Researchers at the Massachusetts Institute of Technology (MIT) have developed the smallest indium gallium arsenide transistor ever built. The compound semiconductor transistor, small enough to be packed into tomorrow’s microchips by the billions, could challenge silicon’s dominance as the workhorse of microchip materials and take Moore’s Law beyond the reach of silicon.
More transistors on a chip yield more power and functions. But shrinking silicon transistors to nanometer size in order to be able to squeeze more of them onto a circuit also diminishes the amount of current, slowing down operating speed. This has led to fears that Moore’s Law — the prediction by Intel founder Gordon E. Moore that the number of transistors on microchips will double every two years — would have to be abandoned. “It is becoming very difficult for silicon transistors to continue to scale down in size and maintain adequate performance, says Jesús del Alamo, the Donner Professor of Science in MIT’s Department of Electrical Engineering and Computer Science (EECS). “One solution is to move to a new material system for the channel in which the electrons travel faster than in silicon. That is what we are exploring.”
The indium gallium arsenide transistor, built in MIT’s Microsystems Technology Laboratories, shows it is possible to build a metal-oxide semiconductor field-effect transistor (MOSFET) with promising logic characteristics that is just 22 nanometers in length, using a silicon alternative capable of producing a larger current at this miniscule scale. “Our transistor features an InGaAs channel (actually a composite of InGaAs and InAs), then it has a thin barrier of InP above, then a gate dielectric of HfO2 and a gate made out of Mo,” del Alamo describes the design. “The transistor is built onto an InP substrate, but in the future it will have to be built on a Si wafer.”
Short gate-length MOSFETs
What is unique and advanced about MIT’s transistor is that it employs the shortest gate-length MOSFETs based on compound semiconductors. Another pioneering research advance is that the transistor’s metal contacts self-align, allowing the contacts to be placed very close to the gate, around 20nm. Transistors consist of three electrodes: the gate, the source and the drain, with the gate controlling the flow of electrons between the other two. The challenge here is that the source and the drain are extremely close together, due to very limited space. Current tools could not accomplish such precision. Thus, having the gate self-align between the other two electrodes most certainly is a breakthrough. “Finally, we use a pure HfO2 dielectric. This also has never been done before for a transistor of this kind,” adds del Alamo. When asked what, in his opinion, the significance of his research success might be for light-based technologies, he replies: “Moore’s law has a bright future!”
MIT’s discovery could blaze the trail for new applications that conceivably would not be feasible with silicon-only technologies. Because compound semiconductors are uniquely suitable for very high-frequency applications and photonics, the integration of III-V MOSFETs and Si could bring benefits beyond just higher density. “It will allow us to integrate photonic devices and logic, and to build high-frequency receivers and amplifiers on the same chip as logic circuits,” del Alamo projects.
In the future, a new generation of alternative microchips will have to have far superior performance to today’s silicon chips. “The key is the ability to integrate billions of transistors on a tiny chip working at high speed and burning little power,” del Alamo explains. “Compound semiconductors might allow that to happen at the point when it becomes very hard for silicon to do it.”
The MIT professor expects this technology to be on the market in as soon as 4 to 6 years. “Industry is pushing hard now on developing compound semiconductor transistors for logic. This could be real relatively soon.” But he also acknowledges that there still is a lot of research to be done, the perhaps most important being “to figure out how to grow high quality III-V layers on Si substrates.”
For the present, del Alamo wants his team to pause and “develop fundamental understanding” of the potentially revolutionary transistors they have at hand. “Then we need to work towards reducing the parasitic resistance. Then we should reduce [the transistors’] overall footprint.” He adds, “…Lots of work ahead!”
Written by Sandra Henderson, Research Editor, Novus Light Technologies Today