Could Indium Gallium Arsenide Dethrone Silicon in the Race for Smaller Transistors?
Recent work by researchers at MIT has led to better understanding of an alloy known as indium gallium arsenide, a room-temperature semiconductor that could lead to faster, denser chips.
Silicon has long been the primary semiconductor for transistors and computer chips. But this dominance isn’t guaranteed to last forever. This is especially true as engineers try to pack more power into smaller and more densely-packed transistors.
Now, MIT researchers have found that an alloy known as indium gallium arsenide (InGaAs) could be the key to unlocking smaller and more efficient transistors. Previously, researchers believed that the performance of InGaAs transistors deteriorated at the small scales required for chip applications, but MIT’s new study shows that this deterioration isn’t an inherent property of the material itself.
A Brief History of Indium Gallium Arsenide Transistors
MIT's new study builds on ongoing research into InGaAs transistors. Back in 2012, a team with MIT’s Microsystems Technology Laboratories announced that they had discovered the smallest InGaAs transistor to date. The 22-nanometer MOSFET was aimed at replacing silicon in computing devices since InGaAs can conduct larger currents at smaller scales.
This announcement came just days after researchers at Purdue University developed a Christmas tree-shaped transistor comprised of three tiny InGaAs nanowires.
Cross-section image of the InGaAs transistor produced by the ECE department at Purdue University. Image used courtesy of Purdue University
Stacked in a vertical structure (which was a novel architecture at the time), the InGaAs transistor was also said to be more compact and energy-efficient than its silicon counterpart.
Indium Gallium Arsenide
Why have researchers at MIT, Purdue, and other institutions turned to InGaAs as we anticipate the end of Moore's law?
InGaAs (sometimes referred to as "gallium indium arsenide, GaInAs") is a III-V compound with properties intermediate between GaAs and InAs. While it's most commonly used as a high-speed, high-sensitivity photodetector for optical fiber telecommunications, it's also a semiconductor at room temperature, making it suitable for applications in electronics.
Electrons are able to pass through InGaAs seamlessly, even at low voltage. This means that InGaAs could enhance a device’s energy efficiency. And because transistors made from the material can process signals quickly, it will, in theory, enable much faster computations.
However, the fact that InGaAs transistors seem to deteriorate at small scales—the scales required for faster and denser chips—throws a spanner in the works.
MIT’s discovery could lead to the material’s (pictured) widespread use in electronics. Image taken from the institution’s previous work involving InGaAs in 2012. Image used courtesy of MIT
In this study, however, the MIT research team found that the material’s small-scale performance issues are partly attributable to oxide trapping, not the material itself as previously thought. This oxide trapping causes electrons to become "stuck" while flowing through a transistor, dampening performance.
“A transistor is supposed to work as a switch. You want to be able to turn a voltage on and have a lot of current,” said Xiaowei Cai, lead author of the study. When electrons are trapped and a voltage is applied, there’s only a very limited amount of channel current, meaning that switching capability is a lot lower than when you’ve got oxide trapping.
InGaAs used in an imaging chip, which is one of its most common applications. Image used courtesy of Imaging Tech Solutions
Indeed, oxide trapping was identified as the cause of performance loss when the MIT researchers studied the frequency dependence of the transistor: the rate at which electric pulses are sent through it. At lower frequencies, the performance of nanoscale InGaAs transistors appeared to be degraded. In contrast, at high frequencies of 1 GHz and above, oxide trapping was no longer an issue.
“When we operate these devices at really high frequency, we noticed that the performance is really good. They’re competitive with silicon technology,” Cai said. The finding could one day serve to boost computing power and efficiency beyond what’s possible with silicon.