UW-Madison helps Imago carry out its big vision for seeing small
Up close, Imago Scientific Instrument's atom probe is a veritable jungle of glinting steel, glowing ports and snaking hoses and wires. This microscope that lets scientists peer at the very atoms that compose matter does so by an intricate process, first plunging specimens to temperatures near absolute zero and then pulling atoms from them one by one. The atoms next fly to a detector, where they are identified and their positions in three dimensions pinpointed — all at the rocketing pace of 50 million atoms per hour.
Known officially as the local electrode atom probe (LEAP) microscope, this remarkable machine is the brainchild of Tom Kelly, a former UW-Madison professor of materials science and engineering, and Imago's founder, chairman and chief technical officer. Kelly may have devised the instrument while at the university, but he is quick to pass on credit to his company's technical team. The 10 atom probes the group has built, shipped and installed at customer locations since 2003 have proven highly reliable, he says; last month, for example, every instrument was operating 100 percent of the time, save one — it was at 96.
On top of this, he adds, each has been delivered on time to customers and fully installed within two weeks as promised.
"For these complicated instruments, this is exceptional," he says. "It's one thing to have good ideas. It's another thing to execute. And execution like this is the domain of engineering."
To find the engineering prowess it relies on to succeed, Madison-based Imago has rarely needed to look any further than its own backyard. Like Kelly himself, most of the company's technical staff hails from either UW-Madison or the Madison Area Technical College. The trend started in 1998 with the company's very first employee, former UW-Madison researcher Tye Gribb. Since then, Imago has tapped over 30 graduates of UW-Madison's science and engineering and MATC's electron microscopy degree programs.
The company also maintains close ties to university faculty, especially those in the College of Engineering departments of materials science and engineering, and electrical and computer engineering.
"Because we have to be very focused on analyzing our customers' samples, we don't get much time to do basic scientific research," says Keith Thompson, Imago's senior applications engineer. "The university really helps to fill that gap. Over a two- to three-year period, we end up getting a lot of important information that we might not otherwise get."
Scientific partnerships with UW-Madison can also contribute again to Imago's own technical staff.
"These collaborations can lead to the development and education of students who then become the next generation of Imago expert engineers and scientists to take the technology to even greater places," says UW-Madison electrical engineering professor John Booske. "It's a very positive and powerful symbiotic relationship."
Making the leap to semiconductors
Booske knows of what he speaks. Thompson, who now helps develop new applications for Imago's atom probe, first began working with the company while employed as a postdoctoral researcher in Booske's lab. The two were perfecting a step in the manufacture of tiny semiconductor devices spanning fewer than a thousand atoms, when they suddenly realized what they faced.
"We saw that the research questions we were pursuing amounted to asking, 'Just where are all the individual atoms and what type of atoms are they?'" says Booske.
At the time, Imago's microscope — originally designed to probe metals — hadn't been rigorously applied to semiconductors. Nonetheless, Booske and Thompson soon recognized it as their best chance to see the three-dimensional positions of atoms within the tiny bits of material they were studying. The pair approached Kelly, who quickly agreed to give Thompson access to Imago's equipment when it wasn't in use. "Basically, nights and weekends," says Thompson, laughing. David Larson, a former graduate student of Kelly's who now serves as Imago's director of applications, also joined the effort.
Because atom probing requires needle-sharp specimens, Thompson's first task was to whittle these into the smooth surface of a semiconductor. Metal samples for LEAP analysis resemble rough-hewn pins — small but easily seen with the naked eye. But the semiconductor samples needed to be much tinier: just 1/200,000th of a centimeter wide and about 1/100th of a centimeter tall.
For help Thompson turned to UW-Madison's Wisconsin Center for Applied Microelectronics (WCAM), a state-of-the-art cleanroom facility that provides equipment and expertise for making microscopic structures of just this type.
"The WCAM facilities are top of the line, and the staff is very knowledgeable and helpful," Thompson says. "They really helped get me going; without the cleanroom, I don't know if this project would have gone very far."
But creating samples proved to be just half the battle. The other major hurdle confronting the team was finding a way to pull atoms from the samples' sharpened ends. Because metals readily conduct electricity, atom probing normally sends an electrical pulse through a metallic specimen to rip atoms from its tip. But semiconductors, as their name implies, conduct electricity only sluggishly. And when Thompson initially tried applying voltage to his samples, they tended to explode.
To solve this problem, Imago figured out how to wrest atoms from semiconductors with a laser beam instead. It announced the advance this summer.
Hired by Imago in April 2004, Thompson now routinely analyzes samples for the company's customers in the semiconductor industry.
"In the past, the complications of getting LEAP to work with semiconductors were so great, people didn't even want to hear about it," he says. "Now, when I get a sample set in, we can generate data right away — it's a two- to three-day turnaround. And I tell people it's all because the foundation was laid on good scientific principles back when I was a postdoc."
The latest joint project between Imago and UW-Madison began this July. Funded by the university's Industrial & Economic Development Research (I&EDR) program, it involves materials science and engineering professors Dane Morgan and Izabela Szlufarska. By harnessing the power of well over one hundred computer processors, these researchers create models depicting the positions of millions of atoms — in essence, simulating in the virtual world what LEAP shows in the real one. The difference is, Morgan and Szlufarska can set their atoms into motion and watch as they shift and move in time.
"With the atom probe microscope, we take snapshots of atomic positions," says Kelly. "We would love to add an element of time to our snapshots."
For example, LEAP might analyze a pair of samples at the start and finish of a certain semiconductor manufacturing step, such as a heat treatment, to see where the atoms originated and where they ended up. Morgan and Szlufarska could then connect the atomic dots, showing through simulation how the atoms traveled from point A to point B.
Modeling could also shed light on a phenomenon already witnessed by Booske and Thompson: the tendency of certain atoms to cluster during semiconductor processing. When this occurs, devices will fail.
"With LEAP, you get a snapshot of clustering, but LEAP can't tell you how the clustering occurred," says Morgan. "With simulations we can see what characteristics and processes within the material caused the clustering — how it happened."
What the researchers gain from the partnership is access to the best atom probe data in the world, says Szlufarska, which they can use to verify their theoretical models. She also appreciates the chance to work on a problem of importance to industry.
As for Kelly, he naturally can't predict exactly where this latest collaboration will go or what it may yield for Imago's bottom-line objective of selling its microscope. Still, he knows from experience what basic science can offer.
"The reason I want to foster this collaboration is this: We make an instrument, but we also need to have answers for potential customers, and understand and be experts in all the ways the atom probe can help them answer questions," he says. "Getting the experimental data is good, but with the additional tools and understanding (brought by Morgan and Szlufarska) we can make it that much better."