BERKELEY ­ In the future the slew of credit cards we all carry will be joined by drug delivery "cards" we can pull out and slap on our arm to give ourselves needed medication, such as a controlled drip of insulin or a brief regimen of antibiotics.

That is the goal of engineers at UC Berkeley, who have developed microscopic mixers and bubble pumps that fit on a patch the size of a thick credit card and can deliver drugs through a needle no larger than a mosquito's snout.

The project is funded partly by the Defense Research Projects Agency, the Berkeley Sensor and Actuator Center, and Becton Dickinson, the world's largest maker of drug delivery devices such as syringes and implantable insulin pumps.

"The government wants a good way to deliver medications in the field, and this provides a very stable way to deliver drugs," says Dorian Liepmann, assistant professor of mechanical engineering at the University of California, Berkeley, and a leader of the research project. "But the device also would be useful for insulin delivery, providing a small continuous dose that would even out the fluctuations in blood sugar and perhaps prevent the circulatory damage common in diabetics."

Liepmann predicts the drug delivery cards would be useful as emergency medicine, or merely as an easy way to deliver a number of drugs, from antibiotics to painkillers. Also, since they deliver drugs continuously and in small doses, they could allow the use of drugs that are too harmful to the liver to deliver by injection today.

"This technology opens up new forms of drug delivery and because of that, makes available a host of new medicines," he says.

Liepmann showed off the new device Wednesday, Mar. 11, during a talk at the 20th annual Industrial Liaison Program conference, a campus event hosted by UC Berkeley's College of Engineering.

The drug delivery system, though still under development, is one of the first practical devices to emerge from a field called microfluidics - the science of moving and mixing fluids in environments no larger than a human hair.

When the field of microelectromechanical systems or MEMS first exploded 10 years ago, many touted the potential for microscopic pumps and drug delivery devices on a chip, but achieving this proved harder than originally thought, Liepmann says. On such a small scale - his mixing chamber contains only 1/100,000 of a cubic centimeter - viscosity plays a proportionately greater role. This precludes the use of turbulence - the type of mixing that occurs when you stir cream into coffee.

"It is a challenge because it's so small that you get only laminar flow, that is, no eddies and swirls. Stirrers don't work, it's like trying to stir molasses," says John Evans, the graduate student who works with Liepmann on the project.

When UC Berkeley researcher Al Pisano, codirector of the Berkeley Sensor and Actuator Center and currently on leave as director of DARPA's MEMS program, first approached Liepmann about a microscopic drug delivery system several years ago, Liepmann immediately thought about a new way of mixing called "chaotic advection" or "laminar mixing."

Ten years ago UC San Diego professor H. Aref argued that you could create chaotic flow fields in simple systems using nothing more than fluid inlets and outlets - known as "sources" and "sinks" in the trade.

"This was a key discovery," Evans notes, "because it suggested that you could generate chaotic flows capable of mixing even if you couldn't fabricate intricate three-dimensional shapes, or generate turbulent flows. These are exactly the constraints we encounter today when designing microfluidic systems."

Evans and Liepmann showed by computer modeling and a simple experiment that this form of mixing can completely mix liquids in 30 seconds that would take 30 minutes to mix by simple molecular diffusion.

"We use a very simple flow field, generated using fluid sources and sinks. But the flow field is designed to be chaotic," Evans says. "In a chaotic flow field, two fluid particles that start out next to one another end up far apart. If this happens for every pair of particles, the sample will become well mixed."

They think they can optimize the technique to achieve mixing in 2-3 seconds, the time it takes to turn on the card and stick it on your upper arm.

"This achieves very fast and controlled mixing," Liepmann says. "If you don't mix in a controlled way, there could be a corner of unmixed liquid, which not only wastes the drug but could have severe medical consequences."

The mixing is accomplished in a small flat chamber 0.6 mm by 1.5 mm and 0.025 mm thick, making a total volume of about 15 nanoliters. Fluid is pumped in and out of four holes in the chamber in a sequence that creates a chaotic flow.

The pump itself uses a bubble as a piston. A microheater is used to evaporate fluid in a dead-end passage. As the bubble forms, fluid is pushed down the channel. When the heater is turned off, the bubble collapses, drawing fluid back up the channel. When this "bubble piston" is combined with new MEMS check valves, fluid can be moved around the device in a controlled manner.

Bubbles generate more force and better displacements than other MEMS actuation methods.

The mixing chamber, reservoirs, pumps and connecting pipes are etched in a silicon wafer using techniques similar to those used in the semiconductor industry. A flat battery would power the heating elements that create the bubbles, and be designed to last only about 24-48 hours. When switched on, the heaters would start the pumping, mixing the chemicals together. The whole device would be about 8 mm thick, the thickness of a soda straw. A prototype series of five mixers, containing over a dozen pumps and valves, fits on a chip only one centimeter on a side.

For diabetics, concentrated insulin would be diluted with water to provide the proper strength drip. With antibiotics, though, freeze-dried medicine would be mixed with water as needed, since many antibiotics, such as the cephalosporins, are unstable after mixing in solution.

For an insulin pump, the eventual goal would be to combine the drug delivery system with a monitor to check blood glucose levels, and a feedback system to adjust the insulin to minimize blood sugar fluctuations. A patient could even have a remote control to alter her insulin intake, if for example she were about to eat a sugary dessert.

"I think in a year we will have a full operating pump," Liepmann says.

Their colleague at Becton Dickinson is Burton Sage. John Evans presented the group's most recent paper on their drug delivery system in January of last year at the IEEE MEMS 97 Conference in Nagoya, Japan.