ARTICLE pubs.acs.org/Langmuir
A Vertical Microfluidic Probe G. V. Kaigala, R. D. Lovchik, U. Drechsler, and E. Delamarche* IBM Research—Zurich, S€aumerstrasse 4, CH-8803 R€uschlikon, Switzerland ABSTRACT: Performing localized chemical events on surfaces is critical for numerous applications. We earlier invented the microfluidic probe (MFP), which circumvented the need to process samples in closed microchannels by hydrodynamically confining liquids that performed chemistries on surfaces (Juncker et al. Nat. Mater. 2005, 4, 622 628). Here we present a new and versatile probe, the vertical MFP (vMFP), which operates in the scanning mode while overcoming earlier challenges that limited the practical implementation of the MFP technology. The key component of the vMFP is the head, a microfluidic device (∼1 cm2 in area) consisting of glass and Si and having microfluidic features fabricated in-plane in the Si layer. The base configuration of the head has two micrometer-size channels that inject/aspirate liquids and terminate at the apex which is ∼1 mm2. In scanning mode, the head is oriented vertically with the apex parallel to the surface with typical spacing of 1 30 μm. Such length scales and using flow rates from nanoliters/second to microliters/second allow chemical events to be performed on surfaces with tens of picoliter quantities of reagents. Before scanning, the head is clipped on a holder for leak-free, low dead volume interface assembly, providing a simple world-to-chip interface. Surfaces are scanned by mounting the holder on a computer-controlled stage having ∼0.1 μm resolution in positioning. We present detailed steps to fabricate vMFP heads having channels with dimensions from 1 μm 1 μm to 50 μm 50 μm for liquid localization over areas of 10 10000 μm2. Additionally, advanced design strategies are described to achieve high yield in fabrication and to support a broad range of applications. These include particulate filters, redundant aperture architectures, inclined flow-paths that service apertures, and multiple channels to enable symmetric flow confinement. We also present a method to characterize flow confinement and estimate the distance between the head and the surface by monitoring the evolution of a solution of fluorescently labeled antibody on an activated glass surface. This flow characterization reveals regimes of operation suitable for different surface topographies. We further integrate the dispensing of immersion liquid to the vMFP head for processing surfaces for extended periods of time (∼60 min). The versatility of the vMFP is exemplified by patterning fluorescently labeled proteins, inactivation of cells using sodium hypochlorite, and staining living NIH fibroblasts with Cellomics. These applications are enabled by the compact design of the head, which provides easy access to the surface, simplifies alignment, and enables processing surfaces having dimensions from the micrometer to the centimeter scale and with large topographical variations. We therefore believe that ease-of-operation, reconfigurability, and conservative use of chemicals by the vMFP will lead to its widespread use by microtechnologists and the chemical and biomedical communities.
1. INTRODUCTION We developed a noncontact microfluidic technology, termed MFP, which can perform local (bio)chemical events on or in close vicinity of surfaces within highly restricted spatial confinements and in liquid environments.1 We now overcome critical challenges limiting the practical implementation of the MFP technology and present them in this paper. We did this by transforming the original technology, which was essentially coplanar to scanned surfaces, into a vMFP technology. Processing of surfaces using low volumes of liquids with spatial confinement is in many contexts a powerful capability. Developing such a technology to be used in conjunction with low volumes of liquids is one of the central goals of the microfluidic research community. It is common to make use of closed microchannels within which numerous applications have been demonstrated.2 However, high hydraulic resistance, the difficulty to introduce samples such as tissues, cells, and mesoscopic objects into microchannels and clogging by particulates and air bubbles strongly limit r 2011 American Chemical Society
the practical use of closed microfluidic systems. Moreover, biological samples are sensitive to their physical environment and are prone to drying artifacts and denaturation. Several strategies have been developed by the research community to overcome the limitations of closed microfluidics in their ability to pattern, process, and analyze samples on surfaces. Inkjet is the most widely used technology to deposit liquids on surfaces; early development of this technology was done by the industry, in particular by IBM, HP, and Canon.3 Inkjet technology is presently limited by several artifacts, including drying effects, the inability to pattern surfaces in liquid environments, and the physics of liquid ejection, which restricts the range of geometrically defined confinements of chemicals on surfaces. Also, the thermal/pressure variations in the droplet ejection may Received: January 27, 2011 Revised: March 10, 2011 Published: April 08, 2011 5686
dx.doi.org/10.1021/la2003639 | Langmuir 2011, 27, 5686–5693
Langmuir result in denaturation of biomolecules employed as ink in inkjet technology. One technique based on microfluidics for processing and interacting with surfaces involves the use of two capillaries placed atop a surface to form a closed fluidic system.4 With this technique chemical reactions are performed on surfaces and applied to electronic wafer-probing,4 patterning fluid lipid membranes onto a glass substrate,5 a droplet-based system to enable localized electrical stimulation, and recording of chemical species.6 While this technique is suitable for certain applications, it is desirable that the delivery/probing tools are not in physical contact with the surface. Another microfluidic approach is the use of micro/nanopipets to process surfaces within liquid environments while not needing physical contact with a surface.7 This involves filling a pipet with electrolytes and submerging the pipets within a liquid medium over a surface. The charged species are delivered on the surface by applying an electric potential. Some examples of the use of nanopipets are in the controlled delivery of DNA and proteins,7 scanning of biological surfaces,8 delivery of individual molecules to cellular compartments,9 and electrochemistry.10,11 A related implementation is scanning electrochemical microscopy to deposit gold on glass implemented in the form of a fountain pen.12 These techniques of applying an electric field need careful design of the electrolyte system, buffers, and biological mediums to enable the electromigration of charged species; designing such a system can be particularly challenging for a complex biological/clinical medium. What is critically needed is a noncontact approach that operates in a liquid environment and is minimally sensitive to the physiochemical properties such as charge and solubility of biological mediums and surface topographies. Other demonstrations of surface processing are dip pen nanolithography (DPN)13,14 and atomic force microscopy (AFM) based approaches. A particularly interesting development of the DPN is the use of feed channels connected to the tip of the AFM to deliver liquids.15 Hollow AFM probes have also been developed for patterning biomolecules on surfaces to scan regions on the millimeter-scale for cellular applications16 and the deposition of oligonucleotides.17 In such demonstrations, the piezo stage is suitable to process hundreds of nanometers of surfaces in the horizontal dimension and only a few nanometers range of variations in the vertical dimension. However, cellular and tissue assays often need processing of the millimeter range (and beyond) in the horizontal and micrometer range in the vertical dimension. Additionally, operating such AFM-based techniques in conjunction with cells/tissues is challenging, because the feedback needed to position the cantilever is confounded by its operation in liquid medium.18 Along these lines is a related demonstration of dispensing reagents to mammalian cells in liquid medium assisted by a polymeric aqueous two-phase system.19 An approach where flow of liquids is created on the surface and confined spatially solely by liquid boundaries is highly desirable. It is for such implementations that the characteristics inherent to microfluidics such as fluid dynamics under low Reynold number conditions can be fully exploited.20 Further, performing chemical and biological assays in this regime of liquid flows can be applied to a mode of noncontact probing/processing of surfaces. To this end, we developed a microfluidic technology, the MFP,5 in which a microfabricated probe head21 is the key component and which interfaces with the surfaces during processing. The MFP head consists of two apertures, one to inject liquid and another to
ARTICLE
Figure 1. vMFP head. (a) A scheme showing a vMFP head having vias, micrometer-size channels, and an apex into which the injection and aspiration channels open to form apertures. (b) Close-up view of channels/apertures showing typical flow rates and dimensions. (c) Photograph of a vMFP head consisting of Si and glass.
aspirate liquid. The injection/aspiration of the processing liquid is performed within an immersion liquid. The immersion liquid assists in the hydrodynamic focusing of the processing liquid at the MFP head. Such an approach overcomes the need to insert samples into closed microchannels. A few examples of what has already been demonstrated with the MFP are the lysis of a single human breast cancer cell followed by collecting and analyzing the mRNA released from the lysate,22 microperfusion of brain tissue,23 and delivery of multiple compounds for pharmacological screening of adhered cells.24 A variant of this approach is a dual pipet for controlled drug infusion.25 While the MFP represents a promising way forward to process surfaces in a liquid environment and operate in the noncontact mode, key challenges central to its practical implementation remains unsolved. Despite recent progress in the various ways described above to probe and process surfaces in liquid environments, there remain several implementation challenges limiting their broader use for chemical and biological applications: (1) solving the chip-toworld interface by bringing the interface of macroscopic devices toward a micrometer scale footprint on the surface of interest, (2) preventing clogging of samples within microchannels, (3) ensuring versatility and reconfigurability for different applications, (4) enabling simplified operation for use of the technique by nonexperts, (5) to be independent of the physicochemical properties such as charge and solubility of the analytes delivered and extracted on or from the surface processed, and (6) the ability to process surfaces typical in biology which are micrometers to centimeters in area, with topographical variations on the micrometer scale. We overcome the above listed challenges by developing the new concept of vMFP, which takes full advantage of the field of microtechnology. Instead of being operated coplanar to the surface, vMFPs are now vertically oriented and used in a scanning mode. This results in a vMFP head with all features in-plane and 5687
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir
ARTICLE
Figure 2. Key components of the vMFP platform. (a) A motorized, high-precision X, Y, Z stage controls the movement of the head and the surface to be processed. The tubing leads to external pumps, and visualization of the surface is performed using an inverted microscope (not shown). (b) Photograph of a custom-built holder that mounts the head and couples to the tubing that connects to the external pumps. (c) Patterns of fluorescently labeled antibodies on a glass surface were formed using a vMFP and visualized with an inverted microscope.
with the flow confinement at the apex of the vMFP head. The base configuration of the vMFP head, as shown in Figure 1, has a single channel for injection of the liquid and another channel for aspiration. In this paper, we present the design and fabrication strategies of vMFP heads, investigate their characteristics, integrate the dispensing of immersion liquid within the vMFP head, and demonstrate their utility by patterning proteins on surfaces and interaction with mammalian cells.
2. MATERIALS AND METHODS 2.1. vMFP Platform. The vMFP platform consists of a head, peripherals for handling liquids, a motorized precision X Y Z stage, and an inverted microscope (Figure 2). The head is a key component of the vMFP platform and is described in detail in the remainder of the paper. The equipment used to hold the head and to scan the substrate is mounted on the stage of a standard inverted microscope (Eclipse TE300, Nikon, Egg, Switzerland). The head itself is housed within a custom-made holder connected to an angle table for tilt correction. The unit for leveling and scanning the substrate consists of an adjustable holder onto which the substrate is clamped and two motorized linear drive units for X and Y positioning. Processing of a substrate is performed by programming the movement of the substrate relative to the vMFP head that is maintained stationary during operation. For the injection and aspiration of liquids, Nemsesys pumps (Cetoni GmbH, Korbussen, Germany) with syringes of 500 μL volume (1705 TLLX, Hamilton, Banaduz, Switzerland) are used. The fluidic connection of the pumps to the head uses standard 1/32 in. transparent tubes with PEEK fittings. The surfaces are monitored using the inverted microscope in the bright field or in fluorescence mode. The patterns on the surface are recorded using a digital camera (CDR-SR100E, Sony, Schlieren, Switzerland). For high-resolution imaging of the surface, an upright microscope (Nikon, Eclipse 90i, Egg, Switzerland) with a cooled, low-noise, high-sensitivity CCD camera (ST-8, SBIG, Santa Barbara, CA) is used. 2.2. Design of vMFP Heads. The vMFP head is a hybrid Si/glass microfluidic device consisting of several elements such as vias,
microchannels (flow paths), filter structures, and an apex that altogether play a critical role in defining the key characteristics of the flow confinement and system response time. Vias interface with the tubing and the external pumping peripherals. The preferred dimension (diameter) of the vias we designed are between 100 and 500 μm: vias having these dimensions are reasonably large to enable simple interfacing with the external fluidic peripherals; have sufficiently low dead volume, resulting in a system response time on the order of seconds; and are easy to fabricate without requiring one to etch high aspect ratio structures. The apex physically supports the flow confinement, comprises apertures and had an area of approximately 1 mm2. We chose this area as our earlier work suggests that the apex (termed “mesa” in the context of planar MFPs1) should be more than 10-fold larger than the spacing between the apertures to ensure that the flow confinement remains unperturbed to conditions such as convection in the immersion liquid. In addition, an area of 1 mm2 is still sufficiently small to easily level the apex and manipulate the head relative to the surface. The microchannels connect the vias to the apex, providing the flow paths to the apertures. For optimal flow confinement and switching times, we designed appropriate width, depth, spacing, and number of channels. We typically used microchannels that are 20 μm deep and 200 μm in width near the vias, which taper to 20 μm at the apertures. These structures have sharp walls and, if needed, can be isotropically etched to form curved walls.26 Structures such as posts/pillars are included along the flow paths to form filter elements to capture particulates. We designed the dimensions and spacing of the posts to ensure that they do not contribute more than 10% of the total resistance of the flow path. Such posts were grouped in areas for filtering particles consecutively in the ranges of 10, 5, and 1 μm. The overall dimension and shape of the head is determined by the functional elements that need to be integrated, the fabrication process, and the application. The heads were typically a rhombus with a surface area of 72 mm2, apex of 1 mm2, and thickness of ∼0.9 mm (Figure 1c). 2.3. Fabrication of the vMFP Heads. Vertical MFP heads were composed of a Si layer (400 μm thick) holding the microfabricated structures that was bonded with a glass layer (500 μm thick) (Figure 3a). 5688
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir
ARTICLE
Figure 3. Fabrication of vMFP heads. (a) Schematic of a vMFP head and close-up view of the apex showing etched structures in Si. (b) Illustration of the process steps and photographs of key stages in the fabrication of vMFP heads. The microchannels and vias were made in a 4 in. Si wafer (Siltronix, Geneva, Switzerland) using standard photolithography with a photoplotted mask (64 000 dpi, Zitzmann GmbH) and deep reactive ion etching (DRIE), as shown in Figure 3b. The microchannels were 20 μm deep, unless mentioned otherwise. A glass wafer (Borofloat 33, SCHOTT AG) was anodically bonded to the processed Si wafer at 450 °C and 1250 V. Channels/apertures of the vMFP heads were loaded with (molten) wax (OCON 199, Logitech GmbH) heated to 80 °C. After dicing, four vMFP heads were simultaneously mounted on a custom-made holder for lapping and polishing of the apex. The vMFP heads were immersed in heptane for 2 h at room temperature to remove the wax. 2.4. Setup To Characterize the Flow Confinement. The flow confinement was characterized by depositing a TRITC-labeled antibody (anti-guinea pig IgG developed in rabbit) (Sigma Aldrich, Milwaukee, WI) at a concentration of 250 μg mL 1 in PBS onto an aldehyde-coated glass slide (Xenobind, Hawthrone, NJ). The immersion liquid was deionized water. The footprint of the antibody patterns was investigated by varying the injection flow rates between 0.5 and 1.5 μL min 1 and with a constant aspiration flow rate of 4 μL min 1. Prior to the experiments, the substrate was leveled, and variations in the vertical dimension of the substrate over an area of 4 mm2 were found to be less than 2 μm. The detailed procedure to level the MFP head with respect to the surface is described elsewhere.21 Briefly, the head was slowly approached to the surface until Newton rings were seen through the inverted microscope. The tilt of the substrate was then adjusted using the screws of the substrate holder until the Newton rings were approximately in the center of the apex. Since there was no active feedback to regulate the distance between the head and the surface, it was necessary to establish the zero position for the z-axis. Once the leveling was completed, the zero position for the z-axis was determined and the flow confinements at various coordinates were established to form multiple
footprints with a 5 s residence time of the head at every location on the substrate. The substrate was then washed with PBS, and images were acquired using a CCD camera. 2.5. Cell Handling and Staining. NIH 3T3 Swiss mouse embryo fibroblasts (ATCC number CRL-1658) were grown on glass slides (Menzel) following the recommendation by ATCC. Briefly, the glass slides used for the cell cultures were washed with ethanol and plasma treated for 30 s using air and a coil power of 200 W. A slab of PDMS was inked with fibronectin (50 μg mL 1 in PBS) for 20 min, rinsed with PBS followed with water, and dried using N2. The inked PDMS was then placed on the plasma-treated glass slide. To hold the cell culture medium, cavities with diameter of 1.5 cm were punched in the PDMS, providing a fibronectin-coated glass bottom. The fibroblasts were seeded at densities between 7.5 104 and 1.25 105 cells cm 2. The cells were incubated for 1 2 days (37 °C, 5% CO2) and a confluent layer of adherent cells was obtained. For selective inactivation of cells, 2.5% sodium hypochlorite (Haenseler AG, Herisau, Switzerland) was used as the processing solution. To visualize the inactivated cells, the glass side was rinsed with trypan blue (Sigma-Aldrich) and with PBS at 37 °C to reduce background signal due to unspecific staining. For staining living cells, the processing solution was Cellomics whole cell stain red (1:20 in PBS, Thermo Sciences).
3. RESULTS AND DISCUSSION 3.1. Design Strategies. We detail below several designs and strategies for the fabrication of vMFP heads, which greatly helped fabricating the heads with high yield. For example, an ubiquitous issue with fabricating microfluidic device is the risk of clogging 5689
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir
ARTICLE
Figure 4. vMFP head design and packaging strategies. (a) Photograph of a processed Si/glass substrate with 33 vMFP heads. (b) Schematic showing the filling of molten wax into the channels servicing several vMFP heads. (c) Unpolished vMFP head with wax filled in the apertures and microchannels. (d) SEM image showing posts with varying dimensions along the flow path to trap particles. (e) Illustration of a vMFP head consisting of an angular flow path to generate in-plane momentum to the processing liquid for manipulation of objects on surfaces. (f) Redundancy in structures and trenches designed as polishing aids to increase the yield in fabrication. (g) Schematic of a head showing three apertures/channels to support a symmetric flow confinement.
microchannels with particulates, particularly during dicing and polishing. We used capillary forces to fill the apertures and microchannels of the heads with molten wax (Figure 4a). This was done on wafer level using channels that connected rows of pairs of apertures and that were fed from a single channel. Dicing the vMFP heads disconnected the apertures from the channels servicing the wax (Figure 4b). Figure 4c shows the apex of a nonpolished vMFP head with wax filled in the apertures. It took only a few minutes of lapping and polishing to yield a vMFP head with a flat apex and, typically, a head was stored with wax in the apertures until it was used. The microchannels linking the apertures to the vias used for liquid injection and aspiration provide ample room for implementing various microfluidic components. Figure 4d shows a series of posts that have diminishing dimensions and spacing in the direction from the via to an aperture. These structures prevent clogging of the apertures by particulates that enter the head with the injection liquid. Other components than filters can be electrodes and heaters for electrochemical detection and temperature optimization of processing liquids, respectively. The in-plane fabrication of microchannels provides interesting possibilities for vMFP heads that are not practical to implement in the case of planar MFPs. For example, apertures can be inclined and serviced with curved microchannels, as in Figure 4e. This opens possibilities such as (1) stretching the footprint of the processing liquid along a surface of interest, (2) giving more in-plane momentum to a processing liquid to move/align objects on a surface, or (3) defining specific shear stress conditions locally on
a surface for studying cell adhesion,27 mechanotransduction,28 and tissue engineering, as a few examples. Polishing small structures such as the apex of one vMFP head (with an area of about 1 mm2) can be challenging, because small changes in the pressure applied to the head during polishing/lapping can dramatically affect the polishing/lapping rate. This is particularly a concern if the head has inclined injection and aspiration microchannels, since the depth of polishing must be stopped exactly where a desired inclination of the channels is obtained. We therefore designed heads having repeated motives at the apex (Figure 4f). Vertical trenches were also added on each side of a motif. As a result, polishing can be stopped and the head inspected using a microscope to verify that polishing/lapping has occurred to the desired depth. If excessive polishing has occurred, only one motif is lost, but not the head, and polishing proceeds using the next motif. Interesting footprint geometries of the processing liquid on a surface can be obtained by having more than two apertures and/ or by implementing a symmetric flow aspiration.29 This may be achieved as shown in Figure 4g by splitting the aspiration flow path and implementing a hydraulic resistance on one of them to equalize flow resistances. 3.2. Flow Confinement Characterization. Characterization of the flow confinement was done by monitoring the fluorescence signal of the deposited TRITC-labeled antibodies on a surface. We systematically changed the distance between the vMFP head relative to the surface and the injection flow rate to find that, irrespective of the ratio of the injection to aspiration 5690
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir
Figure 5. Characterization of the flow confinement supported by a vMFP head with two apertures (apertures, 50 μm 50 μm; spacing, 50 μm). (a) Fluorescently labeled antibodies were deposited on an activated glass with different ratios of flow rates (aspiration flow rate of 4 μL min 1 and injection rate varying between 0.5 and 1.5 μL min 1) and distances of the head from the surface. Multiple patterns were formed using an automated script that operated computer-controlled stages. Uncertainty bars for each flow condition/distance represents 95% confidence on the mean of the antibody coverage (represented here as the area of the antibodies on the glass slide) on the glass slide. (b) Images of patterns created with an injection flow rate of 0.65 μL min 1 and aspiration flow rate of 4 μL min 1. The dotted outline around each pattern serves as a reference and has an area of ∼1.4 104 μm2.
flow rates, the evolution of the flow confinement followed the same trend (Figure 5a). At large distances (greater than 10 μm), the footprint specific to the flow conditions was not much affected by the distance between the head and the surface. Further, a significant increase in the footprint was observed when the head was approximately 10 μm away from the surface. These trends, shown in Figure 5a, can be explained on the basis of the spatial distribution of the confined processing liquid. When the head is sufficiently far, the spatial extension of the processing liquid is largely affected by the injection and aspiration flow rates and is insensitive to the distance between the head and the surface. However, when the head is close to the surface, the hydraulic resistance distribution is altered, thereby affecting the hydrodynamic focusing of the processing solution. This results in stretching of the footprint formed by the processing liquid along the surface, i.e. small changes in the distance are reflected as significant changes in the footprint. These findings reveal a method for easily estimating the distance between the vMFP head and a surface. Taking Figure 5b as an example, where the flow ratio (injection rate/ aspiration rate) was 0.625/4 and the head was about 80 μm from the surface, no antibodies were localized on the surface. The area of the footprint started growing as the head was advanced toward the surface, and at approximately 5 μm distance from the surface there was a significant increase in the footprint. The ability to monitor the footprint to extrapolate the distance between the head and the surface further indicates distinct regimes of operation for the vMFPs. For surface
ARTICLE
Figure 6. Integration of immersion liquid dispensing within a vMFP head. (a) Schematic of a vMFP head having immersion liquid dispensing channels (width, 500 μm; depth, 50 μm). (b) Photograph showing the encapsulation of immersion liquid at the apex of a vMFP head. (c) View of the apex with the outline of the immersion liquid marked with a dotted line. This image was captured in bright field after initiating the dispensing of immersion liquid and aspiration. (d) vMFP head supporting a stable confinement of the processing liquid after significant in-plane movement. (e) Visualization of the flow confinement of fluorescent liquid formed between two apertures in the presence of a black dye (background) as the immersion liquid.
topographies with variations in the vertical dimension on the order of 10 μm or higher, we suggest operating the vMFP with a large spatial distribution created by a high ratio of injection to aspiration flow rates. As an example, for adherent cells on surfaces it is common to have topographies with variations in the vertical dimension on the order of 10 μm. Therefore, the regime of operation with large spatial distributions (e.g., with 1 μL min 1 injection and 4 μL min 1 aspiration flow rate; Figure 5a) of the processing liquid will be appropriate. Similarly, a smaller spatial distribution of the flow confinement (e.g., with 0.75 μL min 1 injection and 4 μL min 1 aspiration flow rate; Figure 5a) is suited for topographies with variations in the range of 1 5 μm in the vertical dimension. 3.3. Integrating Immersion Liquid Dispensing into the vMFP Head. The immersion liquid is central to the flow confinement and local processing of surfaces. In previous work1,21 and preceding sections, the immersion liquid was supplied in hundreds of microliters to milliliter quantities onto the surface of interest prior to operating the vMFP using a pipet. Resupply of immersion liquid was sometimes needed due to its evaporation and fast consumption when high flow rates were used. Aspiration flow rates can reach up to 10 μL min 1, approximately 80% of which is composed of immersion liquid. For these reasons, we developed vMFP heads integrating the immersion liquid dispensing. This integration was obtained by adding a third via to the head, which serviced peripheral microchannels that open on the sidewalls of the vMFP (Figure 6a). Directing the channels for immersion liquid to the sidewalls of the vMFP helps to keep the apex small. These vMFP heads dispense the immersion liquid locally where it is needed while the surrounding area of the substrate can be kept dry. Further, the volume of the immersion liquid encapsulating the apex can be changed by iteratively adjusting the dispensing rate of immersion liquid flow. If the channels providing the 5691
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir
Figure 7. Interaction of vMFPs with fibroblasts. (a) Optical micrograph and corresponding inset of selective inactivation of fibroblasts using a vMFP head (apertures with 50 μm 50 μm dimension and 50 μm spacing) using 2.5% sodium hypochlorite as processing liquid. The immersion liquid was PBS. (b) Image captured in fluorescent mode showing a pattern (“MFP”) written on fibroblasts using a vMFP head with Cellomics as the processing liquid. (c) Detailed view of fibroblasts exposed to Cellomics.
immersion liquid are high on the vMFP head, grooves along the sidewalls of the head can be formed to guide the immersion liquid toward the apex. In order to maintain symmetry in the immersion liquid encapsulation at the apex and to reduce interfaces to the external fluidic peripherals, we designed channels with equal hydraulic resistances and merged the two dispensing channels into a single via (Figure 6a). We tested for the stability of the immersion liquid at the apex of the vMFP head at a flow rate of 3 μL min 1 (Figure 6b). During regular operation, dispensing of the immersion liquid was initiated followed by aspiration, ensuring encapsulation of the entire apex with liquid (Figure 6c). The injection of the processing liquid was then started, which resulted in a flow confinement (Figure 6d,e). We tested the stability of this flow confinement for extended periods (∼60 min) and found that evaporation, scanning in the X and Y directions, convection, and surface characteristics (such as wetability and roughness) had no observable effect on the confinement of the processing liquid. This integration is not only convenient but also opens interesting possibilities, such as performing local electroless deposition of metals,30 eliminating corrosion of some metals due to the formation of a galvanic cell during the fabrication of conductive elements,31 treating fragile materials locally on a surface, and performing reactions at the interface of laminar flows.32 Arguably, a vMFP head having the immersion liquid integrated into it offers a shift in paradigm from processing surfaces using masks and a bulk amount of chemicals toward a much more conservative use of chemicals. 3.4. Localized Chemistry on Cell Cultures. Interacting and performing localized bio(chemical) reactions on cells is highly desirable in biomedical research for numerous applications, such as engineering cellular architectures to create artificial tissues, modulation of stem cell microenvironments during differentiation for regenerative medicine, and dispensing chemicals at
ARTICLE
varying concentrations on cells for screening and drug toxicology studies. For these reasons, we investigated whether the vMFP can be used to inactivate adherent cells and to dispense chemicals on living cells. To this end, heads were designed to match the spatial distributions appropriate to interact with a few cells and sustain flow conditions that do not perturb their position. These head designs support flow confinements ranging from a few (for single cells, Figure 3b) to a few hundred (for cell populations) squared micrometers. In order to inactivate cells, a vMFP head supporting a flow confinement of sodium hypochlorite was positioned 30 μm above a confluent layer of NIH 3T3 mouse fibroblasts. The adherent cells in the path of the head’s trajectory got in contact with sodium hypochlorite and were inactivated (Figure 7a). The average residence time of the vMFP over a fibroblast was ∼1 s, corresponding to 50 nL of sodium hypochlorite dispensed. Inactivated cells were stained using trypan blue and rinsed with PBS; the living cells remained unstained. Alternate indicators of inactivated cells were morphological changes, such as detachment and shrinkage, as seen in Figure 7a. We further investigated local staining of living cells. Cellomics, a whole cell fluorescent stain, was used as the processing liquid. The head migrated at 0.1 mm s 1 and stained adherent fibroblasts along its trajectory. An observation from Figure 7b,c is that the morphology of the stained fibroblasts remained unchanged and the cells continued to be adherent, even after exposure to the flow confinement. Therefore, the flow conditions used (i.e., the sheer forces excreted on the cells) were adequate and did not inactivate or dislodge the fibroblasts. The vMFP trajectories along the adherent cells did not result in sharp interfaces as with patterns formed by the immobilization of fluorescently labeled antibodies on glass (Figure 2). This is likely due to the heterogeneity of the fibroblast population (in shape and distribution), internal diffusion of chemicals within cells, dislodging of cells when inactivated, and perturbations to the flow confinement due to topographical variations. In this work, we did not achieve single cell resolution but were able to interact with a few neighboring cells (two to four cells).
4. CONCLUDING REMARKS The vMFP is a significant step forward in the area of technologies for surface processing and interaction with biological interfaces. The fabrication of vMFP heads is simple and efficient, as long as good design rules are followed. Since the vMFP heads are composed of Si and glass, they are compatible with a large range of chemicals and organics and can be cleaned for reuse. The vMFP is highly versatile, because it localizes chemical events in a simple and interactive manner on surfaces using hydrodynamic forces created by liquids instead of solid walls of microchannels. It exploits laminar flows and physics typical of the microfluidic regime and uses them to process, manipulate, and study liquid liquid and liquid solid interfaces. Unique to the vMFP is the scanning mode of operation; it processes confined spaces, enables easy visualization and alignment with the surface, is chemically resistant, needs only small volumes of reagents/buffers (tens of picoliters), and is able to interact with samples in a noncontact mode. The in-plane design of the vMFP head is well-suited to integrate multiple functional elements such as heaters, electrodes, valves, and sensors. Heating elements can be readily fabricated and used to regulate the temperature of the liquid 5692
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
Langmuir prior to interacting with cells, to catalyze enzymatic reactions, and to increase the rapidity in hybridization of DNA/RNA. Electrodes could be used in interesting ways to perform electrochemical reactions, detection, and analysis. A unique design of flow path geometries can be used to sort and deliver cells, retrieve beads after they capture analytes of interest, and enable reactions at liquid liquid interfaces. Valves along the flow paths can be integrated to perform rapid switching of multiple processing liquids and create low dead-volume environments to deliver and retrieve precious reagents. Further, the glass on the head allows optical access of the injected and aspirated liquids; therefore, it is well-suited to integrate waveguides and sensors for optical or fluorescence detection. We hypothesize that the vMFP can be used for multiplexed staining of tissue samples, performing complex molecular and cell biology procedures including single-cell retrieval/analysis, and as a surgical tool to deliver drugs locally on cancerous cells. The vMFP might therefore spur a diverse and broad range of applications in microtechnology, surface chemistry, biology, and medicine.
’ AUTHOR INFORMATION Corresponding Author
*E-mail:
[email protected].
’ ACKNOWLEDGMENT The authors thank L. Gervais and M. Hitzbleck for discussions. M. Despont and W. Riess are acknowledged for their continuous support. ’ REFERENCES
ARTICLE
(16) Meister, A.; Gabi, M.; Behr, P.; Studer, P.; Voros, J.; Niedermann, P.; Bitterli, J.; Polesel-Maris, J.; Liley, M.; Heinzelmann, H.; Zambelli, T. Nano Lett. 2009, 9, 2501–2507. (17) Descamps, E.; Leichle, T.; Corso, B.; Laurent, S.; Mailley, P.; Nicu, L.; Livache, T.; Bergaud, C. Adv. Mater. 2007, 19, 1816–1821. (18) Oberti, S.; Stemmer, A.; Juncker, D.; Duerig, U.; Schmid, H. Appl. Phys. Lett. 2005, 86, 3. (19) Tavana, H.; Jovic, A.; Mosadegh, B.; Lee, Q. Y.; Liu, X.; Luker, K. E.; Luker, G. D.; Weiss, S. J.; Takayama, S. Nat. Mater. 2009, 8, 736–741. (20) Brody, J. P.; Yager, P.; Goldstein, R. E.; Austin, R. H. Biophys. J. 1996, 71, 3430–3441. (21) Lovchik, R. D.; Drechsler, U.; Delamarche, E. J. Micromech. Microeng. 2009, 19, 8. (22) Shiku, H.; Yamakawa, T.; Nashimoto, Y.; Takahashi, Y.; Torisawa, Y.; Yasukawa, T.; Ito-Sasaki, T.; Yokoo, M.; Abe, H.; Kambara, H.; Matsue, T. Anal. Biochem. 2009, 385, 138–142. (23) Queval, A.; Ghattamaneni, N. R.; Perrault, C. M.; Gill, R.; Mirzaei, M.; McKinney, R. A.; Juncker, D. Lab Chip 2010, 10, 326–334. (24) Ainla, A.; Jansson, E. T.; Stepanyants, N.; Orwar, O.; Jesorka, A. Anal. Chem. 2010, 82, 4529–4536. (25) Feinerman, O.; Moses, E. J. Neurosci. Methods 2003, 127, 75– 84. (26) Madou, M. J. Fundamentals of Microfabrication: The Science of Miniaturization, 2nd ed.; CRC Press: New York, 2001. (27) Lu, H.; Koo, L. Y.; Wang, W. C. M.; Lauffenburger, D. A.; Griffith, L. G.; Jensen, K. F. Anal. Chem. 2004, 76, 5257–5264. (28) Young, E. W. K.; Simmons, C. A. Lab Chip 2010, 10, 143–160. (29) Qasaimeh, M. A.; Juncker, D. Proceeding of the 14th International Conference on Miniaturized Systems for Chemistry and Life Sciences, μTAS 2010, 3 7 October 2010, Groningen 2010, pp 1844 1846. (30) Geissler, M.; Kind, H.; Schmidt-Winkel, P.; Michel, B.; Delamarche, E. Langmuir 2003, 19, 6283–6296. (31) Carvalho, A.; Geissler, M.; Schmid, H.; Michel, B.; Delamarche, E. Langmuir 2002, 18, 2406–2412. (32) Kenis, P. J. A.; Ismagilov, R. F.; Whitesides, G. M. Science 1999, 285, 83–85.
(1) Juncker, D.; Schmid, H.; Delamarche, E. Nat. Mater. 2005, 4, 622–628. (2) Whitesides, G. M. Nature 2006, 442, 368–373. (3) Le, H. P. J. Imaging Sci. Technol. 1998, 42, 49–62. (4) Routenberg, D. A.; Reed, M. A. Lab Chip 2010, 10, 123–127. (5) Smith, K. A.; Gale, B. K.; Conboy, J. C. Anal. Chem. 2008, 80, 7980–7987. (6) Chen, D.; Du, W. B.; Liu, Y.; Liu, W. S.; Kuznetsov, A.; Mendez, F. E.; Philipson, L. H.; Ismagilov, R. F. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 16843–16848. (7) Bruckbauer, A.; Ying, L. M.; Rothery, A. M.; Zhou, D. J.; Shevchuk, A. I.; Abell, C.; Korchev, Y. E.; Klenerman, D. J. Am. Chem. Soc. 2002, 124, 8810–8811. (8) Ying, L. M.; Bruckbauer, A.; Zhou, D. J.; Gorelik, J.; Shevehuk, A.; Lab, M.; Korchev, Y.; Klenerman, D. Phys. Chem. Chem. Phys. 2005, 7, 2859–2866. (9) Bruckbauer, A.; James, P.; Zhou, D. J.; Yoon, J. W.; Excell, D.; Korchev, Y.; Jones, R.; Klenerman, D. Biophys. J. 2007, 93, 3120–3131. (10) Piper, J. D.; Li, C.; Lo, C. J.; Berry, R.; Korchev, Y.; Ying, L. M.; Klenerman, D. J. Am. Chem. Soc. 2008, 130, 10386–10393. (11) Williams, C. G.; Edwards, M. A.; Colley, A. L.; Macpherson, J. V.; Unwin, P. R. Anal. Chem. 2009, 81, 2486–2495. (12) Cortes-Salazar, F.; Lesch, A.; Momotenko, D.; Busnel, J. M.; Wittstock, G.; Girault, H. H. Anal. Methods 2010, 2, 817–823. (13) Piner, R. D.; Zhu, J.; Xu, F.; Hong, S. H.; Mirkin, C. A. Science 1999, 283, 661–663. (14) Ginger, D. S.; Zhang, H.; Mirkin, C. A. Angew. Chem., Int. Ed. 2004, 43, 30–45. (15) Kim, K. H.; Sanedrin, R. G.; Ho, A. M.; Lee, S. W.; Moldovan, N.; Mirkin, C. A.; Espinosa, H. D. Adv. Mater. 2008, 20, 330–334. 5693
dx.doi.org/10.1021/la2003639 |Langmuir 2011, 27, 5686–5693
