Microfluidics and Nanofluidics

, Volume 17, Issue 4, pp 745–750

Syringe-assisted point-of-care micropumping utilizing the gas permeability of polydimethylsiloxane

Research Paper

DOI: 10.1007/s10404-014-1356-4

Cite this article as:
Xu, L., Lee, H. & Oh, K.W. Microfluid Nanofluid (2014) 17: 745. doi:10.1007/s10404-014-1356-4

Abstract

By utilizing the high gas permeability of polydimethylsiloxane (PDMS), a simple syringe-assisted pumping method was introduced. A dead-end microfluidic channel was partially surrounded by an embedded microchamber, with a thin PDMS wall isolating the dead-end channel and the embedded microchamber. A syringe was connected with the microchamber port by a short tube, and the syringe plunger was manually pulled out to generate low pressure inside the microchamber. When sample liquid was loaded in the inlet port, air trapped in the dead-end channel would diffuse into the surrounding microchamber through the PDMS wall, creating an instantaneous pumping of the liquid inside the dead-end channel. By only pulling the syringe manually, a constant low flow with a rate ranging from 0.089 to 4 nl/s was realized as functions of two key parameters: the PDMS wall thickness and the overlap area between the dead-end channel and the surrounded microchamber. This method enabled point-of-care pumping without pre-evacuating the PDMS devices in a bulky vacuum chamber.

Keywords

Point-of-care Polydimethylsiloxane (PDMS) Pump 

1 Introduction

One of the challenging unit operations in point-of-care testing (POCT) diagnostics is autonomous, controllable, and on-demand pumping. Obviously, manual injection of sample liquids with a syringe or a pipette is one of the simplest methods. However, such direct injection method is not suitable for use where constant and controllable pumping is required over the injection time. Currently, most of the pumping methods in POCT are employed passively, mainly based on capillary force (Gervais and Delamarche 2009). Since no external energy is required, the capillary-driven pumping scheme is attractive for many POCT systems. Also based on the capillary effect, alternative ways such as paper-based and textile-based pumping are studied (Nilghaz et al. 2012; Yetisen et al. 2013). However, this pumping method mainly relies on the wetting properties of different testing liquids, which are unstable. Thus, it is difficult to generate stable and steady pumping (Ziegler et al. 2008). Another interesting point-of-care micropumping method is realized by utilizing the elastic deformation of PDMS (Weibel et al. 2007; Li et al. 2012). First, the microfluidic device made of PDMS is deformed by using either thumb or screw. Then, once the external force is removed, the PDMS device will go back to its initial form, which will withdraw liquids flowing inside the channel. In general, the flow rate generated by this pumping method is not linear or constant. In order to keep the liquids flowing inside the channel, the external force has to be repeatedly applied and removed.

Utilizing the high gas permeability of PDMS to achieve pumping has been considered as a promising way. High gas permeability is one of unique properties of PDMS (Merkel et al. 2000). This has allowed easy removal of trapped air bubbles out of microchannels (Ong et al. 2007; Sung et al. 2010). In addition, vacuum-assisted self-powered pumping has been introduced by Hosokawa et al. (2004). Thus, by loading liquid at the inlet port of a dead-end channel in the degassed device, the fluid can be drawn into the channel. However, PDMS devices need to be placed in a vacuum chamber to be pre-vacuumed. This is due to the reabsorption of air inside the vacuumed PDMS devices after air exposure to reach the new equilibrium under the atmosphere. This method was used to perform a simple immunoassay by adding two different types of liquids (Hosokawa et al. 2006). Based on the same principle, a self-powered microfluidic blood analysis system (Dimov et al. 2011; Liang et al. 2011) and a viscometer (Tang and Zheng 2011) have been successfully realized.

Although the vacuum-assisted degas-driven flow is a convenient way to adopt without external pumps, there are still several practical limitations: (1) Before a device can work as a pump, it needs to be stored in a vacuum chamber for more than 30 min to degas the PDMS bulk or in a sealed shrink-wrap vacuum packaging; (2) after the device is exposed to air, its pumping ability decays immediately with time, so the device should be used immediately after air exposure; (3) since pumping ability decays with time, the flow rate generated from the vacuum-assisted flow is nonlinear and uncontrollable; alternatively, instead of pre-evacuating the entire device, a low-pressure source was connected to the chamber and separated with a dead-end channel by a thin PDMS membrane (Mark and Bruce 2006). By controlling the thickness of the membrane, different flow rates were realized. However, the process to make this kind of multilayer sandwich structures is laborious due to precise alignment between the top and bottom layers and careful control of the PDMS membrane thickness and uniformity. Instead of using sandwich structures, a dead-end microchannel surrounded by a vacuumed microchamber in the same layer was employed for PCR test (Trung et al. 2010). In this approach, a constant vacuum was applied by external source and the flow rate is controlled only by varying the PDMS wall thickness between the dead-end channel and surrounded microchamber, which gives a limited range of loading flow rate (0.3–0.7 nl/s).

To overcome these limitations of using the vacuum-assisted pumping, we have designed a simple point-of-care pumping method using a single-layer structure that does not require pre-evacuation of PDMS devices in a vacuum chamber. A dead-end microfluidic channel is partly surrounded by an embedded microchamber, with a thin PDMS wall separating the dead-end channel and the embedded microchamber. A syringe connected with the microchamber port is employed to provide a negative pressure source. Low pressure generated inside the microchamber by pulling out the syringe plunger will draw air trapped in the dead-end channel through the thin PDMS wall, creating an instantaneous pumping of the liquid inside the dead-end channel. The flow rate can be regulated by controlling the wall thickness and the overlap area between the dead-end channel and the surrounded microchamber. Employing only a hand-held syringe, this method allows for on-demand pumping without pre-evacuating the PDMS devices in a bulky vacuum chamber. In this paper, we have systematically investigated the major parameters (e.g., wall thickness, overlap area) that can generate constant and controllable flow rates.

2 Theory

As shown in Fig. 1, a dead-end channel is partly surrounded by an embedded microchamber, with a thin PDMS wall separating the dead-end channel and the embedded microchamber. First, a syringe is connected with the microchamber by a tube. By pulling the plunger, low pressure will be generated inside the microchamber, where cylindrical posts are placed to prevent the collapse of the microchamber at low pressure. Liquid is loaded in the inlet port after a few seconds to allow constant and steady-state air flux. The pressure difference between the dead-end channel and the surrounded microchamber allows air diffusion through the gas-permeable PDMS wall, from inside the channel to the microchamber across the PDMS wall. Therefore, the fluid can be drawn inside the dead-end channel. This approach facilitates a straightforward point-of-care pumping system that does not require pre-evacuation of PDMS devices in a vacuum chamber.
Fig. 1

Schematic illustration of the proposed syringe-assisted point-of-care pumping: a overview and the setup of the proposed device, b enlarged view of the PDMS device (channel is at top surface and boned with glass slide), c cross-sectional view of the device along the A–A′ axis, and d experimental steps with inserted air concentration profile along the A–A′ axis (CChannel stands for the air concentration inside the channel, CPDMS stands for air concentration inside the PDMS wall, CChamber means air concentration inside the chamber, and CATM is the air concentration at atmosphere): Step I connect the microchamber with a syringe by tube; Step II pull the syringe to expand its volume to 200 μl. Therefore, vacuum will be generated in the chamber and air inside the PDMS wall will start diffusing into the chamber, while as inlet is open to atmosphere, its air concentration will not be changed; and Step III add DI water at the inlet after around 5 s. As the inlet is sealed by DI water, the air concentration inside the channel will drop to generate enough pressure to withdraw the DI water inside the channel. Note here that once the generated pressure is enough to pump in the DI water, we assume that the air flux rate equals to DI water flux rate as explained in the theory section. Therefore, the air concentration inside the channel will remain the same

A rough estimate of the characteristic time to allow for the constant and steady-state air flux across the thin PDMW wall can be obtained by examining the diffusion time (tD) across the PDMS wall: tD ≈ w2D−1, where w is the thickness of the PDMS wall and D is the diffusion coefficient of air in PDMS. For example, if w = 50 μm and D = 3.4 × 10−9 m2 s−1 (diffusion coefficient of air in PDMS at atomsphere), the characteristic time to initialize the steady-state air flux is tD1 ≈ 0.74 s. Another characteristic time to diminish the steady-state air flux would depend on the air diffusion from the surface into the microchamber across the thick PDMS layer (e.g., ~5 mm), which is tD2 ≈ 2 h. Therefore, the air flux will be kept steady and constant if devices are operated within tD1 ≪ t ≪ tD2.

According to Fick’s diffusion law (Hosokawa et al. 2004; Tang and Zheng 2011), we can get an approximate value of the steady-state flux F (mol m−2 s−1) of air that diffuses into the microchamber across the thin PDMS wall through the gas-permeable PDMS
$$ F = - D\left. {\frac{\varDelta C(x)}{\varDelta x}} \right|_{x = w} \approx D\frac{C(0) - C(w)}{w} = D\frac{{C_{\text{PDMS}} - C_{\text{Chamber}} }}{w} $$
(1)
where C(x) is the air concentration profile in PDMS across the PDMS wall (Fig. 1c). The initial air concentration inside the dead-end channel and microchamber is CATM = 44 mol m−3 (Merkel et al. 2000; Tang and Zheng 2011). The air concentration saturated in bulk PDMS is known to be CPDMS = 4.89 mol m−3 at atmosphere. The large volume expansion in the syringe (e.g., 1.6 μl to several hundred μl) will initiate a new equilibrium air concentration inside the microchamber: CChamber = 0.35 mol m−3 for the volume expansion to 100 μl. After tD1, the air concentration at x = g will be depleted to C(w) = CChamber, while the air concentration at x = 0 will be C(0) = CPDMS (Fig. 1d). Thereafter, the air flux F will be kept constant if operated within tD1 ≪ t ≪ tD2: the waiting time of 5 s (t ≫ tD1) and the whole pumping time of less than 8 min (t ≪ tD2).
The syringe-assisted air flux causes the degas-driven flow in the dead-end channel. If we assume that the air concentration or the air density inside the channel is unchanged during the pumping process, the volumetric flow rate (Q) and the pumped fluid volume (V) inside the channel can be expressed as
$$ \begin{gathered} Q(t) \approx \,k\frac{F\,S(t)}{{C_{\text{ATM}} }} = \,k\,D\frac{{C_{\text{PDMS}} - C_{\text{Chamber}} }}{{C_{\text{ATM}} }}\frac{\,S(t)}{w} \hfill \\ \Rightarrow V(t) = \,\int {Q(t){\text{d}}t} \hfill \\ \end{gathered} $$
(2)
where k is an empirical factor related to viscous effect of the pumped liquid flow and S(t) is the total surface area that allows air to diffuse into the PDMS bulk that equals the overlap area of the channel and microchamber (which also equals to one side of the verticle area of the PDMS wall). Equation 2 points out that there are two major parameters that can affect the flow rate: the thickness of the PDMS wall (w) and the overlap surface area (S) between the channel and microchamber.

3 Design and fabrication

3.1 Design

In order to study the relationships between the flow rate Q and these two parameters, two sets of devices were designed (Figs. 2a, 3a). In the first set, five different devices were designed with different thickness of the PDMS wall (e.g., 25, 50, 100, 200, and 300 μm), while the overlap area was kept the same (S0 = 0.28 mm2). In the second set, five different devices were made with different overlap areas (e.g., 0.0035, 0.18, 0.52, 0.69, and 0.87 mm2), while the thickness of the PDMS wall was kept unchanged (w = 50 μm). In both two sets, channels and microchambers had the same height (h = 34.5 μm) and all the channels had the same width (d = 100 μm).
Fig. 2

Effect of the thickness of the PDMS wall between the channel and embedded microchamber on syringe-assisted pumping. a Schematic showing devices with varying thickness of the PDMS wall (e.g., 25, 50, 100, 200, and 300 μm), while the overlap area is kept the same (S0 = 0.28 mm2). b Pumped volume profiles versus time with different thickness of the PDMS wall. c Flow rate (QI) versus the thickness of the PDMS wall (w) in Phase I. See Movie S1 for w = 25 μm and Movie S2 for w = 300 μm

Fig. 3

Effect of the overlap area between the channel and embedded microchamber on syringe-assisted pumping. a Schematic showing devices with varying overlap areas (e.g., 0.00345, 0.176, 0.521, 0.693, and 0.866 mm2), while the thickness of the PDMS wall is kept unchanged (w = 50 μm). b,c Pumped volume profiles versus time with different overlap areas. d Flow rate (QI) versus the overlap surface areas (S0) in Phase I. See Movie S3 for S0 = 0.176 mm2 and Movie S4 for S0 = 0.693 mm2

3.2 Microfluidic device fabrication

All devices were fabricated by a standard soft lithography process (Xia and Whitesides 1998). A 3-inch silicon wafer with one side polished (University wafers, South Boston, MA, USA) was submerged into buffered hydrofluoric acid (BHF) at room temperature for 5 min to remove the thin native silicon dioxide layer. After that, the wafer was cleaned by using acetone and methanol, respectively, and then rinsed in deionized water before blown dry by filtered nitrogen gas. After cleaning, the cleaned wafer was placed on a hot plate at 120 °C for 5 min in order to make it completely dehydrated. SU-8 (SU-8 2050, Micro-Chem Corp, Newton, MA, USA) was then spin coated on top of the wafer by using the spin coater (WS-650Mz NPP from Laurell Technologies, North Wales, PA, USA) to the target thickness. After spin coating, soft bake was performed on a leveled hot plate for 3 and 9 min at 65 and 95 °C, respectively. After soft bake, UV photolithography was carried out by using a contact mask aligner. After UV exposure, the post-exposure bake (PEB) was conducted on the leveled hot plate for 2 and 7 min at 65 and 95 °C, respectively, followed by development in SU-8 developer for 5 min. After developing, the wafer was cleaned with isopropyl alcohol. Finally, the wafer was blown dry with filtered nitrogen gas and then placed on the hot plate at 100 °C for 5 min to evaporate any residual of isopropyl alcohol.

A prepolymer of PDMS (Sylgard 184, Dow Corning) and corresponding curing agent was thoroughly mixed at a ratio of 10: 1 (wt/wt). Then, the mixed PDMS was degassed in a vacuum chamber for 20 min to remove all the air bubbles. In order to peel off the PDMS from the wafer mold easily, hexamethyldisilazane (Sigma Aldrich, Saint Louis, MO, USA) was silanized on the surface of the wafer mold in vacuum chamber at room temperature for 30 min. After that, the PDMS mixture was carefully poured onto the wafer mold and then cured at 80 °C for 3 h. In order to make sure the thickness of the PDMS bulk does not affect the experiment, the thickness of all the PDMS devices was made to be more than 1 cm. After PDMS was cured and peeled off from the wafer mold, holes were punched on the PDMS replicas for the connection with the syringe, followed by oxygen plasma treatment for irreversible bonding between PDMS and glass slide at the top surface. Lastly, the device was baked over a hot plate for 48 h at 70 °C to improve the bonding strength and stabilize the surface property of the devices.

3.3 Test setup and procedure

The fabricated microfluidic device was connected through a silicone tube to a glass syringe with maximum volume of 200 μl. The glass syringe was used to generate the low pressure inside the microchamber of the fabricated microfluidic devices. In order to verify that the volume of the syringe does not affect the flow rate, a glass syringe with maximum volume of 100 μl was also used in the same devices. And the results were similar to the glass syringe with maximum volume of 100 μl (the variation in flow rate was within 10 %). Therefore, we adopted the glass syringe with maximum volume of 200 μl in all the following tests. All of the devices were tested three times. First, the syringe was connected and pulled to have volume expansion to 200 μl (as shown in Step I and II in Fig. 1d). After pulling the plunger, DI water was loaded at the inlets after around 5 s (as shown in Step III in Fig. 1d). All the flow processes were recorded by a Nikon stereo-type microscope and camera set. By analyzing the recorded video clips, volume–time curves were plotted (KDS100 W, Fisher Scientific, IL, USA).

4 Results and discussion

As shown in Fig. 2b, before liquid reaches the part of the channel surrounded by the microchamber, the liquid volume V is linear to the pumping time t, which indicates that the flow rate Q remains constant (Phase I). As the flow enters the part of the channel surrounded by the microchamber after a certain time tC, however, the flow rate exponentially decreases (Phase II).

When t < tC (Phase I), the overlap area will be a constant value: S(t) = S0 = 0.28 mm2. Thus, according to Eq. 2, the flow rate QI and VI can be written as
$$ \begin{array}{*{20}c} {Q_{\text{I}} (t) \approx \,k\,\frac{{F\,S_{0} }}{{C_{\text{ATM}} }} = k\,D\frac{{C_{\text{PDMS}} - C_{\text{Chamber}} }}{{C_{\text{ATM}} }}\frac{{\,S_{0} }}{w}} \hfill \\ \quad { \Rightarrow V_{\text{I}} (t) = \,Q_{\text{I}} \,t = \,k\,\frac{{F\,S_{0} }}{{C_{\text{ATM}} }}\,t} \hfill \\ \end{array} \quad \left( {{\text{Phase I}}:t < t_{\text{C}} } \right) $$
(3)
Therefore, as shown in Fig. 2c, the flow rate QI in Phase I was constant at the given thickness of the PDMS wall and proportional to w−1. The empirical factor was k = 0.57 ± 0.03, which indicates pumping power loss of 43 % due to the viscous effect.
On the other hand, when t > tC (Phase II), the overlap area will be reduced and become time dependent because the pulled-in liquid will block the part of the surface where the air can diffuse: \( S\left( t \right) = S_{0} {-}\varDelta S\left( t \right) \), where \( \varDelta S\left( t \right) = 2 \times h \times y\left( t \right) = 2 \times d^{ - 1} \int {Q_{II} \left( t \right){\text{d}}t} \) and h and d are the height and the width of the channel, respectively. Thus, by solving a differential equation according to Eq. 2, the flow rate QII and VII can be written as
$$ \begin{array}{*{20}l} {Q_{\text{II}} (t) \approx \,\,k\,\frac{{F\,S_{0} }}{{C_{\text{ATM}} }}e^{{ - \frac{{t - t_{C} }}{\tau }}} } \hfill \\ { \Rightarrow V_{\text{II}} (t) = \,k\,\frac{{F\,S_{0} }}{{C_{\text{ATM}} }}\,t_{C} + \,\frac{{{\text{d}}S_{0} }}{2}\left( {1 - e^{{ - \frac{{t - t_{C} }}{\tau }}} } \right)} \hfill \\ \end{array} \,\left( {{\text{Phase II}}:t > t_{\text{C}} } \right) $$
(4)
where τ is the time constant, \( \tau = \, (d \times C_{\text{ATM}} )/(2 \times k \times F) \). The time constant τ is proportional to the thickness of the PDMS wall (w). Equation 4 explains the exponential decay of the volumetric flow rate QII(t) and the pumped total volume VII(t) in Phase II, as shown in Fig. 2b.

Similarly, we investigated the syringe-assisted pumping related to the overlap area, while the thickness of the PDMS wall was kept unchanged (w = 50 μm), as shown in Fig. 3a. The pumped total volume was plotted in Fig. 3b. In Phase II, exponential decay was shown with the same time constant τ for different overlap areas. This is due to the fixed PDMS wall thickness, resulting in the same air flux F for all cases. The flow rate of the device with almost zero overlap area (S0 = d × h = 0.0035 mm2) was extremely low (0.089 nl s−1), taking ~9 min to draw in water completely. As only the very end of the channel overlapped with the microchamber, the pumped total volume was linear with time during the whole pumping process (Fig. 3c). In Phase I, as expected from Eq. 3, the flow rate QI was linearly proportional to the overlap surface area (Fig. 3d).

When t ≈ tD2, the syringe-assisted pumping will stop because the pressure difference completely vanishes inside the microchamber due to the air diffusion from outside to the microchamber. However, we can restore the low pressure or vacuum inside the microchamber by simply reconnecting the syringe and pulling the plunger again. Therefore, unlike the vacuum-assisted degas-driven flow, the proposed syringe-assisted method permits instantaneous, recurring, point-of-care pumping. Another advantage of the syringe-assisted pumping is that the air diffusion is bidirectional between the channel and microchamber depending on the polarity of pressure difference. By pushing the plunger, the drawn-in liquid solution can be retrieved from the dead-end channel. Thus, the syringe can supply not only a vacuum source (e.g., low pressure) but also a pressure source (e.g., high pressure) to drive the fluid flow into and from the dead-end channel. This will enable a simple point-of-care pumping system that requires multiple incubations and washing processes.

5 Conclusion

By adjusting the thickness of the PDMS wall and overlap area between the channel and microchamber, we have controlled the flow rate and generated a constant flow before the fluid reaches the part of channel surrounded by the microchamber. By the proposed method, a syringe-assisted, instantaneous, recurring, bidirectional, point-of-care pumping system without external pumps or vacuum chambers has been realized with controllable flow rate ranging from 0.089 to 4 nl s−1.

Acknowledgments

This work was partially supported by grants from NSF (ECCS-1002255 and ECCS-0736501).

Supplementary material

10404_2014_1356_MOESM1_ESM.pdf (5 kb)
Supplementary material 1 (PDF 5 kb)

Supplementary material 2 (WMV 1822 kb)

Supplementary material 3 (WMV 8073 kb)

Supplementary material 4 (WMV 3862 kb)

Supplementary material 5 (WMV 2197 kb)

Copyright information

© Springer-Verlag Berlin Heidelberg 2014

Authors and Affiliations

  1. 1.Sensors and MicroActuators Learning Laboratory (SMALL), Department of Electrical EngineeringUniversity at Buffalo, The State University of New York (SUNY at Buffalo)BuffaloUSA

Personalised recommendations