View Online / Journal Homepage / Table of Contents for this issue

Dynamic Article Links

Lab on a Chip Cite this: Lab Chip, 2012, 12, 4029–4032

COMMUNICATION

www.rsc.org/loc

Electrode-free picoinjection of microfluidic drops{ Brian O’Donovan, Dennis J. Eastburn and Adam R. Abate*

Downloaded by University of California - San Francisco on 13 October 2012 Published on 17 August 2012 on http://pubs.rsc.org | doi:10.1039/C2LC40693D

Received 19th June 2012, Accepted 16th August 2012 DOI: 10.1039/c2lc40693d

Existing methods for injecting reagents into drops utilize metal electrodes integrated into the microfluidic chip, adding cumbersome and error-prone steps to the fabrication process. Here, we introduce a method that uses the injection fluid itself as an electrode by exploiting dissolved electrolytes in the solution. This obviates the need for metal electrodes and reduces fabrication time and complexity while also affording precise control of the injected reagent volumes. Droplet-based microfluidic devices are rapidly becoming an integral tool in the molecular biology laboratory. Use of microfluidic drops as self-contained miniature bioreactors, or ‘‘test tubes’’ for biological experiments, allows for greatly increased throughput while reducing reagent volumes compared to traditional methods.1 One of the most important functions a microfluidic chip must perform to carry out multi-step reactions is the precise and specific addition of reagents to the droplets.2 This can be accomplished by the controlled merging of pairs of droplets through electrocoalescence, whereby an electric field disrupts the droplet interfaces, allowing coalescence in an otherwise stable emulsion.3 This concept is also exploited in the design of the picoinjector, a device that allows for addition of reagents into droplets by applying an electric field as they pass a pressurized channel containing the fluid to be added.4 The original design of the picoinjector, however, requires the fabrication of metal-solder electrodes to apply the electric field, adding complex steps to an otherwise simple device fabrication protocol.5,6 A method for accomplishing picoinjection that eliminated the need for metal electrodes would therefore represent a major improvement over current techniques. In this communication, we introduce a method to picoinject reagents into drops without the use of metal-solder electrodes; instead, the injection fluid itself is used as the conductor through which the electric field is applied. In addition to markedly simplifying device fabrication by obviating the need for electrodes, our process allows more controlled picoinjection: Unlike in previous approaches in which the volume injected is independent of voltage, in our approach, the injection volume depends on the voltage. Consequently, our technique allows injection volume to be Department of Bioengineering and Therapeutic Sciences, California Institute for Quantitative Biosciences, University of California, San Francisco, San Francisco, California, 94158, USA. E-mail: [email protected]; [email protected] { Electronic supplementary information (ESI) available: Expanded description of RT-PCR protocol

This journal is ß The Royal Society of Chemistry 2012

controlled by adjusting the voltage of the applied signal. This is useful because voltage can be adjusted at a rate comparable to the drop rate, allowing injection volume to be adjusted on a drop-todrop basis, and the specific tailoring of the conditions within every microreactor. We fabricate our microfluidic devices in poly(dimethylsiloxane) (PDMS) using soft photolithographic techniques.5 The devices have channel heights of 30 mm, optimal for the picoinjection of water-in-oil droplets that are 50 mm in diameter. The device design is otherwise identical to those described previously,4 except that the channels for the metal-solder electrodes are removed, since they aren’t needed. We also implement a ‘‘Faraday Mote’’ – an empty channel filled with a conducting aqueous solution – that runs between the injection site and the droplet spacer, as shown in Fig. 1B. The mote electrically isolates re-injected drops upstream of the picoinjection site from electric fields emanating from the picoinjector, preventing unintended merging. The emulsion we picoinject consists of monodisperse droplets of 3.8 mM fluorescein sodium salt (C20H10Na2O5) dissolved in MilliQ H2O. The droplets are suspended in a carrier oil of Novec HFE7500 fluorinated oil with 2% (wt/wt) dissolved biocompatible surfactant.7 The picoinjection fluids consist of a dilution series of NaCl ranging from 0 to 500 mM, each containing 3.8 mM fluorescein sodium salt. We chose this range of concentrations to reflect the molarities of dissolved ions present in most biological buffers and reagents. Thus, since in most applications the fluids will already contain the requisite ions, our technique can be used without adding additional reagents to the solutions. Droplets and carrier oil are introduced via syringe pumps (New Era) and spaced using the same carrier oil and surfactant mixture described above (Fig. 1A, B). The picoinjection fluid is contained in a BD Falcon tube. Through the cap of the Falcon tube we submerge a wire electrode into the fluid, as illustrated in Fig. 1A. We seal gaps in the cap with LocTiteTM UV-cured epoxy. The picoinjection fluid is charged using a function generator outputting a 10 kHz sinusoidal signal ranging from 0 to 5 volts. This output is amplified 1000 6 by a Trek 609E-6 model HV-amplifier. The positive output of the amplifier is attached via an alligator clip to the wire submerged in the picoinjection fluid. We attach the ground electrode of the amplifier to the metal needle of a syringe containing a 1 M solution of NaCl, introduced into the Faraday Mote (Fig. 1A). The two electrodes are never in electrical contact and the emulsions exiting the device are collected into separate, electrically isolated containers to prevent a closed circuit or any current flow. Lab Chip, 2012, 12, 4029–4032 | 4029

Downloaded by University of California - San Francisco on 13 October 2012 Published on 17 August 2012 on http://pubs.rsc.org | doi:10.1039/C2LC40693D

View Online

Fig. 1 Schematic of device setup. (A) Drops, spacer oil, and 1 M NaCl are introduced to the PDMS device via syringe pumps. The picoinjection fluid is introduced using an air pressure control pump. Electrodes from the high voltage amplifier are connected to a wire submerged in the picoinjection fluid and to the metal needle of the syringe containing the 1 M NaCl ‘‘Faraday Mote.’’ (B) A magnified view of the droplet spacer and picoinjection site. (C) Further magnified view of the picoinjection site showing the fluid bulge at the injection orifice.

The picoinjected reagent is infused into the device through PE-2 tubing (Scientific Commodities) using an air pressure pump (ControlAir Inc.) controlled by custom LabVIEW software. The injection fluid is pressurized such that the oil/water interface at the picoinjection orifice is in mechanical equilibrium with the droplet channel; the pressure difference across the interface is equal to the Laplace pressure, causing the injection fluid to bulge into the droplet channel without budding off and forming its own drops (Fig. 1C).8,9 For this device, we inject the drops and spacer oil at flow rates of 200 and 400 mL hr21, respectively. At these flow rates, the picoinjection fluid interface is in mechanical equilibrium for an applied pressure of y13 psi. We control the lengths of the tubing carrying the injection fluid and solution serving as a Faraday mote, since longer tubes have higher electrical resistance and may attenuated the AC signal we apply to trigger picoinjection. To picoinject drops with reagent, we re-inject the previously formed monodisperse emulsion into the picoinjection device. The emulsion is introduced at a high volume-fraction such that there is little carrier oil and the drops are packed together. The packed drops travel through a narrowing channel that forces them single file. Additional oil with surfactant is added from two perpendicular channels, spacing the drops evenly, as shown in Fig. 1B. We have also tested our version of picoinjection with a simple T-junction spacer and found that this works too. The droplets then pass the picoinjector, a narrow channel containing the reagent to be added. To trigger picoinjection, we apply the voltage signal to the electrode submerged in the injection fluid, generating an electric field at the picoinjector as the drops pass the injection site. This causes the drops to coalesce with the injection fluid. As they travel past, fluid is injected into them through a liquid bridge formed after the two fluids coalesce.4 The applied signal must have zero offset to prevent electrophoretic migration of charged particles in the solutions. Additionally, the frequency of the signal must be high enough to ensure that during the act of injecting, the sign of the field switches many times between positive and negative, so that the net charge of the fluid added to the droplets is approximately zero. This ensures that the droplets leaving the injector have zero net charge, which is important for ensuring that they remain stable. For our experiments, a 10 kHz signal was applied. 4030 | Lab Chip, 2012, 12, 4029–4032

To analyse the behaviour of our picoinjector, we observe the injection site under a microscope. In the absence of an electric field, a distinct boundary is observed between the droplet and the injection fluid, as shown in Fig. 2A. When a 250 V signal is applied to the picoinjector, the boundary vanishes and droplet coalescence is visible, as demonstrated in Fig. 2B. Thus, electrification of the injection fluid is adequate to trigger picoinjection, demonstrating that electrically-isolated electrodes are not needed. To determine if it is possible to vary the injection volume using the applied voltage, we vary the voltage between 0–5000 V and measure the volume change of the resulting droplets. We quantify injection volume with an optical fluorescence detection setup. As the drops pass a 472 nm wavelength laser focused on the droplet channel y1 cm downstream of the picoinjector, the emitted fluorescence signal from the dissolved fluorescein contained within the drops is amplified by a photomultiplier tube (PMT) and converted to a voltage signal analyzed with LabVIEW FPGA. As the drops pass the laser, their fluorescence signals resemble square waves as a function of time, with amplitudes and widths that correspond to the drop intensity and length, respectively. The drops have a spherical diameter larger than the dimensions of the channel, causing them to be cylindrical in shape. Thus, the drop volume is approximately linear as a function of length. To calculate the volume fractional (Vf) increase, we measure the ratio of the drop length before and after picoinjection. We repeat these

Fig. 2 Bright field microscopy images of the picoinjection site. In the absence of an electric field (A), surfactants prevent coalescence with the injection fluid and a distinct boundary is visible at the droplet/injection fluid interface. When the electric field is applied, the boundary disappears and reagent is injected as the droplet passes (B).

This journal is ß The Royal Society of Chemistry 2012

Downloaded by University of California - San Francisco on 13 October 2012 Published on 17 August 2012 on http://pubs.rsc.org | doi:10.1039/C2LC40693D

View Online

Fig. 3 Volume fraction increase (Vf) of drop size after injection for (A) 100 mM, (B) 50 mM, and (C) 25 mM injection fluids. A stronger electric field more readily ruptures the oil/water interfaces allowing injection over a larger length of the passing droplets, and larger injection volumes. Higher molarities of dissolved electrolytes produce stronger electric fields at the injection site for a given voltage, also increasing injection volume. The error bars represent 1 standard deviation in either direction for .1200 drops sampled at each point.

measurements for a range of applied voltages and molarities of NaCl in the injection fluid. We plot the increase in volume as a function of applied voltage for three representative molarities of injection fluid in Fig. 3. In all cases the injection volume increases with the applied voltage, though this effect is most prominent for the 100 mM injection solution shown in Fig. 3A. We attribute the dependence of the droplet volume on the applied voltage to the observation that the droplets are not perfect cylinders as they travel past the picoinjector. Instead they have a ‘‘bullet’’ shape, with the leading edge having a smaller radius of curvature than the trailing edge, as has been observed previously for drops traveling through microchannels.10 Consequently, as the drops pass the picoinjector, the thickness of the oil layer separating their interface from the bulge of the picoinjection fluid decreases. For an electricallyinduced thin-film instability, the threshold voltage required to rupture the interface depends on the thickness of the film, decreasing as the film gets thinner.11,12 Hence, because the film thickness decreases as the drops pass the picoinjector, the moment of coalescence depends on electric field magnitude: For higher fields it is possible to rupture thicker films, leading to picoinjection at an earlier point; conversely, for lower fields thinner films are ruptured, causing picoinjection to start at a later point. Because the volume injected depends on the duration of picoinjection,4,13 it therefore also depends on applied voltage. This is supported by our data which shows a dependence on applied voltage for all molarities (Fig. 3). We also observe that the curves relating volume injected to applied voltage are lower for lower molarities, as shown for the 50 mM and 25 mM data in Fig. 3B and 3C, respectively. We attribute this to the fact that lower molarity solutions have a lower conductivity, and can thus attenuate the AC signals used to trigger injection, reducing the volume injected for a particular applied voltage. Above 3000 V and 100 mM, the injected volume begins to decrease and the variability in drop size increases. In images of these systems at these voltages, we observe that the picoinjection fluid is no longer held at equilibrium in the picoinjection orifice, but instead wets the channel walls and buds off small drops into the flow channel. We believe at these voltages electrowetting causes the aqueous phase to partially wet the hydrophobic channels, This journal is ß The Royal Society of Chemistry 2012

crawling out of the picoinjector and forming drops in the flow channel.14 To characterize the behavior of our electrode-free picoinjector for all parameters, we measure injection volume as a function of molarity and applied voltage and plot the resulting data on a 2D heat-map, Fig. 4. This data shows that electrification of the injection fluid can trigger injection over a wide range of molarities and applied voltages. It also shows that even for less conducting fluids at lower molarities, picoinjection can be triggered by increasing the voltage to a threshold value that depends on the molarity. This demonstrates that the technique should allow controlled picoinjection for most biological buffers, which commonly have molarities within the tested range. A potential concern with using the injection reagent as the electrode through which picoinjection is triggered is that the electric fields and currents generated by the high-voltage signal may disrupt biomolecules needed for downstream assays. To investigate this issue, we used our picoinjector to prepare droplets for an RT-PCR reaction.{ Drops containing total RNA isolated from an MCF7 human cell line were picoinjected with an RT-PCR reaction mixture containing the enzymes reverse transcriptase (RT)

Fig. 4 Heat map showing injection volume as a function of applied voltage and the molarity of dissolved NaCl in the injection fluid. Arrows/ ticks indicate data points. The injection volume can be adjusted in the range of 0–36 pL with a resolution of y2.6 pL (4% Vf) with 100 V increments of the applied signal. The largest injected volumes were 3000 V with the 100 mM fluid. Increasing electric field above this allows for electrowetting, causing drops to spontaneously form at the picoinjector, adversely affecting injection efficacy and consistency.

Lab Chip, 2012, 12, 4029–4032 | 4031

View Online

Downloaded by University of California - San Francisco on 13 October 2012 Published on 17 August 2012 on http://pubs.rsc.org | doi:10.1039/C2LC40693D

metal electrodes must be included at all sites, greatly complicating device fabrication. Our technique should also enable the execution of more complex assays that are tailored to the specific conditions within each microdrop: with previous methods injection volume could only be adjusted slowly by varying injection pressure or droplet velocity. With our approach, injection volume depends on applied voltage, allowing it to be adjusted electronically. This allows injection volume to be adjusted at rates much faster than with previous methods, and fast enough to keep pace with even the fastest drop rates reported.

Acknowledgements Fig. 5 Ethidium bromide stained 2% agarose gel. Total RNA isolated from an MCF7 human cell line was encapsulated in drops and picoinjected with an RT-PCR reaction mixture either with (+) or without (-) reverse transcriptase (RT) and Taq DNA polymerase. Non-emulsified control reactions were performed in parallel. Both positive control and picoinjected reactions yielded PCR product bands for the 100 bp amplicon, demonstrating that the electric field generated during picoinjection is biologically compatible with DNA, reverse transcriptase, and Taq.

We thank Sean Cater, Adam Sciambi, Allen-Deon Saunders, and Laurens Kraal for assistance with the experimental setup and software development. This work was supported by the Department of Bioengineering and Therapeutic Sciences at UCSF, the California institute for Quantitative Biosciences (QB3), the Bridging the Gap Award from the Rogers Family Foundation.

and Taq DNA polymerase. Negative-control drops were injected with a mixture containing no enzymes. Additional non-emulsified positive and negative control reactions were performed in parallel with the same RT-PCR mixture. Following thermocycling, the emulsions were broken and the amplification products visualized on an ethidium bromide-stained 2% agarose gel. The positive control and picoinjected drops both showed PCR product bands for the expected 100 bp amplicon length, as visible in Fig. 5.By contrast, the negative controls showed no amplification. This experiment demonstrates that applying the triggering signal to the picoinjection fluid is sufficiently biocompatible so as to allow downstream RT-PCR reactions in drops.

References

Conclusions Our method of applying a voltage directly to the picoinjection fluid markedly simplifies the fabrication of microfluidic devices using picoinjection by eliminating the need for metal electrodes. It should also simplify the execution of multiple picoinjections since, with our technique, this requires adding only picoinjection channels at the desired sites of injection; by contrast, with previous methods, picoinjection channels and accompanying

4032 | Lab Chip, 2012, 12, 4029–4032

1 E. Kritikou, Nat. Rev. Genet., 2005, 6, 668. 2 H. Song and D. L. Chen, Ismagilov, Angew. Chem., Int. Ed., 2006, 45, 7336–7356. 3 K. Ahn, J. Agresti, H. Chong, M. Marquez and D. A. Weitz, Appl. Phys. Lett., 2006, 88, 264105. 4 A. R. Abate, T. Hung, P. Mary, J. J. Agresti and D. A. Weitz, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 19163. 5 D. C. Duffy, J. C. McDonald, O. J. A. Schueller and G. M. Whitesides, Anal. Chem., 1998, 70, 4974–4984. 6 A. C. Siegel, D. A. Bruzewicz, D. B. Weibel and G. M. Whitesides, Adv. Mater., 2007, 19, 727–733. 7 C. Holtze, A. C. Rowat, J. J. Agresti, J. B. Hutchison, F. E. Angile`, C. H. J. Schmitz, S. Ko¨ster, H. Duan, K. J. Humphry, R. A. Scanga, J. S. Johnson, D. Pisignano and D. A. Weitz, Lab Chip, 2008, 8, 1632–1639. 8 T. F. Zien, Bull. Math. Biophys., 1969, 31, 681–694. 9 S. A. Vanapalli, A. G. Banpurkar, D. van den Ende, M. H. G. Duits and F. Mugele, Lab Chip, 2009, 9, 982–990. 10 C. Chung, M. Lee, K. Char, K. H. Ahn and S. J. Lee, Microfluid. Nanofluid., 2010, 9(6), 1151–1163. 11 S. Herminghaus, Phys. Rev. Lett., 1999, 83, 2359–2361. 12 C. Priest, S. Herminghaus and R. Seemann, Appl. Phys. Lett., 2006, 89, 134101. 13 P. Mary, A. R. Abate, J. Agresti and D. A. Weitz, Biomicrofluidics, 2011, 5, 024101. 14 F. Malloggi, S. A. Vanapalli, H. Gu, D. Ende and F. J. Mugele, J. Phys.: Condens. Matter, 2007, 19, 462101.

This journal is ß The Royal Society of Chemistry 2012