Anal. Chem. 2003, 75, 4132-4138
A Coverslip Method for Controlled Parallel Sample Introduction into Arrays of (Sub)nanoliter Wells for Quantitative Analysis Robert Moermana,* and Gijs W. K. van Dedema
Kluyver Laboratory of Biotechnology, Delft University of Technology, Delft, The Netherlands
We have developed a straightforward coverslip method that combines rapid sample introduction into arrays of 19.8-55.0-µm-deep microwells (prefilled with reactants) on a chip with complete sealing of these wells to avoid evaporation during optical detection. We used coverslips of 300-1500-µm-thick poly(methyl methacrylate) (PMMA) containing arrays of holes with diameters of 200 and 300 µm, and 800-µm-thick quartz slides containing arrays of holes with diameters of 100 and 200 µm. A coverslip was placed on top of a chip, thereby positioning the holes between the wells. A defined amount of aqueous sample solution was pulled between the cover and silicon nitride coated chip by means of capillary forces, resulting in entrapment of air in all wells. By partly or completely positioning the holes over the wells, air could escape through the holes, resulting in complete filling of the wells with the surrounding liquid within 0.1-0.2 s. Immediately after filling, the slide was rapidly shifted back to its original position and was pressed onto the chip for 5-30 s with a force of 1-2 kg cm-2, resulting in complete sealing of the wells for minutes up to one-half hour. Because the well volumes of an array are identical and the wells are completely filled, reproducible assay volumes and flat menisci are obtained, making this method very suitable for quantitative analysis. Analysis on the nano- and picoliter scale will be commercially feasible and attractive if handling procedures such as dispensing of reagent and sample solutions become very efficient and if compatibility is achieved with available or newly designed detection equipment. Especially in the fields of proteomics, medical diagnostics and high-throughput screening many reagents and sample components are unstable under ambient temperature and humidity conditions, meaning that reagent and sample solutions have to be transferred and delivered to specific areas, such as microwells, as fast as possible. Furthermore, reagent and sample solutions should not be wasted because of liquid handling, especially when reagents are very expensive and when sample volumes are very limited. Moreover, to measure kinetic and endpoint levels quantitatively in defined volumes, liquid handling should be accurate and reproducible. One approach is to integrate * Correspondending author address: Department of Biotechnology, Delft University of Technology, Julianalaan 67, NL-2628 BC Delft, The Netherlands. Phone: + 31 15 278 63 56. Fax: + 31 15 278 23 55. E-mail: R.Moerman@ tnw.tudelft.nl.
4132 Analytical Chemistry, Vol. 75, No. 16, August 15, 2003
liquid handling procedures on-chip1-3 in a closed system to eliminate the strenuous problem of evaporation. However, problems may occur that are associated with the presence of air bubbles, clogging of channels, biocompatibility of materials, and integration of circuit components on-chip. A more flexible and straightforward approach is based on performing analyses in easily accessible open microwells. High-density arrays of uniform microwells4,5 can be fabricated in biocompatible polymer6 or in silicon material using etching technology. First, these microwells have to be prefilled with ultrasmall and defined amounts of different reagent solutions, preferably by using noncontact dispensing technology, such as inkjet printing7,8 or our previously reported electrospray dispensing technique.9-11 Miniaturized electrospray dispensing allows for controlled deposition of reagent solutions in 200-400-µm-wide and 19.8-55.0-µm-deep etched wells in the form of dry layers with diameters ranging between 130 and 350 µm. Using minimum reagent and sample volumes of 100 and 500 pL/well, total reagent and sample volumes of 10 and 50 nL, respectively, are required to perform 100 (bio)chemical assays on 16 mm2. During and after prefilling, the activities of reagents are preserved fully and reproducibly. The traditional immobilization of enzymes and antibodies12-14 offers some interesting advantages, such as regeneration of the immobilized compound and containment of the fluorescent product. However, immobilization requires pretreatment of the chip surface involving washing steps, is not applicable to all chemical compounds, and frequently results in loss of activities (1) Whitesides, G. M.; Stroock, A. D. Phys. Today 2001, June, 42-48. (2) Prins, M. W. J.; Welters, W. J. J.; Weekamp, J. W. Science 2001, 291, 277279. (3) Zhao, B.; Moore, J. S.; Beebe, D. J. Science 2001, 291, 1023-1026. (4) Beyer Hietpas, P.; Ewing, A. G. J. Liq. Chromatogr. 1995, 18, 3557-3576. (5) Bratten, C. D. T.; Cobbold, P. H.; Cooper, J. M. Anal. Chem. 1997, 69, 253-258. (6) Jackman, R. J.; Duffy, D. C.; Ostuni, E.; Willmore, N. D.; Whitesides, G. M. Anal. Chem. 1998, 70, 2280-2287. (7) Lemmo, A. V.; Fisher, J. T.; Geysen, H. M.; Rose, D. J. Anal. Chem. 1997, 69, 543-551. (8) Blanchard, A. P.; Kaiser, R. J.; Hood, L. E. Biosens. Bioelectron. 1996, 11, 687-690. (9) Moerman, R.; Frank, J.; Marijnissen, J. C. M.; Schalkhammer, T. G. M.; van Dedem, G. W. K. Anal. Chem. 2001, 73, 2183-2189. (10) Moerman, R.; van den Doel, L. R.; Picioreanu, S.; Frank, J.; Marijnissen, J. C. M.; van Dedem, G. W. K.; Hjelt, K. T.; Vellekoop, M. J.; Sarro, P. M.; Young, I. T. In Micro- and Nanofabricated Structures and Devices for Biomedical Environmental Applications II; Ferrari, M., Ed.; Proc. SPIE, Prog. Biomed. Opt. 1999, 3606, 119-128. (11) Moerman, R.; Marijnissen, J. C. M.; Frank, J.; Method of the dosed application of a liquid onto a surface; WO00355go; Delft University of Technology, 1999. 10.1021/ac020432n CCC: $25.00
© 2003 American Chemical Society Published on Web 07/11/2003
of enzymes and antibodies. Furthermore, the reaction takes place only at a very thin boundary layer, resulting in a limited amount of fluorescent product. An alternative approach to immobilization would be to prefill the wells with reagents and preservatives in a single step followed by storage of the dry spots at low to very low temperatures to preserve reagent stability. After defrosting and addition of sample solution, the reagents dissolve and mix rapidly in the sample solution by means of convection and diffusion, resulting in controlled kinetics. The crucial missing step in this concept concerns the introduction of a small amount of sample solution into the wells, preferably in a simple and parallel manner, while avoiding carryover and evaporation during sampling and detection. In most cases, the sample consists of a diluted aqueous solution or blood plasma solution, of which 1 nL evaporates within 1 s under normal conditions. Until now, this difficult problem of rapid evaporation of (sub)nanoliter reaction volumes has been tackled by using gel pads,15 adding glycerol to the reaction media,4,16 applying high humidity,14,17 and covering the wells with a mineral oil5 or heptane layer16 or with a microscope slide.6,18,19 However, gel-pad technology is elaborate and still requires a coverslip to prevent evaporation. Although glycerol may slow evaporation, we and others14 noticed that water evaporates from a 70/30% (w/w) water/glycerol mixture; thus, humidity is still required to avoid evaporation during pipetting. Furthermore, addition of glycerol results in a very high background signal when measuring NADH (λexc ) 340 nm, λem ) 465 nm). Covering the wells and, thus, covering the predispensed reagent spots with a hydrophobic heptane layer may result in loss of enzymatic activity. For example, we observed a complete loss of luciferase activity upon adding 20 vol % of heptane. In this paper, we show a new method that is based on a simple multipurpose coverslip that allows rapid and controlled sample introduction into 19.8-55.0-µm-deep wells, followed by sealing the wells for 0.5 h using the same coverslip. This time interval is more than sufficient for performing enzyme rate assays. Optimal filling conditions were investigated as a function of the well depth, well width, hole diameter, and cover thickness. In addition, carryover was studied with Rhodamine B. EXPERIMENTAL SECTION Chemicals. Ethylene glycol (99+%) was from Aldrich (Milwaukee, WI). Brij 35 and D(+) trehalose were from Sigma (St. Louis, MS). Fabrication of Covers and Chips. Coverslips. Poly(methyl methacrylate) (PMMA) covers with thicknesses of 300, 400, 600, 800, 1000, and 1500 µm were cut to a size of 0.8 cm × 1.8 cm. In (12) Grogan, C.; Raiteri, R.; O’Conner, G. M.; Glynn, T. J.; Cunningham, V.; Kane, M.; Charlton, M.; Leech, D. Biosens. Bioelectron. 2002, 17, 201-207. (13) Korri-Youssoufi, H.; Richard, C.; Yassar, A. Mat. Sci. Eng., C 2001, 15, 307-310. (14) Pope, M. R.; Armes, S. P.; Tarcha, P. J. Bioconjugate Chem. 1996, 7 (4), 436-444. (15) Guschin, D., et al. Anal. Biochem. 1997, 250, 203-211. (16) Clark, R. A.; Beyer Hietpas, P.; Ewing, A. G. Anal. Chem. 1997, 69, 259263. (17) Bowyer, W. J.; Clark, M. E.; Ingram, J. L. Anal. Chem. 1992, 64, 459. (18) Gratzl, M.; Yi, C. Anal. Chem. 1993, 65, 2085-2088. (19) van den Doel, L.; Moerman, R.; van Dedem, G. W. K.; Young, I. T.; van Vliet, L. J. In Novel Micro- and Nanotechnologies for Bioengineering Applications; Proc. SPIE, Prog. Biomed. 2002, 4626D, In press.
Figure 1. Image of a chip and PMMA and quartz covers. (A) Chip coated with silicon nitride containing a 10 × 10 array of wells (400 × 400 µm wide, 40 µm deep and 800 µm apart (center to center)). (B) PMMA cover 400 µm thick containing an array of 10 × 10 holes that are 300 µm in diameter and 800 µm apart (center to center). (C) Quartz cover containing an array of 25 holes that are 100 µm in diameter.
the covers, arrays of 25 and 100 holes of 200 and 300 µm in diameter were drilled mechanically, followed by polishing of the contact side (Figure 1B). Arrays of 25 holes of 100 and 200 µm in diameter were “drilled” in quartz plates (Figure 1C) of 1 × 2 cm and 800-µm thick by means of laser ablation (Laserage Technology Corporation, Waukegan, IL). One side (contact side) was polished. For control experiments, Pyrex plate (1 mm thick) was cut into slides of 2 cm × 1 cm: these plates did not contain holes. Substrates. At the Delft Institute of Microelectronics and Submicron technology (DIMES), arrays of 25 round and square wells and arrays of 100 square wells were made in silicon substrates of 2 × 1 cm (500-µm thick) by means of plasma etching. The wells were 200, 300, and 400 µm wide, 200-400 µm apart (edge to edge), and 19.8-55.0µm deep. The chip surface and wells were coated with a 50-nm-thick silicon nitride layer by means of chemical vapor deposition. Coverslip Method. Pipetting and Wetting. As is shown in Figure 2A, the chip was fixed in a lab-made aluminum holder, and the cover was aligned on top of the chip using the tip of the index finger. We used aqueous solutions containing 0, 0.01, and 0.1 wt % of the nonionic surfactant Brij 35. We used cell extract from yeast to observe the transport dynamics of the liquid during wetting, filling, and sealing. In addition, 70 v/v % ethylene glycol and pure ethanol were tested. A defined amount of liquid was pipetted using a single-channel Finnpipet from Labsystems with a pipetting volume ranging from 0.5 to 10 µL. Flexible pipet tips from Bio-Rad with a tip end o.d. of 0.68 mm (Catalogue no. 2239915) allowed us to pipet the liquid against the contact side of cover and chip. This resulted in wetting of the cover and chip surfaces by means of capillary forces. Optimal wetting conditions were determined for quartz and PMMA covers. Well-Filling Characteristics. Degassing and filling of the wells was achieved by manually shifting the cover rapidly to overlap part of each well with part of each hole (“chimney”). Optimal filling of the wells and the speed of filling were determined as a function of well depth, well width, hole diameter, and cover thickness. Analytical Chemistry, Vol. 75, No. 16, August 15, 2003
4133
Figure 2. Principle of filling an array of wells using a coverslip. (A) Photograph of the holder wherein the chip and cover are aligned manually. A line of liquid is pipetted gradually against the short contact side of cover and chip. (B) Image of the wetting process by means of capillary forces. (C) Degassing process, 0.04 s after the wells were partially covered by the holes. The well parts that are filled with liquid have a dark color.
Well Sealing. After degassing and filling, optimal sealing of the wells was investigated for PMMA slides by pressing a cover onto a chip for 5 to 30 s with a force of 1-2 kg cm-2. The pressure was applied with the index fingertip (pressure area of finger tip, ∼1 cm2) that was positioned as close as possible to the target wells (0.5-1-mm space left between well and fingertip) allowing us to observe and record the dissolution and outflow of Rhodamine B. During pressing, a mild stream of air was blown around the chip and cover to remove excess liquid, thus enhancing the sealing process. Carryover. In triplicate measurements, one to three square and round wells (φ ) 400 µm) of the center column of a chip were prefilled in an alternating manner with a 10-3 M solution of Rhodamine B containing 0.5 wt % Brij 35, 0.5 wt % trehalose, and 0.1 M of triethanolamine buffer by means of electrospraying,6 and the outflow of Rhodamine B was monitored during degassing and covering. After alignment, electrospraying was started next to a chip to obtain a stable Taylor cone. The target well was moved underneath the capillary tip within 1 s, and the chip was moved 100 µm upward to initiate spraying. After 1 min of spraying, the table was moved 100 µm downward (spraying stops), 1600 µm sideways, and 100 µm upward (spraying resumes) in a total time period of 0.40 s. We sprayed in a single well for 1 min at a flow rate of 160 pL s-1 to obtain a high enough Rhodamine B concentration to visualize carryover adequately. The applied voltage was 1.20 kV at a spray distance of 350 µm. Electrospray dispensing (ESD), cover experiments, and carryover were followed and recorded with a CCD camera (VC 3031) mounted on a microscope (1629 Z-90, objective 2×, C-mount 20×) using fiber-optic light sources, all from Euromex (Arnhem, The Netherlands). A framegrabber (AGP-V2740 driver) was purchased from Asus Computer International (Newark, NJ). RESULTS Coverslip Method. Pipetting. By using the tip end with a small o.d., we obtained a good contact between the pipetted liquid and contact side of the chip and cover. We gradually deposited the 4134
Analytical Chemistry, Vol. 75, No. 16, August 15, 2003
required amount of sample liquid at the contact side of chip and cover to make sure that all of the liquid was drawn between the chip and cover. With a tip end o.d. of 1 mm, some liquid remained at the chip in the form of a small droplet. Applying a vertical pressure onto the cover during pipetting was kept to a minimum to allow rapid, complete, and homogeneous wetting and to prevent liquid from being drawn into the holes by means of capillary forces. Wetting. When the pipetted liquid made contact with the sides of both the cover and the chip, all pipetted liquid (1.0-4.0 µL) was drawn between the cover and chip by means of capillary forces, as is shown in Figure 2B. To achieve complete wetting within 1-6 s, the PMMA covers were made sufficiently hydrophilic prior to pipetting by putting them in a 5 wt % Brij 35 solution followed by rinsing with purified water. Using a hydrophilic quartz or Pyrex cover, complete wetting occurred within 0.1-0.5 s from the moment wetting started. In some cases, wetting of a PMMA cover was not possible, because the liquid was pinned at the edges of a few holes, probably because the surfaces were not hydrophilic enough locally or the edges of the holes were not smooth enough. In some cases, air crossed the well edge, which showed up as a balloon shape, but this did not affect the wetting or degassing process. In most cases, successful wetting was achieved, meaning that the total area between the cover and chip was wetted, except for the wells (Figures 2B and 3A, side view). The wells remained almost completely filled with air (some corners were slightly wetted, Figure 2B) which is, of course, necessary to prevent premature dissolution of predispensed reagent material. To avoid partial wetting of the wells, it is important to gradually pipet the liquid more than 2 mm away from the first row of wells, allowing the liquid to spread out as a thin layer. So the thickness of the liquid film should not exceed a certain maximum, because this may lead to partial filling of the wells at a premature stage. Wetting a PMMA cover and chip with an ethylene glycol water mixture (70 v/v %) proceeded slowly in a time period of 10-20 s, depending on the pipetted amount. Ethanol wetted the chip and
Figure 3. Schematic diagram illustrating the coverslip method from the side. (A) After pipetting, the wells remain filled with air, and the space between the wells is wetted with sample solution. (B) During the degassing procedure, the air escapes through the holes, and sample solution flows into the wells. (C) After degassing, the wells are completely filled with sample solution, and there is practically no liquid present between cover and chip, thus allowing the wells to seal completely by pressing the cover onto the chip.
PMMA contact surfaces within 1-2 s. Complete entrapment of air in the wells was achieved with both solutions. For comparison, a chip and Pyrex slide (no holes) were wetted according to the above-mentioned procedure, and this also resulted in complete entrapment of air in the wells.
Well-Filling Characteristics. After the surfaces of the microarray and coverslip had been wetted completely (Figure 3A, side view), we started the degassing and filling process (Figure 3B, side view). As is shown in Figure 3C, the optimal situation occurs when all liquid surrounding a well is used to fill that well. In that case the remaining liquid layer between cover and array is zero, meaning that the reagent material is trapped in the wells. We empirically determined the minimum amount of liquid required to fill all wells completely as a function of the well shape and depth. We assume that after wetting, the liquid is distributed homogeneously across the contact area of the chip and cover, resulting in a film thickness (Figure 3A) that is equal to the amount of pipetted liquid divided by the total wetted surface area. The liquid volume surrounding a well is then equal to the film thickness × the wetted surface area surrounding a well. In Table 1, the well filling characteristics (duplicate measurements) are listed as a function of the cover thickness, hole width, and well width. As is shown in Table 1, the required minimum thickness of the liquid layer for rapid and complete filling of square wells is almost linearly correlated with the well depth, which is in agreement with our assumptions. Because nearly all the liquid surrounding a well flows into the well, the remaining liquid layer is reduced to 0-1 µm for square wells and 1-5 µm for round wells. For square wells 400 and 300 µm wide and 19.8, 35.0, and 40.0 µm deep (10 × 10 and 5 × 5 arrays), the degassing and filling process was completed nearly instantaneously (0.5 s
55.0 32.6 19.8
300 300 300
4.9 2.9 1.8
200 200 200
19.7 13.7 8.2
0.5 0.5 0.7
filling time >0.5 s
55.0 35.0 19.8 52.0 52.0 32.6 32.6 19.8 19.8
400 400 400 300 300 300 300 300 300
6.9 4.4 2.4 3.9 3.9 2.3 2.3 1.4 1.4
300, 200b 300, 200 300, 200 200 200 200 200 200 200
16.5 12.1 5.5 16.4 19.7 8.7 10.9 6.5 8.7
1.0c 1.0 1.0 2.1 5.0 0.0d 2.0 1.1d 3.3
filling time >1 s
Square Wells
Round Wells
remarks
pumping motion pumping motion pumping motion
a We assume that the thickness of the liquid film is homogeneous. We also assume that the liquid volume surrounding a well is equal to the film thickness times the wetted surface area surrounding a well, which is equal to the dashed square presented in Figure 2B (the well surface area + hole surface area). b Filling was performed with a PMMA and quartz cover. c Air escaped from the wells in the form of a single bubble. d The cover was pulled to the chip after degassing.
Analytical Chemistry, Vol. 75, No. 16, August 15, 2003
4135
quartz coverslip containing holes of 100 µm in diameter. To degas round wells of 300 and 400 µm wide, we had to pipet a relatively larger amount of liquid between the cover and the chip. Especially for a well depth of 55 µm, degassing proceeded more slowly (0.5-2 s). This was probably caused by the fact that the air escaped in the form of a single bubble, which means that the “air cylinder” present in the well (400 µm in diameter, 70 µm high) had to escape through a hole of 300 µm wide. Wells of 52.0 µm deep and 300 µm in diameter required even more liquid, because the ratio of hole to well diameter (0.67 compared to 0.75 for 400-µm-wide wells) is even more unfavorable. To achieve efficient degassing in these cases, we applied a slightly different method. We pipetted less liquid than required for spontaneous degassing, and we positioned the holes above the wells (degassing did not occur). Then we pressed the cover onto the chip, and as a result, the air bubbles were squeezed through the holes. We noticed that when we released the pressure, the air was pulled into the wells again, so the wells could be degassed and filled with air repeatedly by using this “pumping” mechanism. As is shown in Table 1, the thickness of the remaining liquid layer was reduced from 5 to 2 µm for 300-µm-wide and 52-µm-deep wells by using the pumping method. Pipetting less liquid than listed in Table 1 resulted in slower and incomplete filling of some or all wells. Pipetting more liquid results in a liquid film thickness >0.5 µm, hence, the risk of carryover increases. Well Sealing. After all wells were filled, the cover was manually shifted to its original position and pressed onto the chip. In an optimal situation, another 0.1-0.2 s was required to move the cover to its original position and to apply the pressure. However, because this procedure was performed manually, it took >0.3 s in most cases. Applying a pressure of 1-2 kg cm-2 onto the cover for 30 s resulted in adequate sealing of the wells (no bubble formation) for ∼0.5 h. This can be explained by the fact that the wells are sealed during pressing while the remaining liquid is squeezed out and evaporates via the holes, which is enhanced by blowing with the air pistol. After this, the pressure can be released, because the cover and chip surfaces are strongly attracted toward each other and air cannot enter the wells via the holes. Pressing for 5 s was not sufficient and resulted in the formation of small air bubbles in the wells in a time period of 2-5 min, meaning that too much liquid was still present between the chip and cover. The quality of the cover also affects adequate sealing. Some PMMA covers were slightly scratched around the holes, allowing air to enter the adjacent wells very slowly, resulting in bubble formation in these wells. For comparison, we studied the sealing method by using a very flat Pyrex slide (scratchless). Because now air could not penetrate the bonded chip and cover, water was trapped in the wells for hours, and even at a temperature of 90-95 °C, the water was trapped for 0.5 h, thus indicating the potential strength of the method for a wide variety of chemistries. Carryover. Our intention is to prefill the wells of an array with different reagents (metabolites, enzymes) and to store these chips at +4 to -80 °C. Prior to sample introduction and fluorescence measurement, the chips are defrosted for 5-10 min. After degassing and filling, the reagents dissolve and mix with the 4136
Analytical Chemistry, Vol. 75, No. 16, August 15, 2003
sample molecules, and carryover should be avoided completely (Figure 3, side view). To assess carryover, we filled 400-µm-wide wells of 19.8-55.0 µm with Rhodamine B by means of electrospraying, and in triplicate measurements we monitored the outflow of Rhodamine B during the coverslip procedure (wetting and degassing). Factors Resulting in Carryover. The liquid has to be distributed between cover and chip as fast as possible, preferably within a few seconds, which should be followed as quickly as possible by degassing. If the wetting procedure took more than 5 s, we observed that the outside of the spot was colored darker due to an increase in humidity combined with the hygroscopic property of trehalose and buffer. This is supported by our observation in a few cases that during wetting, small droplets (∼20- to 30-µm diameter) were formed at the cover area above a well. During degassing and filling, this partially wetted Rhodamine B dissolved within 0.1-0.2 s and partly flowed out of the well before the wells were sealed properly. On the other hand, degassing the wells while the liquid was still flowing between cover and chip resulted in immediate carryover of more than 50% of the amount of Rhodamine B. Degassing using covers of 0.6-1.5 mm thick containing holes of 200 µm in diameter resulted in filling of the holes by means of capillary forces during the wetting procedure, which resulted in severe convection of the liquid in and outside the wells; hence, 60-90% of the Rhodamine B flowed out of the wells toward the holes and adjacent wells. The diameter of the deposited reagent spots should be significantly smaller than the well diameter to avoid contact between reagent material and the well wall, because during the pipetting procedure, these well walls are slightly wetted, as is shown in Figure 2B, which results in wetting and carryover of reagent material before degassing is initiated. Optimal Prevention of Carryover. We filled square and round wells 400 µm wide and 19.8, 35.0, 40.0 and 55.0 µm deep with Rhodamine B that was sprayed in the form of spots of ∼300 µm wide, as is shown in Figure 4A. In triplicate measurements, we used covers 300, 400, and 600 µm thick containing holes 300 µm wide, and we pipetted the corresponding minimum amounts of water, as listed in Table 1. The wetting procedure, which was minimized to 3-6 s, was immediately followed by rapid degassing (Figure 4B) and sealing, as is shown in Figure 4C and D. The degassing time required to fill the wells of different depths and the resulting carryover of Rhodamine B to adjacent wells are listed in Table 2. The 19.8-µm-deep wells were filled rapidly; however, they are too shallow (aspect ratio of 1/20), resulting in a flow of 20-40% of the amount of Rhodamine B out of the prefilled wells to adjacent wells. Optimal results were obtained by degassing square wells 35.0 and 40.0 µm deep with an outflow of Rhodamine B of
