Speeding up the manufacture of super small-scale devices

Building a computer chip is a painstaking process. Once a chip is designed, a mask, or template, is created and used to transfer the fine circuit patterns to the surface of a silicon wafer.

(The silicon wafer is the substrate for the transistors, wires, etc, and can produce a handful of chips).

If anything goes wrong with the mask, or if the tiniest change is needed, an entirely new mask must be created.

Scientists at the Oregon Health and Science University Oregon Graduate Institute School of Science & Engineering in the US are working to speed up the chip-making process, reduce manufacturing costs and make chips more flexible so that new masks won't have to be created if mistakes are made that cause the chip not to operate, or to operate poorly.

This shortens the development time for new chips.

With a recent $100,000, one-year grant from the National Science Foundation, four researchers in the school's Department of Electrical and Computer Engineering will build and demonstrate a tool that can create patterns on an extremely small scale (nanotech) for use in electronics, sensors and electrical mechanical systems.

"The current methods for transferring patterns to chips and fabricating masks - photolithography and single electron beam lithography - offer poor resolution and are slow," said Jack McCarthy, a materials scientist and lead investigator on the project.

"We are using a focused ion beam milling machine to develop a prototype source that will allow us to create and focus many individually controlled computerised laser beams with small enough spot sizes to transfer nanoscale patterns directly to the wafer without the use of masks.

"The computer control allows us to easily make pattern changes and modifications."

To build the tool, McCarthy and his colleagues start with a substrate/window made of sapphire. The vacuum side of the window is coated with a relatively thick film of gold.

A focused ion beam (FIB) is then used to cut arrays of cylindrical holes to expose tiny discs of sapphire.

A second layer of much thinner gold is deposited to coat the exposed sapphire. Each one of the discs in the sapphire is the source for an electron beam. So when the laser shines a broad beam of ultraviolet light through the sapphire window onto the array of discs, electrons pop out the vacuum side of the gold film disks, creating many beams.

The beams are generated down a column, focused and reduced to form the pattern with many beams on the work piece.

(The work piece in this case is an electron sensitive polymer (thin film) on a silicon substrate, which allows scientists to create very fine patterns with lines as small as 1/100,000 of a millimetre in width onto silicon wafers. The diameter of a human hair is about one-tenth of a millimetre).

"We are going to study the stability of the electron emission current that is produced by the thin film when the laser hits it," said McCarthy.

"We're also going to test a variety of metals and thin film stacks as electron sources so we can, hopefully, increase the intensity and improve stability.

"We'll also deposit an insulated metal ring around each source so that by applying computer-controlled voltages, we can control the intensity of the beam or shut it off like a light switch with a dimmer," said McCarthy.

"If we can figure out a way to individually control the intensity of each beam with a computer, it will ultimately enable chip makers much more flexibility when patterning chips and speed up the development process for new chips, as well as the manufacturing of chips."

On average, it takes chip makers eight hours to create one mask to transfer one pattern to the multi-layered chip, and it can take up to 25 hours to complete the chip, said McCarthy.

Masks with really fine details - lots of transistors and metals and connecting wires - can take even longer and will become impractical at this scale in the near future.

McCarthy's team includes physicist and surface scientist John Freeouf, and electrical engineers Neil Berglund and Jody House, and several graduate students.

Few university researchers in the US - and none in Oregon - are currently carrying out research on multi-electron beam lithography to pattern chips with such fine detail, said McCarthy.

The OGI researchers' study has profound implications for the microelectronics industry, which will need higher-density chips to provide more memory and smarter, faster processors as we give computers, mobile phones and other electronic devices more tasks.

The research also holds promise for biotechnology in providing electromechanical systems, sensors, processing and memory in smaller and smaller packages for use in such applications as drug delivery systems.

In addition to lithography, this multi-electron beam approach will speed up the testing of finished chips (e-beam review), and improve the imaging and analysis of materials on the atomic scale.

"We hope to be able to create a prototype nanolithography tool that can be used by a variety of microelectronics and biotechnology researchers at OHSU," said McCarthy. "Our small version will be the nuts and bolts needed for a complete production electron lithography machine for handling entire wafers. This research will be the first step toward making a variety of nanotechnology applications practical in development and production. It's really quite amazing."