21 June 2010—This week, at the VLSI Circuits Symposium, in Honolulu, a team of engineers from MIT reported that they have invented an energy-storage chip that overcomes one of the last remaining technical hurdles that have kept ultracapacitors from replacing batteries as the energy-storage device of choice for the tiniest electronics.
Ultracapacitors offer several advantages over batteries: High energy density, rapid recharge, and a virtually unlimited number of charge-discharge cycles are but a few. One of the drawbacks of ultracapacitors is that their voltage decreases along with their state of charge; the voltage in a battery remains relatively stable. By the time an ultracapacitor reaches a 25 percent state of charge, its voltage has dropped by half. (The voltage of a lead-acid battery at this state of charge would decrease only by roughly 5 percent.) Because chips usually operate in a fairly narrow voltage range, such a steep drop would cause failures, such as read-write memory errors.
Leaving one-fourth of an ultracapacitor’s stored charge stranded inside the device is no one’s idea of efficient energy use. So the MIT team came up with a clever way to stave off the voltage drop until almost all the energy is consumed. The circuit they developed rapidly rearranges the configuration of a set of ultracapacitors to get the most charge out of them while maintaining a constant voltage. The 1.3- by 1.4-millimeter chip uses four 2.5-volt, 250-millifarad ultracapacitors connected in parallel. When their state of charge reaches 25 percent and their voltages drop to 1.25 V each—below the preset reference voltage—the energy chip reroutes the ultracapacitors’ connections. This switch makes them two series-connected pairs in parallel. Each pair’s voltage equals the sum of its members, or 2.5 V. And the circuit they are powering is back in business—that is, until so much of the remaining charge has drained away that the two pairs are again supplying less than 1.25 V. The chip’s control system then kicks in again, stacking the pairs so that all four ultracapacitors are in series and the voltage is again at 2.5 V. By the time the voltage drops again to the point where it is incompatible with the circuitry in the device it is supposed to power, 98 percent of the original charge has been used. As the ultracapacitors are recharged, the stacking maneuver is reversed, and the single ”stack” reverts to two series-connected pairs—and subsequently to the original four-in-parallel configuration.
According to William Sanchez, the graduate student who is the leader of the project, one of the few improvements that need to be made is in the efficiency of the device’s DC-to-DC converter. The one used in the experiments reported this week delivers roughly half of the energy from the ultracapacitors to the load. A version of the energy chip Sanchez expects to be delivered from a foundry this summer will be 65 to 85 percent efficient, he says. The goal for a commercial device, he adds, is around 90 percent.
The next stage of the group’s work involves the creation of a tiny implantable medical device powered by the energy chip, which will be used to monitor patients with neurological conditions that cause tremors. The project was inspired by a conversation between Joel L. Dawson, head of the MIT research lab where the work is being done, and Dr. Seward Rutkove, a professor of neurology at Boston’s Beth Israel Deaconess Medical Center. As Dawson recalls it, the gist of Rutkove’s concern was that ”if you’re ill, the last thing you want to do is wear something that announces it to others or is inconvenient for you.” To that end, the team aims to deliver ultracapacitor-powered tremor-measuring gadgets no larger than 2 by 2 by 10 mm by the summer of 2011. Rutkove will then be able to test the devices by letting actual patients wear them. The ultracapacitor circuit that was the focus of the group’s VLSI presentation was tested in a much larger device, measuring 14 by 14 by 20 mm.