Tiny vacuum pumps set to soar
Five years ago, as engineers were developing a Mars-bound vacuum pump the size of a D-cell battery, the U.S. Department of Defense had something even smaller in mind: a chip-scale vacuum pump.
The department’s Defense Advanced Research Projects Agency (DARPA), which issued a challenge in 2008 to develop such a pump, revealed in June that research teams from the University of Michigan, the Massachusetts Institute of Technology and Honeywell International Inc. had each successfully demonstrated a chip-scale vacuum pump. Until that point, credit for the smallest vacuum pump in existence belonged to Creare Inc., an engineering R&D firm in Hanover, N.H., that produced a miniature turbomolecular pump for NASA’s Mars Science Laboratory, otherwise known as the Curiosity rover.
The rotor of Honeywell’s chip-scale vacuum pump, as seen in this silicon disk leaning against a penny, contains about 200,000 turbine blades, each about 250µm wide. Image courtesy Honeywell International.
Creare’s hybrid turbomolecular vacuum pump, said to be capable of producing an ultimate pressure in the 10-8 Torr range, landed on Mars aboard Curiosity on Aug. 6, 2012. The pump provides the rover’s on-board mass spectrometer with the vacuum necessary to analyze ionized samples.
While Creare’s vacuum pump is still the smallest on Mars, back on Earth the smallest vacuum pumps are now about the size of a sugar cube. Each pump can achieve a vacuum pressure of 10-6 Torr within the mass spectrometer’s 1mm3 vacuum chamber while using 10 times less power than its larger predecessor. The breakthrough holds promise for military efforts to detect chemical and biological attacks, according to DARPA.
“The process of creating a vacuum in a room large enough to test a spacecraft, for example, is pretty straightforward,” said Andrei Shkel, who managed DARPA’s chip-scale vacuum pump program. “A sealed room, a large pump and ample power are all that is needed.”
That approach cannot be scaled down to work with “microscale vacuum chambers that are slightly larger than a grain of sand,” Shkel explained. “We had to harness new kinds of physics to develop these pumps, requiring precision and miniaturization techniques that have never been attempted.”
Wei Yang, a principle research scientist and Honeywell’s lead investigator on the DARPA-funded project, likens the chip-scale vacuum pump his team developed to “a bridge between the MEMS world and conventional, but very advanced, precision machining.” Although conventional bearing manufacturing technology and metalmachining technology don’t offer the same level of fabrication efficiency as MEMS, they do provide packaging advantages, Yang said.
“By combining the two capabilities,” he continued, “we use MEMS fabrication to produce the silicon disk, then use precision metalmachining to make some MEMS devices work in a fashion that has never been achieved before.”
As an example, Yang said the blades on Honeywell’s chip-scale vacuum pump move at speeds of about 100m/sec. “Typically, you don’t see that kind of speed in a MEMS device,” he added, noting that each blade is only about 250µm wide with a few microns of tolerance.
While most people are familiar with MEMS capabilities, Yang said the same is not true of its limitations. “For example, high speed has never been achieved in MEMS,” he observed. “The moving parts of MEMS devices are all based on deformation. They don’t have freely moving parts; there are cantilevers, and they vibrate. And vibration doesn’t permit very high speed. You’re going to break things by moving too fast.”
Honeywell’s chip-scale vacuum pump instead uses a 1cm-sized silicon disk, or rotor, that contains about 200,000 turbine blades. To levitate the rotor, Yang’s team used ceramic ball bearings as the driving support mechanism.
Honeywell’s fully assembled vacuum pump prototype with a motor. Image courtesy Honeywell International.
Researchers at the University of Michigan developed this 24-stage microscale rough pump that uses tiny micromachined hexagonal compartments where each element of the array serves as either a pump or a valve. Image courtesy DARPA.
While the rotors are produced via the same processes used to produce MEMS-based accelerometers and gyroscopes, packaging for the complete system definitely differs from that of other MEMS devices. But that’s pretty much the story for MEMS in general, Yang added. “There is no universal approach to packaging.”
With Honeywell’s ball bearing design, Yang said, “we had the task of precision mounting these disks with precision spindles, and that’s a problem we have resolved.” Honeywell uses a combination of precision boring and grinding to produce the mounting structures needed for the pump. And, thanks to state-of-the-art metrology, it’s possible to create accurate structures that are compatible with the MEMS portion of the pump, he added.
“By doing the things we did in this micropump program,” Yang observed, “not only did we achieve a great device with a lot of potential, but we began creating a new process as well—a new kind of MEMS and machining hybrid.”
What’s more, Yang added, the DARPA project successfully removed the last obstacle preventing the miniaturization of systems that require vacuum pumps, such as gas detectors and mass spectrometers. DARPA’s ultimate goal, he noted, is to put a mass spectrometer onto an unmanned aerial vehicle (UAV).
The only obstacle preventing that endeavor, Yang said, was the size of a complete mass spectrometer system, which had ranged from that of a shoebox to a dishwasher. For example, he said, helium mass spectrometers used to detect leaks in the chemical and semiconductor processing industries require so much instrumentation that technicians must wheel the equipment around on a cart.
Creare’s miniature turbomolecular vacuum pump is about the size of a D-cell battery and weighs 150g. Image courtesy Creare.
Although a mass spectrometer by itself is smaller than a sugar cube, the device requires a vacuum pump and an energy source to analyze a gas sample. Typically, Yang said, vacuum pumps used with gas detection systems are the size of a brick, consume up to 100w of power and require a lot of batteries. Plus, he noted, they generally are only equipped to test for one type of gas and must be recalibrated on a regular basis.
With the chip-scale vacuum pump, Yang said it would be possible to create a mass spectrometer system the size of a smart phone that could be used to detect multiple gases and would never require recalibration. In this design iteration, software would be the only thing that would change when switching from analyzing one gas type to another. By using different software-based analysis tools, he noted, “you can monitor 10 different gases with one device.”
Though the chip-scale vacuum pumps developed in the DARPA program are not yet in commercial production, Yang anticipates the technology will find its way onto a UAV within the next 2 to 3 years.
All it will take, Yang concluded, is an additional investment—albeit a significant one. It’s hard to imagine much difficulty attracting investors given that the market for chip-scale vacuum pumps appears ready to soar—literally. µ