Designing Around Nanoscale Transistors

The first semiconductor chip had a single transistor. Today’s chips can have nearly a billion, as individual transistors, packed ever closer together, seem to be shrinking toward the vanishing point. The inevitable progression of Moore’s Law, which predicts that computing power doubles every 18 months, has transformed modern life, but maintaining this pace creates challenges for integrated circuit designers like Ben Calhoun, an assistant professor of electrical and computer engineering.

When transistors were larger, creating uniform transistors was a matter of course. Now that we are producing transistors that are a fraction of a wavelength of the light used to create them, absolute consistency and precision are virtually impossible. Both the transistors’ size and the amount of voltage required to turn them on can vary considerably. And because of their lower turn-on voltage, they do not shut off completely. “They leak power,” notes Calhoun, “draining the power supply without anything to show for it.”

Calhoun’s approach is to design around these issues, for instance, finding ways to harness the power that leaks from these transistors and thus dramatically increase the efficiency of his chips. “I want to push power consumption so low that these chips may be able to scavenge the power to operate from ambient sources such as vibrations or thermal gradients,” he says. Calhoun is designing sub-threshold circuits that use the current that leaks from the transistors to perform other useful operations. This tactic works in situations in which speed is not critical, since low current slows operations.

Lowering the supply voltage also reduces leakage of power in memory cells, which are made of six transistors, but the cells will lose their stored data if the voltage falls too far. Because of the variation in transistors, identifying this failure voltage can be difficult. Calhoun has responded to this issue by designing circuits containing canary cells — essentially cells that are more sensitive to decreases in voltage than the core cells he wants to protect. This approach will let him use the lowest appropriate voltage while limiting the potential for damage.

The applications for Calhoun’s super-high efficiency chips are extensive. They could be used to operate tiny sensors to track the condition of bridges and buildings, in military surveillance systems, or for medical monitoring, an idea that earned Calhoun a Fund for Excellence in Science and Technology (FEST) award. Calhoun also believes they can be used to increase the functionality of radio frequency identification (RFID) tags, now used for electronic toll collection and inventory systems. “By reducing the power consumption of the circuits in RFIDs, you can pack more functionality into the tag, like better encryption,” he says.

Calhoun considers U.Va. a very welcoming place for his research. He is familiar with the University, having earned an undergraduate degree here before going to the Massachusetts Institute of Technology for his doctorate.

Equally important, he appreciates the encouragement and support of his colleagues in the department and in the school, as well as the University’s commitment to science and engineering. “I decided that U.Va. was a place for me to build a career,” he says.