- About MRI
- Research Capabilities
- Facilities & Centers
- Become a User
- For Industry
- News & Events
- Contact Us
It has been over 60 years since the transistor was invented at Bell Labs. Since that time, the workhorse of the computer age has shrunk from the size of a baseball to the point where more than 60 million 2 nm transistors could fit onto the head of a pin, according to an Intel white paper. Bell Labs’ first transistor used germanium as the semiconductor, but germanium was difficult to purify and hard to manufacture. By 1954, Bell Labs had produced the first silicon transistors, and the race was on.
To most Americans at the beginning of the silicon age in the 1950s and early 1960s, a transistor was something akin to a pocket radio you took to the beach to listen to surf music and the Beatles. Computers still filled an entire room and were meant for scientists only. That same Intel paper compares their first microprocessor, introduced in 1971, to recent versions at 32 nm and finds that processor is 4000 times as fast and uses 4000 times less energy. By 2014, Intel expects to begin producing at the 14 nm node, which means that the channel length of the transistor is less than 30 atoms across.
Suman Datta, professor of electrical engineering, spent eight years in the Advanced Transistor Research Group at Intel, in Hillsboro, Oregon, helping to introduce a strained channel silicon transistor technology that went into products at the 130 nm as well as the current 20 nm nodes. Another breakthrough came when his group at Intel found a replacement for silicon dioxide, the standard dielectric in the CMOS transistor, with a transition metal oxide that allowed scaling to continue. The thickness of the gate dielectric had shrunk so much that with silicon dioxide they had only three layers of atoms left; they had literally run out of atoms. "At that point in time, scaling would have come to a screeching halt," Datta says. But by swapping out the silicon dioxide for a transition metal oxide with a higher dielectric constant, they could resume scaling and miniaturization of transistors. Gordon Moore, co-founder of Intel and the author of the observation called Moore’s Law that transistors would drop in size and prices fall correspondingly about every 18 months, emailed the group to congratulate them when the breakthrough was announced, calling it the biggest change in transistor technology since the invention of the transistor at Bell Labs.
Datta was also involved in another big change in transistor technology, the first change from a planar, or flat, architecture to a non-planar or 3D architecture. It was planar technology developed in the 1970s that allowed the first integrated circuits to be built. But as the active parts of the transistor got closer together, current began to flow from the source to the drain even in the off-state. Called leakage current, it could drain the battery in a laptop or cell phone even when they were not in use. They decided they could overcome this problem by making multiple gates to control the flow of current instead of the one gate in planar devices. To do this they flipped the transistor on its side, creating what looked like a fin. This geometry allowed them to wrap the gate around the three sides of the fin. The 3D technology is now in the latest microprocessors used in high-end laptops, desktops, and data centers.
In 2007, Datta left Intel and joined Penn State’s Department of Electrical Engineering to work on more high risk transistor projects, hoping to stay ahead of the technology and possibly lay the groundwork for future transistor and microchip technologies. Penn State’s materials expertise intrigued him, because he wanted to see if he could incorporate more exotic non-silicon materials into a device that could still be incorporated into the familiar CMOS architecture.
"If you go to a store today, you don’t necessarily get microprocessors with faster clock speed," Datta remarks. "What people are looking for is energy efficiency. Can we provide more performance per watt of energy that these transistors consume by using an exotic material such as compound semiconductors?"
That is one of the challenges he and his students have been tackling in Penn State’s Nanofabrication Laboratory cleanroom – building a prototype transistor that can operate on lower voltage than standard CMOS devices yet maintain high performance and power efficiency. They have made substantial progress.
At the International Electron Devices Meeting in December, 2013, Datta’s graduate student Bijesh Rajamohanan, presented a paper on a high frequency, low-power tunneling transistor that could deliver high performance at about half the voltage of standard silicon transistors. Tunnel field effect transistors are considered to be a potential replacement for current CMOS transistors, as device makers search for a way to continue shrinking the size of transistors and packing more transistors into a given area – Moore’s Law. Tunneling is a quantum effect in which electrons are able to cross through a potential barrier, in this case an extremely thin interface between the source and the channel. The researchers tuned the material composition of the indium gallium arsenide/gallium arsenide antimony so that the energy barrier was close to zero, which allowed electrons to tunnel through the barrier when desired. To improve amplification, the researchers moved all the contacts to the same plane at the top surface of the vertical transistor. The researchers from National Institute for Standards and Testing (NIST) and custom wafer manufacturer IQE Inc. were co-presenters.
Another direction his group is following is one that he calls "More than Moore," which means that in the next generation of applications in information technology, people will be interested in interacting more directly with their computer. Called proactive or perceptual computing, this technology would go beyond the keyboard and mouse by creating user-machine interfaces that were not traditionally a part of Moore’s Law. Datta’s group is looking at sensitive microelectromechanical systems (MEMS)-based magnetic sensors directly integrated with CMOS technology that could record and interpret brain signals, possibly allowing a person one day to compose an email with her thoughts. This sounds well into the realm of fantasy; however, Datta says that Microsoft, Samsung, Intel, Google, and several other companies are strongly pursuing perceptual computing.
The third area they are involved in takes them beyond the traditional realms of computing into an area they call "Beyond Moore." Traditional computing is still done using ones and zeroes – digital bits. The brain, however, works differently, Datta explains. "When I see you, I immediately know who you are, where I’ve seen you, what is the context, why we are here. That is done through a very different computing paradigm that Mother Nature has figured out over many millions of years of evolution. So, one of the areas that we are moving strongly into is to try to see whether, with artificial hardware, we can implement what we call this neuromorphic computing paradigm where we don’t do things with ones and zeroes, but do them in an associative sense."
The direction in which this device research is leading in Datta’s group is toward coupled oscillators using complex oxide materials. There is a phenomenon called biological synchronization in which two weakly coupled oscillators will go into resonance when a certain set of patterns is presented to the oscillators. This ability to couple the oscillators’ relative frequency and phase to a pattern could have implications for neuromorphic pattern recognition, for example machine vision. Datta is part of a large team of researchers looking into implementing vision algorithms on new types of hardware. (See the following article "An Expedition in Computing Team Forms to Create a Cognitive Camera".)
"We are interested in taking advantage of the complex oxide expertise resident at Penn State to try to build these novel devices that go beyond just Boolean computing. It’s a new paradigm of computation that folks are very interested in," says Datta.
Datta brings his research experience in industry into the classroom and the lab. He believes that in order to sustain innovation, his students need to have a strong grasp of the fundamentals in physics, chemistry, and mathematics. "Students lose interest and want to know why they can’t skip through the mathematical equations and start building and testing stuff. Because of my unique background at Intel R&D, I can tell them that without mastering the equations I learned as a student I would not have been able to make these contributions later on. I help students understand that we are in the high-tech world where innovation is the only way you survive. Or else you become extinct like the dinosaurs."
Suman Datta, professor of electrical engineering, was part of the Intel team awarded the 2012 SEMI Award for North America for the invention of the high-k plus metal gate transistor. The award is the highest honor conferred by the 2,000 member global industry association for the semiconductor industry. Professor Datta can be contacted at email@example.com.
This Issue's Research Snapshots
Materials Research Institute
Pennsylvania State University
The safety and health efforts of this department are an integral part of our daily operations. In order to ensure the well-being of all those in the department, we will accept no less than full cooperation and participation in our safety practices and procedures.
+ Learn more
Faculty Spotlights are brief biographical sketches of faculty and their research interests.
+ Learn more