Mimicking biological olfaction with very large chemical arrays

Olfactory receptor neurons in the mammalian olfactory system transduce the odor information into electrical signals and project these into the olfactory bulb. In the biological system the huge redundancy and the massive conver- gence of the olfactory receptor neurons to the olfactory bulb are thought to enhance the sensitivity and selectivity of the system. To explore this concept, modular polymeric chemical sensor array consisting of up to 65,536 sensors belonging to tens of different types was built. To interface such a large sen- sor array, a topological array configuration with n rows and m columns was adopted in order to reduce the number of wiring connections, and 16 subarray modules are simultaneously read to make the acquisition speed comparable to biological olfaction. The matrix array configuration is not without conse- quence. It introduces electrical cross-talk which makes the response of each sensor dependent also on the responses of other sensors in the array. An analy- sis of the electrical cross-talk and the solution adopted to compensate for it are here presented together with some discussions on advantages and limitations of the scanning algorithm. Preliminary experiments were carried out on the prototype to test its capability in terms of odor discrimination, classification, and binary mixtures segmentation. The data were analyzed using multivariate methods and the results indicate that the large scale array developed can emu- late several aspects found in biological olfaction.