On-Demand Mixing Droplet Spotter for Preparing Picoliter Droplets on

May 1, 2004 - An on-demand mixing droplet spotter for generating and mixing picoliter droplets has been developed for ultras- mall reaction vessels...
0 downloads 0 Views 149KB Size
Download PDF
Anal. Chem. 2004, 76, 2991-2996

On-Demand Mixing Droplet Spotter for Preparing Picoliter Droplets on Surfaces Osamu Yogi,† Tomonori Kawakami,*,‡ and Akira Mizuno§

Tsukuba Research Laboratory and Central Research Laboratory, Hamamatsu Photonics K.K., 5000 Hirakuchi, Hamakita, Shizuoka 434-8601, Japan, and Surface Molecular Science Division, Institute for Molecular Science, Okazaki National Research Institutes, Myodaiji, Okazaki, Aichi 444-8585, Japan

An on-demand mixing droplet spotter for generating and mixing picoliter droplets has been developed for ultrasmall reaction vessels. The droplets were generated by applying a ∼500-V, ∼2-ms pulsed voltage to the tips of capillary tubes (o.d. ∼20 µm; i.d. ∼12 µm) filled with solution. The mixing process was achieved using electrostatic force. The initial droplet was formed by applying the pulsed voltage between one capillary and the substrate, and the second jet of the other solution was generated from the other capillary and collided with the initial droplet automatically because the electric field lines concentrated on the initial droplet. Using this mixing process, a microarray having a concentration gradient was obtained by spotting ∼6-pL droplets on a surface with a density of one spot per 75 × 75 µm2. Microscale analytical methods have been of interest in biochemical research, drug discovery, environmental chemistry, and other areas where sample availability is often limited. Yet these methods require improved performance in terms of speed, sensitivity, and cost of analysis. Generating droplets smaller than 1 nL is a fundamental technique in modern analytical chemistry. Current methods for generating droplets can be divided into three types: vibrating orifice, extrusion, and electrospray. In the vibrating orifice method, an orifice of a capillary with a liquid stream is vibrated, thus causing the stream to split into droplets. This technique generates droplets smaller than 50 pL, and typically 1-20 pL, one at a time at a fast operating rate (10-100 kHz). The higher the frequency, the smaller the volume of droplets will be. This technique is inconvenient however, for regularly spotting droplets on a surface, for example, fabrication of a microarray, because its operating rate is too fast. The faster the operating rate, the harder it becomes to position droplets regularly on a surface. In the extrusion method, liquid in a container is pushed or extruded from an orifice to generate the droplets. This technique generates one-by-one and on-demand droplets smaller than 1 µL, and typically 30-500 pL, at a moderately fast operation rate (lower than 500 Hz).1-4 The * Corresponding author. E-mail: [email protected]. Tel: +81-53-5840250. Fax: +81-53-584-0260. † Tsukuba Research Laboratory, Hamamatsu Photonics K.K. ‡ Central Research Laboratory, Hamamatsu Photonics K.K. § Okazaki National Research Institutes. (1) Lemmo, A. V.; Fisher, J. T.; Geysen, H. M.; Rose, D. J. Anal. Chem. 1997, 69, 543-551. 10.1021/ac035135c CCC: $27.50 Published on Web 05/01/2004

© 2004 American Chemical Society

droplet volume is basically limited by the size of the orifice. This technique is widely known as a key technology in ink jet printing and has recently been applied in combinatorial chemistry for drug screening,1 micro total analysis systems,2 preparation of DNA chips,3 and sample preparation in matrix-assisted laser description/ ionization time-of-flight mass spectrometry.4 In the electrospray type, a dc voltage is applied to the tip of a capillary supplied with a liquid stream, causing the stream to split into droplets.5 This technique generates one-by-one and on-demand droplets smaller than 10 pL, and typically 30 fL-2 pL, at a moderately fast operating rate of 50 Hz.6 This technique is also used for preparing gas-phase ions in ionization mass spectrometry7 and for laser plasma X-ray generation.8 When fabricating ultrasmall vessels in the future, injecting other solutions into the vessels will prove essential. In one study, droplets of reagent with a volume of ∼0.5 pL were injected into another identical droplet on a substrate having a volume of ∼9 nL for titration using the ink jet technique.9 The mixing operation, however, had some problems in terms of control of the injecting droplet, speed of evaporation, and homogeneity of the mixed droplets. In the other studies using the ink jet technique, droplets of reagent with a volume of 60-180 pL were injected into a nanoto microliter droplet levitated in an acoustic field for sample enrichment10 and chemical reactions.11,12 The method is not preferable to high-throughput analysis, because not more than two levitated droplets could be used at a time. Using an ultrasonic atomizer, a fine mist of reagent was deposited on an array of glycerin droplets with a volume of 1.6 nL and mixed with the (2) Hosokawa, K.; Fujii, T.; Endo, I. Anal. Chem. 1999, 71, 4781-4785. (3) Graves, D. J.; Su, H.-J.; McKenzie, S. E.; Surrey, S.; Fortina, P. Anal. Chem. 1998, 70, 5085-5092. (4) O ¨ nnerfjord, P.; Nilsson, J.; Wallman, L.; Laurell, T.; Marko-Varga, G. Anal. Chem. 1998, 70, 4755-4760. (5) Cloupeau, M.; Prunet-Foch, B. J. Aerosol Sci. 1994, 25, 1021-1036. (6) Yogi, O.; Kawakami, T.; Yamauchi, M.; Ye, J. Y.; Ishikawa, M. Anal. Chem. 2001, 73, 1896-1902. (7) Kebarle, P.; Ho, Y. In Electrospray Ionization Mass Spectrometry; Cole, R. B., Ed.; John Wiley & Sons: New York, 1997; 1-64. (8) Mountford, L. C.; Smith, R. A.; Hutchinson, M. H. R. Rev. Sci. Instrum. 1998, 69, 3780-3788. (9) Litborn, E.; Stjernstro¨m, M.; Roeraade, J. Anal. Chem. 1998, 70, 48474852. (10) Petersson, M.; Nilsson, J.; Wallman, L.; Laurell, T.; Johansson, J.; Nilsson, S. J. Chromatogr., B 1998, 714, 39-46. (11) Eberhardt, R.; Neidhart, B. Fresenius J. Anal. Chem. 1999, 365, 475-479. (12) Santesson, S.; Barinaga-Rementeria Ramı´rez, I.; Viberg, P.; Jergil, B.; Nilsson, S. Anal. Chem. 2004, 76, 303-308.

Analytical Chemistry, Vol. 76, No. 11, June 1, 2004 2991

Figure 1. Block diagram of the on-demand mixing droplet spotter. Key components of the droplet spotter are magnified in the inset.

droplets.13 The operation, however, had some problems in terms of control of mixing ratio in mixed droplets one by one and a low mixing speed due to high viscosity of the glycerin. Inserting a fine glass capillary into a droplet having an oil/water interface, a method that injects a minute sample solution into the droplet, was studied for electrochemical measurements.14,15 In this method, however, it seems to difficult to quickly make mixed droplets and to avoid contamination due to direct contact. The authors of this paper therefore developed a mixing technique that uses electrostatic force. Our system can complete the mixing process within 10 ms before evaporation takes place and can be used with high-viscosity solutions. This paper describes the construction of an on-demand mixing droplet spotter (OMDS) together with characteristics such as the time course of the mixing process, the spatial relationship between the capillaries and the substrate for favorable operation, stability of the mixing volume, and homogeneity of the mixed droplets. Using the OMDS, a microarray having spots with variable dye concentration was prepared to demonstrate the mixing application. EXPERIMENTAL SECTION Overview of the OMDS. Figure 1 shows a block diagram of essential components of the OMDS. These consist of two glass capillaries and mechanical stages for moving the capillaries and substrate. The high-speed video microscope was used for sideview observation of the droplet formation in real time. The power supply for generating pulsed voltages V1 and V2 consists of dc power supply and a field effect transistor (FET) switching device. Height and width of the pulsed voltage were controllable by a computer as described below. The pulsed voltages were connected to the capillaries with a tungsten wire as the anode and the substrate holder as the cathode. Selecting the polarity in this way is effective for avoiding disturbance of droplet formation by corona discharge.5 As the safety precaution for use of high dc voltage supplies, and especially for an unexpected overcurrent to the (13) Gosalia, D. N.; Diamond, S. L. Proc. Natl. Acad. Sci. U.S.A. 2003, 100, 8721-8726. (14) Gratzl, M.; Lu, H.; Matsumoto, T.; Yi, C.; Bright, G. R. Anal. Chem. 1999, 71, 2751-2756. (15) Kashyap, R.; Gratzl, M. Anal. Chem. 1998, 70, 1468-1476.

2992

Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

capillaries while handling, a fuse was connected in series between the power supply and the tungsten electrode to limit dc current to the microampere level. Two computers were used in the OMDS. One computer synchronizes the trigger pulse for driving the power supply with the operation of the mechanical stages. The other computer acquires video images through the video microscope and processes the acquired video images. The core section of the OMDS consists of the capillaries, their holders, the mechanical stages, and a white LED for back-illumination of the tips of the capillaries. In our experiment, this core section was set on a vibration-free table. The ambient temperature was 26 °C, and the relative humidity was ∼55%. Positioning the Capillaries and the Substrate. The capillary was fabricated by thermally extending a glass tube (o.d. 1 mm, i.d. 0.6 mm) using a glass tube puller (PP-830, Narishige, Japan). The outer diameter of the capillary tip thus obtained was 5-40 µm and typically ∼20 µm with the inner diameter always ∼60% of the outer diameter. A tungsten electrode was inserted in a capillary and connected to the power supply through a capillary holder (MEW-M10B, Warner Instrument) equipped with an electrical terminal for the patch-clamp technique. After filling sample solutions, the tips of two capillaries were positioned at the focal point of the microscope horizontally mounted on the vibrationfree table using a manually driven X-Y-Z stage. A substrate was then placed under the tip of capillaries using a motor-driven X-Y stage (SGSP20-35, Sigma Koki). The substrate was set on a vacuum chuck as a substrate holder, which was mounted with a ceramic spacer for insulation on the X-Y stage and used as ground for the cathode. The substrates (20 × 20 mm2 in area and 1 mm thick) were polished quartz with an indium-tin oxide (ITO) layer having low resistance (10 Ω per square). The substrate was connected to the substrate holder and used as a ground electrode. The mixed droplets were prepared on the surface of the ITO layer. Video Microscope for Observation of Mixed-Droplet Formation. Side-view images of the droplets were obtained using a video microscope that consisted of an objective lens (M Plan Apo20x, Mitutoyo, Japan) with a long-working distance (20 mm), high-speed digital camera (Fastcam-X 1280PCI, Photron), and a computer containing a video interface. Light for illuminating the

tip of the capillaries was provided by a white LED (NSPW500BS, Nichia, Japan). Overview of Mixing Process on a Surface. Here we outline the OMDS operating principle. The OMDS uses the phenomena caused by electrostatic force, such as deformation of the meniscus induced by a high-voltage pulse (∼500 V and ∼2 ms).5 Our method of the mixing process consists of six steps described using the schematic illustration in the inset of Figure 1. Step 1: A pulsed voltage V1 is applied between the tip of capillary 1 that contains liquid sample 1 and the substrate surface where droplets are prepared. At this stage, capillary 2 is ground voltage (zero). Step 2: Due to the pulsed voltage V1, the meniscus of liquid sample 1 deforms into a cone pointing at the surface, and a jet of liquid is then ejected from the apex of this cone. The jet is deflected midway between capillary 1 and capillary 2 because the tip of capillary 2 is 0 V. An initial droplet is formed by accumulation of the jet onto the substrate. Step 3: The pulsed voltage V1 is turned off to 0 V and the jet ends. The cone returns to the initial meniscus of the solution. Step 4: The pulsed voltage V2 is applied to the tip of capillary 2, which contains liquid sample 2. The voltage of capillary 1 is zero. Step 5: The meniscus of liquid sample 2 deforms into a conical shape, and a jet of the liquid is ejected from the apex of this cone. The jet is deflected toward the initial droplet and collides with it since the electric field from the tip of capillary 2 is converged onto the initial droplet. The mixing droplet is in this way formed on the surface. Step 6: The substrate is moved to the next position where another droplet is prepared. In our method, the volume of the mixed droplets and its ratio are controllable by the width and height of the pulsed voltages V1 and V2. RESULTS AND DISCUSSION Visualization of Time Course of the Mixing Process. Figure 2 shows the time course of the mixing process using water. The relation of the capillary 1, capillary 2, and substrate positions is shown in Figure 2A. The tips of the capillaries were placed above the substrate with a spacing of 28 µm, and each tip was maintained at a distance of 28 µm. It should be noted that the lower half of the picture in Figure 2 is a mirror image due to the reflection of the substrate. Figure 2A shows the initial stage without applying a voltage (step 6). Figure 2B shows an instant at 1 ms after V1 was applied to the capillary 1. The jet appeared from the tip of capillary 1, and an initial droplet was being formed on the substrate at the middle position of the tips (steps 1 and 2). Figure 2C shows the instant just after V1 was turned off, and the initial droplet having a volume of 1.75 pL was formed (step 3). The volume could be estimated from the side-view profile of the droplet.16 Figure 2D shows 1.5 ms after V2 was on. The jet appeared from the tip of capillary 2 and collided with the initial droplet. A mixed droplet could in this way be obtained on the substrate (steps 4 and 5). The injection volume was 1.43 pL in this operation, because the total mixing volume was 3.18 pL. Panels E and F of Figure 2 show evaporation of the droplet. A water droplet of this size evaporates within a short period of time (∼200 ms). The mixing process of ultrasmall vessels such as droplets has to be performed quickly (16) Several fitting functions, including polynomials, was tested and found that without deformation the side-view profile of a droplet on a surface is well approximated by an arc of a circle.

Figure 2. Time course of the mixing process on a substrate: (A) before applying pulse voltage, (B) on applying V1 (t ) 1 ms) to capillary 1, (C) V1 turned off (t ) 3.5 ms), (D) on applying V2 (t ) 5.5 ms) to capillary 2, (E) the droplet on the way to evaporating (t ) 150 ms), and (F) the droplet completely evaporated (t ) 194 ms). (G) Timing chart of V1 and V2. The applying conditions of V1 and V2 were the same at 600 V and 3 ms. The images A-F correspond to letters A-F in the chart (G). The voltages applied are shown above each capillary as letters of V1, V2, and Gr, which means ground voltage.

Figure 3. (A) Schematic illustration of spatial distribution of capillaries, an initial droplet, and a substrate. (B) Property of the position dini (O) and the height hini (b) of an initial droplet on a substrate as a function of the distance between capillaries Dcap when the distance between the substrate and the capillaries Hcap is fixed at 24 µm.

to avoid evaporation. Humidity should be raised to inhibit the evaporation. Characteristics of the Initial Droplet. Figure 3 shows the property of the initial droplet as a function of Dcap, which is the distance between the two capillaries. In this experiment, Hcap, which is the distance between capillaries and substrate, was fixed Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

2993

Figure 4. Property of the injected volume as a function of V2 (the pulse height, Vp2; the width, ∆t2). (A) Relationship between the injected volume and Vp2 when ∆t2 was fixed at 2 ms. (B) Relationship between the injected volume and ∆t2 when Vp2 was fixed at 500 V.

at 24 µm. The conditions of V1 and V2 were both 500 V and 3 ms. In Figure 3B, the left parameter indicated by (O) shows dini, which is the position of the initial droplet on the substrate, and the right parameter (b) shows hini which is the height of the initial droplet. In reference to Figure 3B, the hini curve indicates that hini tends to increase with a shorter Dcap because of an increased electrostatic force between capillaries 1 and 2 since capillary 2 was 0 V. The curve of dini indicates that dini tends to increase with a shorter Dcap because the deflection of the jet by capillary 2 became larger. In this experimental condition, the mixing process could be made only in a narrow Dcap range of ∼27 µm. The result in the fail 1 area (Dcap < 24 µm) indicates the problem that the jet from capillary 1 was sometimes directly injected into the tip of capillary 2, even when an initial droplet was formed. This is because the electric field between capillaries 1 and 2 became excessively strong. The result in fail 2 (Dcap > 29 µm) indicates a different problem. Namely, the jet from capillary 2 could not reach the initial droplet and so two droplets were formed on a substrate. In this case, because of the increased separation, the electric field between the tip of capillary 2 and the initial droplet deteriorated to a level too low to drive the jet to the initial droplet. As described above, the mixing process could be realized in the limited space distribution of Dcap ) (1.0-1.2)Hcap. In this condition, the CV values were CV(dini) ) 7% and CV(hini) ) 6%, or in other words, the position of the mixed droplet spotting has a data spread of 7% and the volume of the mixed droplet has a spread of ∼20% (1.063 - 1). Characteristics of the Injected Jet. Figure 4 shows the stability and adjustability of the injected volume inside an initial droplet. In this experiment, the condition of V2 (the pulse height, Vp2; the width, ∆t2) was varied to examine these characteristics. The state of pulsed voltage V1 was 500 V and 2 ms. Figure 4A shows the property of the injected volume as a function of Vp2 when ∆t2 was fixed at 2 ms. A favorable range for height was Vp2 ) 480-500 V indicated in Figure 4A as “success”. The injected 2994 Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

Figure 5. Fluorescence distributions from the identical droplet spot. The dashed lines represent fluorescein distributions, and the solid lines represent rhodamine B in (B) and (C). (A) Schematic illustration of a fluorescence image of the droplet shown by polar coordinates. The top view of the droplet is represented by the circle. (B) The case evaluated as nonhomogeneous. The 0.2-pL rhodamine B solution was injected into the initial droplet of fluorescein of 13 pL by the conventional method, which is touching the needle to the initial droplet directly. (C) The case evaluated as homogeneous. The rhodamine B solution of 0.9 pL was injected into the initial droplet of 15 pL by our method.

volume was controlled by a voltage with a CV value of ∼20%. In the case of Vp2 ) 510 V, indicated in Figure 4A as fail 3, the injected jet could not collide with the initial droplet due to an excessive deflecting length. Figure 4B shows the property of the injected volume as a function of ∆t2 with the fixed voltage of Vp2 ) 500 V. The favorable range of the pulse width was ∆t2 ) 1-3 ms. Within this range, the injected volume was linearly controllable with a CV value of ∼20%. At a width of more than 3 ms shown as “nondjustable” in Figure 4B, the injected volume became constant. This saturation is probably caused by accumulation of a positive charge on the droplet surface.6 The positive charge, which is transferred with the liquid from the capillary, remains on the surface because the droplet is pure water with high resistance. Evaluation of the Homogeneity of Mixed Droplets. To confirm the homogeneity of the mixed droplet, making mixed droplets was attempted using a high-viscosity solution consisting of glycerin (95%) and water (5%) containing dye molecules. A homogeneous mixing was assumed to be difficult to achieve in this case, because the solution has a viscosity 600-900 times greater than water.17 Figure 5 shows the fluorescence distributions inside the mixed droplet of fluorescein and rhodamine B. We noticed that a small droplet on a substrate retains a hemispherical

Figure 6. Property of the D value calculated by eq 1 as a function of the droplet volume. The schematic illustration of SL and SR used for D value calculation is also shown in the inset. In this figure, the D values for fluorescein (O) droplets and rhodamine B (b) mixed by our method are shown. The D value for rhodamine B mixed by the conventional method using the needle (9) is shown for comparison.

shape due to surface tension, and the shape is highly symmetric. If the homogeneity inside the mixed droplet is high, the distribution of the components in the droplet also has a highly symmetric distribution. In reference to Figure 5A, the center of the spot is assumed to be the polar coordinate origin. Figure 5B is not a favorable case. The rhodamine B distribution was asymmetric although that of the fluorescein was symmetric. In other words, the concentration of rhodamine B inside the mixed droplet was evaluated as nonhomogeneous. Figure 5C is a favorable case. Each fluorescence distribution was measured as symmetric, so that each concentration was evaluated as homogeneous. Figure 6 shows the relationship between droplet volume and D value that is an index representing the symmetric property and is defined by us as follows. Using the fluorescence distributions as shown in Figure 5B and C, the integrated values of SL and SR are obtained as shown in the inset of Figure 6 at each θi ) i π/n (n ) 36) in Figure 5A. The D value is calculated by the following equation:

D)

1

n

∑δ(θ )

n i)1

i

(1)

where δ(θi) ) |SL - SR|/(SL + SR). The curve of open circle (O), which shows the D value using only droplets of the fluorescein solution, was ∼1%. The D value of 1% can therefore be regarded as homogeneous. The curve of solid circle (b), which shows the D value for various volumes of the rhodamine B solution mixed with the initial droplet of the fluorescein solution of ∼14-pL volume by our mixing method, was ∼1-2%. The conventional mixing method, in which a needle with the rhodamine B solution was touched to the initial droplet,14,15 was also tested for comparison. Its D value (9) for rhodamine B (17) The fluorescein solution has the composition 95% glycerin , 2.5% water, and 2.5% ethanol. The rhodamine B solution has the composition 95% glycerin and 5% water. The fluorescein solution has the dye concentration of 1.25 × 10-4 M and the viscosity of 906 × 10-3 Pa‚s. The rhodamine B solution has the dye concentration of 1.25 × 10-4 M and the viscosity of 616 × 10-3 Pa‚s. To make conductivity of solutions higher, sodium chloride was added at a concentration of 1.1 × 10-2 M.

Figure 7. Fluorescence image of the microarray having spots with variable dye concentrations. (A) Fluorescein image (EX, 450-490 nm; BA, 515-555 nm). (B) Rhodamine B image (EX, 510-560 nm; BA, 590 nm). The index numbers to specify the positions of the spots are indicated at the top and the left of each figure. Each spot was prepared by mixing high-viscosity solutions of fluorescein (1.25 × 10-4 M) and rhodamine B (1.25 × 10-4 M) in various volume-to-volume ratios as follows: (volume of fluorescein solution/volume of rhodamine B solution) ) (∼6 pL/∼0.3 pL) at the position of (1, 1), (2, 1), (3, 2), and (3, 3); (∼3 pL/∼3 pL) at the position of (1, 2), (2, 2) and (2, 3); (0 pL/∼6 pL) at the position of (1, 3); (∼6 pL/0 pL) at the position of (3, 1).

volume was ∼2-6% and was larger than that obtained by our method (1-2%). This comparison indicates that the mixing method using electrostatic force is superior to the needle-touch method in terms of homogeneity inside the mixed droplet. Fabrication of Microarray Having Spots with Variable Dye Concentrations. To demonstrate the capability of the OMDS, a Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

2995

microarray having spots with variable dye concentrations as shown in Figure 7 was fabricated. Each spot was prepared by mixing the fluorescein solution and the rhodamine B solution as described above at various ratios by controlling the pulse width of V1 and V2. Mixed droplets having a volume of ∼6 pL were placed with a density of 1 spot/75 × 075 µm2. Figure 7A shows a fluorescein image of the microarray having spots with various dye concentrations. The spot at lower left is the brightest, and the spot at upper right is the darkest. On the other hand, Figure 7B shows a rhodamine B image of the same microarray. The dye concentration was adjusted so that the spot at upper right is the brightest and that at lower left is the darkest. Based on this observation, the microarray was confirmed to have two concentration gradients for two different fluorescence dyes among the spots. This fabrication demonstrates possible applications of OMDS such as preparation of combinatorial libraries for screening,18 specific ink jet printing consisting of mixed droplets, and on-demand fabrication of electronic circuits using conductive or dielectric material solutions.

CONCLUSION

(18) Takeuchi, T.; Fukuma, D.; Matsui, J. Anal. Chem. 1999, 71, 285-290.

AC035135C

2996

Analytical Chemistry, Vol. 76, No. 11, June 1, 2004

An on-demand mixing droplet spotter was developed to fabricate a specific microarray on a substrate. Improved mixing performance in a droplet was achieved. Components of each spot could be controlled by adjusting the mixing ratio. This system will prove useful in the fields of combinatorial chemistry and screening analysis. The system is expected to improve the speed, sensitivity, and cost of analysis in the future. ACKNOWLEDGMENT The authors thank Professor Saburo Tanaka and Associate Professor Shinji Katsura of Toyohashi University of Science and Technology, and Dr. Mitsuo Hiramatsu of Hamamatsu Photonics K.K. for their discussion and advice. This research was partially supported by the Ministry of Education, Science, Sports and Culture, Grant-in-Aid for Exploratory Research, 15656097, 2003. Received for review September 27, 2003. Accepted February 26, 2004.