While computer chips keep shrinking, the size of gadgets is still limited, in part, by relatively large pieces of quartz that act as pacemakers for electronic circuits. But within the next few months, there may be a less expensive, smaller alternative to quartz. In December, a Sunnyvale, CA-based company, SiTime, is expected to start shipping a silicon-based timekeeper that challenges the 60-year-old quartz technology.
Quartz oscillators are used as timekeepers in virtually all of today’s electronics, says Aaron Partridge, SiTime’s chief technology officer, who calls them “the heartbeat of the system.” Each year, he says, about eight billion quartz crystals–the components of an oscillator that vibrate at a certain frequency–are made.
However, quartz technology is based on a decades-old industry that has changed little since integrated circuits were invented. And while silicon fabrication has enabled integrated circuits to shrink and gain complexity, the fabrication techniques used to make quartz resonators haven’t kept pace in size.
By using silicon instead of quartz, SiTime is able to make a resonator with the width of only a couple of hundred micrometers, says Partridge, while an equivalent quartz resonator has a width of about a millimeter. Moreover, Partridge says, silicon resonators are much less expensive to fabricate.
The idea of abandoning quartz oscillators for silicon ones is not new; researchers at Stanford University, University of Michigan, and the University of California, Berkeley have been working on the technology for decades. For the most part, however, the quality of these silicon systems hasn’t matched that of quartz. In recent years, though, advances in the fabrication of microelectronicmechanical systems (MEMS) have made high-quality silicon oscillators more practical.
Partridge explains that SiTime’s resonators, which are made of single-crystal silicon, rely on a tiny voltage to keep the heart of a device ticking. The resonator is encapsulated and closed off from the environment. When a voltage is applied to the oscillator, the resonator vibrates at a specific frequency. If the mass of the resonator changes during its lifetime, it loses time. Partridge says that even a difference in a single layer of atoms can alter the frequency. So, the SiTime engineers use an ultra-clean technique in which the resonator is fabricated and sealed into a tiny vacuum chamber that’s completely devoid of any extra atoms.
Besides the advantage of size, SiTime’s silicon oscillators are an appealing alternative because of their ability to be tuned to different frequencies. Unlike a quartz crystal, which is fabricated to resonate at a certain frequency throughout its lifetime, a silicon oscillator is capable of vibrating at many different frequencies, depending on the software controlling the circuit. This makes the manufacturing process less costly, according to Partridge, since when a quartz crystal is fabricated, it’s designed to resonate at a single frequency throughout its lifetime. Changing the function of the quartz clock from one that operates a cell phone to one that runs a high-definition television, for example, requires fabricating an entirely different batch.
But SiTime’s silicon resonators don’t require frequency adjusting during fabrication, which makes them easier to manufacture. And when they’re ready to be used, an engineer simply tunes to the desired frequency using software.
SiTime is one of a handful of new companies developing MEMS resonator technology, says Roger Howe, professor of electrical engineering at Stanford and chief scientist at Silicon Clocks, a University of California, Berkeley spin-off. The area is “ripe for commercialization,” he says, thanks to a combination of maturing technologies in the MEMS industry and ultra-clean encapsulation techniques. As a result, a number of startups specializing in MEMS resonators have emerged, including Silicon Clocks and Mobius Microsystems, a University of Michigan spin-off.
SiTime has partnered with major manufacturers of quartz clocks, which will sell its silicon oscillators. But there is more work to be done. Partridge says that the products to ship at the end of the year will satisfy a few applications for quartz oscillators, such as simple microprocessor clocks. But they aren’t good enough to replace, for example, quartz timing mechanisms in GPS devices. Future silicon clocks, he says, will have a higher precision, as his team learns to fine-tune the fabrication techniques.