TOPIC 3:
NATURAL CHEM/BIO TAGS

Two presentations were made in this session, one by Alan Gelperin, Monell Chemical Senses Center, and one by Steven Martin, Sandia National Laboratories.

DESIGNING CHEMICAL SENSOR SYSTEMS FOR ELECTRONIC OLFACTION

Alan Gelperin reviewed recent progress in electronic olfaction technology based on biological models. For example, moths are hypersensitive to a few specific compounds (e.g., pheromones), while a dog’s nose has more general sensitivity. The general-purpose nose can be trained to detect new odors, such as TNT. The artificial nose requires an array of odor sensors, with diverse odor responses (there are some 1,000 different odor receptor classes in mice, 300 in humans), and a computational module for analyzing odor patterns. Hopfield published a paper3 showing that a larger number of different sensor classes in an array (up to ~100) gives a different and richer response than an array with a smaller number of sensor classes. Gelperin felt that an algorithm developed by Hopfield is quietly revolutionizing this field. The algorithm allows a system to recognize a new odor pattern in terms of known odor patterns.

Gelperin focused on organic field effect transistors in which the odor vapor is flowed over a chemically active organic layer between the source and drain of a transistor, and the degree of interaction between the odorant molecules and the active layer is reflected by changes in the current flow. This system has the advantage that the odor can be driven out (to reset the sensor) by reversing the gate voltage rather than having to flow fresh air over the sensor. The organic surface layer should be as thin as possible to maximize the influence of the surface. Another configuration demonstrated for the detection of O2 and CO gases uses changes in current flow through carbon nanotube wires (or nanowires made of other materials) as the sensor.

Special challenges of these systems include the following:

  • Ensuring that the identification of the odor does not depend on concentration;

  • Separating odor “objects” (multiple odors that arrive together);

  • Identifying weak known odors against a background of strong unknown odors;

  • Storing odor patterns for later pattern matches; and

  • Subtracting constant background odors while remaining sensitive to new weak odor inputs.4

3  

J.J.Hopfield 1999. Odor space and olfactory processing: Collective algorithms and neural implementation, Proceedings of the National Academy of Sciences 96 (22):12506–12511.

4  

For example, dogs have to be trained on local background odors for about 2 weeks before they are used to detect land mines in a given region.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 11
Summary of the Sensing and Positioning Technology Workshop of the Committee on Nanotechnology for the Intelligence Community: Interim Report TOPIC 3: NATURAL CHEM/BIO TAGS Two presentations were made in this session, one by Alan Gelperin, Monell Chemical Senses Center, and one by Steven Martin, Sandia National Laboratories. DESIGNING CHEMICAL SENSOR SYSTEMS FOR ELECTRONIC OLFACTION Alan Gelperin reviewed recent progress in electronic olfaction technology based on biological models. For example, moths are hypersensitive to a few specific compounds (e.g., pheromones), while a dog’s nose has more general sensitivity. The general-purpose nose can be trained to detect new odors, such as TNT. The artificial nose requires an array of odor sensors, with diverse odor responses (there are some 1,000 different odor receptor classes in mice, 300 in humans), and a computational module for analyzing odor patterns. Hopfield published a paper3 showing that a larger number of different sensor classes in an array (up to ~100) gives a different and richer response than an array with a smaller number of sensor classes. Gelperin felt that an algorithm developed by Hopfield is quietly revolutionizing this field. The algorithm allows a system to recognize a new odor pattern in terms of known odor patterns. Gelperin focused on organic field effect transistors in which the odor vapor is flowed over a chemically active organic layer between the source and drain of a transistor, and the degree of interaction between the odorant molecules and the active layer is reflected by changes in the current flow. This system has the advantage that the odor can be driven out (to reset the sensor) by reversing the gate voltage rather than having to flow fresh air over the sensor. The organic surface layer should be as thin as possible to maximize the influence of the surface. Another configuration demonstrated for the detection of O2 and CO gases uses changes in current flow through carbon nanotube wires (or nanowires made of other materials) as the sensor. Special challenges of these systems include the following: Ensuring that the identification of the odor does not depend on concentration; Separating odor “objects” (multiple odors that arrive together); Identifying weak known odors against a background of strong unknown odors; Storing odor patterns for later pattern matches; and Subtracting constant background odors while remaining sensitive to new weak odor inputs.4 3   J.J.Hopfield 1999. Odor space and olfactory processing: Collective algorithms and neural implementation, Proceedings of the National Academy of Sciences 96 (22):12506–12511. 4   For example, dogs have to be trained on local background odors for about 2 weeks before they are used to detect land mines in a given region.

OCR for page 11
Summary of the Sensing and Positioning Technology Workshop of the Committee on Nanotechnology for the Intelligence Community: Interim Report An optimal system in the future would have sensor arrays with 50–100 unique sensor types. The array would be small enough to allow one-sniff coverage and would be cheap and easy to replace. It would have low power consumption, interface easily with processing electronics, and have enough processing power to implement fast, new pattern analysis algorithms. Detector replacement is necessary because, like natural olfactory cells, sensor arrays will eventually become poisoned with use. To use artificial nose technology for tracking, robots are being built that can follow a scent using search algorithms inspired by the pheromone-following strategy of a moth—i.e., moving back and forth perpendicular to the wind direction to pick up the scent, then moving upwind toward the source. MICROSENSOR DEVELOPMENT Steven Martin discussed the miniaturization of analytical devices, focusing on the MicroChemLab, a handheld gas chromatograph (GC) that can detect chemical warfare agents or other vapors of interest. Microfabricated components such as sample concentrators, analyte separators, and detectors offer small size, low power, low sample volumes, few (or no) reagents, and rapid analysis (e.g., 2 minutes for MicroChemLab). Disadvantages include high initial costs, less versatility than full-size instruments, less sensitivity, and lower resolution. MicroChemLab uses three microfabricated analysis stages: a sample preconcentrator, a micro gas chromatography stage, and a surface acoustic wave (SAW) detector to provide sensitive detection. The preconcentrator and the SAW detector have nanostructured surfaces. The chromatograph stage uses a thin-film sol-gel coating on the column surface that has tailored porosity to provide separation of analyte chemicals. A system can be designed with multiple parallel GC columns to confirm the analysis and reduce the false alarm rate. Various kinds of detectors (other than the standard SAW detector) are possible. In principle, it is possible to use the MicroChemLab as a reader for a signal encoded as a chemical mixture. The retention time window is used to form individual bits: An eluted peak within a given window is a 1; no peak is a 0. If detectors with different specificities are used, further information can be encoded. PANEL 3 DISCUSSION Initial discussion focused on the information requirements for electronic noses. These include training, pattern recognition algorithms, a database for matching patterns, and an algorithm to determine the statistics of matching. Gelperin mentioned an experiment in which an artificial nose was tested to see if it could identify grocery store produce by its odor. The rapidity of response is critical—the system must identify the odor within about 2 seconds. In addition to speed, reliability and effective background subtraction are key features. There are trade-offs to be made among these characteristics. Gelperin indicated that the prototype device had trouble distinguishing between some types of produce. A question was raised as to whether these artificial systems really mimic natural systems. For example, the eye does not raster over a pattern—it processes certain recognition elements. Do we understand this preprocessing step for olfaction? Gelperin agreed, but felt that the Hopfield model is a good place to start. It was commented that one should focus not just on the size of the array but also on the manner in which the sensors are clustered and on the time derivative of their responses. Another question was raised about whether we understand the vomeronasal organ in mice, where neurons respond to specific pheromones. If we can design an artificial nose whose sensitivity is specific to a given compound, such as TNT, then our analysis is simplified.

OCR for page 11
Summary of the Sensing and Positioning Technology Workshop of the Committee on Nanotechnology for the Intelligence Community: Interim Report Can we distinguish one human from another on the basis of scent? There is ongoing research to determine what molecules are different from person to person and what level of difference can be identified. Laboratory tests have been able to distinguish differences in the scents of fraternal twins but not identical twins. Quantitative studies on sensitivity and recognition are very recent. It was noted that natural systems (e.g., a dog’s nose) use cascade amplification mechanisms to increase sensitivity; do we need similar amplification in artificial noses? Optically based sensor systems are already within a factor of 10 of the sensitivity of a dog’s nose. We would like amplification, but we are getting close to the sensitivity of natural systems without it. Some researchers are looking at ion channel amplification as one mechanism for improving sensitivity.

OCR for page 11
Summary of the Sensing and Positioning Technology Workshop of the Committee on Nanotechnology for the Intelligence Community: Interim Report This page intentionally left blank.