Willson has been a pioneer in developing technologies that enable the manufacturing of smaller, faster, and more efficient microelectronic components. One of his latest projects, funded by Intel, was to advance a technique known as pitch-division lithography. If the technique could be perfected, it could double the number of transistors that Intel could fit on one of their microchips, and it could do so without requiring massive new investment in equipment.
The lab had already overcome many of the obstacles by the time Wang arrived, but there was a key compound that needed to be synthesized, a “two stage photobase generator.” So far all efforts to synthesize it, including those by an outside chemical company, had failed. So Willson did what he often does in such situations. He offered a cash prize.
Wang, who is now 20 and set to graduate in May, won it.
The Foundation: Writing with Light
To understand why it was important that Wang successfully synthesized a two stage photobase generator, it helps to begin with the process for making transistors, which is called photolithography.
Photolithography begins when you shine light through a stencil-like mask onto what’s basically a cake with a top layer of icing. The cake is the silicon wafer that will eventually become the transistor. The icing is a polymer called a “photoresist,” which is designed to generate acid when struck by light.
That acid makes it soluble, so it can be removed with some processing. The remaining photoresist, which wasn’t touched by light, protects the parts of the silicon that need to remain pure.
“To make a transistor you have to induce impurities in the silicon,” explains Wang, “and it’s an incredibly delicate process. You have to have different regions with different impurities, and if they don’t go in exactly the right spot it won’t work. So what this process is helping you do is cover some spots you don’t want to touch and manipulate the spots that are exposed.”
For the past few decades researchers have been able to double the amount of transistors on a chip every 18 to 24 months—what’s known as Moore’s Law—by tinkering with three of the four basic elements in the process. They’ve improved the lens used to shine the light. They’ve reduced the wavelength of the light. And they’ve used masks with finer and finer slits. Each advance has meant smaller and smaller transistors.
The gains to be harvested in these areas, however, are now running up against hard physical limits.
“Ultimately you can only make features as small as your light allows,” says Wang. “At some point you’re trying to draw thin lines with a dull pencil. That’s the point we’ve reached, at the physical limits of this technique. If we want to continue making smaller features we either have to change to a new technique or we have to play with another element in the process. So what we’re asking is: Can you do something new in the chemistry of the photoresist?”
Writing Two Lines Where Once There Was One
Pitch-division lithography exploits a subtlety of what happens to light when it passes through the mask. It gets diffracted a bit, so it ends up hitting the photoresist at different intensities.
“Think of it as one wavelength of a sine curve,” says Wang. “So you peak in the middle and decrease on either side.”
In standard photolithography this doesn’t matter much because all you’re trying to do is induce one chemical change, which happens in the region exposed to the peak intensity light.
What if, however, you could create a photoresist that generates the optimal amount of acid not when there is peak light—since there’s only one peak per exposure—but when there is an intermediate intensity of light, since that intermediate intensity occurs twice per exposure? Then when the photoresist is treated, two areas would be removed, and you’d functionally double the amount of features you could create in the same area of the silicon.
To achieve this profile, Willson’s lab turned to one of those principles of chemistry that kids learn in grade school. Acids and bases neutralize each other. They designed a photoresist that contains both light-sensitive acid-generators and light-sensitive base generators. But they made sure the base-generators were less sensitive to light. With this difference in sensitivities, different intensities of light should generate varying amounts of acid and base during a given exposure.
Under peak light conditions, a lot of acid is generated but so is a lot of base. There’s enough base, in fact, to wholly neutralize the acid. Then it’s as if the photoresist hasn’t received any exposure.
In the intermediate zones, however, a bunch of acid is generated but the base is generating too slowly to fully quench the acid. And in those two regions, separated by the area where the acid and base had cancelled out, the acid concentrations will render the photoresist soluble.
“So you have a region in the beginning where it is being developed away,” says Wang, “then at the peak where there’s a lot of acid and base it’s not, then it is.”
When Wang arrived in Willson’s lab, this basic technique had already been developed. They were able draw two lines on the silicon from only one on the mask. The problem was that the lines were rough at the edges.
“We called it a low chemical contrast,” says Wang. “It’s a result of the fact that in the intermediate regions you still get generation of base. It’s slow but there is a little and you’re partially quenching the acid, so the amount of acid is not increasing as quickly as you’d get in a conventional resist. If you’re not changing the boundary between soluble and not soluble as fast, it’s going to be kind of blurry.”
Sharpening the Lines
What needed to be made was a compound that took a little while, after being exposed to light, to begin generating base. Then the acid concentration could rise more dramatically, and there’d be more chemical contrast, which would lead to sharper lines.
“What we hypothesized,” says Wang, “was that you could get this delay if you could move from first-order to second-order reaction kinetics.”
What that’s meant, in practice, has been trying to synthesize a base generator that starts out in a latent form. The first photon that hits it wouldn’t trigger the generation of base. Instead it would change it to its active form by, for instance, cleaving away one part of the compound. Then the next photon would trigger the process of generating base.
After many (many) failures, Wang finally succeeded during fall 2012. The trick, he says, wasn’t knowing the final chemical structure he wanted. That was known all along. The trick was getting there, figuring out how to facilitate each step in a multi-step process. In particular there was one solvent, toluene, that helped facilitate the key reaction where the rest of the solvents he’d tried didn’t work. It was some luck, some science, but mostly persistence.
“That’s the thing about chemistry,” says Wang. “You can know the structure you want. You can be familiar with reactions that should, in theory, produce it. But strange things happen. Synthesis is very sensitive to conditions, and you don’t always understand how or why. So you keep trying different things, holding different variables constant while changing others, and you hope that you find a way through.”
Wang’s initial results were good enough to win the prize from Willson. His molecule responded to light in precisely the way he was hoping. The next step is embed the molecule in a fully formed photoresist, then test it again. Willson has sent it out to a colleague at a semiconductor company to do that, and they’re waiting on the results.
If those results are good, the lab can refine the process until one day, perhaps, they’ll hand it over to Intel, where it could help build the next generation of chips. By that point, if it comes, Wang is likely to be in graduate school, earning his doctorate in organic chemistry. If it doesn’t come, for whatever reason, Wang is still likely to be in graduate school, tinkering around with molecules, hoping that good luck, skill, insight, and persistence will combine to lead to further discoveries.
“It’s the challenge I like,” he says, “and the complexity, and being able to use my hands, to be at the lab bench manipulating things. There are so many different factors to keep in mind, different potential routes to get to the final structure you’re trying to create. It’s fascinating.”