Key Concept Turns 30 — and is Still Going Strong

Success in chemical amplification based on committment and willingness to try new things.
25 February 2020
By Hank Hogan
Symposium Chair Will Conley of Cymer, John Maltabes, Chris Mack of Fractilia Corp, and Symposium Chair Kafai Lai of IBM
From left to right: Symposium Chair Will Conley of Cymer, John Maltabes, Chris Mack of Fractilia Corp, and Symposium Chair Kafai Lai of IBM T.J. Watson Research Center.

At the start of the 2020 SPIE Advanced Lithography conference, John Maltabes strode onto stage and on behalf of his co-authors accepted a plaque for "1X deep-UV lithography with chemical amplification for 1-micron DRAM production." Maltbes was lead author of the 11 listed on the paper, which was published in 1990.

SPIE celebrated that 30-year old paper because it described the first commercial application of chemically amplified resist. It's a concept that underpins technology advances over the decades since.

"It really opened the door for lithography after that. Because without that, we wouldn't have 193-nm [lithography]. We wouldn't have EUV [lithography]," said Maltabes. When the paper appeared, he was a senior associate engineer at IBM. He's now a program manager at Applied Materials.

Using a chemically amplified resist solved a key problem confronting the semiconductor industry: a lack of photons. In semiconductor photolithography, photons, particles of light, pass through a mask and expose resist sitting on a wafer. The resist hardens as a result and with further processing becomes durable enough to withstand corrosive chemicals. As a result, the pattern on the mask is transferred into materials on the wafer.

Following this patterning, another layer is grown or deposited on the chip. Repeating the deposition-photolithography-etch process over and over again creates the layers of insulators and conductors that make up an integrated circuit.

Now, Moore's Law calls for increasing chip performance at falling prices. Traditionally, that's been accomplished by steadily shrinking chip feature sizes — while cutting the cost of manufacturing a transistor. Pulling off that feat has meant that the wavelength of photolithographic photons must be shorter and shorter. This has happened as lithography has moved from 365 nanometers in the 1970s and 1980s to 248-nm in the 1990 paper to the 13.5 nm of EUV just entering high volume production today.

But in the 1980s, the industry faced a problem. The light source of choice, a mercury vapor lamp, put out too few photons at 248 nm, producing roughly a thirtieth of what was needed to expose the resist then in use. There were other light sources available, but they were more expensive.

Enter chemically amplified resist, which was developed by a team at IBM that included C. Grant Wilson, Jean Fréchet, Hiroshi Ito, and others. Wilson and Ito were among the co-authors of the 1990 paper.

In a chemically amplified resist, a single photon initiates a chain reaction. This multiplies the photon's impact and makes it do the work previously done by many in hardening the resist. That same approach underlies the resists that have appeared since the publication of the 1990 paper.

Getting the chemically amplified resist to work in high-volume manufacturing took engineering, according to Maltabes. The process involved cross-pollination of ideas and expertise.

For instance, initially the photolithographic process was inconsistent, with results that were good one time and bad the next. There was no obvious reason for this variability based on the chemistry involved, which was understood from a mechanistic point of view.

"But the thing we didn't realize was all the environmental factors," Maltabes recalled.

The group solved that problem by installing filters in the production cleanroom to remove airborne molecular contaminants. The team developed this solution because of interactions between groups involved in semiconductor and disk drive manufacturing as well as individual chance encounters.

Another issue that arose early on was that the chemically amplified resist was too sensitive. That made it difficult, if not impossible, to avoid overexposure and lead to an inability to print a clear image of what was on the mask. A different chemical formulation lead to a resist that give some process margin, an outcome that Maltabes likened to racing a car.

"You don't want to drive a car at the limit of the speedometer and the engine revolution. You want to drive it somewhere in the middle," he said.

As such tales demonstrate, the success and eventual dominance of the chemically amplified resist concept was not assured. In part, the group behind the 1990 paper prevailed because they were committed to the method and were willing to try new things.

Maltabes specifically mentioned this human side of technology advances in talking about the background of the 1990 paper. It was, he said, a talented team.

"I was so fortunate to be able to work with all these people," he said.

Hank Hogan is a science writer based in Reno, Nevada.

Enjoy this article?
Get similar news in your inbox
Get more stories from SPIE
Recent News
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?