- Email: [email protected]
“Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device
Accepted Manuscript Title: “Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device Authors: Terence G. Henares, Kentaro Yamada, Shunsuke Takaki, Koji Suzuki, Daniel Citterio PII: DOI: Reference:
S0925-4005(17)30096-5 http://dx.doi.org/doi:10.1016/j.snb.2017.01.088 SNB 21613
To appear in:
Sensors and Actuators B
Received date: Revised date: Accepted date:
24-11-2016 9-1-2017 12-1-2017
Please cite this article as: Terence G.Henares, Kentaro Yamada, Shunsuke Takaki, Koji Suzuki, Daniel Citterio, “Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device, Sensors and Actuators B: Chemical http://dx.doi.org/10.1016/j.snb.2017.01.088 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Highlights
Sample liquid transport by free surface flow on a paper device is demonstrated.
Channel width and surface tension of sample liquid are dominant factors.
Charged polymeric nanoparticles can immobilize water soluble colorimetric dyes.
No inter-assay cross-contamination by reagent diffusion for at least 60 min.
Paper-based analytical devices are entirely fabricated by inkjet printing.
“Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device
Terence G. Henares,‡ Kentaro Yamada,‡ Shunsuke Takaki, Koji Suzuki and Daniel Citterio*
Department of Applied Chemistry, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan. Fax: +81 45 566 1568; Tel: +81 45 566 1568
‡ These authors contributed equally to this work.
* Corresponding author: [email protected].
Abstract With “drop-slip” (DS) bulk liquid flow on a fully inkjet-printed microfluidic paper-based analytical device (PAD), a new method for the rapid transport of sample liquid is presented. The main driving force for DS flow is the slip flow of fluid on wetted porous cellulose acting as a lubricated surface, which is predominantly influenced by the width of the hydrophilic channel and the surface tension of the sample liquid. The application of DS flow is demonstrated by a model colorimetric metal assay (Zn2+, Cu2+, Fe2+) with inkjet-deposited indicators on a PAD with optimized channel dimensions of 2 mm width and 110 m height. The presence of bulk liquid on the entire device does not result in any mixing of assay components across adjacent sensing regions connected by microfluidic channels. DS flow is a useful alternative sample liquid transport method for PADs, allowing to perform multiple fully independent assays with enlarged sample volumes without requiring any significant variation in device design and fabrication.
Keywords: Microfluidic paper-based analytical device, Free surface flow, Inkjet printing
1. Introduction Since the first introduction of microfluidic paper-based analytical devices (PADs) in 2007 by Whitesides and co-workers,[1] they have attracted attention as alternative analytical platforms enabling low-cost, easy-to-use, and multiplexed (bio)chemical assays.[2–6] Paper (most commonly filter paper or chromatography paper), a sustainable and inexpensive material, is the backbone of these analytical devices. The hydrophilicity, porosity and high surface-to-volume ratio of the cellulosic network are the primary properties amenable for microfluidic applications. Fluid transport in paper microchannels is accomplished by capillary wicking through the porous paper network, eliminating the need for tubing and bulky pump systems. However, the PADs reliant on the capillarity-driven lateral flow format suffer from one or several of the following disadvantages: 1) long analysis times derived from relatively slow capillary wicking speeds, 2) evaporation loss of sample liquid transported in open microfluidic paper channels, and 3) heterogeneous color development due to the washing-away effect of colorimetric signalling compounds by the sample liquid. So far, faster sample liquid flow or reduction of sample evaporation on PADs have been primarily achieved either by utilizing accelerated liquid transportation in hollow channel structures[7–12] or by printing a toner layer to cover channels.[13] Although these approaches are generally advantageous in terms of physical strength of the device and prevention of contamination,
increased complexity of the device fabrication process (e.g. stacking and alignment of multiple layers, extra toner printing step) is inevitable. In this paper, we introduce the fast bulk liquid flow on top of hydrophilic channels as an alternative method of liquid transport for PADs. Such faster sample liquid transportation involves a “drop-slip” (DS) flow, where sample liquid is added onto a microfluidic channel in a continuous drop-wise manner. The first introduced sample drop pre-wets the paper microchannel via capillary forces, followed by free surface flow of second (or later) drops of sample on a lubricated cellulosic surface. Sample transportation based on free surface flow leads to improvement of PAD assays in two aspects: 1) much faster sample flow speed thanks to the aid of pressure originating from spreading of the bulging sample fluid; and 2) larger amount of transported sample liquid resulting in higher sensitivity and reduced influence of evaporation. Although bulk liquid flow has already been realized on a TiO2/UV-treated superhydrophilic paper platform,[14] the applicability of DS flow-based liquid transport to untreated cellulosic microfluidic channels has to the best of our knowledge not been demonstrated. Herein, the colorimetric detection of Cu2+, Zn2+, and Fe2+ has been performed as model chemical assays to demonstrate the feasibility of simultaneous independent sensing using a DS flow-based PAD. Zn2+ and Cu2+ detection was carried out with Zincon as the indicator, of which the reactivity toward Zn2+ is tunable by pH. For the Fe2+ assay, Ferene S was employed together with masking
agents to eliminate interferences from Cu2+. These models allow to verify the possibility of on-device sample pre-treatment (pH control and masking of interfering compounds) on PADs operated in DS flow mode. Among a variety of approaches to patterning of paper with flow channels, the wax or inkjet-printing methods have become the most frequently used for their accessibility and rapidity. In particular, inkjet-printing is known to be the only standard printing technology capable of fabricating entire PADs (i.e. applicable to both channel patterning and the deposition of assay reagents).[3, 15– 18] We showcase the strength of the inkjet-printing fabrication approach by depositing all components of the PAD with desktop-type inkjet printers. A “fully inkjet-printed PAD” relying on drop-slip flow liquid transport has been developed by depositing UV-curable ink for patterning, polymeric nanoparticles for immobilization of indicators, and chromogenic ion indicators and different sample pre-treatment ink solutions by means of piezoelectrically and thermally actuated desktop inkjet printers.
2. Experimental 2.1. Chemicals Reagents of the highest grade commercially available were used for this work. 2-Aminoethylmethacrylate hydrochloride (AEMH), cetyltrimethylammonium chloride (CTMA-Cl)
(25% in water), 3-(2-pyridyl)-5,6-di(2-furyl)-1,2,4-triazine-5’,5’’-disulfonic acid disodium salt (Ferene S), and tetramethylammonium hydroxide pentahydrate (TMAOH) were purchased from Sigma-Aldrich
(St.
Louis,
MO).
Styrene,
octadecyl
acrylate
(ODA),
2,2'-azobis(2-methylbutyronitrile) (V-59), copper (II) chloride, zinc acetate, iron (II) chloride, ammonium acetate, L(+)-ascorbic acid and thiourea were obtained from Wako Chemicals (Osaka, Japan). 1-(2-hydroxycarbonyl-phenyl)-5-(2-hydroxy-5-sulfophenyl)-3-phenylformazan sodium salt (Zincon) and N-tris(hydroxymethyl) methyl-3-aminopropanesulfonic acid (TAPS) were acquired from
Dojindo
(Kumamoto,
Japan).
1,10-Bis(acryloyloxy)decane
(DDA)
and
2,2-dimethoxy-2-phenylacetophenone (BDK) were purchased from Tokyo Chemical Industry TCI (Tokyo, Japan). All the reagents were used without further purification except for styrene, which was vacuum distilled (10 mm Hg, 62˚C) to remove the polymerization inhibitor. Also, ODA and DDA were treated with 0.1 wt% NaOH to remove the MEHQ stabilizer. All aqueous solutions were prepared using ultrapure (> 18 M cm) deionized water.
2.2. Materials and instrumentation Whatman 540 filter paper (GE Healthcare, Buckinghamshire, UK) was used as substrate for PAD fabrication. A piezoelectrically actuated EPSON PX-105 inkjet printer (Seiko Epson, Suwa, Japan) with custom refillable ink cartridges (Daiko, Hachioji, Japan) was used to print the
UV-curable and cationic nanoparticle inks. On the other hand, a thermally actuated Canon iP2700 inkjet printer with its original ink cartridges was used to print the chromogenic ion indicators, masking agent and pH buffer inks. UV curing was done under a Hg-Xe lamp (Lightingcure LC-6, Hamamatsu Photonics, Hamamatsu, Japan), and a DVM2500 optical microscope (Leica, Wetzlar, Germany) was used to analyze cross-sections of PADs.
2.3. Fabrication of fully inkjet-printed PADs Fully inkjet-printed PADs (Figs. 1a and 1b) were fabricated by sequentially printing UV-curable ink, cationic nanoparticle, chromogenic ion indicator and pre-treatment buffer inks. The detailed compositions of all printing inks are given in Tables S-1 and S-2 of the supplementary material. The cationic nanoparticles were synthesized as previously reported[19] (details available in the supplementary material). Printing patterns were designed with Microsoft PowerPoint software. The flow channels defined by hydrophobic barriers were obtained by printing UV-curable ink as previously described by our group.[20] In short, black outlines of 0% transparency were printed on the topside of the filter paper. Next, the ink was allowed to penetrate into the paper for 20 s, before the substrate was placed upside down on a cooled plate (10˚C) for 1 min to retard further diffusion of the liquid ink, followed by 15 min of UV irradiation on the printed side. Hydrophobic barriers on the backside of the paper were obtained analogously, with the exception of printing the UV-curable ink
as black areas of 60% transparency. The Hg-Xe lamp used for UV-curing was placed at a distance from the filter paper surface to yield a power of 4 mW cm−2 (measured at 365 nm). Figure 1a depicts the PAD design in which each ion sensing zone undergoes multiple deposition of different chemically functional inks. First, all sensing regions were subjected to 1 printing cycle of cationic nanoparticle ink using the Epson PX-105 printer. Next, aqueous inks of assay reagents were sequentially inkjet-printed by the Canon iP2700 printer (printing graphics and detailed printer settings are available in Fig. S-1 in the supplementary material). Briefly, Zincon and Ferene S indicator inks were simultaneously printed, followed by deposition of TAPS/TMAOH buffer (pH 8.5) on region 1 in 6 printing cycles using the cyan, magenta, and yellow ink reservoirs, respectively. Finally, an ink containing ammonium acetate (pH buffer), thiourea (masking agent) and ascorbic acid (reducing agent) was dispensed on region 3 from the black cartridge. For region 2, only cationic nanoparticle and Zincon inks were deposited, without any pH buffer.
2.4. Colorimetric ion detection A schematic representation of the operational procedure is demonstrated in Figure 2. Sample solution (80 L) containing metal ions was introduced into the inlet port in a drop-by-drop manner. Absorption of the initial liquid droplets immediately saturated the porous filter paper, allowing the slipping of the following drops over the paper surface and a complete liquid coverage of each
sensing zone. After a 5 min incubation time, the excess sample liquid was removed by the aid of a Kimwipe tissue paper. Finally, the PAD was left to dry at room temperature for about 5 min, before colorimetric measurement was performed by scanning with a 9000F MARK II color scanner (Canon, Tokyo, Japan). The Image J software (NIH, Bethesda, MD, USA) was used to acquire numerical color intensity values on the RGB scale.
3. Results and discussion 3.1. “Drop-slip” flow on fully inkjet-printed PAD The “drop-slip” (DS) fluid flow on a wetted porous cellulose surface resulted in the concurrent lateral capillary flow and bulk fluid flow (Fig. 2). The cellulose fiber network inside the patterned channels is first saturated by liquid via capillary action and then serves as a lubricated surface where slipping of the bulk liquid follows. The occurrence of DS flow is determined by the channel width as demonstrated in Figure 3a. A simple PAD design has been used to study the liquid flow behavior on channels of various widths (0.3–3.5 mm). The results show that channels of 1.5 mm or larger widths result in DS flow (15 L aqueous dye solution dropped into a 14 mm long channel), while in more narrow channels (width ≤ 1.0 mm), liquid transport is limited to capillary flow wicking. Although the width of the hydrophilic paper channel plays a dominant role in the liquid flow behavior, other factors have to be considered. As shown in Table S-3, the surface tension and the depth of the
hydrophilic channel influence the threshold width for the occurrence of DS flow, but viscosity was practically irrelevant at least up to a value of 5 cP. The hydrophilic channel depth was modulated by varying the printing density of the UV curable ink on the backside of the filter paper (Fig. S-2). The results indicate that 1) large channel width, 2) low liquid surface tension, and 3) deep hydrophilic channels promote DS flow. Occurrence or prevention of DS flow is determined by competing negative and positive driving forces. Figure 3b schematically illustrates the bulk liquid front on a lubricated channel. At the boundary of the sample liquid and the hydrophobic barrier, interfacial tension is generated in the opposite direction to the bulk liquid spreading (indicated as in the bottom left of Fig. 3b), resulting in a negative effect on DS flow. On the other hand, both negative and positive driving forces for DS flow occur at the center of the lubricated channel (bottom right of Fig. 3b, where L is the surface tension of the sample solution, S is the interfacial tension between the gas phase and the lubricated channel, and LS is the interfacial tension between the sample solution and the lubricated channel). Herein, positive driving forces are dominant over the negative ones, since the liquid spreads on the lubricated paper surface without forming a spherical droplet. Thus, the locally-generated positive driving force is expressed by the following equation: 𝛾𝑠 − 𝛾𝐿 cos 𝜃𝑘 − 𝛾𝐿𝑆 > 0 where k is the position of interest along a coordinate axis perpendicular to the channel direction. As a
consequence, the total driving force for DS flow occurrence within a channel of width w is represented as follows: ∑𝑤 𝑘=0(𝛾𝑆 − 𝛾𝐿 cos 𝜃𝑘 − 𝛾𝐿𝑆 ) − 2𝜎 > 0
(2)
Increased channel width results in a larger contribution of positive driving forces (eq. (1)), leading to DS flow by overcoming the channel width independent interfacial tension at the liquid-hydrophobic barrier boundaries (2). Lower liquid surface tension contributes to the occurrence of the DS flow by decreasing L and values, which work as negative factors. Finally, the hydrophilic channel depth also has some effect on LS, as concluded from the fact that too shallow channels adversely influence DS flow occurrence (Table S-3 in the supplementary material). Although not investigated in this study, DS flow occurrence on substrates other than paper is expected to be dependent on the surface tension of the sample liquid, the surface wettability of the substrate, and the roughness of the substrate. It is postulated that DS flow-based sample transportation is possible even in the absence of capillary forces, provided that the substrate surface is sufficiently wettable by the deposited liquid (i.e. low contact angles). However, it should be noted that the use of non-paper substrates might eliminate several advantageous aspects associated with paper as an analytical platform, including its high surface-to-volume ratio useful for assay reagent immobilization, low-cost, and disposability. It is noteworthy to mention that fluid handling techniques have negligible influence on the
outcome of DS flow behavior. Dye solution (200 L) was introduced dropwise onto a PAD with 50 mm channel length, 110 m channel depth, and multiple varying channel width structures. Variation of the liquid dropping rate from 20 to 60 drops per minute did not alter the minimal channel width (1.5 mm) required for DS flow (refer to Video S-1 for the case of 20 drops per minute rate). Moreover, even at increased volume of single droplets (constant total volume), achieved by cutting off the head of the pipette tip, DS flow occurred only at channel widths of more than 1.5 mm (Video S-2). As shown in Figure S-3, the fluid front progression speed in DS flow is dependent on the condition of liquid introduction, such as the dropping rate or the volume of single drops. However, it is clear that the DS flow system allows much faster sample fluid transportation than traditional capillary force-driven liquid flow (see the supplementary material and Figure S-3 for the detailed experimental setup and results). On a fully inkjet-printed PAD with multiple channels, the liquid transported by DS flow filled the entire device within less than 10 seconds (Figure 4 and Video S-3). Even though the solution filled the center sensing zone first and there was a short delay in liquid arrival in the remaining sensing regions, this gap of a few seconds is assumed to have an insignificant effect on the outcome of ion sensing.
3.2. Immobilization of colorimetric sensing reagents Lateral capillary driven flow is the most common fluid transport operating in PADs. When
water soluble colorimetric sensing reagents adsorbed on the cellulose material are exposed to a lateral fluid flow within the cellulose network, they tend to dissolve and to be carried away along the direction of the flow. As a consequence, increased color reagent density evolves on the periphery of the sensing zone or a gradient in color intensity is often observed. Such a non-uniform color distribution is detrimental to reproducible quantitative colorimetric data processing, giving rise to measurement errors. For this reason, simple means to immobilize water soluble sensing reagents are of high interest. This issue has been previously addressed by depositing various nanomaterials (silica nanoparticles,[21] Fe3O4 magnetic nanoparticles,[22] multiwalled carbon nanotubes,[22] graphene oxide[22]) or polymers with charged residues[23] together with colorimetric assay reagents. In the current work, immobilization of water soluble metal ion indicators plays an even more important role, because of increased DS bulk liquid flow and a resident sample liquid droplet on top of the sensing area during the incubation period. For this purpose, cationically charged polystyrene-based core/shell nanoparticles were selected. In contrast to non-particulate polymers,[23] the nanoparticles used here are water insoluble, which makes them suitable for extended contact with sample liquid. The cellulose network-anchored cationic nanoparticles are expected to enable the electrostatic binding of the anionic chromogenic ion indicators, preventing their washout into the sample solution during sensing. The interaction of the indicators with the nanoparticles was confirmed from the fact that the ultrafiltrate of dye-particle mixtures exhibited no color (detailed experimental procedure and
results are available in the supplementary material Figure S-4). The synthesized nanoparticles have a measured average diameter of about 110 nm (Fig. S-5) and a 1 wt% aqueous suspension has a -potential of 64 mV, confirming their cationic charge, making them less likely to aggregate inside the printer cartridge. For an ink containing nanoparticles to be ejected by an inkjet printer, the particle size should be around 100 times smaller than the nozzle diameter, which in the case of the used Epson printer is about 20 m. The measured particle size of 110 nm is much less than the required 200 nm, thus, successful printing of the nanomaterial on a cellulosic substrate was achieved. The addition of 0.1% Triton X-100 to the 1 wt% cationic nanoparticle suspension lowered the ink surface tension and enhanced its printability, leading to a stable (at least 6 months without removing the ink cartridge from the printer), clog-free ink formulation. It is important to mention that this ink is recommended for the piezoelectrically actuated (Epson) rather than the thermally actuated (Canon) printer, since the latter showed more incidences of nozzle clogging with particle containing inks.
3.3. Colorimetric model assays The DS fluid flow concept was applied to a colorimetric metal cation detection assay model. For this purpose, Zincon (responsive to Zn2+ and Cu2+) and Ferene S (responsive to Fe2+) were chosen as colorimetric indicators. The detection of those ions with the selected chromogenic indicators
requires different sample pre-treatment and pH conditions for each respective cation. Zincon responds to the presence of Zn2+ only at comparably high pH values between 8.5−9.5.[24] On the other hand, the working pH range for Cu2+ detection is broader (5.5−9.5)[24] than that of Zn2+. The Ferene S-based Fe2+ detection system requires a masking agent for Cu2+ or Cu+ to prevent the interference during the Fe2+-Ferene S complex formation.[25] A thermally actuated (Canon) printer was used to deposit the chromogenic ion indicator and pre-treatment reagent inks (pH buffer, masking agent), since this type of printer is compatible with inks with similar surface tension and viscosity to water. Therefore, the chromogenic ion indicator and buffer inks did not require any additives to adjust surface tension or viscosity. Proper print color settings in the graphic software allowed the simultaneous, but individually addressable ejection of ink in color printing mode without interfusion of other inks as experimentally verified and shown in Figure S-6. In all cases, chromogenic ion indicators (Zincon or Ferene S) were inkjet-printed on paper sensing regions in 6 cycles to obtain sufficient color intensity. As anticipated, a sensing region with printed Zincon did not respond to changes in Zn2+ ion concentrations in the absence of printed buffer (pH 8.5) (Fig. S-7a), and without the printed thiourea masking agent, interference from Cu2+ (20 M) in an Fe2+ assay was observed (Fig. S-7b). Therefore, buffer (for Zn2+ sensing) and masking agent (for Fe2+ sensing) inks were printed in 6 cycles into the respective sensing regions.
For system evaluation, the colorimetric response of the printed indicators and additives on paper to metals was first evaluated by independently printed isolated single spot tests. Images showing metal concentration-dependent color changes and calibration plots based on the above-described printing conditions are shown in Fig. S-8. For quantitative evaluation of colorimetric signals, the hue parameter or the red intensity were employed for Zincon- and Ferene S-based systems, respectively. The hue is a parameter mostly reflecting the color regardless of its intensity[26] and thus, superior in quantifying the colorimetric response of Zincon, which changes its color from pink to blue depending on the amount of Cu2+ or Zn2+. On the other hand, upon complexation with Fe2+, Ferene S turns from a colorless state, where the hue parameter is not defined, to blue color. Since this change from colorless (white paper substrate) to blue corresponds to a decrease of red and green intensities in the RGB color space, the changes in red color intensity have been adapted as the quantification signal for Fe2+ sensing. Figure 5a schematically illustrates the mechanism of ion capture on sensing zones with Cu2+ detection as an example. The metal ions present in the sample solution form complexes with the printed chromogenic ion indicators, giving rise to a metal concentration-dependent color change. Actual photos of a sensing zone with inkjet deposited cationic nanoparticles and Zincon on filter paper in the presence or absence of Cu2+ are shown in Figure 5b. Sample solutions directly dropped on inkjet-printed ion sensing spots were allowed to incubate for 5 min, before the excess liquid was
removed with the aid of tissue paper. A change in color from pink to blue was observed in the presence of Cu2+. The uniform color distribution across the circular sensing zone implies that the cationic nanoparticle-anionic ion indicator combination is effective in keeping the sensing components in place over the entire incubation period, while capturing metal ions from the solution.
3.4. Study of inter-region reagent contamination in multichannel µPADs In contrast to the conventional use of PADs, where capillary flow-driven wetting out of the paper results in unidirectional sample transport, the application of DS bulk fluid flow involving the presence of bulk sample liquid on the entire PAD could lead to unwanted diffusion-controlled migration of solutes. Based on this fact, cross-contamination by assay components deposited on discrete sensing regions interconnected by microfluidic channels has to be taken into account. For this reason, the absence of diffusion-driven mixing of printed reagents from different sensing regions was experimentally investigated making use of the strongly pH-dependent Zn2+-response of Zincon, with the incubation time extended up to 60 min. First, the cationic nanoparticle-Zincon system was sequentially printed onto three sensing regions. Next, the pH 8.5 buffer ink was printed into sensing regions A and C, while no buffer was deposited on sensing region B (Fig. 6). As a consequence, both sensing regions A and C are expected to respond to the presence of Zn2+ (alkaline pH condition), while region B should not respond to this cation due to the unfavourable pH condition. The
experiment revealed that even at extended incubation times of up to 60 min, no colorimetric response is observed in sensing region B. This clearly demonstrates that the diffusion of pH buffer components between neighbouring sensing zones is not an issue, since any contamination of zone B with pH 8.5 buffer would have resulted in a colorimetric response to the presence of Zn2+. With the adopted 5 min incubation time discussed above, cross-contamination by diffusive reagent exchange between printed sensing regions can be excluded. To further look at the dynamic behavior of pre-deposited components on paper, the assay reagents were replaced by standard dye-based color inks (cyan, magenta, yellow) as visually observable water-soluble quasi-reagents. Time-course observation of the inkjet-printed color inks after application of 80 L of pure water (Video S-4 for a movie of the first 5 min, Fig. S-9 for photographs up to 80 min after water addition) revealed the elution of dye components into the bulk water, as well as their diffusion. However, in no case was dye-spreading observed beyond the branches leading to single sensing regions. This double experimental confirmation allows the conclusion that chemical cross-talk between sensing regions is not an issue when working with PADs in DS flow operation mode and that multiple sensing regions maintain their independent response characteristics.
3.5. “Drop-slip” flow-implemented colorimetric PAD
Based on the fact that chemical assays requiring different conditions are feasible by using DS flow on a PAD, colorimetric response data to Zn2+, Cu2+, and Fe2+ was acquired on a multi-ion PAD. The results in Figure 7 show the colorimetric response after 5 min incubation obtained from regions 1, 2, 3 (Fig. 1b), respectively. Region 1 with printed TAPS/TMAOH buffer (pH 8.5) shows concentration-dependent response to both Zn2+ and Cu2+ (Fig. 7a), because of the satisfactory assay condition for those metals. As expected, only Cu2+ is detected in region 2 (Fig. 7b), where the basic buffer component is absent. Finally, region 3 responds only to Fe2+ without significant interference from Cu2+ at least up to 50 M (Fig. 7c). It is important to note that chemical assays requiring different conditions can be independently carried out in each sensing spot with no interference from adjacent regions. In addition, good agreement with the result of spot tests has been found (Fig. S-10), which confirms the independent response of each sensing zone. The identical response profiles observed for the spot tests and the multi-ion PAD also demonstrate that analyte loss during sample transport (e.g. due to evaporation of sample liquid) is not an issue in the bulk DS flow PAD operation mode.
Conclusions The concept of “drop-slip” fluid flow on unmodified cellulosic microfluidic channels has been demonstrated for the first time. The approach is taking advantage of a lubricated (wetted) layer of
porous cellulose to spontaneously induce a bulk liquid flow on the surface of a multi-channel PAD. Fluid motion along patterned channels of hydrophilic cellulose is the result of a combination of lateral capillary flow and bulk fluid slip flow, resulting in the rapid distribution of the sample liquid into multiple sensing regions. Colorimetric response within as short as 5 min has been achieved. Since larger sample volumes are required compared to conventional purely capillary flow-driven PADs, devices operated in bulk surface flow mode might not be applicable to all analytical tasks targeted by PADs. However, in many cases, there is no shortage of available sample volume. In these situations, working with “drop-slip” bulk flow will contribute to drastically shortened analysis time, while improving sensitivity due to the increased amount of analyte reaching the sensing zone. In contrast to other methods of sample flow rate increase on PADs, the present approach does require neither a significant variation in device design and fabrication, nor in the amount of colorimetric reagents. Therefore, the bulk liquid transport method demonstrated here is expected to be applicable to various types of PADs with minimal adaptation to the original device.
Acknowledgements The authors thank Prof. Koichi Asakura for helpful discussion, and Dr. Yuki Hiruta for measuring the -potential of the nanoparticle inks. This work was financially supported by the Medical Research and Development Programs Focused on Technology Transfer: Development of Advanced
Measurement and Analysis Systems (SENTAN) by the Japan Agency for Medical Research and Development (AMED).
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version.
References [1] A.W. Martinez, S.T. Phillips, M.J. Butte, G.M. Whitesides, Patterned paper as a platform for inexpensive, low-volume, portable bioassays, Angew. Chem. Int. Ed. 46 (2007) 1318–1320. [2] D.M. Cate, J.A. Adkins, J. Mettakoonpitak, C.S. Henry, Recent Developments in Paper-Based Microfluidic Devices, Anal. Chem. 87 (2015) 19–41. [3] K. Yamada, T.G. Henares, K. Suzuki, D. Citterio, Paper-Based Inkjet-Printed Microfluidic Analytical Devices, Angew. Chem. Int. Ed. 54 (2015) 5294–5310. [4] Y. He, Y. Wu, J.-Z. Fu, W.-B. Wu, Fabrication of paper-based microfluidic analysis devices: a review, RSC Adv. 5 (2015) 78109–78127. [5] A.K. Yetisen, M.S. Akram, C.R. Lowe, Paper-based microfluidic point-of-care diagnostic devices, Lab Chip 13 (2013) 2210–2251. [6] S. Ahmed, M.-P.N. Bui, A. Abbas, Paper-based chemical and biological sensors: Engineering
aspects, Biosens. Bioelectron. 77 (2016) 249–263. [7] C. Renault, X. Li, S.E. Fosdick, R.M. Crooks, Hollow-channel paper analytical devices, Anal. Chem. 85 (2013) 7976–7979. [8] S. Jahanshahi-Anbuhi, P. Chavan, C. Sicard, V. Leung, S.Z. Hossain, R. Pelton, J.D. Brennan, C.D. Filipe, Creating fast flow channels in paper fluidic devices to control timing of sequential reactions, Lab Chip 12 (2012) 5079–5085. [9] A.C. Glavan, R.V. Martinez, E.J. Maxwell, A.B. Subramaniam, R.M. Nunes, S. Soh, G.M. Whitesides, Rapid fabrication of pressure-driven open-channel microfluidic devices in omniphobic RF paper, Lab Chip 13 (2013) 2922–2930. [10]
J.-H. Shin, G.-J. Lee, W. Kim, S. Choi, A stand-alone pressure-driven 3D microfluidic
chemical sensing analytic device, Sens. Actuators B: Chem. 230 (2016) 380–387. [11]
C.K. Camplisson, K.M. Schilling, W.L. Pedrotti, H.A. Stone, A.W. Martinez, Two-ply
channels for faster wicking in paper-based microfluidic devices, Lab Chip 15 (2015) 4461–4466. [12]
E.T. Silva, M. Santhiago, L.T. Kubota, W.K. Coltro, F.R. Souza, Triboelectric effect as a
new strategy for sealing and controlling the flow in paper-based devices, Lab Chip 15 (2015) 1651– 1655. [13]
K.M. Schilling, A.L. Lepore, J.A. Kurian, A.W. Martinez, Fully enclosed microfluidic
paper-based analytical devices, Anal. Chem. 84 (2012) 1579–1585.
[14]
J. Songok, M. Tuominen, H. Teisala, J. Haapanen, J. Mäkelä, J. Kuusipalo, M. Toivakka,
Paper-Based Microfluidics: Fabrication Technique and Dynamics of Capillary-Driven Surface Flow, ACS Appl. Mater. & Interfaces 6 (2014) 20060–20066. [15]
K. Abe, K. Suzuki, D. Citterio, Inkjet-printed microfluidic multianalyte chemical sensing
paper, Anal. Chem. 80 (2008) 6928–6934. [16]
K. Maejima, S. Tomikawa, K. Suzuki, D. Citterio, Inkjet printing: An integrated and green
chemical approach to microfluidic paper-based analytical devices, RSC Adv. 3 (2013) 9258–9263. [17]
S.Z. Hossain, R.E. Luckham, A.M. Smith, J.M. Lebert, L.M. Davies, R.H. Pelton, C.D.
Filipe, J.D. Brennan, Development of a bioactive paper sensor for detection of neurotoxins using piezoelectric inkjet printing of sol−gel-derived bioinks, Anal. Chem. 81 (2009) 5474–5483. [18]
X. Li, J. Tian, G. Garnier, W. Shen, Fabrication of paper-based microfluidic sensors by
printing, Colloids Surf. B 76 (2010) 564–570. [19]
A. Musyanovych, R. Rossmanith, C. Tontsch, K. Landfester, Effect of hydrophilic
comonomer and surfactant type on the colloidal stability and size distribution of carboxyl-and amino-functionalized polystyrene particles prepared by miniemulsion polymerization, Langmuir 23 (2007) 5367–5376. [20]
K. Yamada, S. Takaki, N. Komuro, K. Suzuki, D. Citterio, An antibody-free microfluidic
paper-based analytical device for the determination of tear fluid lactoferrin by fluorescence
sensitization of Tb3+, Analyst 139 (2014) 1637–1643. [21]
E. Evans, E.F.M. Gabriel, T.E. Benavidez, W.K.T. Coltro, C.D. Garcia, Modification of
microfluidic paper-based devices with silica nanoparticles, Analyst 139 (2014) 5560–5567. [22]
F. Figueredo, P.T. Garcia, E. Cortón, W.K. Coltro, Enhanced Analytical Performance of
Paper Microfluidic Devices by Using Fe3O4 Nanoparticles, MWCNT and Graphene Oxide, ACS Appl. Mater. Interfaces 8 (2015) 11–15. [23]
M.M. Mentele, J. Cunningham, K. Koehler, J. Volckens, C.S. Henry, Microfluidic
Paper-Based Analytical Device for Particulate Metals, Anal. Chem. 84 (2012) 4474–4480. [24]
C.E. Säbel, J.M. Neureuther, S. Siemann, A spectrophotometric method for the
determination of zinc, copper, and cobalt ions in metalloproteins using Zincon, Anal. Biochem. 397 (2010) 218–226. [25] S. Eskelinen, M. Haikonen, S. Räisänen, Ferene-S as the chromogen for serum iron determinations, Scand. J. Clin. Lab. Investig. 43 (1983) 453–455. [26] K. Cantrell, M.M. Erenas, I. de Orbe-Payá, L.F. Capitán-Vallvey, Anal. Chem. 82 (2010) 531– 542.
Figure Captions Figure 1 a) Fabrication of fully inkjet-printed PAD; b) Actual PAD dimensions and summary of inkjet printing conditions. Figure 2 Operational procedure for the fully inkjet-printed PAD with “drop-slip” fluid handling technique. Figure 3 a) Optimization of channel width; b) Schematic illustration of the bulk liquid front on a lubricated channel. “Negative” and “positive” surface tension contributions to DS flow occurrence are indicated by red and blue arrows, respectively. Figure 4 Time course of DS fluid transport on a fully inkjet-printed PAD. Figure 5 (a) Mechanism of ion capture illustrated on the example of Zincon and Cu2+. (b) Images of fully inkjet-printed ion sensing zone before and after exposure to Cu2+. Figure 6 Cross-contamination study with sequentially inkjet-printed cationic nanoparticle-Zincon for various incubation times. Regions A and C contain printed 0.4 M TAPS/TMAOH buffer (pH 8.5) for zinc sensing. The schematic at the top of the figure illustrates the expected response behavior in the absence or presence of cross-contamination between sensing zones. Figure 7 Response of each ion sensing region after 5 min incubation with different metal ion concentrations; a), b), and c) correspond to regions 1, 2, and 3 as described in Figure 1.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Biography Terence G. Henares received his PhD from the University of Hyogo in 2008. After postdoctoral studies at the Population Council Center for Biomedical Research in New York and at Osaka Prefectural University, he became a Project Associate Professor at Keio University. His research interests are in the development of microfluidic analytical devices. Kentaro Yamada received his B. Eng. and M. Eng. from Keio University in 2013 and 2014, respectively. He is currently carrying out PhD research, focusing on the development of inkjet-printed microfluidic paper-based analytical devices. Shunsuke Takaki received his B. Eng. and M. Eng. from Keio University in 2012 and 2014, respectively. His research topic is the detection of metals using inkjet-printed paper devices. Koji Suzuki received his PhD in 1982 from Keio University. He then became a faculty member as a Research Associate, Assistant Professor in 1988, Associate Professor in 1993, and full Professor in 1998. From 1990 to 1992, he was a Guest Professor at the Swiss Federal Institute of Technology (ETH), Zurich. Since 2015, he serves as the President of the Japan Society for Analytical Chemistry (JSAC). His research focuses on chemical and biochemical sensors based on functional molecule creation. Daniel Citterio received his doctoral degree in Chemistry in 1998 from the Swiss Federal Institute of Technology (ETH) in Zurich. After postdoctoral research at Keio University, he became a Research Associate at ETH in 2002. Following post-graduate studies and work at Ciba Specialty Chemicals, he returned to Keio University in 2006. In 2009 he became a tenured Associate Professor and in 2014 Professor in Analytical Chemistry. His research interests are the development of low-cost analytical devices and the development of functional molecules for chemical sensing.
S0925-4005(17)30096-5 http://dx.doi.org/doi:10.1016/j.snb.2017.01.088 SNB 21613
To appear in:
Sensors and Actuators B
Received date: Revised date: Accepted date:
24-11-2016 9-1-2017 12-1-2017
Please cite this article as: Terence G.Henares, Kentaro Yamada, Shunsuke Takaki, Koji Suzuki, Daniel Citterio, “Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device, Sensors and Actuators B: Chemical http://dx.doi.org/10.1016/j.snb.2017.01.088 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Highlights
Sample liquid transport by free surface flow on a paper device is demonstrated.
Channel width and surface tension of sample liquid are dominant factors.
Charged polymeric nanoparticles can immobilize water soluble colorimetric dyes.
No inter-assay cross-contamination by reagent diffusion for at least 60 min.
Paper-based analytical devices are entirely fabricated by inkjet printing.
“Drop-slip” bulk sample flow on fully inkjet-printed microfluidic paper-based analytical device
Terence G. Henares,‡ Kentaro Yamada,‡ Shunsuke Takaki, Koji Suzuki and Daniel Citterio*
Department of Applied Chemistry, Keio University, 3-14-1 Hiyoshi, Kohoku-ku, Yokohama 223-8522, Japan. Fax: +81 45 566 1568; Tel: +81 45 566 1568
‡ These authors contributed equally to this work.
* Corresponding author: [email protected].
Abstract With “drop-slip” (DS) bulk liquid flow on a fully inkjet-printed microfluidic paper-based analytical device (PAD), a new method for the rapid transport of sample liquid is presented. The main driving force for DS flow is the slip flow of fluid on wetted porous cellulose acting as a lubricated surface, which is predominantly influenced by the width of the hydrophilic channel and the surface tension of the sample liquid. The application of DS flow is demonstrated by a model colorimetric metal assay (Zn2+, Cu2+, Fe2+) with inkjet-deposited indicators on a PAD with optimized channel dimensions of 2 mm width and 110 m height. The presence of bulk liquid on the entire device does not result in any mixing of assay components across adjacent sensing regions connected by microfluidic channels. DS flow is a useful alternative sample liquid transport method for PADs, allowing to perform multiple fully independent assays with enlarged sample volumes without requiring any significant variation in device design and fabrication.
Keywords: Microfluidic paper-based analytical device, Free surface flow, Inkjet printing
1. Introduction Since the first introduction of microfluidic paper-based analytical devices (PADs) in 2007 by Whitesides and co-workers,[1] they have attracted attention as alternative analytical platforms enabling low-cost, easy-to-use, and multiplexed (bio)chemical assays.[2–6] Paper (most commonly filter paper or chromatography paper), a sustainable and inexpensive material, is the backbone of these analytical devices. The hydrophilicity, porosity and high surface-to-volume ratio of the cellulosic network are the primary properties amenable for microfluidic applications. Fluid transport in paper microchannels is accomplished by capillary wicking through the porous paper network, eliminating the need for tubing and bulky pump systems. However, the PADs reliant on the capillarity-driven lateral flow format suffer from one or several of the following disadvantages: 1) long analysis times derived from relatively slow capillary wicking speeds, 2) evaporation loss of sample liquid transported in open microfluidic paper channels, and 3) heterogeneous color development due to the washing-away effect of colorimetric signalling compounds by the sample liquid. So far, faster sample liquid flow or reduction of sample evaporation on PADs have been primarily achieved either by utilizing accelerated liquid transportation in hollow channel structures[7–12] or by printing a toner layer to cover channels.[13] Although these approaches are generally advantageous in terms of physical strength of the device and prevention of contamination,
increased complexity of the device fabrication process (e.g. stacking and alignment of multiple layers, extra toner printing step) is inevitable. In this paper, we introduce the fast bulk liquid flow on top of hydrophilic channels as an alternative method of liquid transport for PADs. Such faster sample liquid transportation involves a “drop-slip” (DS) flow, where sample liquid is added onto a microfluidic channel in a continuous drop-wise manner. The first introduced sample drop pre-wets the paper microchannel via capillary forces, followed by free surface flow of second (or later) drops of sample on a lubricated cellulosic surface. Sample transportation based on free surface flow leads to improvement of PAD assays in two aspects: 1) much faster sample flow speed thanks to the aid of pressure originating from spreading of the bulging sample fluid; and 2) larger amount of transported sample liquid resulting in higher sensitivity and reduced influence of evaporation. Although bulk liquid flow has already been realized on a TiO2/UV-treated superhydrophilic paper platform,[14] the applicability of DS flow-based liquid transport to untreated cellulosic microfluidic channels has to the best of our knowledge not been demonstrated. Herein, the colorimetric detection of Cu2+, Zn2+, and Fe2+ has been performed as model chemical assays to demonstrate the feasibility of simultaneous independent sensing using a DS flow-based PAD. Zn2+ and Cu2+ detection was carried out with Zincon as the indicator, of which the reactivity toward Zn2+ is tunable by pH. For the Fe2+ assay, Ferene S was employed together with masking
agents to eliminate interferences from Cu2+. These models allow to verify the possibility of on-device sample pre-treatment (pH control and masking of interfering compounds) on PADs operated in DS flow mode. Among a variety of approaches to patterning of paper with flow channels, the wax or inkjet-printing methods have become the most frequently used for their accessibility and rapidity. In particular, inkjet-printing is known to be the only standard printing technology capable of fabricating entire PADs (i.e. applicable to both channel patterning and the deposition of assay reagents).[3, 15– 18] We showcase the strength of the inkjet-printing fabrication approach by depositing all components of the PAD with desktop-type inkjet printers. A “fully inkjet-printed PAD” relying on drop-slip flow liquid transport has been developed by depositing UV-curable ink for patterning, polymeric nanoparticles for immobilization of indicators, and chromogenic ion indicators and different sample pre-treatment ink solutions by means of piezoelectrically and thermally actuated desktop inkjet printers.
2. Experimental 2.1. Chemicals Reagents of the highest grade commercially available were used for this work. 2-Aminoethylmethacrylate hydrochloride (AEMH), cetyltrimethylammonium chloride (CTMA-Cl)
(25% in water), 3-(2-pyridyl)-5,6-di(2-furyl)-1,2,4-triazine-5’,5’’-disulfonic acid disodium salt (Ferene S), and tetramethylammonium hydroxide pentahydrate (TMAOH) were purchased from Sigma-Aldrich
(St.
Louis,
MO).
Styrene,
octadecyl
acrylate
(ODA),
2,2'-azobis(2-methylbutyronitrile) (V-59), copper (II) chloride, zinc acetate, iron (II) chloride, ammonium acetate, L(+)-ascorbic acid and thiourea were obtained from Wako Chemicals (Osaka, Japan). 1-(2-hydroxycarbonyl-phenyl)-5-(2-hydroxy-5-sulfophenyl)-3-phenylformazan sodium salt (Zincon) and N-tris(hydroxymethyl) methyl-3-aminopropanesulfonic acid (TAPS) were acquired from
Dojindo
(Kumamoto,
Japan).
1,10-Bis(acryloyloxy)decane
(DDA)
and
2,2-dimethoxy-2-phenylacetophenone (BDK) were purchased from Tokyo Chemical Industry TCI (Tokyo, Japan). All the reagents were used without further purification except for styrene, which was vacuum distilled (10 mm Hg, 62˚C) to remove the polymerization inhibitor. Also, ODA and DDA were treated with 0.1 wt% NaOH to remove the MEHQ stabilizer. All aqueous solutions were prepared using ultrapure (> 18 M cm) deionized water.
2.2. Materials and instrumentation Whatman 540 filter paper (GE Healthcare, Buckinghamshire, UK) was used as substrate for PAD fabrication. A piezoelectrically actuated EPSON PX-105 inkjet printer (Seiko Epson, Suwa, Japan) with custom refillable ink cartridges (Daiko, Hachioji, Japan) was used to print the
UV-curable and cationic nanoparticle inks. On the other hand, a thermally actuated Canon iP2700 inkjet printer with its original ink cartridges was used to print the chromogenic ion indicators, masking agent and pH buffer inks. UV curing was done under a Hg-Xe lamp (Lightingcure LC-6, Hamamatsu Photonics, Hamamatsu, Japan), and a DVM2500 optical microscope (Leica, Wetzlar, Germany) was used to analyze cross-sections of PADs.
2.3. Fabrication of fully inkjet-printed PADs Fully inkjet-printed PADs (Figs. 1a and 1b) were fabricated by sequentially printing UV-curable ink, cationic nanoparticle, chromogenic ion indicator and pre-treatment buffer inks. The detailed compositions of all printing inks are given in Tables S-1 and S-2 of the supplementary material. The cationic nanoparticles were synthesized as previously reported[19] (details available in the supplementary material). Printing patterns were designed with Microsoft PowerPoint software. The flow channels defined by hydrophobic barriers were obtained by printing UV-curable ink as previously described by our group.[20] In short, black outlines of 0% transparency were printed on the topside of the filter paper. Next, the ink was allowed to penetrate into the paper for 20 s, before the substrate was placed upside down on a cooled plate (10˚C) for 1 min to retard further diffusion of the liquid ink, followed by 15 min of UV irradiation on the printed side. Hydrophobic barriers on the backside of the paper were obtained analogously, with the exception of printing the UV-curable ink
as black areas of 60% transparency. The Hg-Xe lamp used for UV-curing was placed at a distance from the filter paper surface to yield a power of 4 mW cm−2 (measured at 365 nm). Figure 1a depicts the PAD design in which each ion sensing zone undergoes multiple deposition of different chemically functional inks. First, all sensing regions were subjected to 1 printing cycle of cationic nanoparticle ink using the Epson PX-105 printer. Next, aqueous inks of assay reagents were sequentially inkjet-printed by the Canon iP2700 printer (printing graphics and detailed printer settings are available in Fig. S-1 in the supplementary material). Briefly, Zincon and Ferene S indicator inks were simultaneously printed, followed by deposition of TAPS/TMAOH buffer (pH 8.5) on region 1 in 6 printing cycles using the cyan, magenta, and yellow ink reservoirs, respectively. Finally, an ink containing ammonium acetate (pH buffer), thiourea (masking agent) and ascorbic acid (reducing agent) was dispensed on region 3 from the black cartridge. For region 2, only cationic nanoparticle and Zincon inks were deposited, without any pH buffer.
2.4. Colorimetric ion detection A schematic representation of the operational procedure is demonstrated in Figure 2. Sample solution (80 L) containing metal ions was introduced into the inlet port in a drop-by-drop manner. Absorption of the initial liquid droplets immediately saturated the porous filter paper, allowing the slipping of the following drops over the paper surface and a complete liquid coverage of each
sensing zone. After a 5 min incubation time, the excess sample liquid was removed by the aid of a Kimwipe tissue paper. Finally, the PAD was left to dry at room temperature for about 5 min, before colorimetric measurement was performed by scanning with a 9000F MARK II color scanner (Canon, Tokyo, Japan). The Image J software (NIH, Bethesda, MD, USA) was used to acquire numerical color intensity values on the RGB scale.
3. Results and discussion 3.1. “Drop-slip” flow on fully inkjet-printed PAD The “drop-slip” (DS) fluid flow on a wetted porous cellulose surface resulted in the concurrent lateral capillary flow and bulk fluid flow (Fig. 2). The cellulose fiber network inside the patterned channels is first saturated by liquid via capillary action and then serves as a lubricated surface where slipping of the bulk liquid follows. The occurrence of DS flow is determined by the channel width as demonstrated in Figure 3a. A simple PAD design has been used to study the liquid flow behavior on channels of various widths (0.3–3.5 mm). The results show that channels of 1.5 mm or larger widths result in DS flow (15 L aqueous dye solution dropped into a 14 mm long channel), while in more narrow channels (width ≤ 1.0 mm), liquid transport is limited to capillary flow wicking. Although the width of the hydrophilic paper channel plays a dominant role in the liquid flow behavior, other factors have to be considered. As shown in Table S-3, the surface tension and the depth of the
hydrophilic channel influence the threshold width for the occurrence of DS flow, but viscosity was practically irrelevant at least up to a value of 5 cP. The hydrophilic channel depth was modulated by varying the printing density of the UV curable ink on the backside of the filter paper (Fig. S-2). The results indicate that 1) large channel width, 2) low liquid surface tension, and 3) deep hydrophilic channels promote DS flow. Occurrence or prevention of DS flow is determined by competing negative and positive driving forces. Figure 3b schematically illustrates the bulk liquid front on a lubricated channel. At the boundary of the sample liquid and the hydrophobic barrier, interfacial tension is generated in the opposite direction to the bulk liquid spreading (indicated as in the bottom left of Fig. 3b), resulting in a negative effect on DS flow. On the other hand, both negative and positive driving forces for DS flow occur at the center of the lubricated channel (bottom right of Fig. 3b, where L is the surface tension of the sample solution, S is the interfacial tension between the gas phase and the lubricated channel, and LS is the interfacial tension between the sample solution and the lubricated channel). Herein, positive driving forces are dominant over the negative ones, since the liquid spreads on the lubricated paper surface without forming a spherical droplet. Thus, the locally-generated positive driving force is expressed by the following equation: 𝛾𝑠 − 𝛾𝐿 cos 𝜃𝑘 − 𝛾𝐿𝑆 > 0 where k is the position of interest along a coordinate axis perpendicular to the channel direction. As a
consequence, the total driving force for DS flow occurrence within a channel of width w is represented as follows: ∑𝑤 𝑘=0(𝛾𝑆 − 𝛾𝐿 cos 𝜃𝑘 − 𝛾𝐿𝑆 ) − 2𝜎 > 0
(2)
Increased channel width results in a larger contribution of positive driving forces (eq. (1)), leading to DS flow by overcoming the channel width independent interfacial tension at the liquid-hydrophobic barrier boundaries (2). Lower liquid surface tension contributes to the occurrence of the DS flow by decreasing L and values, which work as negative factors. Finally, the hydrophilic channel depth also has some effect on LS, as concluded from the fact that too shallow channels adversely influence DS flow occurrence (Table S-3 in the supplementary material). Although not investigated in this study, DS flow occurrence on substrates other than paper is expected to be dependent on the surface tension of the sample liquid, the surface wettability of the substrate, and the roughness of the substrate. It is postulated that DS flow-based sample transportation is possible even in the absence of capillary forces, provided that the substrate surface is sufficiently wettable by the deposited liquid (i.e. low contact angles). However, it should be noted that the use of non-paper substrates might eliminate several advantageous aspects associated with paper as an analytical platform, including its high surface-to-volume ratio useful for assay reagent immobilization, low-cost, and disposability. It is noteworthy to mention that fluid handling techniques have negligible influence on the
outcome of DS flow behavior. Dye solution (200 L) was introduced dropwise onto a PAD with 50 mm channel length, 110 m channel depth, and multiple varying channel width structures. Variation of the liquid dropping rate from 20 to 60 drops per minute did not alter the minimal channel width (1.5 mm) required for DS flow (refer to Video S-1 for the case of 20 drops per minute rate). Moreover, even at increased volume of single droplets (constant total volume), achieved by cutting off the head of the pipette tip, DS flow occurred only at channel widths of more than 1.5 mm (Video S-2). As shown in Figure S-3, the fluid front progression speed in DS flow is dependent on the condition of liquid introduction, such as the dropping rate or the volume of single drops. However, it is clear that the DS flow system allows much faster sample fluid transportation than traditional capillary force-driven liquid flow (see the supplementary material and Figure S-3 for the detailed experimental setup and results). On a fully inkjet-printed PAD with multiple channels, the liquid transported by DS flow filled the entire device within less than 10 seconds (Figure 4 and Video S-3). Even though the solution filled the center sensing zone first and there was a short delay in liquid arrival in the remaining sensing regions, this gap of a few seconds is assumed to have an insignificant effect on the outcome of ion sensing.
3.2. Immobilization of colorimetric sensing reagents Lateral capillary driven flow is the most common fluid transport operating in PADs. When
water soluble colorimetric sensing reagents adsorbed on the cellulose material are exposed to a lateral fluid flow within the cellulose network, they tend to dissolve and to be carried away along the direction of the flow. As a consequence, increased color reagent density evolves on the periphery of the sensing zone or a gradient in color intensity is often observed. Such a non-uniform color distribution is detrimental to reproducible quantitative colorimetric data processing, giving rise to measurement errors. For this reason, simple means to immobilize water soluble sensing reagents are of high interest. This issue has been previously addressed by depositing various nanomaterials (silica nanoparticles,[21] Fe3O4 magnetic nanoparticles,[22] multiwalled carbon nanotubes,[22] graphene oxide[22]) or polymers with charged residues[23] together with colorimetric assay reagents. In the current work, immobilization of water soluble metal ion indicators plays an even more important role, because of increased DS bulk liquid flow and a resident sample liquid droplet on top of the sensing area during the incubation period. For this purpose, cationically charged polystyrene-based core/shell nanoparticles were selected. In contrast to non-particulate polymers,[23] the nanoparticles used here are water insoluble, which makes them suitable for extended contact with sample liquid. The cellulose network-anchored cationic nanoparticles are expected to enable the electrostatic binding of the anionic chromogenic ion indicators, preventing their washout into the sample solution during sensing. The interaction of the indicators with the nanoparticles was confirmed from the fact that the ultrafiltrate of dye-particle mixtures exhibited no color (detailed experimental procedure and
results are available in the supplementary material Figure S-4). The synthesized nanoparticles have a measured average diameter of about 110 nm (Fig. S-5) and a 1 wt% aqueous suspension has a -potential of 64 mV, confirming their cationic charge, making them less likely to aggregate inside the printer cartridge. For an ink containing nanoparticles to be ejected by an inkjet printer, the particle size should be around 100 times smaller than the nozzle diameter, which in the case of the used Epson printer is about 20 m. The measured particle size of 110 nm is much less than the required 200 nm, thus, successful printing of the nanomaterial on a cellulosic substrate was achieved. The addition of 0.1% Triton X-100 to the 1 wt% cationic nanoparticle suspension lowered the ink surface tension and enhanced its printability, leading to a stable (at least 6 months without removing the ink cartridge from the printer), clog-free ink formulation. It is important to mention that this ink is recommended for the piezoelectrically actuated (Epson) rather than the thermally actuated (Canon) printer, since the latter showed more incidences of nozzle clogging with particle containing inks.
3.3. Colorimetric model assays The DS fluid flow concept was applied to a colorimetric metal cation detection assay model. For this purpose, Zincon (responsive to Zn2+ and Cu2+) and Ferene S (responsive to Fe2+) were chosen as colorimetric indicators. The detection of those ions with the selected chromogenic indicators
requires different sample pre-treatment and pH conditions for each respective cation. Zincon responds to the presence of Zn2+ only at comparably high pH values between 8.5−9.5.[24] On the other hand, the working pH range for Cu2+ detection is broader (5.5−9.5)[24] than that of Zn2+. The Ferene S-based Fe2+ detection system requires a masking agent for Cu2+ or Cu+ to prevent the interference during the Fe2+-Ferene S complex formation.[25] A thermally actuated (Canon) printer was used to deposit the chromogenic ion indicator and pre-treatment reagent inks (pH buffer, masking agent), since this type of printer is compatible with inks with similar surface tension and viscosity to water. Therefore, the chromogenic ion indicator and buffer inks did not require any additives to adjust surface tension or viscosity. Proper print color settings in the graphic software allowed the simultaneous, but individually addressable ejection of ink in color printing mode without interfusion of other inks as experimentally verified and shown in Figure S-6. In all cases, chromogenic ion indicators (Zincon or Ferene S) were inkjet-printed on paper sensing regions in 6 cycles to obtain sufficient color intensity. As anticipated, a sensing region with printed Zincon did not respond to changes in Zn2+ ion concentrations in the absence of printed buffer (pH 8.5) (Fig. S-7a), and without the printed thiourea masking agent, interference from Cu2+ (20 M) in an Fe2+ assay was observed (Fig. S-7b). Therefore, buffer (for Zn2+ sensing) and masking agent (for Fe2+ sensing) inks were printed in 6 cycles into the respective sensing regions.
For system evaluation, the colorimetric response of the printed indicators and additives on paper to metals was first evaluated by independently printed isolated single spot tests. Images showing metal concentration-dependent color changes and calibration plots based on the above-described printing conditions are shown in Fig. S-8. For quantitative evaluation of colorimetric signals, the hue parameter or the red intensity were employed for Zincon- and Ferene S-based systems, respectively. The hue is a parameter mostly reflecting the color regardless of its intensity[26] and thus, superior in quantifying the colorimetric response of Zincon, which changes its color from pink to blue depending on the amount of Cu2+ or Zn2+. On the other hand, upon complexation with Fe2+, Ferene S turns from a colorless state, where the hue parameter is not defined, to blue color. Since this change from colorless (white paper substrate) to blue corresponds to a decrease of red and green intensities in the RGB color space, the changes in red color intensity have been adapted as the quantification signal for Fe2+ sensing. Figure 5a schematically illustrates the mechanism of ion capture on sensing zones with Cu2+ detection as an example. The metal ions present in the sample solution form complexes with the printed chromogenic ion indicators, giving rise to a metal concentration-dependent color change. Actual photos of a sensing zone with inkjet deposited cationic nanoparticles and Zincon on filter paper in the presence or absence of Cu2+ are shown in Figure 5b. Sample solutions directly dropped on inkjet-printed ion sensing spots were allowed to incubate for 5 min, before the excess liquid was
removed with the aid of tissue paper. A change in color from pink to blue was observed in the presence of Cu2+. The uniform color distribution across the circular sensing zone implies that the cationic nanoparticle-anionic ion indicator combination is effective in keeping the sensing components in place over the entire incubation period, while capturing metal ions from the solution.
3.4. Study of inter-region reagent contamination in multichannel µPADs In contrast to the conventional use of PADs, where capillary flow-driven wetting out of the paper results in unidirectional sample transport, the application of DS bulk fluid flow involving the presence of bulk sample liquid on the entire PAD could lead to unwanted diffusion-controlled migration of solutes. Based on this fact, cross-contamination by assay components deposited on discrete sensing regions interconnected by microfluidic channels has to be taken into account. For this reason, the absence of diffusion-driven mixing of printed reagents from different sensing regions was experimentally investigated making use of the strongly pH-dependent Zn2+-response of Zincon, with the incubation time extended up to 60 min. First, the cationic nanoparticle-Zincon system was sequentially printed onto three sensing regions. Next, the pH 8.5 buffer ink was printed into sensing regions A and C, while no buffer was deposited on sensing region B (Fig. 6). As a consequence, both sensing regions A and C are expected to respond to the presence of Zn2+ (alkaline pH condition), while region B should not respond to this cation due to the unfavourable pH condition. The
experiment revealed that even at extended incubation times of up to 60 min, no colorimetric response is observed in sensing region B. This clearly demonstrates that the diffusion of pH buffer components between neighbouring sensing zones is not an issue, since any contamination of zone B with pH 8.5 buffer would have resulted in a colorimetric response to the presence of Zn2+. With the adopted 5 min incubation time discussed above, cross-contamination by diffusive reagent exchange between printed sensing regions can be excluded. To further look at the dynamic behavior of pre-deposited components on paper, the assay reagents were replaced by standard dye-based color inks (cyan, magenta, yellow) as visually observable water-soluble quasi-reagents. Time-course observation of the inkjet-printed color inks after application of 80 L of pure water (Video S-4 for a movie of the first 5 min, Fig. S-9 for photographs up to 80 min after water addition) revealed the elution of dye components into the bulk water, as well as their diffusion. However, in no case was dye-spreading observed beyond the branches leading to single sensing regions. This double experimental confirmation allows the conclusion that chemical cross-talk between sensing regions is not an issue when working with PADs in DS flow operation mode and that multiple sensing regions maintain their independent response characteristics.
3.5. “Drop-slip” flow-implemented colorimetric PAD
Based on the fact that chemical assays requiring different conditions are feasible by using DS flow on a PAD, colorimetric response data to Zn2+, Cu2+, and Fe2+ was acquired on a multi-ion PAD. The results in Figure 7 show the colorimetric response after 5 min incubation obtained from regions 1, 2, 3 (Fig. 1b), respectively. Region 1 with printed TAPS/TMAOH buffer (pH 8.5) shows concentration-dependent response to both Zn2+ and Cu2+ (Fig. 7a), because of the satisfactory assay condition for those metals. As expected, only Cu2+ is detected in region 2 (Fig. 7b), where the basic buffer component is absent. Finally, region 3 responds only to Fe2+ without significant interference from Cu2+ at least up to 50 M (Fig. 7c). It is important to note that chemical assays requiring different conditions can be independently carried out in each sensing spot with no interference from adjacent regions. In addition, good agreement with the result of spot tests has been found (Fig. S-10), which confirms the independent response of each sensing zone. The identical response profiles observed for the spot tests and the multi-ion PAD also demonstrate that analyte loss during sample transport (e.g. due to evaporation of sample liquid) is not an issue in the bulk DS flow PAD operation mode.
Conclusions The concept of “drop-slip” fluid flow on unmodified cellulosic microfluidic channels has been demonstrated for the first time. The approach is taking advantage of a lubricated (wetted) layer of
porous cellulose to spontaneously induce a bulk liquid flow on the surface of a multi-channel PAD. Fluid motion along patterned channels of hydrophilic cellulose is the result of a combination of lateral capillary flow and bulk fluid slip flow, resulting in the rapid distribution of the sample liquid into multiple sensing regions. Colorimetric response within as short as 5 min has been achieved. Since larger sample volumes are required compared to conventional purely capillary flow-driven PADs, devices operated in bulk surface flow mode might not be applicable to all analytical tasks targeted by PADs. However, in many cases, there is no shortage of available sample volume. In these situations, working with “drop-slip” bulk flow will contribute to drastically shortened analysis time, while improving sensitivity due to the increased amount of analyte reaching the sensing zone. In contrast to other methods of sample flow rate increase on PADs, the present approach does require neither a significant variation in device design and fabrication, nor in the amount of colorimetric reagents. Therefore, the bulk liquid transport method demonstrated here is expected to be applicable to various types of PADs with minimal adaptation to the original device.
Acknowledgements The authors thank Prof. Koichi Asakura for helpful discussion, and Dr. Yuki Hiruta for measuring the -potential of the nanoparticle inks. This work was financially supported by the Medical Research and Development Programs Focused on Technology Transfer: Development of Advanced
Measurement and Analysis Systems (SENTAN) by the Japan Agency for Medical Research and Development (AMED).
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version.
References [1] A.W. Martinez, S.T. Phillips, M.J. Butte, G.M. Whitesides, Patterned paper as a platform for inexpensive, low-volume, portable bioassays, Angew. Chem. Int. Ed. 46 (2007) 1318–1320. [2] D.M. Cate, J.A. Adkins, J. Mettakoonpitak, C.S. Henry, Recent Developments in Paper-Based Microfluidic Devices, Anal. Chem. 87 (2015) 19–41. [3] K. Yamada, T.G. Henares, K. Suzuki, D. Citterio, Paper-Based Inkjet-Printed Microfluidic Analytical Devices, Angew. Chem. Int. Ed. 54 (2015) 5294–5310. [4] Y. He, Y. Wu, J.-Z. Fu, W.-B. Wu, Fabrication of paper-based microfluidic analysis devices: a review, RSC Adv. 5 (2015) 78109–78127. [5] A.K. Yetisen, M.S. Akram, C.R. Lowe, Paper-based microfluidic point-of-care diagnostic devices, Lab Chip 13 (2013) 2210–2251. [6] S. Ahmed, M.-P.N. Bui, A. Abbas, Paper-based chemical and biological sensors: Engineering
aspects, Biosens. Bioelectron. 77 (2016) 249–263. [7] C. Renault, X. Li, S.E. Fosdick, R.M. Crooks, Hollow-channel paper analytical devices, Anal. Chem. 85 (2013) 7976–7979. [8] S. Jahanshahi-Anbuhi, P. Chavan, C. Sicard, V. Leung, S.Z. Hossain, R. Pelton, J.D. Brennan, C.D. Filipe, Creating fast flow channels in paper fluidic devices to control timing of sequential reactions, Lab Chip 12 (2012) 5079–5085. [9] A.C. Glavan, R.V. Martinez, E.J. Maxwell, A.B. Subramaniam, R.M. Nunes, S. Soh, G.M. Whitesides, Rapid fabrication of pressure-driven open-channel microfluidic devices in omniphobic RF paper, Lab Chip 13 (2013) 2922–2930. [10]
J.-H. Shin, G.-J. Lee, W. Kim, S. Choi, A stand-alone pressure-driven 3D microfluidic
chemical sensing analytic device, Sens. Actuators B: Chem. 230 (2016) 380–387. [11]
C.K. Camplisson, K.M. Schilling, W.L. Pedrotti, H.A. Stone, A.W. Martinez, Two-ply
channels for faster wicking in paper-based microfluidic devices, Lab Chip 15 (2015) 4461–4466. [12]
E.T. Silva, M. Santhiago, L.T. Kubota, W.K. Coltro, F.R. Souza, Triboelectric effect as a
new strategy for sealing and controlling the flow in paper-based devices, Lab Chip 15 (2015) 1651– 1655. [13]
K.M. Schilling, A.L. Lepore, J.A. Kurian, A.W. Martinez, Fully enclosed microfluidic
paper-based analytical devices, Anal. Chem. 84 (2012) 1579–1585.
[14]
J. Songok, M. Tuominen, H. Teisala, J. Haapanen, J. Mäkelä, J. Kuusipalo, M. Toivakka,
Paper-Based Microfluidics: Fabrication Technique and Dynamics of Capillary-Driven Surface Flow, ACS Appl. Mater. & Interfaces 6 (2014) 20060–20066. [15]
K. Abe, K. Suzuki, D. Citterio, Inkjet-printed microfluidic multianalyte chemical sensing
paper, Anal. Chem. 80 (2008) 6928–6934. [16]
K. Maejima, S. Tomikawa, K. Suzuki, D. Citterio, Inkjet printing: An integrated and green
chemical approach to microfluidic paper-based analytical devices, RSC Adv. 3 (2013) 9258–9263. [17]
S.Z. Hossain, R.E. Luckham, A.M. Smith, J.M. Lebert, L.M. Davies, R.H. Pelton, C.D.
Filipe, J.D. Brennan, Development of a bioactive paper sensor for detection of neurotoxins using piezoelectric inkjet printing of sol−gel-derived bioinks, Anal. Chem. 81 (2009) 5474–5483. [18]
X. Li, J. Tian, G. Garnier, W. Shen, Fabrication of paper-based microfluidic sensors by
printing, Colloids Surf. B 76 (2010) 564–570. [19]
A. Musyanovych, R. Rossmanith, C. Tontsch, K. Landfester, Effect of hydrophilic
comonomer and surfactant type on the colloidal stability and size distribution of carboxyl-and amino-functionalized polystyrene particles prepared by miniemulsion polymerization, Langmuir 23 (2007) 5367–5376. [20]
K. Yamada, S. Takaki, N. Komuro, K. Suzuki, D. Citterio, An antibody-free microfluidic
paper-based analytical device for the determination of tear fluid lactoferrin by fluorescence
sensitization of Tb3+, Analyst 139 (2014) 1637–1643. [21]
E. Evans, E.F.M. Gabriel, T.E. Benavidez, W.K.T. Coltro, C.D. Garcia, Modification of
microfluidic paper-based devices with silica nanoparticles, Analyst 139 (2014) 5560–5567. [22]
F. Figueredo, P.T. Garcia, E. Cortón, W.K. Coltro, Enhanced Analytical Performance of
Paper Microfluidic Devices by Using Fe3O4 Nanoparticles, MWCNT and Graphene Oxide, ACS Appl. Mater. Interfaces 8 (2015) 11–15. [23]
M.M. Mentele, J. Cunningham, K. Koehler, J. Volckens, C.S. Henry, Microfluidic
Paper-Based Analytical Device for Particulate Metals, Anal. Chem. 84 (2012) 4474–4480. [24]
C.E. Säbel, J.M. Neureuther, S. Siemann, A spectrophotometric method for the
determination of zinc, copper, and cobalt ions in metalloproteins using Zincon, Anal. Biochem. 397 (2010) 218–226. [25] S. Eskelinen, M. Haikonen, S. Räisänen, Ferene-S as the chromogen for serum iron determinations, Scand. J. Clin. Lab. Investig. 43 (1983) 453–455. [26] K. Cantrell, M.M. Erenas, I. de Orbe-Payá, L.F. Capitán-Vallvey, Anal. Chem. 82 (2010) 531– 542.
Figure Captions Figure 1 a) Fabrication of fully inkjet-printed PAD; b) Actual PAD dimensions and summary of inkjet printing conditions. Figure 2 Operational procedure for the fully inkjet-printed PAD with “drop-slip” fluid handling technique. Figure 3 a) Optimization of channel width; b) Schematic illustration of the bulk liquid front on a lubricated channel. “Negative” and “positive” surface tension contributions to DS flow occurrence are indicated by red and blue arrows, respectively. Figure 4 Time course of DS fluid transport on a fully inkjet-printed PAD. Figure 5 (a) Mechanism of ion capture illustrated on the example of Zincon and Cu2+. (b) Images of fully inkjet-printed ion sensing zone before and after exposure to Cu2+. Figure 6 Cross-contamination study with sequentially inkjet-printed cationic nanoparticle-Zincon for various incubation times. Regions A and C contain printed 0.4 M TAPS/TMAOH buffer (pH 8.5) for zinc sensing. The schematic at the top of the figure illustrates the expected response behavior in the absence or presence of cross-contamination between sensing zones. Figure 7 Response of each ion sensing region after 5 min incubation with different metal ion concentrations; a), b), and c) correspond to regions 1, 2, and 3 as described in Figure 1.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Biography Terence G. Henares received his PhD from the University of Hyogo in 2008. After postdoctoral studies at the Population Council Center for Biomedical Research in New York and at Osaka Prefectural University, he became a Project Associate Professor at Keio University. His research interests are in the development of microfluidic analytical devices. Kentaro Yamada received his B. Eng. and M. Eng. from Keio University in 2013 and 2014, respectively. He is currently carrying out PhD research, focusing on the development of inkjet-printed microfluidic paper-based analytical devices. Shunsuke Takaki received his B. Eng. and M. Eng. from Keio University in 2012 and 2014, respectively. His research topic is the detection of metals using inkjet-printed paper devices. Koji Suzuki received his PhD in 1982 from Keio University. He then became a faculty member as a Research Associate, Assistant Professor in 1988, Associate Professor in 1993, and full Professor in 1998. From 1990 to 1992, he was a Guest Professor at the Swiss Federal Institute of Technology (ETH), Zurich. Since 2015, he serves as the President of the Japan Society for Analytical Chemistry (JSAC). His research focuses on chemical and biochemical sensors based on functional molecule creation. Daniel Citterio received his doctoral degree in Chemistry in 1998 from the Swiss Federal Institute of Technology (ETH) in Zurich. After postdoctoral research at Keio University, he became a Research Associate at ETH in 2002. Following post-graduate studies and work at Ciba Specialty Chemicals, he returned to Keio University in 2006. In 2009 he became a tenured Associate Professor and in 2014 Professor in Analytical Chemistry. His research interests are the development of low-cost analytical devices and the development of functional molecules for chemical sensing.