Electronic nose vapor sensors are printed arrays of transistors that can detect ambient chemicals and odors and then alert a consumer if the internal contents of a medicine bottle or bottle of wine have changed. Thus, if acetic acid has begun to build, sensors will send a signal that the wine is going sour before you inadvertently give it the thumbs up at tasting.
One of the key features is that the sensors should be cheap to manufacture, allowing them to be inserted into wine bottles. An integrated nose--which would include a sensor array and a silicon chip--could cost a few dimes to manufacture now and drop to less than 5 cents over time. A nose based on all printed semiconductors would be even cheaper and could be feasible over time.
All-silicon noses exist now, but they cost hundreds of dollars.
"Our ultimate goal is to go to fully printed systems," Vivek Subramanian, a professor of electronic engineering at UC Berkeley, said during a presentation at thetaking place in San Francisco. "Wine spoilage is a pretty good application because there are few false positives."
The sensors work by interacting with the environment. When odors or chemicals come into contact with the surface of the sensors, transistors embedded in the sensors react with the stimuli and absorb electrical carriers. The reaction causes films inside the sensor to swell or change their state in other ways. This change is then amplified and propagated through a network.
Each sensor contains a wide variety of transistors. Structurally, the transistors are similar, but they are made up of different compounds and detect different environmental stimuli.
Unlike most semiconductors, which are made of silicon, these are organic semiconductors, which are made out of carbon. The switch to carbon brings a couple of distinct advantages. One, chemists can form a lot of compounds with carbon, which makes it possible to create the variety of sensing transistors needed to detect a large number of environmental agents.
Two, organic semiconductors can be printed directly onto thin polymer sheets, which is a lot less costly than making silicon transistors. With silicon, engineers blow hot metallic vapors onto wafers, scrape away excess materials, and then introduce more vapors.
"The key point is that it is entirely additive, which dramatically reduces the steps in processing, and that is where you get lower costs," Subramanian said. Inkjet printing also allows designers to place unique sensing transistors on the polymer sheet relatively easily. Thus, several thousand alcohol-detecting sensing elements can be sprayed onto a sheet at once and arranged next to transistors for detecting acids in the same way that a single printer can precisely spray different colored inks.
In silicon chips, creating multi-sensing arrays with differing transistors would take several steps.
Gold is used to construct the transistor gates. Gold in its bulk state melts at 1,000 degrees Celsius. Nanoparticles of gold, however, melt at 100 degrees Celsius. Thus, the gold can be melted without destroying the underlying polymer sheet. The reactive films in the transistor need to be around 20 nanometers thick. (Materials at the nanoscale--which is defined as particles measuring 100 nanometers or less--oftenthan their bulk counterparts.)
If carbon is such a great material for making transistors, why don't more manufacturers use it today? Environmental stimuli and stress can wear them out. Electrons move slowly too.
"The performance is frankly lousy" compared with silicon transistors, Subramanian said. But for cheap sensors, organic transistors work fast enough.
Experiments conducted by Subramanian and his students showed that carbon-silicon integrated noses can detect turning in wine, but a substantial amount of work lies ahead for developing full carbon sensors.