Stopped at the border
A separate nanotechnology clean area captures two contrasting cleanroom processes.
By Hank Hogan
John Weaver is facing potential contamination-control problems.
So, as facility manager of the Brick Nanotechnology Center at Purdue University (West Lafayette, Ind.), Weaver is taking steps to combat the problems—a difficult set of tasks that arise when technologies merge and their contamination suppression techniques collide.
The nanotechnology center is building a 25,000-square-foot, mixed-use cleanroom that will be completed this year and occupied next April. Weaver says the cleanroom will be aimed at research—some in microelectromechanical systems (MEMS), and some in the marriage of biology and MEMS, or BioMEMs.
That's where the contamination-control conflicts emerge.
"When you clean the wall in a biological cleanroom, you want to leave a residue that will kill any kind of organics," Weaver says. "In a microelectronics cleanroom, you want to leave no residue whatsoever when you're cleaning the wall because residues will particulate. So, you have a little bit of cross-purpose there."
The strategy adopted by the Nanotechnology Center is to divide and conquer. Biological processes will be done in a separate clean area that adjoins a semiconductor cleanroom. The two technologies will be able to pass parts back and forth as needed.
Cleaning out holes
While quartz, glass, plastics and other materials are currently the BioMEMS substrate of choice, there is also interest in silicon. But mixing silicon circuitry and finely etched surface features with wet biochemistry is a challenge that doesn't have a commercial solution. What's more, many biomedical applications involve disposable elements and so place a premium on cost. Due to these factors, some BioMEMS experts believe that hybrid packages—the electronics on one substrate and the chemistry on another—will be the norm. In that scenario, the evolving technology won't have much impact on cleanrooms and contamination-control techniques.
But other researchers think that the need for low cost, high performance and very small devices will drive the increasing integration of silicon and biology. In some cases, such as a detector aimed at a specific type of virus or bacteria, coating a MEMS structure with antibodies may be required to make the scheme work.
That's the case with products from start-up Protiveris Inc. (Rockville, Md.), which is commercializing BioMEMS technology from Oak Ridge National Laboratory. The company is coating silicon cantilevers on one side with protein, antibody, antigen or DNA segments so that they will bend when exposed to a solution of complementary molecules. (See related story, page 5.) The stated goal is to provide a tool to facilitate protein research and the discovery of new pharmaceuticals.
A split personality
As for what a silicon-based BioMEMS manufacturing facility based on non-sterile manufacturing might look like, one possibility is the split personality of the Brick Nanotechnology Center's new cleanroom. Facility manager Weaver notes that there will be two clean areas—one for the biological side and the other for semiconductor-based operations. Gloveboxes in the wall will allow materials to pass between the two. The filtered air systems for each section will be separate but the water from the semiconductor side will be used to feed the biological side.
Behind this split-personality solution are the different contamination-control needs of the two technologies. Gloveboxes are used because devices must be moved from one area to another, and these minienvironments are relatively easy to seal. The water system method is because the biological side requires a higher degree of sterilization, while the air-handling arrangement is to keep contaminants out of the biological side.
That particular contamination concern also affects the construction of the sub fab that runs beneath both areas. The bio cleanroom floor is, as Weaver puts it, pristine and not penetrated. The main reason is that sealing a floor is difficult. By avoiding such breaches, Weaver says, "There are no little cracks or crevices where bacteria can grow."
MEMS are a 20-year-old technology that borrow heavily from both semiconductor and cleanroom technology, with notable differences and issues in contamination-control protocol, more so when fabricating BioMEMS. (See "Contamination control issues for labs-on-a-chip," above.)
Aaron Fleischman, co-director of the BioMEMS laboratory at The Cleveland Clinic Foundation (Cleveland, Ohio), who is involved in researching this area, says of future projects: "Sterilization downstream is a big issue, whether or not it's going to be heat sterilized or gamma irradiated."
Fleischman says that electronics based on run-of-the-mill CMOS processing can't stand up to gamma radiation, a fairly common sterilization technique in the medical field. Chemical sterilization could be used, but Fleischman says there are no conclusive compatibility studies of such processes and MEMS materials.
There could also be problems with high aspect ratio MEMS devices, which might have holes that would be very deep when compared to their diameter. Liquids would have trouble reaching the bottom of the holes and rooting out bacteria.
For these and other reasons, Fleischman says that sterile manufacturing, if it can be achieved, would be ideal. Sterile manufacturing, however, would likely require changes in cleanrooms and contamination control.
HANK HOGAN is a special correspondent to CleanRooms, living in Austin, Texas. He can be reached at: email@example.com
Contamination control issues for labs-on-a-chip
Born two decades ago, when IBM researcher Kurt Peterson noted that silicon had potentially useful mechanical properties, microelectromechanical systems—MEMS—today form the basis for a multibillion-dollar industry. Companies such as Texas Instruments Inc. (Dallas, Texas), Analog Devices Inc. (Norwood, Mass), Omron Corp. (Kyoto, Japan) and others make micromachined gyros, pressure sensors, ink jet nozzles, micromirrors, and other devices. Analog Devices claims that they have shipped more than 100 million MEMS accelerometers.
In-Stat/MDR, (Scottsdale, Ariz.), a market research firm, predicted last August that MEMS revenues would grow 16 percent a year over the next few years. In-Stat senior analyst and MEMS market expert Marlene Bourne predicted that trend would lead to sales of $7.9 billion in 2007.
Much of commercial MEMS is silicon-based and borrows freely from semiconductor manufacturing. The same goes for its contamination-control procedures, with a few notable differences. For example, MEMS devices may make use of materials harmful to semiconductor circuits and so their construction may be confined to a particular part of a common cleanroom. Micromachined products also require care beyond that which is given to semiconductors.
"We have a standard cleanroom, but extend cleanroom practices to the probe/trim floor and the assembly area," says Bob Sulouff, director of business development for the micromachined products division at Analog Devices.
The reason is that many MEMS devices move. Accelerometers, for instance, may detect motion by the change in capacitance due to the flexing of interlaced silicon fingers. Micromirrors may pivot about an axis in order to turn a pixel on or off.
Such movement means the devices cannot be sealed underneath a protective passivation layer, as is common for integrated circuits. Hence, MEMS devices are at risk from the dust and contamination resulting from the dicing of a wafer into individual units.
Once that dicing is complete, the die can be contaminated during assembly into packages. And even after that, there can be contamination. Some devices, such as accelerometers and micromirrors, can be sealed and protected in hermetic packages. Pressure sensors, on the other hand, must interact with the outside world.
Manufacturers get around these problems, notes Sulouff, by working with vendors to improve the cleanliness of tools used in dicing and assembly. They also use tape and other physical barriers to ensure that particles don't land on die and gum up the micromachines. Sensors and other devices that are exposed can be protected through the use of a barrier, such as a clean and well-controlled oil.
One problem arises because MEMS parts are too clean and need a bit of dirtying up. Sticky friction, or stiction, is a tendency for MEMS parts to stick together or slide poorly, which leads to long-term performance drift or outright failure due to the ultraclean surface of the parts and their small size.
The solution is a surface coating. Proprietary versions have been developed by Texas Instruments and Analog Devices. Start-up Microsurfaces, Inc. (Minneapolis, Minn.) supplies its own brand. The Microsurfaces' coating is a monolayer 1 to 2-nm thick that binds to the surface and prevents stiction. According to the company's Xiaoyang Zhu, the coating also controls surface physical and chemical properties, thereby making the device less susceptible to contamination.
Analog Devices' etch sinks, which are used in the manufacture of MEMS devices, are automated sinks that contain acids, bases and solvents while controlling temperature, etchant filtering, time, and mechanical movement.
Jeff Borenstein, director of the biomedical engineering center at The Charles Stark Draper Laboratory Inc. (Cambridge, Mass.), was originally involved in Draper's MEMS fabrication. Draper develops devices such as gyros and accelerometers, fabricates limited numbers through prototype production, and then transfers them elsewhere for manufacturing.
Despite all of the manufacturing and contamination challenges faced by MEMS, Borenstein notes how incredibly the tiny devices work. "One thing that struck me when I got involved with MEMS was how sturdy the devices really are. When they're done and packaged, they're very sturdy, very robust, and very reliable," he says.
But the same can't be said for BioMEMS. There are not many manufacturers of these devices, nor many units being made—yet. According to a February study by In-Stat, microfluidic sales will grow from $1.7 billion in 2003 to $2.7 billion in 2008. While most of that is and will be in ink jet nozzles, some is and will be in drug delivery, lab-on-a-chip, and other micromachined BioMEMs applications. In-Stat's Bourne notes that sales of lab-on-a-chip and micromachined units ran about $84 million in 2003 and will grow to $264 million by 2007.
While commercial MEMS is overwhelmingly silicon-based, the same isn't true for BioMEMS. Caliper Technologies Corp. (Hopkinton, Mass.), for example, constructs its microfluidic LabChip devices, used for drug discover and other applications, out of quartz and glass. According to Dave DeGrasse, senior director of the company's Mountain View operations, quartz and glass are biocompatible and transparent. Plastic is another possibility but requires higher than current volumes to be cost-effective.
Today, Caliper's manufacturing process starts with what are essentially semiconductor mask blanks, which are put into an ISO Class 6 ballroom-style cleanroom with about 25 percent HEPA filter coverage. In a lithography area that's ISO Class 5, Caliper prints 3 to 5-µm wide features on the substrates, which are then etched to produce channels that are roughly 50-µm wide, 10-µm deep, and possibly hundreds or thousands of microns long. A top plate is thermally sealed to the etched bottom plate in the ISO Class 5 area, and then diced into chips and eventually assembled into systems.
"The contamination issues that we face are very similar to what a semiconductor company would face," notes DeGrasse. The only difference is that the feature size is larger and, hence, the contamination-control needs are less stringent.—HH