Inkjet-Printed Microfluidic Multianalyte Chemical Sensing Paper

Aug 13, 2008 - The conversion into digital color data was performed on an Apple Macintosh personal computer (Apple Inc., Cupertino, CA) using the Digi...
0 downloads 11 Views 463KB Size
Download PDF
Anal. Chem. 2008, 80, 6928–6934

Inkjet-Printed Microfluidic Multianalyte Chemical Sensing Paper Koji Abe, Koji Suzuki, and Daniel Citterio* Department of Applied Chemistry, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan This paper presents an inkjet printing method for the fabrication of entire microfluidic multianalyte chemical sensing devices made from paper suitable for quantitative analysis, requiring only a single printing apparatus. An inkjet printing device is used for the fabrication of threedimensional hydrophilic microfluidic patterns (550-µmwide flow channels) and sensing areas (1.5 mm × 1.5 mm squares) on filter paper, by inkjet etching, and thereby locally dissolving a hydrophobic poly(styrene) layer obtained by soaking of the filter paper in a 1 wt % solution of poly(styrene) in toluene. In a second step, the same inkjet printing device is used to print “chemical sensing inks”, comprising the necessary reagents for colorimetric analytical assays, into well-defined areas of the patterned microfluidic paper devices. The arrangement of the patterns, printed inks, and sensing areas was optimized to obtain homogeneous color responses. The results are “all-inkjet-printed” chemical sensing devices for the simultaneous determination of pH, total protein, and glucose in clinically relevant concentration ranges for urine analysis (0.46-46 µM for human serum albumin, 2.8-28.0 mM for glucose, and pH 5-9). Quantitative data are obtained by digital color analysis in the L*a*b* color space by means of a color scanner and a simple computer program. Inkjet technology is no longer only an office printing technology, but has gradually become a versatile tool for various industrial fabrication processes for accurately depositing very small quantities (tens of picoliters) of materials at defined spots on the surface of a wide variety of substrates. So far, the technology has been mostly applied in plastic electronics and for the manufacture of polymer light-emitting diodes.1-4 Inkjet printing is also on the way to become a time-saving, low-cost alternative to photolithography, because it allows the direct patterning of a substrate surface,5,6 compared to the multistep process of photolithography. It does not rely on the use of a specific mask, and cleanroom * To whom correspondence should be addressed. E-mail: citterio@ applc.keio.ac.jp. Phone: +81-45-566-1566. Fax: +81-45-566-1566. (1) Sele, C. W.; von Werne, T.; Friend, R. H.; Sirringhaus, H. Adv. Mater. 2005, 17, 997–1001. (2) Sirringhaus, H.; Kawase, T.; Friend, R. H.; Shimoda, T.; Inbasekaran, M.; Wu, W.; Woo, E. P. Science 2000, 290, 2123–2126. (3) de Gans, B. J.; Duineveld, P. C.; Schubert, U. S. Adv. Mater. 2004, 16, 203–213. (4) Yoshioka, Y.; Calvert, P. D.; Jabbour, G. E. Macromol. Rapid Commun. 2005, 26, 238–246. (5) de Gans, B. J.; Hoeppener, S.; Schubert, U. S. Adv. Mater. 2006, 18, 910– 914.

6928

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

facilities are not a prerequisite. Furthermore, inkjet printing has become an alternative arraying technique for biological sample arrays, which, in contrast to contact-based methods such as pin spotting systems, has the advantages of reduced contamination and no risk of substrate damage.7,8 Despite of all these advantages, its flexibility, low cost, and ease of highly parallel mass production,9 the potential of inkjet printing is still not fully used in the field of analytical chemistry and, to our best knowledge, has so far been limited to the fabrication of electrochemical sensing devices.10-13 On the other hand, microfluidic devices incorporating various chemical functions have been demonstrated to be suitable tools for simultaneous multianalyte optical sensing.14,15 In general, microfluidic devices that function without external equipment and reagents are adequately inexpensive to be commonly used, including in developing countries, in emergency first aid situations, and in home health-care settings, in contrast to conventional medical diagnostic technologies that were designed for airconditioned laboratories, highly trained personnel, a constant supply of reagents, and considerable volumes of sample.16-19 Patterned microfluidic devices require only low sample volumes, which is an important criterion when working with sparsely (6) de Gans, B. J.; Hoeppener, S.; Schubert, U. S. J. Mater. Chem. 2007, 17, 3045–3050. (7) Allain, L. R.; Stratis-Cullum, D. N.; Vo-Dinh, T. Anal. Chim. Acta 2004, 518, 77–85. (8) Hasenbank, M. S.; Edwards, T.; Fu, E.; Garzon, R.; Kosar, T. F.; Look, M.; Mashadi-Hossein, A.; Yager, P. Anal. Chim. Acta 2008, 611, 80–88. (9) Schubert, U. S. Macromol. Rapid Commun. 2005, 26, 237. (10) Li, B.; Santhanam, S.; Schultz, L.; Jeffries-El, M.; Iovu, M. C.; Sauve, G.; Cooper, J.; Zhang, R.; Revelli, J. C.; Kusne, A. G.; Snyder, J. L.; Kowalewski, T.; Weiss, L. E.; McCullough, R. D.; Fedder, G. K.; Lambeth, D. N. Sens. Actuators, B 2007, 123, 651–660. (11) Cho, H.; Parameswaran, M.; Yu, H.-Z. Sens. Actuators, B 2007, 123, 749– 756. (12) Setti, L.; Fraleoni-Morgera, A.; Ballarin, B.; Filippini, A.; Frascaro, D.; Piana, C. Biosens. Bioelectron. 2005, 20, 2019–2026. (13) Setti, L.; Fraleoni-Morgera, A.; Mencarelli, I.; Filippini, A.; Ballarin, B.; Di Biase, M. Sens. Actuators, B 2007, 126, 252–257. (14) Henares, T. G.; Takaishi, M.; Yoshida, N.; Terabe, S.; Mizutani, F.; Sekizawa, R.; Hisamoto, H. Anal. Chem. 2007, 79, 908–915. (15) Hisamoto, H.; Nakashima, Y.; Kitamura, C.; Funano, S.i.; Yasuoka, M.; Morishima, K.; Kikutani, Y.; Kitamori, T.; Terabe, S. Anal. Chem. 2004, 76, 3222–3228. (16) Sia, S. K.; Linder, V.; Parviz, B. A.; Siegel, A.; Whitesides, G. M. Angew. Chem., Int. Ed. 2004, 43, 498–502. (17) Rodriguez, W. R.; Christodoulides, N.; Floriano, P. N.; Graham, S.; Mohanty, S.; Dixon, M.; Hsiang, M.; Peter, T.; Zavahir, S.; Thior, I.; Romanovicz, D.; Bernard, B.; Goodey, A. P.; Walker, B. D.; McDevitt, J. T. Plos Med. 2005, 2, 663–672. (18) Whitesides, G. M. Nature 2006, 442, 368–373. (19) Yager, P.; Edwards, T.; Fu, E.; Helton, K.; Nelson, K.; Tam, M. R.; Weigl, B. H. Nature 2006, 442, 412–418. 10.1021/ac800604v CCC: $40.75  2008 American Chemical Society Published on Web 08/13/2008

available clinical samples, or relatively high-cost sensing reagents, or both. Since inkjet printing is suitable for both the creation of patterned substrates and the dispensing of chemical reagents, we regard it as a promising alternative technology for the fabrication of simple patterned microfluidic sensing devices for simultaneous multianalyte detection, in particular for multianalyte optical sensing devices that require the immobilization of several types of functional dye materials on a single patterned substrate. Modern inkjet printing systems can be equipped with several print heads that are available for the parallel dispensing of multiple liquids. Whitesides and co-workers have shown the benefits of using photolithographically patterned paper for performing a simple bioassay.20 Compared to the well-established, paper-based, dipstick assays, patterned paper allows working with much smaller sample volumes and prevents the leaking of reagents into the sample. Very recently, the Whitesides group has also demonstrated a plotting method for the creation of flow channels and sensing areas on paper substrates.21 The presented method is very cost optimized and can be realized at places with almost no infrastructure at all, making it highly suitable for the local on-site fabrication in developing countries, including very remote locations. On the other hand, an alternative philosophy to cost reduction of simple diagnostic devices and to a reasonable cost-performance balance for the emergency first aid and home health-care sector could be the inexpensive mass production based on industry-scale inkjet printing, which would simultaneously allow us to overcome potential limitations of the plotting method. While the plotter allows the fabrication of fluidic patterns, it is not readily adoptable to the dispensing of the reagents required for chemical sensing. In contrast to the “on-demand” inkjet printing technology that allows the highly reproducible control of not only the place but also the time and the amount of liquid deposition, the plotting technique is a continuous line drawing method, with the amount of dispensed liquid being more difficult to control and, therefore, limiting the resolution of plotted structures to ∼1 mm. Finally, felt tip-based x,y-plotters are gradually becoming difficult to obtain on the market, since they are replaced by inkjet printing devices. In this paper, we demonstrate the applicability of modern inkjet printing technology for the fabrication of microfluidic single-use optical sensing devices for quantitative analysis on the example of patterned sensing paper for the simultaneous determination of pH, total protein, and glucose. The two principal fabrication steps, the creation of a microfluidic pattern and the immobilization of the reagents required for chemical sensing by an analyte-specific colorimetric reaction, are realized on the same inkjet printing apparatus, resulting in a quasi “all-inkjet-printed” microfluidic system. Quantitative information is obtained by digital color analysis of the devices after sample application. EXPERIMENTAL SECTION Materials. All chemical reagents for the preparation of aqueous test samples (pH buffers, glucose solutions, human serum albumin (HSA) solutions), poly(styrene) (Mw 280 000; Tg 100 °C), and organic solvents were purchased from commercial suppliers (20) Martinez, A. W.; Phillips, S. T.; Butte, M. J.; Whitesides, G. M. Angew. Chem., Int. Ed. 2007, 46, 1318–1320. (21) Bruzewicz, D. A.; Reches, M.; Whitesides, G. M. Anal. Chem. 2008, 80, 6116-6121.

(Wako Pure Chemical Industries, Sigma-Aldrich, Junsei Chemical) and used without further purification. Filter paper (Advantec No. 2) was obtained from Toyo Roshi Co., Ltd. (Tokyo, Japan). Glucose oxidase (GOx) and horseradish peroxidase (HRP) were purchased from Tokyo Chemical Industry Co., Ltd. (Tokyo, Japan). Instruments. A PicoJet-2000 device from Microjet (Shiojiri, Nagano, Japan) was used as a piezo-driven inkjet printer. The system was equipped with a MD-K-130 dispenser print head from Microdrop Technologies (Norderstedt, Germany) with a nozzle diameter of 50 µm. The driving signal was a rectangular pulse (width 50 µs, frequency 200 Hz). The pulse amplitude was set between 70 and 78 V. The pulse amplitude and width were chosen to stabilize droplet formation, which was monitored by a built-in video camera. To measure the dimensions of the created patterns and to monitor the sample flow in the patterned paper, a Keyence (Osaka, Japan) VHX-600 digital color microscope was used. A color image scanner (model CS 9950FV) from Canon Inc. was used to record the images of the patterned sensing paper before and after analyte application. The conversion into digital color data was performed on an Apple Macintosh personal computer (Apple Inc., Cupertino, CA) using the Digital Color Meter software (Ver. 3.6.1) provided with the Mac OS X (Ver. 10.5) operating system. For mathematical curve fitting of the experimental color data, the Igor Pro 4.01 software package (WaveMetrics, Lake Oswegeo, OR) was used. Preparation of the Patterned Paper. Filter paper (10 cm × 8 cm) was soaked in a 1.0 wt % solution of poly(styrene) in toluene for 2 h and then allowed to dry at room temperature for 15 min. Then, the hydrophobic poly(styrene)-modified paper was placed on the stage of the inkjet printer, and a hydrophilic pattern was etched by the ejection of toluene droplets in a manner similar to that previously reported in the literature for glass substrates.5 Ten piezopulses per spot with a distance of 150 µm between spots were applied onto the poly(styrene)-modified paper to locally dissolve the polymer material and to re-expose the hydrophilic paper. This print cycle was repeated for overall 5, 10, and 30 times, respectively. In order to visualize the dissolution of the hydrophobic polymer by inkjet etching, 0.005 wt % of the hydrophobic dye Fat Brown RR were mixed into the poly(styrene) soaking solution. For the fabrication of the hydrophilic sensor patterns used for the actual chemical sensing, 10 print cycles were chosen. In this case, no dye was added to the poly(styrene) soaking solution. Preparation of the Chemical Sensing Inks. For the pH assay, a pH-responsive ink was prepared by dissolving 0.5 mg of thymol blue (TB), 1.2 mg of methyl red (MR), 6 mg of bromothymol blue (BTB), and 10 mg of phenolphthalein in 10 mL of 95:5 (v/v) ethanol/water. Then, 0.01 mol/L NaOH solution was added dropwise to the mixed indicator solution until the color changed to green. For the protein assay, a protein-sensitive ink was prepared by mixing 3.3 mM tetrabromophenol blue (TBPB) in 95:5 (v/v) ethanol/water with citrate buffer solution (pH 3.0) in a 1:2 volume ratio. For the glucose assay, a glucose-sensitive ink was obtained by first dissolving 340 units of GOx and 136 units of HRP in 10 mL of citrate buffer solution (pH 6.0). Then, the GOx/HRP solution was mixed with 7.5 mg of o-toluidine dissolved in 10 mL of ethanol in a 1:1 volume ratio. Finally, 0.075 wt % sodium L-ascorbate was dissolved in the obtained solution as a stabilizer. Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

6929

Figure 1. Schematic representation of the fabrication process of the inkjet-printed microfluidic multianalyte chemical sensing paper featuring microfluidic channels connecting a central sample inlet area with three different sensing areas and a reference area. Steps 2 (patterning) and 3 (chemical sensing reagent application) are performed on the same inkjet printing apparatus (the pen symbol indicates the use of the inkjet printer).

Printing of the Chemical Sensing Inks. Each of the chemical sensing inks was printed in defined separate reagent areas located either in the hydrophilic channels or in the sensing areas of the patterned paper. Ten piezopulses/spot with a distance of 150 µm between spots were applied to form the chemical reagent areas. In the hydrophilic channels (pH and protein assays), 24 print cycles of 9 spots arranged in a line were applied. In the square sensing areas (glucose assay), 5 print cycles of 9 × 9 spots, printed once from top to bottom and once from bottom to top, were applied (resulting in overall 10 cycles/spot). Sample Measurements. The performance of the patterned sensing papers was verified by applying 4.5 µL of mixed sample solutions of protein (HSA) and glucose buffered at different pH values in the clinically relevant concentration ranges for urine analysis (0.46-46 µM for HSA, 2.8-28.0 mM for glucose, and pH 5-9) to the central sample inlet area of the sensor. The color changes in the chemical sensing areas were monitored by the naked eye and measured digitally with specific software (described above) after recording a color scan of the sensing pattern 10 min after sample application. Small pieces of red and yellow vinyl tape were placed at the edges of the sensing paper to act as a reference of constant color for the scanner autocalibration occurring before each scan. Data Processing. The CIE 1976 L*a*b* color coordinate system was applied to transform the scanned images of the sensing patterns into numerical color values. The software allows the averaging of color values over a freely selectable square area. To create a calibration curve, single color coordinate values (a* for pH, b* for protein, and L* for glucose) were plotted against the logarithm of the corresponding concentration. Curve fitting was performed using a sigmoidal fitting function. RESULTS AND DISCUSSION Patterning of the Sensing Paper. The procedure of the fabrication of the patterned chemical sensing paper using an inkjet printer is outlined in Figure 1. In a preparatory step, a piece of filter paper was soaked in a solution of a highly hydrophobic polymer in an organic solvent. The use of a hydrophobic polymer was required in this approach, since the polymer functions as the barrier to confine the aqueous sample in the final sensing paper. Poly(styrene) dissolved in toluene was selected because of its high hydrophobicity and its suitability for inkjet etching. In order to obtain an absolutely hydrophobic paper in all three dimensions, 6930

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

Figure 2. Patterning of paper by inkjet etching of poly(styrene) with toluene: (a) outline of the printing pattern (interspot distance 150 µm), and (b) inkjet etched pattern colored with a color ink to visualize the structure. Fifteen identical patterns were printed onto a single 10 cm × 8 cm filter paper.

we relied on a dip coating process, rather than on spin coating that would have led to a surface coating without hydrophobically modifying the internal areas of the paper. For this reason, a relatively long soaking time of 2 h was found to be most suitable. The three-dimensional hydrophobicity was essential in order to confine the chemical inks and the aqueous samples in well-defined hydrophilic channels and sensing areas of a fixed volume. A surface-only modification, as would have been obtained with spin coating, would allow an uncontrolled spreading of printed chemical inks and of the sample solutions inside the paper, by passing under the hydrophobic cover layer. In the second step, toluene was used as the “ink” and ejected onto the hydrophobic paper using the inkjet printer in an outline as shown in Figure 2a. The organic solvent locally dissolves the hydrophobic polymer material at well-defined and controlled places. The dissolved excess polymer is accumulated at the outside of the liquid droplets according to the “coffee ring effect” upon toluene evaporation22,23 and taken up by the surrounding paper. The repeatedly applied toluene droplets re-expose the originally hydrophilic filter paper structure, resulting in hydrophilic flow channels and sensing areas, surrounded by the remaining hydrophobic polymer. To visualize the inkjet etching process, the hydrophobic polymer was doped with the hydrophobic dye Fat Brown RR (Figure S-1a, Supporting Information, SI). To investigate the influence of the amount of printed toluene on the inkjet-etched patterns, the print cycle was repeated for overall 5, 10, and 30 times, respectively, and the dimensions of the created patterns were measured. The reproducibility of the printing process as a function of the number of printing cycles was evaluated by comparing the results obtained for four patterns fabricated under identical printing conditions. Table 1 summarizes the widths of the channels, sensing areas, and central sample inlet areas, as well as the corresponding standard deviations. The patterns fabricated in 10 printing cycles showed the best reproducibility with no statistically significant differences between the single patterns, while larger variations were observed for those obtained after 5 or 30 printing cycles. Interestingly, a too large number of printing cycles has a negative effect on the quality of the patterns. It should be noted however, that the observed variations are not only caused by the actual printing process, but are to a certain degree also influenced by the roughness of the filter paper observed in the microscopic images. The functioning of the inkjet(22) Deegan, R. D.; Bakajin, O.; Dupont, T. F.; Huber, G.; Nagel, S. R.; Witten, T. A. Nature 1997, 389, 827–829. (23) Soltman, D.; Subramanian, V. Langmuir 2008, 24, 2224–2231.

Table 1. Average Widths of the Hydrophilic Flow Channels, Square Sensing Areas, and Square Central Sample Inlet Areas Measured for Four Patterns as a Function of the Number of Toluene Printing Cyclesa average width (mm) printing cycles 5 10 30

pattern number sensing areas (n ) 24)b flow channels (n ) 12)d sample inlet area (n ) 6)c sensing areas (n ) 24) flow channels (n ) 12) sample inlet area (n ) 6) sensing areas (n ) 24) flow channels (n ) 12) sample inlet area (n ) 6)

1 1.77 ± 0.08c 0.42 ± 0.04 1.75 ± 0.05 1.82 ± 0.05 0.45 ± 0.05 1.82 ± 0.04 2.08 ± 0.08 0.60 ± 0.05 2.08 ± 0.05

2 1.77 ± 0.06 0.40 ± 0.04 1.79 ± 0.07 1.85 ± 0.04 0.46 ± 0.04 1.82 ± 0.04 1.85 ± 0.05 0.45 ± 0.04 1.60 ± 0.56

3 1.79 ± 0.06 0.41 ± 0.07 1.81 ± 0.03 1.84 ± 0.05 0.47 ± 0.05 1.82 ± 0.06 2.02 ± 0.05 0.61 ± 0.04 1.76 ± 0.58

4 1.79 ± 0.05 0.44 ± 0.04 1.58 ± 0.58 1.83 ± 0.07 0.46 ± 0.05 1.84 ± 0.04 1.91 ± 0.07 0.57 ± 0.05 1.69 ± 0.56

a The corresponding sensing patterns are shown in Figure S-1a, Supporting Information. b The width of each area was measured at three positions in a horizontal and at three positions in a vertical direction (6 measurements/area). c ±1 σ. d The width of each of the four channels connecting the sample inlet with the sensing areas was measured at three positions.

etched patterns as microfluidic systems for liquid transport was evaluated by applying 4 µL of a colored water sample to the central sample inlet area (Figure S-1b, SI). In the case of 5 printing cycles, the liquid flow did not reach the sensing areas, while the flow was not reproducible in the case of 30 printing cycles. The best results were observed in the case of 10 printing cycles, where all tested patterns were entirely and homogeneously wetted. On the basis of all of these results, 10 repeated toluene printing cycles were regarded as the best condition for the patterning of the sensing paper for all further experiments. Additionally, the working of the “coffee ring effect” was confirmed by the accumulation of the dissolved excess polymer at the edges of the printed channels and sensing areas as visualized by the increased concentration of the doping dye displaced together with the polymer (Figure S-2, SI). Figure 2b shows a single microfluidic pattern on paper (fabricated by 10 toluene printing cycles) after absorbing by capillary action 4 µL of water colored with a purple dye. The dimension of the pattern was selected to allow the use of minimal sample volumes, with the sensing areas being sufficiently large for visual inspection. The flow of the hydrophilic sample initially placed into the central area of the patterned paper is strictly confined to the flow channels and the sensing area without any leaking. Furthermore, a liquid volume as small as 4 µL was sufficient to homogeneously wet and color the entire patterned device. The selected number of toluene printing cycles prevented the complete dissolution of the polymer throughout the thickness of the paper layer, as concluded from the observation that no colored sample was leaking through the bottom of the paper. Chemical Sensing Ink Printing. In the third step of the overall fabrication process, the three types of chemical sensing inks were printed into the previously etched hydrophilic channels or sensing areas (Figure 1). For the pH-sensitive ink, a mixture of four differently colored pH-indicators (TB, MR, BTB, and phenolphthalein) with different color transition points were selected, in order to allow pH sensing in the physiologically relevant pH range of urine between pH 5 and 9. For the sensing of total protein concentration in urine, an ink comprising the pHindicator TBPB was selected. This dye undergoes a color change from yellow to blue upon complexation with proteins, based on

the so-called “protein error” of a pH-indicator.24-27 The colorimetric sensing of glucose relies on the oxidation of glucose by GOx, followed by the HRP-catalyzed oxidation of the colorless reagent o-toluidine with the released H2O2 to form a blue chromophore.28,29 Small amounts of ascorbic acid were added as stabilizer to prevent the development of color in the absence of glucose. Prior to printing, the ink compositions were optimized in terms of color changes observable by the naked human eye. For this purpose, the performance of the chemical inks was evaluated by manual application to untreated filter paper, followed by the addition of an aqueous sample solution. In the case of the glucose sensing ink, the amount of added ascorbic acid was stepwise increased until no color changes were observed upon the addition of glucose samples below the clinically relevant urine glucose concentration. Ethanol was added as cosolvent to water for all chemical sensing inks in order to overcome solubility problems. Furthermore, the presence of ethanol positively influenced the printability by lowering the surface tension of the ink and prevented the spreading of the printed ink in the hydrophilic patterns after printing by enhancing the evaporation rate. The inkjet printing of the chemical sensing inks onto the microfluidic patterned paper sensors is a very reproducible process, resulting in sensing areas with uniform amounts of sensing reagents, as demonstrated on the example of the pH- and protein-sensitive inks (Figure S-3, SI). To fabricate sensors for the simultaneous multianalyte sensing of pH, total protein, and glucose, each of the three chemical sensing inks was first printed into a separate square-shaped sensing area located at the end of a microfluidic channel, leaving the fourth sensing area free for reference purposes (Figure S-4, SI). Five microliters of mixed samples of HSA and glucose at various pH values in the clinically relevant range for urine analysis were injected into the sensors. In all cases, color changes depending on the concentration of each analyte were observed in the sensing areas (Figure S-5, SI). However, these color changes were lacking uniformity, making it difficult for the human observer to judge the final color and, Suzuki, Y. Anal. Sci. 2005, 21, 83–88. Bracken, J. S.; Klotz, I. M. Am. J. Clin. Pathol. 1953, 23, 1055–1058. Scheurlen, P. G. Clin. Chim. Acta 1959, 4, 760–766. Doumas, B. T.; Watson, W. A.; Biggs, H. G. Clin. Chim. Acta 1971, 31, 87–96. (28) Free, H. M.; Collins, G. F.; Free, A. H. Clin. Chem. 1960, 6, 352–361. (29) Comer, J. P. Anal. Chem. 1956, 28, 1748–1750. (24) (25) (26) (27)

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

6931

Figure 4. Optimized inkjet fabricated microfluidic multianalyte sensor before sample application. The dotted squares indicating the sensing areas have been added for illustrative purposes only and form no part of the sensor pattern.

Figure 3. Different outlines of chemical sensing ink (proteinsensitive) prints and sensing areas: (a) squares printed inside square sensing areas; (b) triangles printed into triangular sensing areas; (c) lines printed inside the microfluidic channel immediately prior to the exit into square sensing areas; and (d) squares printed with triplicate ink amount (compared to a) inside square sensing areas before (1) and after (2) application of 4 µL of 46 µM aqueous HSA solution.

therefore, limiting the usefulness of the sensor for practical applications. To overcome these limitations, several alternative approaches including different geometrical arrangements and increased sensing ink printing cycles were evaluated. The various evaluated setups are shown in Figure 3, using the protein-sensitive chemical sensing ink and 46 µM HSA in water as test sample. Panels a-1 and a-2 in Figure 3 show the original printing outline of the chemical sensing ink in square-shaped sensing areas before and after application of 4 µL of sample. In Panels b-1 and b-2, triangular printings on triangular sensing areas are shown in order to investigate the influence of sensing area geometry on the spreading of ink and sample. In Panels d-1 and d-2, square printings with triplicate amounts of chemical ink compared to Panels a-1 and a-2 are shown. Also, with triangular shapes or increased ink amounts, the lack of uniformity of the color change still exists, although some improvements over the original outline (Panels a-1 and b-1) can be observed. The best results were achieved with the configuration shown in Panels c-1 and c-2, where the chemical sensing inks were printed into the patterned hydrophilic channel just in front of the sensing area. As a result, the reagents were continuously dissolved, transported, and spread into the initially empty square sensing area with the capillary force driven flow of the sample. With this approach, the most uniform color changes were observed in the case of pH and protein sensing areas. On the other hand, this outline was not suitable for glucose sensing. For glucose samples, no color 6932

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

changes at all were observed in the sensing areas. This was attributed to the fact that the aqueous samples, passing only once over the reagent area, are not able to sufficiently transport all the reagents (GOx, HRP, o-toluidine) in the required ratios for the glucose assay into the sensing areas. Therefore, the flow channel printing method was adopted for the pH and protein sensing inks, while the glucose assay ink was directly printed into the sensing areas (Figure 4). Sample Measurements. The performance of the patterned sensing papers was verified by applying 4.5 µL of mixed samples of HSA and glucose at various pH values in the clinically relevant range for urine analysis. The results are shown in Figure 5, which color images were obtained by scanning 10 min after sample application. While the color changes in the pH and total protein sensing areas were complete ∼2.5 min after inlet of the sample, the bienzyme-coupled chromogenic reaction in the glucose sensing areas required ∼10 min for the complete color development. The reproducibility of the color changes was evaluated by triplicate measurements of selected samples (Figure S-6, SI). Our research group has earlier demonstrated that digital color analysis is a useful tool to obtain more quantitative results compared to the simple visual inspection of color changes.30,31 In the present approach, a very simple form of digital color analysis was performed, requiring only a standard office type color scanner and color analysis software. For the color transitions observed in the three different sensing areas, the uniform L*a*b* color space proposed by the Commission Internationale de I’Eclairage (CIE) in 1976 was regarded as most suitable. The color changes observed for the pH sensing corresponded to significant changes along the a*-axis (Figure S-7a, SI), while the total protein sensing areas underwent color transitions from yellow to blue, corresponding to the direction of the b*-axis (Figure S-7b, SI). In the case of glucose sensing, where an intense color only develops after contact with glucose samples, focus was set on the L*-axis, representing the lightness of the color (white, L* ) 0; black, L* ) 100). Based on these observations, calibration curves for the three different analytes were obtained (Figure 6), demonstrating the potential for basic quantitative analysis by very simple means. In the case of the pH and the total protein assays, which show (30) Suzuki, K.; Hirayama, E.; Sugiyama, T.; Yasuda, K.; Okabe, H.; Citterio, D. Anal. Chem. 2002, 74, 5766–5773. (31) Hirayama, E.; Sugiyama, T.; Hisamoto, H.; Suzuki, K. Anal. Chem. 2000, 72, 465–474.

Figure 5. Multianalyte sensing of pH (top sensing area), HSA (left sensing area), and glucose (right sensing area) with the inkjetfabricated microfluidic multianalyte sensors (pH ) 6.0 (a), 6.5 (b), 7.0 (c), 7.5 (d), 8.0 (e), and 9.0 (f); HSA [µM] ) 0.8 (a), 1.1 (b), 2.8 (c), 4.6 (d), 15.0 (e), and 45.0 (f); glucose [mM] ) 2.9 (a), 4.1 (b), 5.5 (c), 9.1 (d), 14.0 (e), and 28.0 (f)). Color scans were recorded 10 min after application of mixed HSA and glucose samples, buffered at various pH values.

homogeneous distribution of the color over the entire square sensing area, the mean digital color values over the entire sensing area were used for the calibration curves. For the glucose assay, which resulted in less homogeneous color distribution in the sensing areas, the digital color values were averaged over a smaller area of most intense color at the far edges of the sensing areas. A sigmoidal curve fit to the experimentally obtained data was performed to obtain the calibration curves. Simple mathematical functions with no derivation from chemical equilibria data were applied. A very good curve fit with small errors was observed for the pH sensor (Figure 6a). In the case of the total protein assay, satisfying results were obtained as well, although some deviation from the fitting curve was observed at lower protein concentrations, and the standard deviations for triplicate measurements were larger than for the pH sensor (Figure 6b). The least satisfying results were achieved with the glucose sensor (Figure 6c). The latter is partially attributed to the inhomogeneous color distribution inside the sensing area, a problem that was overcome for the pH and the total protein sensors by the optimized outline of chemical sensing ink printing and sensing areas, as described above. In this work, we have been able to demonstrate that a double enzyme colorimetric assay can be implemented into an inkjet-printed sensing system. However, for future applications, significant improvements are still required. Approaches could include different geometrical arrangements for the glucose sensing area, the use of different colorimetric indicator systems, or both.

Figure 6. Calibration curves obtained for (a) pH, (b) HSA, and (c) glucose by L*a*b* digital color analysis of color scanned images of the patterned sensing papers recorded 10 min after sample application. Experimentally determined values (b) and mathematically fitted curves (solid line; sigmoidal fitting function); the error bars indicate the standard deviations for samples measured in triplicate.

CONCLUSION We have demonstrated the applicability of the inkjet printing technology for micropatterning of well-defined channels and sensing areas on paper, as well as for immobilizing chemical and biochemical reagents that are vital components of multiplexed microassays in the biomedical or environmental analytical fields. The fabricated hydrophobic/hydrophilic structure of the patterned paper prevents the uncontrolled spreading and dilution of the samples as well as of the chemical sensing inks. The adsorbed liquid volume is predetermined and remains constant independently of the applied sample volume, because of the fixed volume of the three-dimensionally etched channel system. This guarantees a uniform concentration of the sensing reagents after dissolution in the sample liquid. At the same time, the samples are transported on the paper substrate by capillary action, making the use of an external pumping system unnecessary. The system, although Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

6933

being very simple in terms of applied materials, allows reproducible quantitative analysis also with very small sample volumes. The inkjet etching technology, applied to re-expose hydrophilic patterns on fully hydrophobic polymer soaked paper substrates, allows fabricating flow patterns with resolutions in the several hundred micrometer range. The most advantageous feature of the presented inkjet printing method is that it allows the fabrication of entire microfluidic multianalyte chemical sensing devices requiring only a single printing apparatus relying solely on the inkjet printing technology. There is no requirement for expensive additional equipment, no transfer of the substrate in between different process steps, and no need for curing of polymers or development of photoresists. This newly developed concept will enable the industry-scale mass production of simple chemical sensing devices with well-balanced cost and performance.

6934

Analytical Chemistry, Vol. 80, No. 18, September 15, 2008

ACKNOWLEDGMENT We thank Yushi Heta of Keio University for his advice concerning the geometrical optimization of the microfluidic patterns. SUPPORTING INFORMATION AVAILABLE Additional figures as mentioned in the text (influence of toluene printing cycles on inkjet etching, demonstration of the coffee ring effect, reproducibility of the printing procedure, sensing paper with square-shape printed chemical sensing inks, color changes after sample application, triplicate sample analysis, and a*b* plots). This material is available free of charge via the Internet at http://pubs.acs.org. Received for review March 25, 2008. Accepted July 18, 2008. AC800604V