MADISON, Wis. The University of Wisconsin is using digital imaging technology in a novel attempt to simplify the fabrication of custom chips. Rather than creating a set of photomasks to define micron-scale features, a team of semiconductor engineers and molecular biologists hit on the idea of using a Texas Instruments Inc. micromirror array chip the Digital Light Processor to create a "virtual mask" for performing photolithography. The technique was devised as an economical way to create custom chemical-analysis arrays for a gene sequencing project.
Instead of shining light through a mask to expose only selected portions of a substrate, an array of 480,000 tiny aluminum mirrors reflects light onto just the areas of the substrate that would be illuminated through a mask. Thus, instead of cutting as many as 100 physical masks per chip, the TI micromirrors can be instantly loaded with the data for each, configuring its mirrors to shine in exactly the same pattern as the masks. The technique probably would not be appropriate for mass semiconductor manufacturing, but is ideal for producing a small number of custom chips in a research project.
"By patterning the light with micromirrors instead of making masks, we have eliminated all the expense and time-consuming tasks involved in making a set of masks for a chip we can make a new chip in about eight hours using our virtual masks," said University of Wisconsin researcher Roland Green. Green's research team includes Yongjian Yue of the department of electrical and computer engineering and Clark Nelson of the Biotechnology Center, as well as a group of genetics specialists.
The research project uses combinatorial drug analysis techniques that have evolved from semiconductor manufacturing techniques. The idea is to synthesize arrays of closely packed variations on a specific formula which can then be tested en masse for compliance with desired design specifications. Each mask can be configured to selectively add new molecules to an array of previously defined samples.
Since many variations of the basic molecule are typically synthesized in these massive arrays, the number of masks required to create all the desired variations on the basic formula varies sometimes up to 100 separate masks per chip. However, the object is not to produce thousands of chips, but just a few for a specific experiment. Thus the standard manufacturing methods based on masks are not really the best techniques for combinatorial drug research.
After fabrication of the array, scientists resort to trial-and-error tests on each sample. Because the samples are so tightly packed, they can all be tested in parallel, thereby saving time. Drug discoveries, DNA identification, peptide protein synthesis and complex carbohydrate investigations have all benefited from this massive parallelism.
A person's blood, for instance, can be simultaneously applied to all samples in a screening, with only those samples that match a molecule in the blood showing a reaction (signaled by turning on a florescent atom for visual inspection).
Maskless Array Synthesis (MAS), as the UW-Madison group has dubbed the new virtual mask technology, is currently used only to create arrays of certain organic molecules principally DNA for the human genome project. But in principle this massively parallel molecular assembly method could also be applied to the synthesis of other materials.
"This technology gives researchers the ability to make any array of synthetic compounds, any time, and right there on their own benchtop," said Wisconsin professor Michael Sussman.
In fact, there is no theoretical boundary to using array synthesis, except the chemicals needed to build up the molecules one layer at a time. Usually a photolithography step preps the array elements for the next chemical addition. The chemicals are selectively added to the array of samples by coating the chips with an organic material that releases a specific molecule at exposed array sites.
"We have already perfected the chemistry for building organic materials like DNA, carbohydrates and proteins like peptides, but it is only a matter of time until the chemistry for building other types of synthetic materials is perfected using this method," Green said.
Today, researchers testing synthetic materials must hire a company to fabricate the masks to create the arrays of thousands of tightly packed samples. After fabrication, the researcher uses the array to test all variations of the molecule simultaneously, after which a robot scanner identifies which elements of the array passed the test. By going to MAS for virtual masks, this whole process can be done in the lab, without having to contract with an outside company.
"I like to compare MAS to desktop publishing, because it takes a job that used to require off-site mask cutting and has put it on the desktop. Instead of sending your data off somewhere and waiting several weeks to get your chip back, it takes just eight hours to make your own unique chip in your own lab," said Franco Cerrina, a UW-Madison professor of electrical and computer engineering.
The Wisconsin Alumni Research Foundation, a not-for-profit corporation that manages intellectual property of UW researchers, has applied for a patent on MAS. The MAS technology has also been licensed to a new company, NimbleGen Systems LLC, formed in cooperation with the university and based in Madison. Three UW researchers participated in the founding of NimbleGen.
Currently, the only place to send data for the cutting of the masks necessary to fabricate massive arrays of closely packed variations of a specific molecule is Affymetrix Inc. (Santa Clara, Calif.). The company offers a service for what it terms "GeneChips" arrays of up to 500,000 elements, each containing a unique molecular sample.
Affymetrix recently acquired a line of laboratory equipment from Genetic Microsystems that provides researchers with the ability to fabricate their own user-defined arrays without having to build a GeneChip. Affymetrix will still market the GeneChip as the technology of choice for "disposable" high-volume applications, like blood testing for identifying known genes. The new "chipless" lab equipment solution will permit researchers to make low-volume test microarrays for known or unknown genes.
As to the MAS desktop chip-making technology, Rava, senior vice president and chief technology officer of Affymetrix, said "been there, done that.
"We believe that the standard semiconductor manufacturing technologies provide a scalable process for making large numbers of high-quality arrays," Rava said. "Micromirrors and other technologies for manufacturing GeneChip arrays may have niche applications for creating small numbers of custom arrays, but researchers who wish to have the flexibility to make a small number of arrays don't need to make chips. They can uses tools like those we agreed to acquire from Genetic MicroSystems."