Fahlman says as the number of these infinitesimally small transistors are squeezed into smaller packages, more heat must be removed.
“We know that laptops, for example, are running really, really hot,” he says. “My MacPro just about burns my lap.”
Overcrowded chips need new technology
To make chips more energy efficient and to save consumers from hot laptops, Fahlman says computer scientists need a different material than silicon in the gate-insulating layer where heat-generating electrons flow on computer chips.
He says the natural silicon-based layer is already spread too thin – less than a few monolayers – to accommodate today’s small, fast machines. Since the layer’s limit has been reached, the electrons are escaping, wasting power and generating heat.
Fahlman believes that advances in nanotechnology, a field of applied science that controls matter on an atomic and molecular scale, will help alleviate the problem. To research this hypothesis, he and a group of undergraduate chemistry students built an atomic layer deposition (ALD) system to aid them in the process – one of fewer than 20 such machines in the nation.
The chamber is designed to produce thin films of any metal oxide or nitride – semiconductor materials that control electron flow – without contaminants, particulates, or residue.
Chemistry graduate student Michael Lubitz, who helped Fahlman build the chamber as an undergraduate, says his involvement with the project has given him nanotechnology experience and the assurance that he chose the right career path. He currently is working in Fahlman’s lab to record the conductive properties of thin films.
“Dr. Fahlman’s research piqued my interest. The ALD project was overwhelming at first, but it ended up being my favorite,” Lupitz says. “The ALD has really opened up research opportunities for me and other CMU students.”
Fahlman based his ALD system on the instrument he used when conducting his doctoral research at Rice University; however, this system offers significantly greater control over film thickness and purity. It gives Fahlman and his students an edge in the race to help solve the problem with silicon. Based on his current research, hafnium oxide is the most promising alternative to silicon, so he and his students are attempting to design compounds that will serve as precursors to deposit very thin layers of that material.
Getting ahead of industry
Fahlman holds up a silicon wafer in his Engineering and Technology Building office. The wafer is reflective and has the appearance of a small cosmetic mirror on this foundation of the computer chip.
In a laboratory connected to his office, undergraduate and graduate students are working on similar wafers. First, they remove the natural forming silicon dioxide (SiO2) from the silicon wafer by carefully dipping the silicon chip in diluted hydrofluoric acid solution. Then they put the silicon chips in the ALD machine chamber to grow a more efficient gate-insulating layer such as hafnium oxide (HfO2) or another yet undiscovered high-K dielectric.
“My students and I are looking to improve the gate-insulating layer so that we can keep shrinking the layer down and putting on more transistors without building up the heat,” Fahlman says.
They began their research in 2008, but Fahlman says that he is already proud of what they have accomplished just by building the system.
“There aren’t many universities that have the capacity to conduct this research,” he says. “CMU does. We are increasing our national prominence. This is a project that Intel is most likely working on.”
The next ‘little’ thing
Fahlman says that creating an alternative gate-insulating layer is just the first step in solving current computer chip limitations. He believes that the technology world eventually will need to discover or create an alternative to silicon. But experts say that may not happen for at least another decade – too long by some measures.
“Even once the gate insulator problem is solved, other developments will need to be included in order to progress Moore’s Law,” Fahlman says. “Most likely, this will include the use of nanostructures and perhaps entirely new paradigms such as quantum and optical computing devices.”
As a whole, the computer industry is actively pursuing an alternative to silicon dioxide gate insulators.
“CMU is working hard to solve this problem ahead of the industry.”
• By Sarah A. Chuby