Professor Harold Craighead looks on as graduate student Stephen Turner adds a DNA sample to a laboratory setup for testing an experimental biochip

In a science fiction movie, a doctor puts a drop of blood into a small box and instantly reads out a complete DNA analysis. In real life, machines that analyze DNA are about the size of a refrigerator. And hundreds of them, working for the past 10 years, haven't been able to read the information in the equivalent of one person's DNA.

In the basement of Clark Hall, a group led by Harold Craighead, professor of applied and engineering physics, is developing biochips that could be a step toward the science fiction ideal.

The goal is to "sequence" DNA, finding the order in which four repeating bases are arranged on the chain. A global effort is under way to sequence all the DNA in the human genome by 2005. But, says Craighead, only a small percentage has been completed because the process being used, gel electrophoresis, is too slow. The process uses multiple copies of the DNA chain cut into pieces of varying length. The base on one end of each piece is labeled with a radioactive or fluorescent marker, and the fragments are placed at the beginning of a column of gelatinous material. When an electric field is applied, the pieces migrate toward the far end. Short pieces move farther than long pieces, and a map is formed showing how the lengths are distributed. From this, all the places the tagged base appears in the original chain can be calculated.

Craighead's group hopes to replace the organic gel with a tiny solid-state device made with the same techniques that can carve tiny electronic circuits on a silicon chip.

One approach to such a device uses microscopic passages so narrow that they slow down the movement of DNA fragments, in the fashion of a gooey organic gel. Graduate student Stephen Turner, working with scientists at Princeton University, has manufactured a sort of forest of vertical pillars, 100 nanometers (nm) thick and 100 nm apart, comparable to the diameter of a DNA chain. (A nanometer is one-billionth of a meter.) This smallest such "artificial gel" yet made was fabricated by the Cornell Nanofabrication Facility's electron-beam lithography and materials processing tools.

Turner is running DNA samples through the chips to see how fragments of different lengths can be identified. He mounts the chips between two microscope slides, glues small reservoirs to each end to hold a water-DNA mixture, places the slides on a microscope stage and applies an electric field, then watches and measures what happens. "Our measurements give us the first chance to test theories of DNA motion without having to guess size and placement of the obstacles in the gel," says Turner.

Another way to sequence DNA might be to clip off one base at a time and identify it. "Nobody has done this yet, but we're working on one component of such a system," Craighead says. A device being tested by graduate student Mathieu Foquet consists of microscopic capillaries that pass between tiny optical wave guides (basically light tunnels) made of silicon dioxide glass. If each of the four bases in a DNA sample can be tagged to fluoresce in a different color, they can be identified as they pass the wave guides.

A third possibility, being tried by graduate student Jongyoon Han, is a chip with openings of varying diameters that will sort DNA fragments by size, somewhat the way a coin counter separates dimes and quarters.

Some day, Craighead says, we may look back on refrigerator-sized DNA sequencers as the primitive ancestors of small, fast, inexpensive solid-state instruments that rapidly sequence genes.