC detector, all others detect the flux of column flow. Retention time is determined by using the derivative of the detector output. The sensor frequency determines the amount of analyte detected.
Figure 4 depicts the process of vapor identification and recognition for the GC/SAW electronic nose, employing mushroom vapors as an example. The vapor is analyzed after five seconds of sampling, and VaporPrint® images and chemical chromatograms are formed. The status of selected sensor alarms is posted in less than one minute time.
Figure 4. Pattern Recognition Process including the setting of alarms on individual sensor elements
The system has the ability to form a sensor array with alarms, and this feature enables monitoring of only specific analytes of interest, e.g.. explosives or drugs of abuse. For the mushroom example, five sulfur compounds were selected as the five sensor vectors that have to be monitored. Then, a simple array of sensor meters is displayed, representing a mushroom nose. Through a simple software click, another set of sensors can be loaded to obtain a marijuana nose.
A number of electronic nose pattern recognition algorithms were evaluated, based upon sliding sets of correlations employing known compound patterns related with complex fragrances. The aim was to determine the best "pattern recognition algorithm". So far, nothing has been found which can approach the performance of a human operator.
Figure 5 shows the situation, demonstrated in the pictures of plant leaves. A human can easily and immediately identify the marijuana leaf, while the task can be long, daunting and tedious for computation algorithms. For similar reasons, only trained humans are deployed as screeners at airport security checkpoints.
Figure 5. The human is an optimal pattern recognizer
However, humans must be trained to recognize VaporPrint® patterns, since they excel if they are properly trained.
VaporPrint® images became richer with more volatile and complex compounds as found in perfumes and food items. Figure 6 shows some representative images from drugs of abuse, infectious bacteria and flammable fuels. It has been proven that human operators can recognize food smells or certain images because they resembled common shapes.
Figure 6. VaporPrints® for common infectious diseases, explosives, drugs of abuse and Flammable Fuels
Law Enforcement Applications and Requirements
Specific detectors are employed by traditional drug and bomb detectors, designed to filter out interfering substances and obtain a YES or NO detection of specific chemicals under diverse conditions. However, an electronic nose is designed to quantify and characterize various types of smells universally, including those from drugs of abuse and explosives.
Law enforcement agencies make use of this universality feature for a wide diversity of applications. As an instance of general drug enforcement, Table 1, downloaded from the web site of the Los Angeles police narcotics department [Ref. 6], shows the various compounds that need to be detected.
Table I. List of Illegal Drugs
|Inhaled Household Products
Other than the requirements of narcotics officers, there are also needs for forensic experts, crime scene investigators, bomb squad officers and general health and safety officers. Additionally, there are also specialized United States (US) government agencies such as the Department of Defense and the US Customs. The US Customs has a list of contraband covering more than four pages, and their number is not shrinking. The use of electronic nose for the military services is unlimited.
Training is imparted to law enforcement officers on situational awareness, and in this respect, the electronic nose enhances the officer’s ability. The usefulness of recognizing cigarette smoke and that of the illegal drug marijuana can be considered as an example.
While the two odors are complex, their VaporPrint® images are noticeably different and easily recognizable. The capability to rapidly evaluate the odors present in a given situation could give the enforcement officers other information such as that open alcoholic substances were being consumed or that a weapon had recently been fired.
Figure 7. Identify and discriminate between different sources of smoke.
However, cost and performance has been the main barrier in the use of an electronic nose. Law enforcement officers are not inclined to deal with complex instruments and even more complicated software. For an instrument to be useful, its cost must fall within the budget of the enforcement agencies.
Previous sensor array systems were expensive, slow, frequently unstable over time, and not specific, and did not possess sufficient sensitivity. In contrast, the operation time for GC/SAW nose is less than one minute, and the instrument has a high degree of specificity and sensitivity. The electronic nose is stable, and utilizes a low-cost solid state sensor technology and retains calibration over time.
Figure 8. Commercially available Electronic Nose
Commercial Availability of Electronic Nose
The Electronic Sensor Technology, Inc. (EST), Newbury Park, CA, began the first commercial production of a GC/SAW electronic nose or chemical vapor analyzer in 1997. This electronic nose performed VaporPrint® imaging, flash chromatography, and sensor arrays with user-defined alarm levels. Easily operated with a single hand, this instrument delivered chromatograms in less than one minute, employing a patented SAW sensor.
The electronic nose can be operated in the laboratory as well as on site. The instrument is robust in construction and stable over time, and can be operated in a temperature range of 0 °C-40 °C. The system is equipped with a Pentium computer with pre-installed PCAnywhere® for remote operation and Office97® software. Add-ons such as low-cost GPS receivers to accurately record the location of each measurement are available as options.
Figure 9. Laboratory style Fast GC/SAW
For environmental monitoring of PCBs in soil and VOCs in water, the GC/SAW electronic nose is the first to receive validation from the US EPA. Moreover, the White House Office of National Drug Control Policy (ONDCP) has validated this instrument’s performance to detect heroin and cocaine vapors. For US government and law enforcement efforts involving drug interdiction, the GC/SAW electronic nose can be bought under Section 1122 from the Government Services Administration (GSA).
In 1998, EST launched a portable, bench top version of the GC/SAW electronic nose, for the benefit of forensic experts. Parts per trillion (ppt) sensitivity is obtained for semi-volatile compounds, and parts per billion (ppb) sensitivity for volatile compounds. The electronic nose is operable over a 1,000,000 to 1 dynamic range of vapor concentrations, since the GC/SAW employs a variable sample time preconcentrator with electronically variable sensitivity in the SAW sensor.
The commercially-available electronic nose is a new investigative tool in the hands of law enforcement officers. The electronic nose can provide a visually recognizable image of specific vapor mixtures (fragrances) consisting of possibly hundreds of different chemical species. The new electronic nose is quick in operation (operation time is less than one minute), and operable over a wide range of vapor concentrations.
It is simple to use and calibrate, and has picogram sensitivity. The instrument can detect as well as measure many different, and sometimes, complex fragrances. This is accomplished by employing pattern recognition and a visual fragrance pattern, called a VaporPrint® derived from an integrating solid-state detector. A VaporPrint® image enables viewing and recognizing a complex ambient environment as part of a previously-learned image set. Quick assessment of unknown vapor or smell is possible by using the ability of the law enforcement officer to recognize visual patterns.
The sensitivity and speed of the electronic nose has been validated for drugs of abuse by the ONDCP and for environmental monitoring by the US EPA. Moreover, the new electronic nose is GSA-listed, and approved for State and Local Law Enforcement use under the ONDCP/GSA “1122” program. The instrument can be made to adapt and learn to recognize new vapors, thus resulting in a new useful tool for forensic laboratories, US customs inspections and many other Federal and State law enforcement agencies.
- H.T. Nagle, S. Schiffman and R. Guitierrez-Osuna, “The How and Why of Electronic Noses”, IEEE Spectrum, pg. 22-33, September 1998.
- E. J. Staples, “Dioxin/Furan Detection and Analysis Using A SAW Based Electronic Nose”, Proceedings of the 1998 IEEE International Ultrasonics Symposium, October 1998, Sendai, Japan.
- E. J. Staples, Method and Apparatus for Analyzing Vapor Elements, U.S. Patent Pending
- P. Keller, R.T. Kouzes, L.J. Kangas, “Three Neural Network Bassed Sensor Sstem for Environmental Monitoring”, Proceedings IEEE Electro94 Conference, Boston, MA, USA, 10-12 May 1994.
- J. Gardner and P. Bartlett, Electronic Noses: Principles and Applications, Oxford University Press, November 1998.
This information has been sourced, reviewed and adapted from materials provided by Electronic Sensor Technology.
For more information on this source, please visit Electronic Sensor Technology.