The Center for Polymer Microfabrication is designing processes for manufacturing microfluidic chips. Pictured here is a chip fabricated by the center’s tailor-made production machines. Image: Melinda Hale |
In
the not-too-distant future, plastic chips the size of flash cards may quickly
and accurately diagnose diseases such as AIDS and cancer, as well as detect
toxins and pathogens in the environment. Such lab-on-a-chip technology—known as
microfluidics—works by flowing fluid such as blood through microscopic channels
etched into a polymer’s surface. Scientists have devised ways to manipulate the
flow at micro- and nanoscales to detect certain molecules or markers that
signal disease.
Microfluidic
devices have the potential to be fast, cheap, and portable diagnostic tools.
But for the most part, the technology hasn’t yet made it to the marketplace.
While scientists have made successful prototypes in the laboratory,
microfluidic devices—particularly for clinical use—have yet to be manufactured
on a wider scale.
Massachusetts
Institute of Technology’s (MIT) David Hardt is working to move microfluidics
from the laboratory to the factory. Hardt heads the Center for Polymer
Microfabrication—a multidisciplinary research group funded by the Singapore-MIT
Alliance—which is designing manufacturing processes for microfluidics from the
ground up. The group is analyzing the behavior of polymers under factory
conditions, building new tools and machines to make polymer-based chips at
production levels, and designing quality-control processes to check a chip’s
integrity at submicron scales—all while minimizing the cost of manufacturing.
“These
are devices that people want to make by the millions, for a few pennies each,”
says Hardt, the Ralph E. and Eloise F. Cross Professor of Mechanical
Engineering at MIT. “The material cost is close to zero, there’s not enough
plastic here to send a bill for. So you have to get the manufacturing cost
down.”
Micromachines
Hardt and his colleagues found that in making microfluidic chips, many research
groups and startups have adopted equipment mainly from the semiconductor industry.
Hardt says this equipment—such as nanoindenting and bonding machines—is
incredibly expensive, and was never designed to work on polymer-based
materials. Instead, Hardt’s team looked for ways to design cheaper equipment
that’s better suited to work with polymers.
The
group focused on an imprinting technique called microembossing, in which a
polymer is heated, then stamped with a pattern of tiny channels. In experiments
with existing machines, the researchers discovered a flaw in the embossing
process: When they tried to disengage the stamping tool from the cooled chip,
much of the plastic ripped out with it.
To
prevent embossing failures in a manufacturing setting, the team studied the
interactions between the cooling polymer and the embossing tool, measuring the
mechanical forces between the two. The researchers then used the measurements
to build embossing machines specifically designed to minimize polymer “stickiness.” In experiments, the group found that the machines fabricated
chips quickly and accurately, “at very low cost,” Hardt says. “In many cases it
makes sense to build your own equipment for the task at hand,” he adds.
In
addition to building microfluidic equipment, Hardt and his team are coming up
with innovative quality-control techniques. Unlike automobile parts on an
assembly line that can be quickly inspected with the naked eye, microfluidic
chips carry tiny features, some of which can only be seen with a
high-resolution microscope. Checking every feature on even one chip is a
time-intensive exercise.
Hardt
and his colleagues came up with a fast and reliable way to gauge the “health”
of a chip’s production process. Instead of checking whether every channel on a
chip has been embossed, the group added an extra feature—a tiny X—to the chip
pattern. They designed the feature to be more difficult to emboss than the rest
of the chip. Hardt says how sharply the X is stamped is a good indication of
whether the rest of the chip has been rendered accurately.
Jumpstarting an industry
The group’s ultimate goal is to change how manufacturing is done. Typically, an
industry builds up its production processes gradually, making adjustments and
improvements over time. Hardt says the semiconductor industry is a prime
example of manufacturing’s iterative process.
“Now
what they do in manufacturing is impossibly difficult, but it’s been a series
of small incremental improvements over years,” Hardt says. “We’re trying to
jumpstart that and not wait until industry identifies all these problems when
they’re trying to make a product.”
The
group is now investigating ways to design a “self-correcting factory” in which
products are automatically tested. If the product doesn’t work, Hardt envisions
the manufacturing process changing in response, adjusting settings on machines
to correct the process. For example, the team is looking for ways to evaluate
how fluid flows through a manufactured chip. The point at which two fluids mix
within a chip should be exactly the same in every chip produced. If that mixing
point drifts from chip to chip, Hardt and his colleagues have developed
algorithms that adjust equipment to correct the drift.
“We’re
at the stage where we’d like industry to know what we’re doing,” Hardt says. “We’ve been sort of laboring in the vineyard for years, and now we have this
base, and it could get to the point where we’re ahead of the group.”