Modeling Electronic Olfaction1

Caltech's electronic nose consists of an array of 20 electrically conducting carbon black loaded polymers, all fitted within a small space. The carbon black gives the sensors conducting properties; the polymer matrix provides sensitivity to vapor bound compounds (odorants). Odorant caused swelling of the polymer decreases the number of conducting pathways through the carbon black percolation network, increasing its resistivity. The change in resistivity is linearly related to the concentration of the various odorants. Each of the sensors is broadly tuned to a given region in solubility space. By combining the responses of all sensors with principal component analysis Caltechís electronic nose is capable of resolving specific odorants form complex mixtures, and provide an indication of their concentration at levels superior than the human nose. Combined with robotics the electronic nose is capable of imparting directional control to a robot even in the presence of a mild wind. Search and target acquisition strategies have been modeled after the behavior of rats in control laboratory experiments. The MSC has succeeded in developing a molecular based model of electronic olfaction useful for the design of new polymer sensors for specific odorants. (Sensors: Nate Lewis, robotics: Rod Goodman, Behavioral Biology: Jim Bower, Modeling: William A. Goddard III).

Figure 1. A conducting carbon black loaded polymer swells with an odorant changing its resistivity.

DR/R = Ro exp(-DEk/RT)

exp(- Sbi [Hi(k) - Hi(n)])

Where DR/R is the change in resistivity upon exposing polymer sensor n to odorant k, Ro is sensorís resistivity in air, DEk is a diffusion barrier proportional to the molecular volume of the odorant, Hi(k) and Hi(n) respectively the three components of odorant and polymer solubility parameters (electrostatic, dispersion, and hydrogen bond). These components are calculated using first principles molecular dynamics (MD) in a suitable force field2.

1.- M. Blanco, M. Belmares, N. Lewis, W.A. Goddard III, in preparation

2.- S. L. Mayo, B. D. Olafson, and W. A. Goddard III J. Phys. Chem. 94, 8897 (1990)

Figure 2. MD condensed phase yields the three component solubility parameters Hi for polymer and odorant molecules.

Figure 3. Theory vs. measured changes in resistivity for seven polymeric sensors and 24 odorant molecules.