Tape underlayment rotary-node (TURN) valves for simple on-chip microfluidic flow control
- First Online:
- Cite this article as:
- Markov, D.A., Manuel, S., Shor, L.M. et al. Biomed Microdevices (2010) 12: 135. doi:10.1007/s10544-009-9368-7
- 191 Views
We describe a simple and reliable fabrication method for producing multiple, manually activated microfluidic control valves in polydimethylsiloxane (PDMS) devices. These screwdriver-actuated valves reside directly on the microfluidic chip and can provide both simple on/off operation as well as graded control of fluid flow. The fabrication procedure can be easily implemented in any soft lithography lab and requires only two specialized tools—a hot-glue gun and a machined brass mold. To facilitate use in multi-valve fluidic systems, the mold is designed to produce a linear tape that contains a series of plastic rotary nodes with small stainless steel machine screws that form individual valves which can be easily separated for applications when only single valves are required. The tape and its valves are placed on the surface of a partially cured thin PDMS microchannel device while the PDMS is still on the soft-lithographic master, with the master providing alignment marks for the tape. The tape is permanently affixed to the microchannel device by pouring an over-layer of PDMS, to form a full-thickness device with the tape as an enclosed underlayment. The advantages of these Tape Underlayment Rotary-Node (TURN) valves include parallel fabrication of multiple valves, low risk of damaging a microfluidic device during valve installation, high torque, elimination of stripped threads, the capabilities of TURN hydraulic actuators, and facile customization of TURN molds. We have utilized these valves to control microfluidic flow, to control the onset of molecular diffusion, and to manipulate channel connectivity. Practical applications of TURN valves include control of loading and chemokine release in chemotaxis assay devices, flow in microfluidic bioreactors, and channel connectivity in microfluidic devices intended to study competition and predator/prey relationships among microbes.
KeywordsMicrofluidicsOn-chip valveFlow controlGradient formationBacteriaProtozoaMicrobiological predationValve fabricationHydraulic valveValve arraysMulti-channel closure
The simplest microfluidic devices that have been fabricated from PDMS using soft lithography feature uncomplicated microchannel architectures filled with quiescent or flowing liquids controlled either by hydrostatic pressure or external syringe pumps and valves (Beebe et al. 2002; Cho et al. 2003; Duffy et al. 1998; Georgescu et al. 2008; Liu et al. 2008; Walker et al. 2005). The most intricate devices use banks of pneumatically controlled on-chip valves and pumps to control complex bioreactors and microchannel networks (Gomez-Sjoberg et al. 2007; Thorsen et al. 2002; Unger et al. 2000). As microfluidic devices become more widely accepted as practical tools in the scientific laboratory, there is a need for devices of intermediate complexity that perform a particular function well, but at low cost in terms of time, money, and expertise both to fabricate and operate. The full potential of adopting microfluidic technology will not be realized as long as compact, elegant PDMS devices must be controlled by bulky syringe pumps or solenoid valve banks and computerized controllers costing hundreds to thousands times more than the devices themselves. These issues are especially pertinent for applications involving large numbers of devices used for experiments in space-restricted environments, such as cell-culture hoods and incubators, particularly when they involve scientists, technicians, and students who are not microfluidic experts.
Researchers at the Vanderbilt Institute for Integrative Biosystems Research and Education (VIIBRE) are developing a family of low-cost microfluidic devices such as bioreactors (Prokop et al. 2004) and nanophysiometers (Werdich et al. 2004) for investigating the chemotactic and morphogenic behavior of eukaryotic cells (Georgescu et al. 2008; Liu et al. 2008), as well as the ecology and behavior of individual microbes (Wang et al. 2005, 2008). Many of these devices target the long-term observation of a small number of cells subjected to precisely controlled microenvironmental conditions, including chemical gradients of nutrients or signaling molecules, which influence the proliferation, morphology, and developmental fate of individual cells. These applications often require a valve to be opened or closed once or twice during the time course of the experiment, which may be up to a week. Therefore, there is a need for an alternative to pneumatic valves and off-chip controllers.
Numerous valve technologies compatible with microfluidics have been recently developed and described in the literature (Easley et al. 2006; Grover et al. 2003; Gu et al. 2004; Kovacs 1998; Weibel et al. 2005). However, most previously proposed technologies are bulky or complex to fabricate, require external actuation and power source maintenance, or are not compatible with tall (greater than 20 μm) microfluidic channels. Our goal is to provide a valve that can be easily integrated into microfluidic devices, is simple to fabricate and operate, does not require external power connections, and is capable of reversibly closing a variety of microfluidic channels.
A low-cost valve to compress and block a microfluidic channel was proposed by Weibel et al. (2005). Their design incorporates a small diameter mechanical screw that is inserted into a blind hole drilled into the PDMS structure directly above a microfluidic channel. When this screw is turned clockwise, the underlying channel is collapsed; subsequent counter-clockwise screw action will allow resilient forces in the PDMS to restore fluid flow in the channel. However, fabrication of the Weibel et al. design requires precisely controlled and time-consuming drilling of previously molded PDMS structures to incorporate each individual valve. This approach is not well suited for mass production of microfluidic designs that require a large number of valves, because a single error in alignment or drilling depth can render an entire device useless. Ges et al. have advanced this approach by inserting an oversized threaded insert with a screw into a blind hole drilled into the PDMS to compress the channel located beneath the hole (Ges et al. 2008; Ges and Baudenbacher 2008). While this eliminates the problem of the screw threads pulling out of the PDMS, it still requires drilling a blind hole to a precise depth into the PDMS, with the concomitant risk of damaging the device. This embedding technique also is not well suited for parallel fabrication of large numbers of valves on a single device.
To overcome these difficulties and to make fabrication and operation more user-friendly, we take a simple and robust approach. Our valves consist of a valve body, which is easily fabricated using a simple injection molding method and a small captive stainless steel 0–80 machine screw (1.5 mm diameter with 3.15 threads/mm). The assembled valves are aligned with the underlying microfluidic network and embedded into the device during a two-step PDMS casting procedure. Valves are fabricated as a strip for ease of alignment and placement, which gives rise to the acronym TURN (Tape Underlayment Rotary-Node) valve. Valves can be placed as strips by matching the underlying microfluidic network to the valve spacing in the mold, or cut out and used individually.
2.1 TURN valve fabrication
In the molding process, molding screws are first threaded all the way into the mold from the back so that screw threads protrude slightly above its top surface. Then, a light coat of aerosol mold release compound (Smooth-On, Inc.) is applied. Next, thermosetting plastic glue (HS-300, Adhesive Solutions Co. LLC) (FDA approved for food packaging) is delivered by an 80 W hot-glue gun (GR-24 Pro, Stanley Tools) to injection-mold the valve bodies. Glue is injected into the mold recesses sequentially in one motion, without breaking the glue flow, so as to fill the mold and encapsulate the protruding screws. Molten glue completely conforms to the shape of the mold and the screw threads, providing excellent gripping force in the finished product. After the molten plastic is solidified (typically 10 min), the molding screws are partially unscrewed. Excess glue is trimmed with a razor blade to form a smooth surface flush with the top surface of the brass mold. Molding screws are then completely backed out, and a strip, or tape, of valve bodies is removed from the mold.
Valve fabrication is completed by inserting into the valve bodies a different set of 0–80 screws with polished ends that will act as valve actuators. These are screwed into the valve bodies to be exactly co-planar with the bottom of the valve assembly. Polishing of the screw ends is important to avoid galling of the PDMS surface and eventual rupture of the fluidic channels during valve engagement. Finally, valve tapes, with screws inserted, are rinsed with isopropyl alcohol, dried, and stored under sterile conditions for later assembly into PDMS microfluidic devices.
2.2 Valve incorporation into microfluidic devices
The fabrication of the microfluidic devices used in these experiments is straightforward and is widely described in the literature (Whitesides et al. 2001). Briefly, SU-8 series photoresist (MicroChem) is spin-coated to the desired feature height on the surface of a silicon wafer. A standard contact print photolithographic process is then used to selectively expose photoresist to ultraviolet light. Following development of unexposed photoresist, and hard-baking the polymerized device pattern, a positive relief mold is created. The height of the microfluidic network depends on a particular application and hence the choice of SU-8 resist. We routinely fabricate devices ranging from 1 μm to >300 μm tall.
Next, a 1 mm-thick layer of 10:1 mixture of PDMS and curing agent is poured onto the mold to create the deformable layer of PDMS upon which the TURN valve operates. The whole assembly is then placed in a 70°C oven (DX 400, Yamato) and allowed to partially cure until the surface is semi-rigid, but still sticky (~25 min). The mold is then removed from the oven and strips of TURN valves are carefully placed on top of the partially cured PDMS in the appropriate locations utilizing tweezers and a low-power stereoscope. The alignment marks built into the photolithographically fabricated master greatly simplify the process of precisely positioning the valves on top of the thin layer of PDMS still rigidly supported by the silicon wafer, as does the rigid linear arrangement of a row of valves in the tape that allows many valves to be installed in a single operation. When all valves are correctly positioned, an additional ~3 mm of PDMS is poured over the entire device to completely encapsulate the valve assembly. After overnight curing, the microfluidic device molded in PDMS and containing the embedded valves is removed from the silicon wafer, access holes are punched for the external fluidic lines, and the whole device is irreversibly bonded to the microscope slide or cover glass by activating the PDMS and the glass surface via a 30 s low pressure O2 plasma treatment (PDC-32G, Harrick Co.).
3 Results and discussion
3.1 Controlling multiple flows
3.2 Time release of diffusive gradients
3.3 Microbial behavior studies
Another interesting application of TURN valves enables experimenters to control connectivity within microfluidic networks. Typical microbial habitats have complex microscale physical, chemical, and biological structure. Channel dimensions and connectivity affect microbial migration and dispersal rates, susceptibility of bacteria to predation, community stability, and net microbial productivity (Anderson and Domsch 1995; Fenchel 1969; Ranjard and Richaume 2001; Wang et al. 2005, 2008).
For example, some of our work has focused on interactions between bacteria and protozoa predators within microfluidic devices (Wang et al. 2005, 2008). Protozoan grazing may enhance net bacterial activities via release of limiting nutrients, or through physical disturbance of microbial habitats (Glud and Fenchel 1999; Hahn and Hofle 2001; Mattison et al. 2005), leading to enhanced rates of bacterial biodegradation of crude oil, polycyclic aromatic hydrocarbons, and coal tar (Madsen et al. 1991; Rogerson and Berger 1983; Tso and Taghon 2006). However, efforts to understand the impact of habitat structure on microbial interactions have been limited by the inability to control the dispersal of microorganisms in microstructured habitats. The combination of microfluidics and TURN valves can provide a powerful yet simple tool that overcomes such problems.
Similar “breakthrough” experiments were performed using Escherichia coli (E. coli), a motile 1 µm rod-shaped gram-negative bacterium. This organism was transfected with a green fluorescent protein (GFP) plasmid, greatly assisting optical detection even at low magnification. In devices consisting of several habitats in series with microfluidic channels and a TURN valve, bacteria readily dispersed across open TURN valves. The concentration of bacteria used in these experiments was quite high, approximately 109 cells/ml suspended in Luna-Bertani broth with 50 µg/mL of kanamycin. Within several hours, the bacteria had dispersed throughout the entire open region of the device. We used the crossed device shown in Fig. 5(a) to demonstrate the bacterial impermeability of a closed TURN valve. Bacteria were introduced into one fluidic leg of the closed cross. After a seven-day incubation period the bacteria spread evenly throughout the inoculated channel but were not able to penetrate the closed TURN valve. (Even at high magnification, no fluorescence from GFP bacteria was observed either visually or with long integration times on a cooled CCD camera (Q-Color 5). Figure 7(b) shows a combination of fluorescent and white light illumination pictures of the closed valve with bacteria aggregating around the collapsed channel ceiling but unable to penetrate the closed valve into the remaining three channels.
3.4 Hydraulic valve
During our experiments with TURN valves, we identified a need to close multiple channels with a single actuator. A single TURN valve can reliably close two parallel unconnected 100 µm wide channels that are directly under the center of the TURN valve. However it would be useful to have a more general approach to control multiple independent channels with one single actuator. We have developed a hydraulic technique for controlling multiple shallow channel closures with a single TURN valve. Quake et al. (Unger et al. 2000) have shown that by applying a pneumatic pressure to the gas- or fluid-filled channels that are located above or below the fluidic network and are separated from it by a thin extensible membrane, it is possible to block the flow in that fluidic network. Even though devices with very complex architecture and functions can be achieved using that approach, such systems require an external gas tank or a compressor and computer-controlled pressure and bleed valves, and often constrain routing of microfluidic and pressure channels through the devices (Gomez-Sjoberg et al. 2007). The external connections can be very useful but require substantial investment in experiment design and infrastructure, and are not well suited for the use of large numbers of separate devices in an incubator, particularly if the valves need to be opened or closed only once or twice in the lifetime of the device.
We have also demonstrated that it is possible to use a low-cost stepping motor and driver (Digi-Key MTSD-V1) and flexible plastic shafts, such as those used in radio-controlled model airplanes, to operate TURN valves automatically or remotely. While this may prove useful for automated loading of large numbers of devices, in this paper we emphasize the minimalist approach of using a small screwdriver to actuate the TURN valve by hand.
We have extended the concept of a simple mechanical screw valve system in a manner consistent with ease of soft lithographic microfluidic device manufacture and operation. TURN valves do not require tedious fabrication and use of photocurable polymers, and they can be easily implemented in any laboratory that uses soft lithography. These valves do not require expensive external actuation and control parts and are capable of blocking a variety of microfluidic channels, including those with tall aspect ratios and intersections. They can be used individually or in a long tape or strip, and when employed in passive diffusion devices, can aid in generating adjustable, spatially varying concentration gradients. They provide an extremely tight, cell-impermeable seal and can separate protozoa and bacteria in experimental devices for studying microbial population dynamics. Lastly, the TURN valve concept can be used as an efficient and extremely inexpensive actuator in hydraulic systems to close multiple channels simultaneously. TURN valve devices are compact, inexpensive, easy to use, and can be readily moved into different locations or in and out of an incubator without concern for any external pressure connections, pressure leaks, and power sources.
Information regarding the design, construction, and availability of CNC molds for the valve bodies and the corresponding mask layouts is available at http://www.vanderbilt.edu/viibre/technologies.html.
This work has been supported in part by National Institutes of Health Grants U01AI061223 and U54CA113007, National Science Foundation Grants 0120453 and 0649883, DOD CDMRP/BCRP grant BC061791, the Vanderbilt Institute for Integrative Biosystems Research and Education, and the Searle Systems Biology and Bioengineering Undergraduate Research Experience. We are especially indebted to John Fellenstein and Robert Patchin for the fabrication of the brass molds. We thank Adit Dhummakupt for assistance with the protozoa work and Allison Price and Don Berry for their editorial assistance in the preparation of this manuscript.