Micromachines 2014, 5, 50-65; doi:10.

3390/mi5010050
OPEN ACCESS

micromachines
ISSN 2072-666X
www.mdpi.com/journal/micromachines
Article

Dynamics of Microvalve Operations in Integrated Microfluidics

Alan T. H. Lau, Hon Ming Yip, Kathy C. C. Ng, Xin Cui and Raymond H. W. Lam *

Department of Mechanical and Biomedical Engineering, City University of Hong Kong,

Hong Kong; E-Mails: [email protected] (A.T.H.L.);
[email protected] (H.M.Y.); [email protected] (K.C.C.N.);
[email protected] (X.C.)

* Author to whom correspondence should be addressed; E-Mail: [email protected];

Tel.: +852-3442-8577; Fax: +852-3442-0172.

Received: 15 December 2013; in revised form: 20 January 2014 / Accepted: 21 January 2014 /
Published: 10 February 2014

Abstract: Pneumatic microvalves are widely used key components for automating liquid
manipulation and flow control in microfluidics for more than one decade. Due to their
robust operations and the ease of fabrication, tremendous microfluidic systems have been
developed with the multiple microvalves for higher throughput and extended
functionalities. Therefore, operation performance of the microvalves in the integrated
microfluidic devices is crucial to the related applications, in fields such as micro-flows,
cell analyses, drug discovery, and physical/chemical detections. It has been reported that
operation performance of the microvalves are highly sensitive to the device configuration
and pressurization scheme. This implies the further development of integrated
microfluidics with a larger number of the valves may suffer the problems of undetermined
microvalve behaviors during operations, which can become an unavoidable hurdle in the
device design and optimization processes. Herein, we characterize responses of the
individual microvalves for different operation configurations, e.g., membrane thicknesses
and driving pressures. We investigate also the effects in microfluidics integrated with the
more valves, through experiments, modeling and simulations. We show that dynamics of
the microvalves is indeed influenced by the configurations, levels of design complexity and
positions in the devices. Overall, taken dynamics of the microvalve responses into
considerations, this work provides insights and guidelines for better designs of integrated
microfluidics for the future applications requiring higher throughput and improved
operation performance.
Micromachines 2014, 5 51

Keywords: microfluidic; valve; lumped model; integration; response

1. Introduction

Since the past two decades, miniaturization of the fluidic operations had become the great demand
for applications requiring precise control of liquid manipulations [1,2]. Microfluidic systems typically
combined with individual components such as valves, pumps [3], mixers [4,5], filters/sorters [6],
sensors and heaters [7]. The related technologies [1,8] have already been commercialized into highly
automated products, e.g., Fluidigm, South San Francisco, CA, USA [9]. Multilayer soft lithography for
PDMS emerged as a ground-breaking technology for numerous biological and chemical applications,
such as protein crystallization [10], blood diagnostics [11], chemical synthesis [10], bio-molecular
separations [12–14], enzymatic assays [15,16], immuno-hybridization reactions [17], large-scale
single-molecule assays [18], self-contained cell culture [19], cell sorting [20], cell-based screening [21],
drug screening [22], self-sustainable microfluidic cell cultures [4,23], and oxygenated microenvironments
for mammalian and bacterial cells [24]. Thorsen et al. [25] demonstrated a microfluidic platform as
a milestone of the large-scale integration, which contained thousands of pneumatic microvalves and
hundreds of individually addressable microchambers. A microfabricated emulsion generator
array [1,18,26,27] has been established for high-throughput single-cell analyses. A microfluidic
platform integrated with microarray technology for cellular viral-host proteomic interaction analysis
has also been reported recently [28].
There have been increasing interests in applying dynamics of the pneumatic microvalves in
automated microfluidic operations. For example, an array of microvalves was configured to control
pulsed sample flows for a micro-scale flow injection analysis [29]. Due to elasticity of the valve
membrane materials (polydimethylsiloxane (PDMS)) [30], the applied pressure in an overhead
actuation channel can modulate flow rates in the working flow microchannels by defining the
membrane deflection and the corresponding fluidic resistance [31]. Moreover, the pneumatic rotary
pumps were applied to mix a defined amount of medium in closed microchambers [23] based on the
Taylor dispersion effect [32]. While the pneumatic microvalve has proven as an effective fluidic
actuation component, the essential precise controls of membrane deflections highly depend on the
pneumatic pump configurations, such as the valve dimensions and pressure sources [4]. Indeed,
considering a microvalve gates an underneath flow channel based on deflection of the elastomeric
valve membrane, responses of the microvalve should rely on physical effects such as the membrane
stiffness and the fluidic resistance in liquid flows. The microvalves in microfluidic devices can have
distinct levels of complexity (e.g., quantities and arrangements), and therefore physical characteristics
of the individual microvalves can vary for different device designs. Together, there is necessity for the
detailed analysis of microvalve operations for different degrees of microfluidic integration. The
universal strategy converting the microvalve-driven microfluidic devices into mathematical models
would be extremely helpful in predicting the microvalve responses, and further provide directions to
optimize overall performance of the devices.
Micromachines 2014, 5 52

Typically, microfluidic devices fabricated by soft lithography have been modeled as networks
consisting of multiple lumped elements, which can be described by simple analytical relation
individually [33]. For example, Bourouia and Grandchamp modeled vibration-based micropumps by
defining non-linear mechanical properties for the silicon flat membranes and check-valves [34].
Mastrangelo’s group has established multiple microfluidic devices using the lumped element strategy
to regulate chemical concentrations defined by difference schemes [35], e.g., the “concentration
digital-to-analog converter” employed multiple inputs as a binary number to generate discretized
chemical concentrations [36], the “pulse code modulator” defining the chemical concentration by
generating a flowing stream of solute-solvent plugs with variable lengths and mixing the liquids along
a long microchannel [37], and a microchannel network as a band-pass filter for the variation of
chemical concentrations in a solution [38]. In addition, Hong and Pan described a flow rate-controlled
switch as a lumped element for the design of micro-flow regulation devices [39].
In this work, we systematically investigate dynamic responses of the pneumatic microvalves in the
integrated microfluidics conditions. Though a few studies on the pneumatic microvalves have been
reported previously [31,40,41], their focuses were not related to the large-scale integration aspect.
Based on the multilayer soft lithography [3], we first design and fabricate microfluidic devices
consisting of a manageable number of microvalves, which can represent the major physical settings of
the general valve-based microfluidics. We quantify membrane stiffnesses of the individual
microvalves under both static and stepwise pressure actuations via the experiments under different
operation conditions. On the other hand, we simplify the system characteristics into mathematical
models by considering only dominant physical factors in the multiple-valve microfluidic devices, i.e.,
membrane stiffnesses and fluidic resistances in channels. We then adopt the experimental stiffness
values and fit them in the models to predict responses of every microvalve in the devices. Comparing
these predictions with the corresponding experimental measurements can validate the proposed
modeling strategy for the general integrated microfluidics.

2. Methods

2.1. Fabrication

Two molds were fabricated for microstructures required in the microvalve devices based on
photolithography as shown in Figure 1. To fabricate the mold for the control layer (blue patterns in
Figure 1a), SU-8 negative photoresist (SU-8 25, Microchem, Newton, MA, USA) photoresist was
spin-coated on a silicon wafer (~2400 rpm) for a thickness of 20 μm, prebaked at 65 °C for 2 min and
soft-baked at 95 °C for 5 min on a hot plate, followed by ultraviolet (UV) exposure using a mask
including the desired micro-patterns for 40 s (4 times of 10-second exposures). Post exposure bake was
carried out at 65 °C for 1 min and at 95 °C for 2 min. SU-8 developer (Microchem) was then applied to
remove the unexposed photoresist. To enhance mechanical properties of the cross-linked photoresist,
the mold was baked again on a hot plate at 65 °C for 30 min.
For the second mold, the fabrication of SU-8 on a silicon wafer was adopted again for narrow
channels (width: 20 µm) in the flow layers (black patterns in Figure 1a), with a thickness of ~5 µm
(SU-8, Microchem). AZ4620 photoresist (AZ Electronic Materials, Branchburg, NJ, USA) was then
Micromachines 2014, 5 53

patterned on the mold for the rest flow channels (red patterns in Figure 1a) using photolithography.
Briefly, hexamethyldisilazane (Sigma-aldrich, St. Louis, MO, USA) was first applied on the patterned
SU-8/silicon wafer in order to promote adhesion of AZ4620 on silicon. AZ4620 was spin-coated at
1400 rpm for 30 s, baked at 95 °C for 30 min and exposed under UV light for 80 s (4 times of
20-s exposures). After developing the photoresist, the mold was reflowed on a hot plate for 1 min at
140 °C, which is at the temperature region for glass-liquid transition of the photoresist. The reflow
process can create rounded profiles of the mold structures and of the microchannel cross sections in the
PDMS devices. To facilitate the release of PDMS layers from substrates, the two molds were silanized
with a high-molecular-weight trichloro-perfluorooctyl saline (Sigma-Aldrich) in a chemical fume hood
for 2 h before the downstream device manufacturing processes.

Figure 1. (a) Design layout of the microvalve devices. Upper-right insets: a folded flow
channel structure and a region overlapping with flow channels and valve chambers. The
channels/chambers were filled with color dyes for better visualization. Lower-right inset:
Individual microvalve structure labeled with key configuration parameters. This structure
only shows the two spin-coated polydimethylsiloxane (PDMS) layers, whereas the upper
thick PDMS layer and the underneath glass slide are skipped. Scale bar in inset: 100 µm;
(b) Fabricated microfluidic device containing 100 microvalves.

Afterwards, PDMS pre-polymer (Sylgard 184, Dow Corning, Midland, MI, USA) was prepared by
mixing the monomer and the curing agent with a 10:1 volumetric ratio using a glass rod. To remove air
bubbles trapped in PDMS during mixing, vacuum was applied to degas the pre-polymer in a vacuum
bell jar. The PDMS was then poured onto the mold fabricated with the control channel patterns and
baked in an oven at 80 °C for 20 min. The PDMS substrates with the control channel patterns were cut
from the mold with a traversing blade. The substrates were then peeled off and punched at the channel
openings using a hole-chopper (Cat# 15071, Harris Uni-Core, Ted Pella, Inc., Redding, CA, USA) for
gas pressure connections. On the other hand, PDMS pre-polymer was spin-coated for 60 s on the
Micromachines 2014, 5 54

flow-channel AZ/silicon mold at a specified rotational speed (2400, 2600, 2800 or 3200 rpm) for
different thicknesses (34, 28, 25 or 18 µm), and on glass slides at 3000 rpm for a thickness of ~20 µm
for the PDMS base layer (HB in Figure 1a). The spin-coated PDMS layer with the mold partially cured
in an oven at 80 °C for 9 min, while the PDMS layers on glass slides were baked at 80 °C for >2 h.
Afterwards, the control-channel PDMS substrates were subsequently aligned and placed onto the
flow-channel PDMS layer under a stereo microscope (JSZ-6S, Yee Mau Industrial Co., Hong Kong).
The bound PDMS substrates were baked again in an oven at for 2 h at 80 °C such that further curing of
the PDMS interface could induce permanent PDMS bonding. The lower flow-channel layer was then
chopped along the device boundaries using a razor blade. After peeling off the PDMS substrates,
flow-channel inlets and outlets were generated by punching holes through the substrates. Oxygen
plasma treatment (PDC-002, Harrick Plasma, Ithaca, NY, USA) with power 30 W for 30 s was applied
to the multilayer PDMS substrates and the PDMS/glass slides, in order to activate surfaces on both the
channel sides. Subsequently, permanent seal of the flow channels was achieved by bonding the
surfaces together. To finish the fabrication as shown in Figure 1b, the devices were baked in an oven
overnight to ensure thorough PDMS curing.

2.2. Measurement of Valve Responses

We observed dynamics of the microvalves under an inverted optical microscope (PW-BDS500EPI,

Proway Optics and Electronics, Ningbo, China) equipped with a cooled CCD digital camera
(TCC5.01CE, Tucsen Imaging Technology, Fuzhou, China). Fluidic connections for both the control
and flow channels were achieved by inserting stainless steel adaptors (New England Small Tube,
Litchfield, NH, USA) and Tygon tubing (Cole-Parmer, Vernon Hills, IL, USA). To prepare the
experiments, we injected red dye (McCormick, Hunt Valley, MD, USA) into the flow channels using
syringes for better visualization, while distilled water was pressurized into the control channels
at ~5 psi.
In this work, measurements of the valve responses included two major aspects: the membrane
stiffness and the transient response. We investigated the membrane stiffness by recording microscopic
images of the membrane deflections under different static pressures along the control channels. For the
dynamic responses, we applied a step-signal to an electro-fluidic valve (LHDA1211111H,
Lee Products, Buckinghamshire, UK) to switch the driving pressures for the microvalve actuation or
deactuation, while the signal was simultaneously fed also to a light emitting diode (LED) for
visualizing the command. We used a high-speed (1000 frames per second) digital camera (EX-FH20,
Casio, Tokyo, Japan) to capture both the LED illumination and the deformation of the microvalves
through an eyepiece on the microscope. The LED has only negligible time delay (<70 ns) comparing
to the frame-refreshing time in the videos (1 ms). Afterwards, we extracted snapshots of the captured
videos and analyzed for response time of the microvalves under different operation configurations.

2.3. Membrane Deformation Simulation

Mechanical deformations of single valves with defined configurations were computed by

commercial finite element analysis software (COMSOL Multiphysics 4.3, COMSOL, Burlington, MA,
USA). A two-dimensional computational model with HM = 25 µm was constructed for the valve
Micromachines 2014, 5 55

“center” cross-section as indicated in the inset of Figure 1a. Because of the large deformation during
the valve operation, we adopted the Neo-Hookean hyperelastic model [42] (Lamé constant
µ = 678.6 kPa; Laméconstant λ = 1.0714 MPa) with the density ρ = 920 kg/m3, the Young’s modulus
E = 2.2 MPa [43], and the Poisson’s ratio γ = 0.4 [44], in order to describe the mechanical behaviors of
PDMS. For the material domains, triangular meshes were generated with an average element size of
~0.2 µm2; and all peripheries of the flow channel region were set as contact pairs. The boundary
conditions were defined according to operations of the single valves in the practical implementation.

2.4. Simulation of Lumped Circuit Models for Multiple-Valve Devices

We simplified physical characteristics of the devices with different numbers of multiple

microvalves as lumped circuit models to simulate the dynamic responses of all the microvalves.
Subsequently, we expressed each lumped model as a series of time-dependent differential equations.
(Details are described later in the Section Lumped Modeling for Prediction of Valve Dynamics.) We
wrote MATLAB scripts (version R2010b, Mathworks, Natick, MA, USA) to numerically solve these
equations by a finite different scheme similar to the Crank-Nicolson method, which discretizes the
time-space as an average of forward and backward Euler approximations to achieve second-order
convergence in the time-space.

2.5. Statistics

p-Values were calculated using the two-tail Student’s t-test to identify whether there was
a significant change for two selected sets of experimental data. Since the data sets in the current study
had independent sample sizes and standard deviations, we adopted a standard t-test scheme, also
known as the Welch’s t-test. In all comparisons, we defined a significant change by p < 0.05.

3. Results and Discussion

3.1. Device Configuration

We designed and repeatedly manufactured microfluidic devices to analyze the functional responses
in the microfluidics integrated with multiple pneumatic microvalves. These devices were fabricated
with the two-layer soft lithography [3] with a control channel layer overhead the flow channels (Figure 1).
We considered the microfluidic designs consisting of multiple pneumatic microvalves (1 to 100
microvalves) connected in serial. In essence, these testing devices should reflect major dynamics of the
typical integrated microfluidics, which has analogous device configurations with multiple microvalves
in the control channels. The microvalves were designed to gate parallel microchannels sharing the
same liquid inlet and outlet (Figure 1b).
Here, we considered fixed dimensions of the upper valve chambers (length: LV = 150 µm and width:
WV = 200 µm) and the flow channels (width: W = 100 µm, center height: H and radius of surface
curvature: R) as shown in Figure 1a. We calculated the cross-section geometry of a reflowed channel
(Figure 1a) by correlating a relation that the cross-section area after reflow was unchanged:
 R2 / 2  W  R  H  / 2  WH o (1)
Micromachines 2014, 5 56

where R is the radius of curvature, θ is the arc angle of the channel surface profile, and H o is the
height of channel structures on the mold before reflow. With the measurable dimensions (W and H o )
of the fabricated microchannels, we further applied the geometric relations to solve for R and H:
tan( / 2)  W / 2  R  H  (2)
and
R  W 2 / (8H )  H / 2 (3)
The microfluidic devices used in this study had W = 100 μm and H o = 10 μm and hence we
obtained that R = 92.14 µm and H = 14.75 µm. In our studies, we investigated the device operations by
considering different thicknesses of the PDMS membrane layers (HM) and applied pressures on the
membranes (p).

3.2. Individual Membrane Deformation

Characterization of the valve deformation was performed by both experimental and computational
studies. We fabricated the single-valve devices with different spin-coated PDMS thickness (HM) and
recorded microscopic images of the microvalves for different driving pressures. Figure 2a shows that
the color of a microvalve region in a channel injected with a red dye changed greatly under a larger
driving pressure. Considering that the intensity level should reflect the amount of liquid and hence the
channel volume in the microvalve region, we wrote MATLAB scripts to quantify the corresponding
membrane deflections using the intensity changes in the entire areas (white box in Figure 2a) and the
center slices (black box in Figure 2a) of the valve regions indicated in Figure 2a (upper left).
Additionally, we performed simulations to validate the correlations between the membrane deflections
and the intensity changes. Because the membrane deflections at the center cross-sections (Figure 1a)
can be simplified as two-dimensional models based on the geometric symmetry, we constructed
numerical models representing the “center” cross-sections for and defined the boundary conditions
accordingly. Figure 2b shows a model consisting of a control channel, a flow channel and a base; and
their dimensions matches the device configuration adopted in this work. We then imposed compressive
pressures on the membrane tops of the models, ranging from 0 psi to 12 psi, until the valve closure was
obtained. We compared the simulated membrane deflections (Figure 2b) with the experimental
intensities under the sample device conditions with reasonable agreement (Figure 2c, left plot). This
correlation implies that the intensity changes over the entire valve areas should reflect the overall
membrane deflections. In particular, Figure 2c (right plot) indicates that even when the valve closure is
obtained at ~6.5 psi (without any red dye at the “center region”), the membranes can further deflect
under higher driving pressures (the “entire valve” curve).
Micromachines 2014, 5 57

Figure 2. (a) Photographs of a representative microvalve under different driving pressures.

Scale bar: 50 µm; (b) Simulations of deformation at the “center” cross-section in a
microvalve (HM = 25 µm) under different levels of a driving pressure (p). Sample stress
profiles of the model where p = 3 psi, 4.5 psi and 12 psi are provided here; (c) Comparison
of the flow channel area at the center cross-section between experimental and simulation
values (left), and color intensities of microvalve regions in the photographs mentioned in
sub-figure a (right). Each data point was obtained from measurements of at least 3 devices.
Error bars represent the standard deviations.

3.3. Dynamics of Individual Microvalves

We investigated dynamics of the microvalve operations fabricated with different spin-coated PDMS
layer thicknesses (18, 25, 28 and 34 µm). We examined transient responses at the microvalve regions,
whose underneath flow channels were injected with a red dye for visualization, under different levels
of driving pressures ranging from 8 psi to 20 psi. In each experiment, we characterized the microvalve
dynamics in terms of attachment time, closure time and restoring time. The attachment time of
a microvalve was defined as the time required for a deflecting membrane under a driving pressure first
Micromachines 2014, 5 58

contacted with the bottom of the flow channel. The closure time was defined as the period taken for a
microvalve completely blocked the underneath flow channel. As aforementioned in Section 2.2, the
reference times for both membrane attachment and closure were indicated by the lighting of a
neighboring LED. The restoring time was defined as the duration required for a fully deflected
membrane to return to its open state (with ~80% of the channel undeflected volume) after the driving
pressure was released, indicated by the de-illumination of the LED. As demonstration, Figure 3a shows
representative time-lapsed images of membrane deflections in a microvalve area (layer thickness
HM = 25 µm) during a constant driving pressure (12 psi) was applied (left) and then released (right).
Once after a microvalve was pressurized, the red color intensity at the valve region reduced (8 ms) and
the valve center turned to transparent at the attachment time (~12 ms). The expanding transparent area
indicated further membrane deflection and then the channel was fully blocked at the closure time
(~16 ms). The transparent region continuously expanded until the membrane reached its maximum
deflection (~32 ms). We subsequently released the pressure to record the restoring time (~40 ms),
which was longer than the closure time for every microvalve in all our measurements involved in this
work. This observation may be explained by the fact that the restoring time included an additional
period for a membrane deforms from its maximum deflection state to the channel-closure state,
comparing to the time for valve closure.

Figure 3. (a) Time-lapsed micrographs of a microvalve when it was pressurized and

de-actuated. Scale bar: 50 µm; (b) Plots of attachment time, closure time and restoring time
against the driving pressure of single-valve devices fabricated with different membrane
thicknesses. Error bars indicate standard deviations of the data points, each obtained
from 3–5 fabricated devices. Standard deviations smaller than the data symbol height are
not shown.
Micromachines 2014, 5 59

Selection of the configuration parameters is important for the valve operations. In particular,
complete valve closure cannot be achieved by an insufficient driving pressure below a cutoff level,
which increases with the membrane thickness. In the experiments, we could not obtain the
attachment/closure of microvalves with a spin-coated PDMS thickness of 34 µm driven by a pressure
of 8 psi as indicated in Figure 3b. In addition, ultra-thin spin-coated PDMS layers (<16 µm based on
our experiments) were not stiff enough to restore to their undeformed states once pressurized.

3.4. Dynamics of Multiple-Valve Operations

To study the dynamics of microvalves in integrated microfluidics, we performed experiments on the

microfluidic devices, in which a shared control line linked up the microvalves overhead an array of
parallel microchannnels, as described in Device configuration and Figure 1. We fabricated various
devices consisting of different numbers of microvalves (1, 3, 10, 30 and 100) and compared the
operation characteristics of microvalves at different positions and in different device types. In these
sets of experiments, we were interested in the variations on valve responses mainly caused by the
device complexity or, more specifically, the number of microvalves in control lines; therefore we fixed
the spin-coated PDMS layer thickness (25 µm) and the driving pressure (12 psi). For each device
condition, we performed measurements on multiple individually fabricated devices (n = 9–18) to
ensure the repeatability. Our results (Figure 4) indicate that the devices containing more microvalves
along the control lines caused the significantly slower responses (i.e., longer attachment, closure and
restoring times). Moreover, in order to identify the microvalve positions in a device containing X
microvalves, we describe here the valve positions with consecutive indices, with the valve closest to
the pressure source as “position 1” whereas the valve located the end of a control line with “position X”.
We analyzed the response times for microvalves with selected position indices as shown in Figure 4 and
demonstrated that the responses of microvalves in a device were position-dependent, i.e., a valve
located further from the pressure source (larger position index) in general responded slower along the
control line. For instance, the microvalve at position 100 in the 100-valve device could have a
restoring time of >120 ms, which was about three times of the restoring time for a single-valve device.

3.5. Lumped Modeling for Prediction of Valve Dynamics

It has been demonstrated in the previous sections that operation performance of individual
microvalves varies in different microfluidic designs and configurations. In the general cases, additional
microvalves along a control line would slow down responses of all the microvalves. A modeling
technique predicting the essential dynamic behaviors for every valve in an integrated device can
facilitate analyses of the further integration of microfluidics, which involves parallel operations of
a large number of the microvalves. For this purpose, we lumped and converted dominating physical
effects of microvalves in our testing devices as circuit models with equivalent dynamic responses, by
expressing gas/liquid pressure as voltage, fluid flow rate as current, fluidic resistance as resistors, and
membrane stiffness as capacitors. The resultant circuit model of an X-valve device consisting of
multiple branches of capacitors and resistors is shown in Figure 5a. Considering that the flow of
a rectangular channel in either the control or the flow layer has a sufficiently low Reynolds number
(Re << 1) that viscous effects dominates, fluidic resistance (RI) of an individual channel is [45].
Micromachines 2014, 5 60

12 LI  192 H I 
tanh[(2n  1)WI / (2 H I )] 
RI  1  5   (4)
WI H I 3   WI n 0 (2n  1)5 
where μ is liquid viscosity (~10−3 Pa s for water), WI is channel width, HI is channel height, and LI is
total channel length. For the microfluidic devices used in this study, we considered the resistances
along microchannels at different locations: the pressure inlet channel RVIN (=2.4 × 104 Ns/m3),
the bridging channel linking up neighboring valves RVG (=3.07 × 103 Ns/m3), the inlet channels
RIN (=8.4 × 105 Ns/m3) and outlet channels ROUT (=RIN) in the flow layer. On the other hand, we
expressed the membrane stiffness as the equivalent capacitance CM as a function of the driving
pressure (with a unit of volumetric deformation per pressure), i.e., CM = AV HO(1 − VC)/p, where
AV (=WLV in Figure 1a) is membrane area, p is net downward pressure, HO is height of channel
structure on mold before reflow as defined previously, and VC is normalized volume a flow channel
section underneath a membrane, ranging from 0 (closure) to 1 (fully open state). Since Figure 2c
indicates that both the normalized total flow channel volume in microvalve regions tended to reduce
exponentially with the driving pressures, we approximated the channel volume as VC = e−kP, where k is
a constant. As shown in Figure 5b (the plot on the left), we obtained k = 0.22 psi−1 (or 3.191 × 10−2 kPa−1)
by fitting k for the least square errors of VC with the experimental mean valves reported in Figure 2c.
These experimental results (Figure 2) indicate also that the downward pressure levels for the microvalve
attachment and closure were ~3.5 psi (equivalent to VC ~ 0.463) and ~6.5 psi (VC ~ 0.239),
respectively. Therefore, we adopted:
CM  WLV  H o 1  e kP  / p (5)

for the membranes in order to describe such nonlinear physical behavior (Figure 5b, right).
To describe essential states in the models, our computation recorded transient profiles of the
“nodal” pressures above and below the X membranes and hence there were 2X state-variables in total
(Figure 5a). Applying the conservation of mass for liquid flows (or the nominal Kirchoff’s current law
in the models) at all the 2X nodes, we obtained the interconnecting relation for each node position i as
the followings:
(VS − V1)/RVIN = (V1 − V2)/RVG + CM d(V1 − VX+1)/dt for i = 1 (6)
(Vi-1 − Vi)/RVG = (Vi+1 − Vi)/RVG + CM d(Vi − VX+i)/dt for i = 2, 3, …, X – 1 (7)
(VX-1 − VX)/RVG = CM d(VX − V2X)/dt for i = X (8)
(VS − V1)/RVIN = (V1 − V2)/RVG + VX+1 (1/RIN + 1/ROUT) for i = X + 1 (9)
(Vi-1 − Vi)/RVG = (Vi+1 − Vi)/RVG + Vi (1/RIN + 1/ROUT) for i = X + 2, X + 3, …, 2X – 1 (10)
(VX-1 − VX)/RVG = V2X (1/RIN + 1/ROUT) for i = 2X (11)
We then converted each model into 2X equations expressed in the form ΨVS = ΘV + ΦdV/dt, where
VS is the device driving pressure, Ψ is a 2X × 1 vector mapping effects of VS to the 2X states,
V = [V1, V2, …, V2X]T is a vector including all the state variables, and Θ and Φ are time-variant
2X × 2X matrices describing the rest relations in the model. It should be mentioned that in this study
we considered all the multiple-valve devices with the common spin-coated PDMS thickness
(HM = 25 µm) and level of the pressure actuation (12 psi).
Micromachines 2014, 5 61

Figure 4. Bar charts showing response times of microfluidic devices with different
numbers of valves connected in serial. (a) Attachment time; (b) closure time; and
(c) restoring time of selected valve positions are listed for devices with 1, 3, 10, 30 and 100
microvalves. Error bars represent the standard deviations. Hashes indicate significant time
increments (p < 0.05) of the valves closest to the gas inlets (“position 1”) in different
multi-valve devices relative to the single-valve ones. Each asterisk indicates a significant
time increment (p < 0.05) of a microvalve comparing to the valve at “position 1” of the
same device type.
Micromachines 2014, 5 62

Figure 5. (a) Lumped circuit model of a multiple-valve device. The driving pressure VS
was set as square-wave inputs according to the operation. CM is equivalent capacitance for
the PDMS membrane. RVIN, RVG, RIN and ROUT are fluidic resistances of different channels
in the device; (b) Plots of channel volume in the valve region (left) and membrane stiffness
(right) as functions of the net downward pressure acting on the membrane; (c) Transient VC
of a microvalve (valve position “1” in a 100-valve device) under the actuation and
deactuation of VS; (d) Attachment, closure and restoring times of the single-valve devices.
Only average values of the experimental results from Figure 3 are shown here;
(e) Comparison of the responses times between the experimental values and the predictions
using the circuit models. The bars indicate averages of the experimental values from Figure 4.

4. Conclusions

In this research, we investigated dynamic responses of the pneumatic microvalve operations in

integrated microfluidics. Through measurements on the various representative single-/multiple-valve
designs and operation conditions, we examined in detailed the effects of different configurations
parameters on the attachment, closure and restoring times of the microvalves. We demonstrated that
the slower responses in the valve attachment and closure were caused by a thicker membrane layer or
an insufficient driving pressure. A longer restoring time was observed under an excessive driving
pressure. Thus, a driving pressure should be chosen with an appropriate range for the short overall
response period. The microfluidic devices integrated with more microvalves will have the slower
responses. This relation indicates also the expected response retardation in the largely integrated
microfluidic devices. In particular, the response times can become highly inconsistent between
different valves (i.e., larger the valve “position”, longer the response times) in highly integrated
microfluidics with ~100 valves in the control lines. Further, our simulation results based on lumped
circuit models have shown reasonable agreement with the experimental data. These comparisons have
Micromachines 2014, 5 63

validated that the modeling strategy is effective for predicting the valve behaviors in general integrated
microfluidics. Applying this analysis strategy in the device designs can help determine the allowable
level of integration (e.g., the number of valves along each control line) and further optimize the
devices according to the target operation performance. We anticipate that this strategy can become an
essential tool supporting the rapid development and integration of microfluidics, and the device
commercialization processes.

Acknowledgments

We thank the financial supports from City University of Hong Kong (project# 7200267 and
9610212), Croucher Foundation (start-up allowance), and Early Career Scheme of Hong Kong
Research Grant Council (project# RGC124212).

Conflicts of Interest

The authors declare no conflicts of interest.

References

1. Whitesides, G.M. The origins and the future of microfluidics. Nature 2006, 442, 368–373.
2. Ho, C.-M.; Tai, Y.-C. Micro-electro-mechanical-systems (MEMS) and fluid flows. Annu. Rev.
Fluid Mech. 1998, 30, 579–612.
3. Unger, M.A.; Chou, H.-P.; Thorsen, T.; Scherer, A.; Quake, S.R. Monolithic microfabricated
valves and pumps by multilayer soft lithography. Science 2000, 288, 113–116.
4. Hong, J.W.; Studer, V.; Hang, G.; Anderson, W.F.; Quake, S.R. A nanoliter-scale nucleic acid
processor with parallel architecture. Nat. Biotechnol. 2004, 22, 435–439.
5. Stroock, A.D.; Dertinger, S.K.; Ajdari, A.; Mezic, I.; Stone, H.A.; Whitesides, G.M.
Chaotic mixer for microchannels. Science 2002, 295, 647–651.
6. Fu, A.Y.; Chou, H.P.; Spence, C.; Arnold, F.H.; Quake, S.R. An integrated microfabricated cell
sorter. Anal. Chem. 2002, 74, 2451–2457.
7. Wu, J.; Cao, W.; Wen, W.; Chang, D.C.; Sheng, P. Polydimethylsiloxane microfluidic
chip with integrated microheater and thermal sensor. Biomicrofluidics 2009, 3, 012005;
doi:10.1063/1.3058587.
8. Melin, J.; Quake, S.R. Microfluidic large-scale integration: The evolution of design rules for
biological automation. Annu. Rev. Biophys. Biomol. Struct. 2007, 36, 213–231.
9. Smith, C. Tools for drug discovery: Tools of the trade. Nature 2007, 446, 219–222.
10. Lau, B.T.; Baitz, C.A.; Dong, X.P.; Hansen, C.L. A complete microfluidic screening platform for
rational protein crystallization. J. Am. Chem. Soc. 2007, 129, 454–455.
11. Paik, S.-J.; Byun, S.; Lim, J.-M.; Park, Y.; Lee, A.; Chung, S.; Chang, J.; Chun, K.; Cho, D.D.
In-plane single-crystal-silicon microneedles for minimally invasive microfluid systems.
Sens. Actuators A 2004, 114, 276–284.
12. Hadd, A.G.; Jacobson, S.C.; Ramsey, J.M. Microfluidic assays of acetylcholinesterase inhibitors.
Anal. Chem. 1999, 71, 5206–5212.
Micromachines 2014, 5 64

13. Harrison, D.J. Micromachining a miniaturized capillary electrophoresis-base. Science 1993, 261,
895–895.
14. Li, P.C.H.; Harrison, D.J. Transport, manipulation, and reaction of biological cells on-chip using
electrokinetic effects. Anal. Chem. 1997, 69, 1564–1568.
15. Hadd, A.G.; Raymond, D.E.; Halliwell, J.W.; Jacobson, S.C.; Ramsey, J.M. Microchip device for
performing enzyme assays. Anal. Chem. 1997, 69, 3407–3412.
16. Lagally, E.T.; Medintz, I.; Mathies, R.A. Single-molecule DNA amplification and analysis in an
integrated microfluidic device. Anal. Chem. 2001, 73, 565–570.
17. Wang, J.; Ibáñez, A.; Chatrathi, M.P.; Escarpa, A. Electrochemical enzyme immunoassays on
microchip platforms. Anal. Chem. 2001, 73, 5323–5327.
18. Zeng, Y.; Novak, R.; Shuga, J.; Smith, M.T.; Mathies, R.A. High-performance single cell genetic
analysis using microfluidic emulsion generator arrays. Anal. Chem. 2010, 82, 3183–3190.
19. Hung, P.J.; Lee, P.J.; Sabounchi, P.; Lin, R.; Lee, L.P. Continuous perfusion microfluidic cell
culture array for high-throughput cell-based assays. Biotechnol. Bioeng. 2005, 89, 1–8.
20. Fu, A.Y.; Spence, C.; Scherer, A.; Arnold, F.H.; Quake, S.R. A microfabricated
fluorescence-activated cell sorter. Nat. Biotechnol. 1999, 17, 1109–1111.
21. Kim, M.S.; Ju, H.Y.; Park, J.-K. A microfluidic platform for 3-dimensional cell culture and
cell-based assays. Biomed. Microdevices 2007, 9, 25–34.
22. Tian, J.; Gong, H.; Sheng, N.; Zhou, X.; Gulari, E.; Gao, X.; Church, G. Accurate multiplex gene
synthesis from programmable DNA microchips. Nature 2004, 432, 1050–1054.
23. Gomez-Sjoberg, R.; Leyrat, A.A.; Pirone, D.M.; Chen, C.S.; Quake, S.R. Versatile, fully
automated, microfluidic cell culture system. Anal. Chem. 2007, 79, 8557–8563.
24. Lam, R.H.; Kim, M.C.; Thorsen, T. Culturing aerobic and anaerobic bacteria and mammalian
cells with a microfluidic differential oxygenator. Anal. Chem. 2009, 81, 5918–5924.
25. Thorsen, T.; Maerkl, S.J.; Quake, S.R. Microfluidic large-scale integration. Science 2002, 298,
580–584.
26. Ottesen, E.A.; Hong, J.W.; Quake, S.R.; Leadbetter, J.R. Microfluidic digital PCR enables
multigene analysis of individual environmental bacteria. Science 2006, 314, 1464–1467.
27. Marcus, J.S.; Anderson, W.F.; Quake, S.R. Microfluidic single-cell mRNA isolation and analysis.
Anal. Chem. 2006, 78, 3084–3089.
28. Ben-Ari, Y.; Glick, Y.; Kipper, S.; Schwartz, N.; Barbiro-Michaely, E.; Gerber, D. Microfluidic
large scale integration of viral-host interaction analysis. Lab Chip 2013, 13, 2202–2209.
29. Hosokawa, K.; Maeda, R. A pneumatically-actuated three-way microvalve fabricated with
polydimethylsiloxane using the membrane transfer technique. J. Micromech. Microeng. 2000,
10, 415; doi:10.1088/0960-1317/10/3/317.
30. Kartalov, E.P.; Scherer, A.; Quake, S.R.; Taylor, C.R.; Anderson, W.F. Experimentally validated
quantitative linear model for the device physics of elastomeric microfluidic valves. J. Appl. Phys.
2007, 101, 64505; doi:10.1063/1.2511688.
31. Goulpeau, J.; Trouchet, D.; Ajdari, A.; Tabeling, P. Experimental study and modeling of
polydimethylsiloxane peristaltic micropumps. J. Appl. Phys. 2005, 98, doi:10.1063/1.1947893.
32. Hou-Pu, C.; Unger, M.A.; Quake, S.R. A microfabricated rotary pump. Biomed. Microdevices
2001, 3, 323–323.
Micromachines 2014, 5 65

33. El-Ali, J.; Sorger, P.K.; Jensen, K.F. Cells on chips. Nature 2006, 442, 403–411.
34. Bourouina, T.; Grandchamp, J.-P. Modeling micropumps with electrical equivalent networks.
J. Micromech. Microeng. 1996, 6, 398; doi:10.1088/0960-1317/6/4/006.
35. Zeng, Y.; Azizi, F.; Mastrangelo, C. Behavioral Modeling of Solute Tracking in Microfluidics.
In Proceedings of the IEEE Behavioral Modeling and Simulation Workshop 2009, San José, CA,
USA, 17–18 September 2009; pp. 1–6.
36. Chen, L.; Azizi, F.; Mastrangelo, C.H. Generation of dynamic chemical signals with microfluidic
C-DACs. Lab Chip 2007, 7, 850–855.
37. Azizi, F.; Mastrangelo, C.H. Generation of dynamic chemical signals with pulse code modulators.
Lab Chip 2008, 8, 907–912.
38. Xie, Y.; Wang, Y.; Chen, L.; Mastrangelo, C.H. Fourier microfluidics. Lab Chip 2008, 8,
779–785.
39. Hong, L.; Pan, T. Three-dimensional surface microfluidics enabled by spatiotemporal control of
elastic fluidic interface. Lab Chip 2010, 10, 3271–3276.
40. Studer, V.; Hang, G.; Pandolfi, A.; Ortiz, M.; French Anderson, W.; Quake, S.R. Scaling
properties of a low-actuation pressure microfluidic valve. J. Appl. Phys. 2004, 95, 393–398.
41. Pandolfi, A.; Ortiz, M. Numerical Analysis of Elastomeric Fluidic Microvalves. Sens. Lett. 2008,
6, 43–48.
42. Basar, Y.; Ding, Y. Shear deformation models for large-strain shell analysis. Int. J. Solids Struct.
1997, 34, 1687–1708.
43. Khanafer, K.; Duprey, A.; Schlicht, M.; Berguer, R. Effects of strain rate, mixing ratio, and
stress-strain definition on the mechanical behavior of the polydimethylsiloxane (PDMS) material
as related to its biological applications. Biomed. Microdevices 2009, 11, 503–508.
44. Xia, Y.; Kim, E.; Zhao, X.-M.; Rogers, J.A.; Prentiss, M.; Whitesides, G.M. Complex optical
surfaces formed by replica molding against elastomeric masters. Science 1996, 273, 347–349.
45. Joekar-Niasar, V.; Schotting, R.; Leijnse, A. Analytical solution of electrohydrodynamic flow and
transport in rectangular channels: Inclusion of double layer effects. Comput. Geosci. 2013, 17,
497–513.

© 2014 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article
distributed under the terms and conditions of the Creative Commons Attribution license
(http://creativecommons.org/licenses/by/3.0/).