Microfluidics - Introduction to BioMEMS - page 150

he Quake microvalves have the following limitations:
Pattern design versatility : he microchannel height and width cannot be designed as
separate parameters because the photoresist relow process melts the whole volume of
the photoresist. his means that lines of diferent width (but originally designed to be
the same height) end up having diferent heights, and lines designed to be of diferent
heights (but the same width) end up having very similar heights. Although devices
with microchannels of various heights are possible (e.g., optimizing the photoresist
relow), valving of the shallowest channels is diicult because it involves bending a
wider gap of PDMS (because the master is evenly covered by spin-coating with PDMS).
Microscopy : he rounded roof produces a rounded water-PDMS interface that acts
as a lens, which severely distorts images in transmitted-light microscopy modes such
as phase-contrast microscopy—the most prevailing mode of observation of live cells
in culture to date. In practice, the only viable cell imaging option is luorescence
microscopy (in which light is emitted by the sample, so the light is not required to
go through the sample if using an inverted microscope). Unfortunately, luorescence
microscopy comes at a cost—in terms of supplies (dyes) and equipment (UV lamp and
ilters) and because it is limited by the amount of luorescence signal it can collect, as
the signal fades with time and with light exposure (photobleaching).
Fluid dynamics modeling : Unlike for rectangular cross-sections, the equations gov-
erning low in a rounded roof channel do not have an analytical solution, and thus
inite-element modeling is the only option to predict shear stress, low resistance,
and others. he cross-section of the microchannel needs to be characterized for each
pattern design if inite-element modeling is to be undertaken. Even then, the curved
geometry of the channel is not straightforward to input into most modeling packages,
which are standardized for rectangular walls and circular pipes. hus, modeling of
low in microchannels made by photoresist relow is, at the very least, cumbersome
and highly computation-intensive for complex channel architectures. For microlu-
idic cellular studies, in which knowledge of shear stress is paramount, not performing
such low simulations can amount to walking barefoot on gravel in the dark—it can
get painful.
Metering : he volume of a chamber created between two valves is diicult to deter-
mine (although it can be characterized for each device by luorescence microscopy).
his volumetric uncertainty can be important when using the chambers as chemical
reactors, although it is possible to implement a metering scheme.
Portability : he valves are open at rest, requiring energy to close. his implies that
the device cannot be “unplugged” and transported to another location (for example,
a microscopy facility) if the separation between volumes is to be maintained (e.g., in
chemical reactions).
In sum, the Quake microvalves are not optimal in terms of microscopy, compatibility with
cells, modeling, metering, and portability. Nevertheless, despite their limitations, without any
doubt, they have revolutionized the ield of microluidics.
3.8.1.5 PDMS Microvalve: The “Doormat” Design
In 2000, just three months ater Stephen Quake submitted his manuscript to Science and while
it was still in press, Kazuo Hosokawa and Ruytaro Maeda of Japan's AIST (National Institute of
Advanced Industrial Science and Technology) in Tsukuba submitted an entirely diferent PDMS
microvalve design ( Figure 3.39 ). Here, the PDMS membrane is contact-transferred as a ilm,
“sandwiched” between the control channel and the luidic channel. he valve is then formed by
a small pad of PDMS membrane suspended over a control channel and it is positioned under the