At the heart of modern electronics are transistors, which act like valves to direct the flow of electrons. Now researchers at the University of California at Berkeley have created the first transistors that electrically control molecules instead. By connecting them to microscopic test tubes and petri dishes, these nanoscale transistors could lead to labs-on-a-chip that work without moving parts.

Much as 30-ton computers shrank over decades to microchip size, investigators are now miniaturizing labs to run millions of experiments simultaneously and dramatically speed analysis of DNA, proteins and other molecules. Although valves and pumps exist to control flow in microfluidic channels, they are not easy to miniaturize further for use on nanometer levels, says Berkeley mechanical engineer Arun Majumdar.

Instead of relying on mechanical manipulations, Majumdar and his colleagues speculated that silicon transistors might electrically control ions dissolved in fluids as well as they could electrons. Prior attempts to control ions by charging the surfaces of microfluidic channels, however, showed that ions quickly migrated to channel walls and canceled out the voltage, shielding the rest of the liquid from further electric manipulation.

With Peidong Yang, Rohit Karnik, Rong Fan and their co-workers, Majumdar found that channels less than 100 nanometers across are tiny enough to enable electric fields to breach this shielding. After constructing a 35-nanometer-high channel between two silica plates and filling it with potassium chloride saltwater, they demonstrated that voltage applied across this nanofluidic transistor could switch potassium ion flow on and off. Matching results were seen with transistors made from silica tubes 10 to 100 nanometers wide. "It's very good basic science and a clever idea. They elegantly take advantage of a physical effect that only dominates in very small channels," remarks Stanford University biophysicist Stephen Quake.

Most biomolecules are electrically charged, and the transistor could manipulate DNA fragments effectively. Majumdar envisions nanofluidic transistors that could rapidly sort the slew of molecules in cells by their mass and charge, thereby helping to purify DNA for sequencing or look for markers of disease.

Currently the transistors work on femto-liter (10-15 liter) amounts of fluid or less, roughly one one-hundredth the volume of a red blood cell. They could in theory prove sensitive enough to detect and manipulate single biomolecules, for exquisitely sensitive bioweapons detectors or "a lab for a single molecule, where you trap them and then study their behavior with light, force or any stimulus you want," Majumdar says.

Silicon offers the opportunity to build conventional and nanofluidic transistors onto the same chip for computerized control of chemical and biological processing. In early prototypes the voltage needed to switch ion flow on and off was 75 volts, far too high to incorporate into modern integrated circuits. But Majumdar explains they dropped the switching voltage to a sufficiently low one volt by thinning the channel walls. The team hopes to link nanofluidic transistors together into an integrated circuit within the year as the next step to harnessing massive numbers of transistors in parallel.