"It takes so long to test for food poisoning today that by the time the test is finished, much of the tainted food has already been delivered and consumed. We reduce the testing time to a matter of minutes," said Rashid Bashir, a researcher on the project. The chemical aspect of the design was handled by Michael Ladisch, and Arun Bhunia provided expertise in food science.

It will still take a few years to engineer practical end-user commercial biochips, but by then they may be applicable to dozens of pathogens. "Soldiers could use the chip to identify airborne biological weapons, physicians could use it to diagnose common diseases, and farmers could even use it to identify crop diseases," said Bashir. Pharmaceutical developers could use it to identify the beneficial biochemicals in folk remedies.

The key is protein matching. "A protein is like a lock that only a single key will open. We bond a protein to the biochip with electrostatic attraction that can only be opened by the molecule we're testing for," said Bashir.

The first prototype-application biochip will be developed to detect the deadly pathogen Listeria monocytogenes in foods. According to Bashir, statistics from the Centers for Disease Control and Prevention reported roughly 2,500 Listeria cases, 20 percent of them fatal, for 1999.

Inside the biochip, a silicon substrate has been sculpted to channel a flow of chemicals from an input port and over the tiny electrodes to which the protein identifier has been attached. When a "key" molecule comes along and binds to a protein identifier, the impedance of the electrode below it changes, allowing a computer attached to the chip to read out the triggered protein identifier.

The chip comprises wells connected by channels, etched from oxide-coated silicon wafers and sealed with a glass cover through which fluorescent marker molecules can be observed.

The cavities on the lab prototype ranged from 80 square microns to 530 x 850 microns. Channel widths ranged from 20 to 100 microns, but all were etched to a depth of 10 microns. Both bare and oxide-insulated flat platinum electrodes were tried as sensing posts for each cavity.

Liquids containing the sample to be tested were pumped into the channels through tubes connected to the chip, for which top and edge connections were tested on the prototype. Voltage-assisted adsorption immobilized the protein identifier on the surface of the electrode, and impedance changes upon binding were observed to change the electrochemically induced potential caused by enzymatically catalyzed reactions on the electrode surface.

For the demonstration, Bashir chose to attach the protein avidin to the chip. Avidin binds only with biotin. When the chip was presented with biotin at its input port, it immediately captured the "key" molecule and unlocked its protein identifier, changing the impedance of the electrode. Bashir also verified with a microscope the presence of the specific fluorescene-labeled biotin molecules sent into the chip.

Bhunia is compiling a list of protein "locks" and the specific molecules that act as keys to unlock them. For instance, for the Listeria food poisoning application the team plans to tackle as its first commercial chip, the specific antibodies from a rabbit immune to Listeria will be used.

"What we have is high-tech litmus paper that will instantly tell us the presence of any specific molecules and their concentrations," said Bashir. "The chips will be cheap enough to be one-use disposable and will probably come in a little carrier like a game cartridge to plug into a handheld unit." Users could test for a chemical by inserting the cartridge containing the biochip for that substance. Since each chip has thousands of separate cells for protein detectors, the number of cells triggered during the measurement period would indicate the concentration of that substance.

Once a substance is detected, the chip is removed, but it can be reinserted at any time to verify the concentration of the substance detected during a specific recording period. Chips can't be reused, because the agents that flush out the detected molecules would also have to flush out the proteins to which they are attached.

"Instead of putting the pump on the chip, thereby making it expensive, we will use a macro-sized pump in our handheld unit to deliver samples to the disposable chip," said Bashir.