Microchemical Journal 193 (2023) 109044 Contents lists available at ScienceDirect Microchemical Journal journal homepage: www.elsevier.com/locate/microc A novel integrated platform enabling simultaneous microextraction and chemical analysis on-chip Elia Santigosa-Murillo a, Alexandre Moreno b, María Ramos-Payán c, Juan Manuel Ríos b, Xavier Muñoz-Berbel b, d, María Muñoz a, *, Pablo Giménez-Gómez b, 1, * a Department of Analytical Chemistry, Universitat Autónoma de Barcelona, 08193 Bellaterra, Barcelona, Spain b Instituto de Microelectrónica de Barcelona (IMB-CNM, CSIC), 08193 Bellaterra, Barcelona, Spain c Department of Analytical Chemistry, Faculty of Chemistry, University of Seville, 41012 Seville, Spain d CIBER de Bioingeniería, Biomateriales y Biomedicina (CIBER-BBN), Spain A R T I C L E I N F O A B S T R A C T Keywords: The nature and size of biological, pharmaceutical or environmental analytes complicates their extraction and Electromembrane microextraction detection outside of laboratories and near the site of interest by the current chromatographic methods because Electrochemical impedance spectroscopy they require the combination of bulky extraction and detection methods. In order to solve this inefficient detection centralized control, a ground-breaking and miniaturized proof of concept platform is developed in this work. The Biological, pharmaceutical or environmental analytes detection platform integrates for the very first time an electro-membrane extraction process and an accurate analyte Thin-film microelectrodes quantification method in the same device, by using electrochemical impedance spectroscopy (EIS) as analytical technique. The microfluidic flow cell, including the microfluidic components, is fabricated in polymeric mate­ rials by rapid prototyping techniques. It comprises a four-electrode platinum thin-film chip that enables the control of the microextraction and the full characterization of the process, i.e., extraction efficiency determi­ nation, at the same time. The microfluidic system has been simulated by using computational tools, enabling an accurate prediction of the effect of the different experimental conditions in the microextraction efficiency. The platform has been validated in the microextraction of the nonsteroidal anti-inflammatory drug ketoprofen in a range from 0.5 ppm to 6 ppm. The predicted microextraction efficiency values obtained by EIS were compared with those calculated from the high-performance liquid chromatography coupled with a diode array detector (HPLC-DAD), showing an excellent agreement. This validates the high potential of this integrated and minia­ turized platform for the simultaneous extraction by electro-membrane and also the analysis within the platform, solving one of the of most important limitations of current systems. 1. Introduction The sample processing and the analyte detection method are the key factors in miniaturized systems because of the complexity of real sam­ Lab-on-a-chips (LOCs) implement multiple functions in a low-cost, ples matrices [3]. For the sample processing, different microextraction miniaturised and automated system, widely improving the perfor­ techniques [4] have been adapted to microfluidic platforms to take mance of traditional laboratory assays in portability and robustness [1]. advantage of the LOC functionalities, like the liquid phase micro­ They are simple-to-operate (even for non-trained personnel) and safe, extraction (LPME) and the electromembrane extraction (EME). In LPME, minimizing the users’ exposure to chemicals and human errors. More­ the neutral analytes diffuse through a supported liquid membrane [5], over, their low sample and reagents consumption reduces the waste while in the EME, the electric field generated between two electrodes production and the assay time, making them ideal for in-field analysis, placed on each of the liquid phases (the donor and the acceptor) allows and enabling the application of on-time corrective actions in very short the charged analytes to move from the donor to the acceptor phase [6]. times, a key factor in environmental control, clinical analysis or quality Regarding the chemical analysis, there is an increasing demand on the control in the food industry [2]. ever-smaller portable analytical devices to monitor in a decentralised * Corresponding authors. E-mail addresses: maria.munoz@uab.cat (M. Muñoz), pablo.gimenez-gomez@mmk.su.se (P. Giménez-Gómez). 1 Present address: Department of Materials and Environmental Chemistry, Stockholm University, 106 91Stockholm, Sweden. https://doi.org/10.1016/j.microc.2023.109044 Received 28 April 2023; Received in revised form 28 June 2023; Accepted 30 June 2023 Available online 1 July 2023 0026-265X/© 2023 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 manner (i.e., environmental monitoring, biochemical detection or clin­ studied and presented in this work as a proof of concept for the very first ical diagnostics [7–10]), without the need for sample collection and time. This proof-of-concept platform is fabricated by multi-layered transport to the laboratory to speed-up the analysis and shorten PMMA and it integrated two silicon chips with four platinum micro­ decision-making at lower cost. electrodes placed face-to-face in the microchannels respectively Such current trend towards the integration of microextraction and immersed in the donor and in the acceptor solution. The multielectrode chemical analysis in a single LOC [11–13] began with the first molecular system enabled to properly study the option to use different areas and separation using microchannels proposed by Sato et al. in 2000 [14]. electrodes to evaluate the microextraction through chemical analysis of From there, many other microfluidic systems have been developed both phases and the membrane by EIS, meaning in a fully character­ aiming to improve the low efficiency and reproducibility of this initial ization and control of the process. The system has been fully studied by system. In 2010, the first EME-based microfluidic system integrating a using computational software and EIS as characterization technique in supported liquid membrane and two platinum threads as electrodes in a terms of microfluidic design and microelectrodes configuration. The polymethyl methacrylate (PMMA) platform was published [15]. predicted microextraction efficiencies by ESI were compared to those Although it had high extraction efficiencies, its single-use design limited obtained from the HPLC-DAD analysis. The microfluidic flow cell has its applicability. After that, other systems coupled to HPLC-DAD have been fully characterised to perfectly understand and control the been developed with the aim of miniaturizing the channels used for the microextraction process for the selective determination of the ketopro­ extraction of different biological, pharmaceutical or environmental fen (KTP) [28,29], a non-steroidal anti-inflammatory drug, in a robust analytes [16–18] to reduce the analysis time, but they had low enrich­ and reproducible way, obtaining promising results. ment factors, which did not allow to detect the very low concentrations required in environmental monitoring. In 2015, a new system improving 2. Experimental the enrichment factor of the extraction with novel microchannel ge­ ometries and working in stationary conditions was developed [18], but 2.1. Reagents and solutions the assay time and the processed sample volume was very high. At that time, the integration of LPME and EME in microfluidics allowed the KTP, 1-octanol, formic acid, ammonium chloride, sodium phosphate simultaneous extraction of analytes of similar nature, but the selective dibasic, methanol and chloride acid, were purchased from Sigma­ microextraction of analytes of different nature was still a challenge to –Aldrich (Madrid, Spain). All the solutions used in this work were pre­ solve. In 2016, the first system trying to solve this challenge was pro­ pared in deionizing water, except where otherwise noted. A stock posed [19]. The system (working in continuous mode) integrated a methanol-based solution of 100 mg/L of KTP was prepared every LPME process into microfluidics, resulting in higher extraction effi­ week and stored at 4 ◦ C in the fridge. For the assays, water-based so­ ciencies and enrichment factors. However, the technique did not allow a lutions of KTP were prepared every day from this stock solution. To good extraction when using analytes with different structure, acid-basic mechanically clean and electrochemically activate the microelectrodes, nature and polarity. This limitation was solved by Yamini et al. [20] by ethanol 96%, 6 M sulphuric acid and 0.1 M potassium nitrate were connecting two miniaturised EME devices and by using an acceptor prepared and stored at room temperature. solution adjusted with two different pH to favour the selective analyte extraction. Although the extraction was good, the system required 2.2. Equipment sample preconcentration to operate in stationary flow, consuming important sample volumes (1,0 mL) and requiring large extraction times For HPLC-based separation, a liquid chromatograph (Agilent 1100 (more than 40 min). During the last five years, the first microchips series, Barcelona, Spain) equipped with a binary pump system (G1312A, integrating LPME and EME [21] or EME-EME [22,23] for the separation Agilent, Barcelona, Spain) and an autosampler injector (G1313A 5 µL, of basic/acid analytes with high efficiency extraction (around 100 %) Agilent, Barcelona, Spain) was used. The set-up for the KTP analysis was have been developed. These microchips-based devices can obtain completed with a column (LiChroCART1 75–4 Purosphere STAR RP- reproducible results with high robustness and sensitivity in a minia­ 18e, 3 mm, 75 mm × 4.0 mm of inner diameter, VWR, Barcelona, turized way, consuming very low volumes of reagents and samples. Spain) preceded by a guard column (Kromasil1 100 Å, C18, 5 mm, 20 Moreover, the high miniaturisation enables to shorten the analysis time mm × 4.6 mm of inner diameter, Scharlab S.L., Barcelona, Spain). and improve the process efficiency. In addition, the mass transfer is To electrochemically activate the microelectrodes, an Autolab elec­ enhanced by the use of microchannels, and the use of electrodes in EME trochemical workstation (PGSTAT-100 potentiostat–galvanostat, Eco­ microsystems allow selective microextraction by adjusting the direction chemie, Utrecht, The Netherlands) controlled with the NOVA advanced and the magnitude of the power supply. electrochemical software was used. The electrochemical cell (platinum But, despite of the advantages mentioned above of microchips-based microelectrodes were used as working electrodes) was completed with a systems, all of the reported systems still require the combination of the commercial platinum electrode as counter electrode (Radiometer device with traditional equipment for the chemical analysis (HPLC- Analytical, Lyon, France) and a commercial Ag/AgCl (3 M KCl) double DAD), limiting their fully portability and their on-line application for in junction reference electrode (Orion 92–02-00, Thermo Fisher Scientific situ analysis. At this point, electrochemical impedance spectroscopy Inc., Beverly, USA). (EIS) technique has demonstrated a high potential to characterize the Impedance measurements were performed with a 1260 Solartron solutions and the membrane integrated into non-miniaturised EME Impedance Analyzer (Solartron Analytical, Farnborough, UK), systems [24–26]. In 2020, the very first EME-based millifluidic device controlled with the ZPlot® software. impedance characterization was validated [27] and applied to the The polymeric microfluidic flow cell was manufactured by using a extraction of parabens with very high efficiencies in urine samples CO2 laser cutter (Epilog Mini 24, Epilog Laser, United States), followed (86%-92%), opening a new challenge to the monitoring of processes in by a solvent assisted bonding under pressure and temperature with a real time. Nevertheless, this system used platinum threads as electrodes, manual 2-column laboratory press (PW 10H-HKP300, Paul-Otto Weber and presented two important limitations: (1) the reproducibility and GmbH, Remshalden, Germany). reusability of the system was very low; and (2) it did not work as a real multiplexed system for the simultaneous extraction and chemical anal­ 2.3. Design and fabrication of the microfluidic flow cell ysis of the microsystem, limiting their potential application in decen­ tralised assays. Silicon chips already developed by the group and applied to lactic To solve the current challenge of simultaneous microextraction and and malic detection [30] were used in this case, which containing four chemical analysis, a pioneering miniaturised microfluidic platform is independent in-parallel platinum (Pt) microelectrodes (i.e., one of 2 mm 2 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 × 2.5 mm and three of 1 mm × 2.5 mm, separated each other 0.6 mm). 2.4. Computational simulations The chips also included four contact pads located 2.9 mm away from the microelectrodes to enable the electrical connection between the mi­ The effect of the variation of the main design parameters involved in croelectrodes to the external equipment once implemented in the the performance of the microfluidic flow cell was evaluated by using the polymeric microfluidic flow cell for extraction and chemical analysis of computational COMSOL Multiphysics® 6.0 software (from now referred analytes. as COMSOL). In all the cases, a Tertiary Current Distribution (Nernst- The microfluidic flow cell (30 mm × 30 mm × 11.25 mm) consisted Planck Interface) was used as the physics interface to carry out the of two symmetrical structures, each one containing three independent simulations. This interface enabled to simulate the potential distribution PMMA layers, which chemically bonded through a solvent-assisted of the electrochemical cell by considering the transport of the ions and process (7.5 kN, 80 ◦ C) with methacrylic acid (Fig. 1A). The first layer other phenomena as diffusion, convection or migration transport of (layer 1; 5 mm-thick) contained two 5 mm-diameter cylindrical holes for uncharged species, by considering the Nernst-Planck equation. positioning the fitting threads working as fluidic inlet and outlet, which A 2D design of the microfluidic flow cell was employed for the were connected to external Teflon tubes (0.5-mm inner diameter, simulations aiming to generate a model able to predict the kinetics and Teknokroma, Barcelona, Spain). The layer 1 also contained a hole (9.4 final concentration of KTP extracted in ammonium buffered solution. mm × 2.5 mm) for embedding the spring-loader connector which The simulations considered the following conditions and extraction enabled the electrical connection between the chip positioned in the mechanism: (i) a 10 mM ammonium buffer water solution at pH 10.5 other symmetrical structure and the external potentionstat-galvanostat. with KTP (donor solution) and a 10 mM ammonium buffer water solu­ The second layer (layer 2; 500 µm-thick) contained a hole (11 mm × 9 tion at pH 10.5 without KTP (acceptor solution), were inoculated mm × 0.5 mm) with the footprint of the silicon chip for its correct through the inlets of the microfluidic flow cell placed on the top layer, as positioning and alignment with the spring-loaded connector and the is schematised in Fig. 2A; (ii) The KTP was spontaneously dissociated in microfluidic channels from the other PMMA layers. Besides, this layer the donor solution, forming negative charged ions; (iii) The positive contained two holes of 1 mm-diameter aligned with the fitting threads potential applied between the microelectrodes in both sides of the from layer 1, to ensure properly fluidic connection between layers (from microchannel induced the migration of negatively-charged KTP mole­ the microfluidic channel in layer 3 to the inlet/outlet in layer 1). Finally, cules through the 25 μm-thick membrane doped with 1-octanol, from layer 3 defined the microfluidic channel (22 mm × 3 mm), which was the donor to the acceptor phase solution; and (iv) the microextraction aligned with the four platinum microelectrodes of the chip (embedded in efficiency was determined by comparing the quantity of the KTP present layer 2) and with the microfluidic inlet and outlet included in layer 1. in the outlet of the acceptor solution, with the quantity of the KTP The thickness of layer 3 was initially set at 175 µm, although several inoculated in the inlet of the donor solution. thicknesses between 175 µm and 380 µm were evaluated in the opti­ Different microelectrode configurations and working conditions mization of the microfluidic flow cell performance. An image of the were evaluated using COMSOL and compared in terms of micro­ three layers assembled with the silicon chip is illustrated in Fig. 1B. extraction efficiency percentage. The extraction efficiency was defined Both symmetrical structures were assembled with the microelec­ as the percentage of KTP in the outlet of the acceptor solution in relation trodes of the two chips facing each other, and incorporating a porous to that present in the inlet of the donor solution (microextraction effi­ membrane (25 μm thickness, 55% porosity, and 0.21 μm × 0.05 μm ciency = (KTP concentration in the acceptor solution outlet / KTP pores, Celgard 2500, Celgard, Charlotte, NC, USA) between both concentration in the donor solution inlet) × 100). And the micro­ microchannels. The symmetrical structures were clamped with six 3-mm extraction efficiency percent variation was defined as the difference diameter screws to avoid fluidic leakage (Fig. 1C). Fig. 1D shows an between the efficiency of one case of study versus the first case, divided image of the final microfluidic flow cell. In the characterization of the by the value obtained in the first case, in percentage terms. The influ­ microfluidic flow cell, the solutions were inoculated into the system with ence in the microextraction efficiency of the applied potential, the flow two 5 mL syringes connected to two external syringe pumps (Nemesys rate, the thickness of the microfluidic channels, the total area and B101-02B, Cetoni GmbH, Korbussen, Germany). positioning of the microelectrodes was evaluated (details of the micro­ electrodes in Fig. 2B). An initial KTP concentration in the inlet of the donor solution of 3 ppm was considered in all simulations. Fig. 1. A) Schematic layer by layer view of one of the parts of the microfluidic flow cell; B) Detailed image of the bonded three layers forming one of the parts of the microfluidic device; C) Schematic view of both parts assembled; D) Detailed image of the microfluidic flow cell. 3 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 Fig. 2. A) 3-D scheme of the microfluidic flow cell cross section; B) Detailed scheme detailing the position and the 3-D view of the layers 2 and 3 including the chip and the microchannel; C) Cross-section of the microfluidic performance of the microfluidic flow cell. In COMSOL simulations, the theoretical model and the microelec­ applied between the microelectrodes placed in both microchannels to trodes labelling presented in Fig. 2C were employed. The influence in extract the KTP from the donor to the acceptor solution through the the microextraction efficiency of the applied potential, the flow rate and electro-membrane; and (vi) 12-μL extracts were collected in the outlet of the microchannels thickness were evaluated with the same model, the acceptor solution after the microextraction process to determine by varying the parameter under study, while keeping constant the other HPLC-DAD the microextraction efficiency (as is described in Section two. In all cases, the microextraction potential was applied between the “Chromatographic measurements”). microelectrodes 2.4, 2.3 and 2.2 in the acceptor and the microelectrodes The influence of the extraction conditions was also evaluated in the 1.4, 1.3 and 1.2 in the donor phase. This ensured a high microelectrode experimental system, namely the microextraction potential (from + 3 V area for the microextraction process (three microelectrodes from each to + 7 V), the flow rate in the donor and acceptor microchannel (be­ chip), while leaving one of the microelectrodes of each chip free (mi­ tween 3 µL/min and 5 µL/min); the microchannel thickness (from 175 croelectrodes 2.1 and 1.1) for the EIS chemical analysis between the µm to 380 µm); and the microelectrodes configuration by varying the electro-membrane. number (total area) and the position of the connected microelectrodes. The position and total area of the microelectrodes used to apply the During the evaluation of each experimental condition, the other pa­ microextraction potential was also studied in detail to quantify their rameters were kept constant. influence in the microextraction efficiency percent variation. The following combinations of microelectrodes (according to the scheme in 2.6. Chemical analysis by impedance of the microfluidic flow cell Fig. 2C) were tested: (i) the 2.4 + 2.3 + 2.2 and the 1.4 + 1.3 + 1.2; (ii) the 2.4 + 2.3 and the 1.4 + 1.3; (iii) the 2.4 and the 1.4; (iv) the 2.4 and The microfluidic flow cell was calibrated with seven KTP concen­ the 1.3; (v) the 2.4 and the 1.2; (vi) the 2.3 and the 1.4; (vii) the 2.3 and trations by EIS analysis in the range from 0 ppm to 5 ppm to obtain the the 1.3; (viii) the 2.3 and the 1.2; (ix) the 2.2 and the 1.4; (x) the 2.2 and limit of detection (LOD), the sensibility, the reproducibility and the the 1.3; (xi) the 2.2 and the 1.2. In this study, the flow rate, the potential linear range for the chemical analysis of the molecule. A potential of + 7 and the microchannels thickness were kept constant during all the V was applied between the microelectrodes placed in the acceptor (i.e., simulations. 2.2, 2.3 and 2.4) and the donor microchannels (i.e., 1.2, 1.3 and 1.4). Both the acceptor and the donor solution were injected at a flow rate of 2.5. Validation of the simulation studies 3 µL/min. The acceptor extracts were collected for 4 min (total volume of 12 μL), measured by HPLC-DAD and compared with EIS data. EIS The influence of the conditions from the COMSOL simulations were measurements were performed after the microextraction with the mi­ also experimentally validated by a series of assays of microextraction croelectrodes 1.1 (donor microchannel) and 2.1 (acceptor micro­ with the manufactured microanalytical flow cell. The process was as channel). EIS experimental conditions were as follows: Open Circuit follows: (i) the porous membrane was doped with 1-octanol and allo­ Potential, OCP; 10 mV AC amplitude from 1 MHz to 10 Hz; 10 points per cated between both microfluidic channels placed in each of the poly­ decade). During the EIS measurements, the potential and the flow rate meric structures of the microfluidic flow cell as is detailed in Fig. 1A; (ii) were stopped to avoid interference in the impedimetric determination. the two symmetrical structures forming the microfluidic flow cell were Finally, the influence of the microextraction time on the micro­ assembled by using the screws; (iii) both the acceptor (10 mM ammo­ extraction efficiency was studied in the range from 30 s to 180 s. The nium buffer solution at pH 10.5) and the donor solution (10 mM microextraction process was evaluated with the following conditions: ammonium buffer solution at pH 10.5 containing KTP) were flowed +7 V, 175 µm-thick microchannels and 3 µL/min, using the microelec­ inside the microchannels by using the fluidic inlets; (iv) a potential was trodes 2.2 and 2.1 in the acceptor, and 1.2 and 1.1 in the donor 4 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 microchannel, respectively, and with the donor solution initially con­ Ni = − Di ▿ci − zi um,i Fci ▿∅l + ci u = Ji + ci u (3) taining 1 ppm of KTP. EIS measurements were performed at previous experimental conditions (i.e., OCP; 10 mV AC amplitude from 1 MHz to which depends on the concentration (ci), the valence (zi), the ion 10 Hz; 10 points per decade) by using two different microelectrodes coefficient of diffusion (Di) and the mobility (um, i) of the ion, the configurations: (i) 1.1 (donor microchannel) vs. 2.1 (acceptor micro­ Faraday constant (F), the electrolyte potential (∅l), the velocity (u) and channel) to analyse the impedance in the electro-membrane; (ii) and 2.1 the molar flow related to the convective transport (Ji). (acceptor microchannel) vs. 2.2 (acceptor microchannel) to monitor the The mass transport considered the transport of mass from the donor changes in the acceptor phase. As before, 12 μL of the acceptor solution to the acceptor microchannel by diffusion and migration. The diffusion, were collected and measured by HPLC-DAD to calculate the micro­ corresponding to the spontaneous transport of diluted species from the extraction efficiency. microchannel with a higher concentration to that with a lower one, considered only the mass transfer coefficient to simplify the calculations 2.7. Chromatographic measurements [31]. Only small changes of concentration near the microchannels and the membrane interfaces were assumed. The HPLC-DAD was used as the golden standard to validate all the The migration considered the movement of charged particles in the experimental assays described before. Briefly, 5 µL extracts were generated electric field, and they were based on the Nernst-Einstein measured by using a mobile phase consisting of 0.1% formic acid at pH equation (eq. (4): 2.6 (component A) and methanol (component B). The isocratic method um,i = Di/RT (4) was performed with 30% component B for 5 min, at a flow rate of 0.5 mL/min and at 25 ◦ C. The KTP wavelength used for the diode-array The migration was directly proportional to the Di, dependent on the detection (DAD) was 255 nm, with a retention time of 4.8 min. electric field gradient applied between the microelectrodes in both sides Consecutive extracts were injected after 3 min to properly re-equilibrate of the membrane, and inversely proportional to the Planck constant (R) the column to the initial conditions. and the temperature (T). The microextraction efficiency (EE) for KTP in the HPLC-DAD was The water-based electrolyte charge conservation considered the calculated according to the following equation (eq. (1): proton and hydroxyl transport from water ionisation automatically in the computational model. Therefore, the initial pH value in the simu­ na Va C a EE = x100% = x100 (1) lation only depended on the concentration of the previously defined n0 Vo Co constant species. Regarding the ion exchange membrane, the same mass Where, na is the number of moles found in the acceptor phase, no is transport mechanisms were considered with minimal modifications. the number of initial moles in the donor phase. Va and Vo are the vol­ Concretely, the coefficient of diffusion of the species though the mem­ umes used for the acceptor and the donor phase, respectively. Ca is the brane was adjusted (1.0 × 10-9 m2/s for both the NH± - 4 and the Cl ions, concentration of the KTP in the outlet of the acceptor solution and Co is 8.0 × 10 m /s for the KTP , 9.3 × 10 m /s for the H , and 5.3 × 10-9 -10 2 - -9 2 ± the initial concentration of the KTP in the inlet of the donor solution. m2/s for the OH–) according to bibliography [32], and the electrolyte volume fraction (ε), which is related to the membrane porosity (porous 3. Results and discussion matrix properties − 25 μm thickness, 55% porosity, and 0.21 μm × 0.05 μm pores -), was set at a value of 0.41. A free-triangular user-controller 3.1. Computational validation mesh, with an element size normal calibrated for fluid dynamics (maximum element size of 0.0169 mm, minimum element size of 7.5 × The influence of the microextraction conditions in the designed 10-4 mm, maximum element growth rate of 1.15, curvature factor of 0.3 microfluidic flow cell were firstly simulated with COMSOL. A 2-dimen­ and resolution of narrow regions of 1) was selected for the simulations. sional model approximation of the microfluidic flow cell was used to Finally, a stationary study model as solver was applied to compute the reduce the complexity of the calculations, and therefore the simulation study flow and the chemical composition at the equilibrium for the time (Figure S1, in the Supplementary Information -SI-). It is important chemical species. to highlight that the tertiary current distribution physics used in the A general view of the results obtained from the simulations are simulations enabled to define the parameter “out-of-plane thickness”. shown in Figure S1 (in the SI). Figure S1A (in the SI) represents the This parameter defined the thickness of the geometry perpendicular to electrolyte potential distribution generated in the microfluidic flow cell the two-dimensional cross-section (microchannel thickness), instead of after the application of a + 7 V potential between the microelectrodes being a 2-dimensional model. The following parameters were defined 1.4, 1.3 and 1.2 (acceptor microchannel) and the microelectrodes 2.4, and set constant along all the simulations: diffusion coefficients for NH± 4 2.3 and 2.2 (donor microchannel). The surface colour and the grey lines and Cl- of 1.0 × 10-9 m2/s; 298.5 K of temperature; NH4Cl concentration represent the electrolyte potential and the electrolyte density current in the inlet of 0.05 mol/dm3; cell length of 0.0068 m; membrane width vectors generated throughout the microfluidic flow cell, respectively. of 0.025 mm; microchannel thickness (out-of-plane thickness) of 2.5 Simulation results showed high electrolyte potentials (red colour cor­ mm; and KTP concentration in the inlet of 1.2 × 10-5 mol/dm3. responding to + 7 V) on the acceptor microelectrodes applying the Regarding the tertiary current distribution definition, best results were positive potential and low electrolyte potential (dark blue colour) on the obtained considering: (i) a mass transport of the dissolved species based donor microelectrodes. Intermediate areas between both microelec­ on both diffusion and migration effects; (ii) water-based with electro- trodes presented potentials between + 2 V and + 6 V, with increasing neutrality as the charge conservation model; and (iii) three species as values when approaching to the acceptor microelectrodes, as was ex­ - - dependent variables (NH± 4 and Cl from the buffer and KTP from the pected. Regarding to the electrolyte current density vectors, which were analyte). described as the sum of fluxes of all the ions in the electrolyte, The following assumptions were considered in the simulations. First, perpendicular vectors meaned that the current density was maximum the general mass balance for the diluted species in the electrolytes was along the surface of the face-to-face microelectrodes applying the defined by the following equation (eq. (2): microextraction potential. Curved vectors in the microchannels areas without microelectrodes also demonstrated the existence of an ions flow dci /dt + ▿ × Ni = Ri,tot (2) along the microchannel during microextraction process. This potential The equation was directly proportional to the total flow of the species and current density distribution favoured the migration of negatively- in the electrolyte (N), which is defined by the Nernst-Planck equation charged KTP ions in ammonium buffer from the donor (initial KTP (eq. (3): 5 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 concentration = 3 ppm) to the acceptor solution (initial KTP concen­ and the ionic migration associated to it. tration = 0 ppm). The distribution of the KTP ions throughout the As a summary, and considering all the computational simulations, microchannels is illustrated in Figure S1B (in the SI). Red-coloured areas the microfluidic flow cell should obtain high microextraction effi­ corresponded to high KTP concentration (3 ppm), and blue-coloured ciencies by applying high microextraction potentials in microelectrode areas to a KTP concentration of 0 ppm. Intermediate values of KTP arrays facing each other with a high total area, and by using low flow concentration were represented with different colours, where warmer rates and thin microchannels. The configuration corresponding to a ones corresponded to higher KTP concentrations. three-microelectrodes facing each other was considered the best archi­ A small KTP concentration (blue colour) was found close to the tecture for efficient microextraction with the microanalytical flow cell. acceptor inlet since mass transport at this point was mainly dependent on diffusion. Higher KTP concentrations were observed along the 3.2. Experimental microextraction results with the microfluidic flow cell acceptor microchannel, with warmer colours closer to the outlet. The result confirmed that mass transport was dominated by migration in Computational simulations were validated experimentally with the response to the electric field gradient generated between the micro­ microextraction system implemented in the set-up illustrated in Fig. 3A electrodes. This validated the suitability of the design and the micro­ by applying the experimental conditions detailed in the section 2.6. Both fluidic flow cell model to study microextraction processes. the donor and the acceptor solutions were flowed from the right (inlets) to the left (outlets) side of the microfluidic flow cell, as it is shown in 3.1.1. Effect of the applied potential. Fig. 3B. The microextraction efficiency was studied at: (i) micro­ The influence of several key parameters in the microextraction effi­ extraction potentials in the range from + 3 V to + 7 V (Fig. 5C); (ii) flow ciency was studied, concretely the applied potential, the flow rate, the rates between 3 µL/min and 5 µL/min (Fig. 5D); (iii) microchannel microchannels thickness and the number and position of the micro­ thicknesses in the range from 175 µm to 380 µm (Fig. 5E); and (iv) electrodes, as described in Section 2.4. As shown in Figure S2A (in the different microelectrodes configurations (Fig. 5F). In agreement with SI), it was a direct correlation between the applied potential and the computational simulations, higher microextraction efficiencies were microextraction efficiency. These results were in agreement with the obtained when applying high microextraction potentials (+7 V) in a Nernst-Planck equation, where the flow of charged species is linearly configuration involving large microelectrode areas of microelectrodes related to the potential in the electrolyte. According to that, the facing each other (microelectrodes configuration i in Fig. 5F), low flow microextraction potential should be as high as possible, but avoiding rates (3 µL/min) and thin microchannels (175 µm). The low micro­ electrolysis and other problems associated to the use of large potentials. extraction efficiency obtained of 43% at optimal conditions may be associated with some limitations of the experimental set-up. Concretely, 3.1.2. Effect of the flow rate. the applied potential could not be increased over + 7 V because of The microextraction efficiency decreased with the flow rate damage of the metallic surface and electrolytic corrosion; the flow rate (Figure S2B, in the SI), as was expected from the Nernst-Einstein equa­ could not be decreased below 3 µL/min for the high pressures generated tion. According to that equation, in migration-dominated systems the in the microchannels during the assays; the microchannel thickness migration depends linearly on the microextraction time (i.e., residence could not be reduced below 175 µm for the lack of available time of the species in a fluidic system). Therefore, the flow rate should be manufacturing systems; the location, the area and the number of elec­ kept as small as possible to guarantee a high microextraction effi­ trodes could not be modified since dependant on the chip design. ciencies. When comparing the influence of both parameters in the Despite of these experimental limitations, the reproducibility and microextraction of charged molecules, it was evident that the influence repeatability of the microextraction process was very high (three of the potential rate was higher than the flow rate, since the slope of the different systems were used for of each studied condition were used the linear trendline of the former was 27 % higher than the latter. same day, obtaining a relative standard deviation lower than 5%), demonstrating the potential of the proposed microfluidic flow cell for an 3.1.3. Effect of the microchannels thickness. accurate microextraction. In the next assays, the conditions providing The variation in the microextraction efficiency with the micro­ the optimal microextraction efficiencies were employed to fully validate channels thickness is illustrated in Figure S2C (in the SI), which pre­ the presented microanalytical flow system for simultaneous micro­ senting a polynomial decay. This behaviour was in agreement with the extraction and chemical analysis. second term of the Nernst–Planck equation (see eq. (3), in which ▽∅l is inversely proportional to the distance between microelectrodes [33]. 3.3. Analytical assessment of the microfluidic flow cell Therefore, this parameter was found to be the most critical one since small changes in thickness resulted in a large variation in micro­ The analytical capacities of the microfluidic flow cell for the detec­ extraction efficiency. tion of KTP were evaluated under optimal microextraction conditions in presence of seven concentrations of KTP (i.e., 0.3 ppm, 0.5 ppm, 1.0 3.1.4. Effect of the number of electrodes. ppm, 2.0 ppm, 3.0 ppm, 4.0 ppm and 5-0.0 ppm). The relationship be­ Finally, the influence of the number (total area) and the position of tween the KTP concentration in the analytical solution and the the microelectrodes in the microextraction was also studied. Results are impedimetric signal was determined in situ as follows (see Fig. 4A): (1) illustrated in Figure S2D (in the SI), where the plot in the left shows the the donor and the acceptor solutions were flowed at 3 µL/min for 3 min effect of the total area of the microelectrodes in the microextraction to properly fill the microchannels with the solutions; (2) then, a + 7 V process and the one in the right the effect of the electrode position. A microextraction potential was applied between the microelectrodes in linear correlation was observed between the total area magnitude and the acceptor (microelectrodes 2.2 + 2.3 + 2.4) and the donor (micro­ the microextraction efficiency, as expected when considering migration electrodes 1.2 + 1.3 + 1.4) microchannels during 4 min; and (3) after processes in microfluidic systems operating at constant flow. The mi­ the microextraction process, impedance measurements were performed croelectrodes position also influenced the microextraction efficiency. between the microelectrodes 1.1 (donor) and 2.1 (acceptor) to correlate Configurations with the microelectrodes in both microchannels faced the recorded signal with the KTP concentration in the solution. Based on each other (configuration vii -electrodes 2.3 and configuration 1.3- and a previous publication of the group [27], the equivalent circuit pre­ xi -electrodes 2.2 and 1.2-) presented lower microextraction efficiencies sented in the Fig. 4B was used to characterize the proposed microana­ than the others. The increasing of the microextraction efficiency when lytical flow system. The circuit consisted of four components: the an electrode in one channel is displaced relative to an electrode on the resistance of the electrolyte solution (RS), the double layer capacitance other channel was related to the increase of the current density vectors (Cdl) -here represented by a constant phase element to consider the 6 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 Fig. 3. A) Image of the set-up used during the experimental measurements with the microfluidic flow cell; B) Scheme of the experimental procedure followed to carry out the assays; C) Experimental microextraction efficiencies obtained for a potential of + 3 V, +5 V and + 7 V; D) Experimental microextraction efficiencies obtained for a flow rate of 5 µL/min, 4 µL/min and 3 µL/min; E) Experimental microextraction efficiencies obtained for a microchannels thickness of 175 µm, 250 µm and 380 µm; F) Experimental microextraction efficiencies obtained for different combinations of microelectrodes: (i) 1.2 + 1.3 + 1.4 + 2.2 + 2.3 + 2.4; (ii) 1.3 + 1.4 + 2.3 + 2.4; (iii) 1.4 + 2.4; (iv) 1.3 + 2.4; and (xi) 1.2 + 2.2. Each bar represents the mean efficiency percentage value of three replicates recorded consecutively under the same conditions, with the error bars being the corresponding standard deviation. roughness of the microelectrode surface-, the capacitance of the mem­ inversely proportional linear correlation in the range between 0.5 ppm brane (CM) and the resistance of the membrane (RM). The RM was and 5.0 ppm. The lowest KTP concentration (0.3 ppm) showed a high RM experimentally determined (see Figure S3, in the SI) as the diameter of value in comparison to the other concentrations which does not fall the semicircle of the plot from the Nyquist diagram for each concen­ within the linear adjustment, meaning that the resistance of the mem­ tration of KTP in the solution (Fig. 4C). As can be seen in Fig. 4C, the brane for concentrations below 0.5 ppm of KTP is very high and the flux diameter of the semicircle decreased (and therefore the resistance of the of the ions is not as favoured as for higher KTP concentrations. The membrane) when increasing the KTP concentration, following an sensitivity of the measurement was − 11120.7 ± 778.5 Ω/ppm 7 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 Fig. 4. A) Scheme of the experimental procedure and the electrical connections to carry out the impedimetric tests; B) Equivalent circuit of the impedance spectra; C) Nyquist plots for different KTP concentrations. (coefficient of correlation -R2- = 0.996), with a LOD (calculated using and conditions (see Fig. 3F, microelectrodes combination ii), demon­ the 3σ IUPAC criterion and the lowest concentration distinguishable strating again the high reproducibility of the system. from zero − 0.3 ppm-) of 0.12 ± 0.01 ppm (Figure S4, in the SI), and a These results demonstrates that the proposed microfluidic flow cell relative standard deviation (R.S.D.) of the sensitivity lower than 4% (n could be used to properly monitor on time the amount of the KTP = 3). extracted from the donor through impedance measurement, as well as Impedimetric detection of KTP considered the change in the medium the quantification of KTP extracts of unknow concentration by inter­ conductivity (i.e., Rs) associated to the presence of charged KTP con­ polation in the calibration curve thanks to the high repeatability of the centration. The simplicity and rapidness of the impedance measurement measurement system. This confirmed that the proposed proof of concept allowed continuous KTP detection and was used to monitor micro­ microanalytical flow system could be used for both analytes micro­ extraction efficiency in real time. The real time analysis of micro­ extraction and analysis, avoiding the need for sample collection and extraction is presented in Fig. 5A, while the two microelectrode transport to the laboratory for analysis with standard methods (HPLC- configurations used in the impedance measurements is schematized in DAD) and the associated costs. Fig. 5B. Experimentally, two electrode configurations were employed for impedance analysis, namely (i) two faced microelectrodes located in 4. Conclusions different microchannels (1.1 + 1.2 and 2.1 + 2.2, corresponding to the microelectrodes combination ii) and (ii) two serial microelectrodes both A proof of concept of a new microfluidic platform integrating mi­ located in the acceptor microchannel (2.1 + 2.2) for precise analysis of croelectrodes and an EME-based microextraction process in a fast pro­ Rs in the acceptor phase. To study the microextraction efficiency of a 1 totyped microfluidic flow cell has been studied and validated by using ppm KTP solution over time, a + 7 V microextraction potential was EIS. Two silicon chips with 4 platinum microelectrodes placed on both applied in the system during different times between 0 s and 180 s (0 s, sides of the doped membrane enable, for the very first time, the simul­ 30 s, 60 s, 90 s, 120 s and 180 s). The acceptor solution (9 µL) was taneous monitoring of the microextraction process and the full chemical collected after each microextraction to be analysed by HPLC-DAD for characterization of the system by using impedance technique. comparison with impedance data. Between consecutive measurements, Although some microfluidic systems in the literature have allowed the system was totally cleaned by flowing fresh solutions in both the their on-line coupling to analysis instruments (e.g., HPLC-DAD), they acceptor and the donor microchannels. The aim of this assay was to entail a great analysis time and hinder their portability. These chal­ determine the optimal microextraction time by monitoring micro­ lenges have been addressed with the promising operational results of the extraction efficiency in real time. Results showed important differences developed compact microfluidic flow cell. The performed on-time between the impedimetric response of both microelectrode configura­ microextraction and EIS characterization of KTP in this work drasti­ tions. In the case of faced microelectrodes, RM presented similar mag­ cally reduces the sample consumption and the microextraction times of nitudes over time (Fig. 5C), whereas a clear decrease in resistance was analytes in comparison to current methods, resulting in a perfect obtained when employing only microelectrodes in the acceptor micro­ candidate with unique characteristics to develop the next generation of channel (Fig. 5D). There was a clear correlation between impedance on-site and remote analytical control systems. data from the second configuration (serial microelectrodes in the The presented proof of concept platform presented here could be acceptor microchannel) and chromatographic measurements, which applied to analytes of different chemical nature and from different presented a linear increase of the microextraction efficiency for the 1 matrices, by adapting the design and using more advanced and reliable ppm KTP solution with the microextraction time (Fig. 5E). The value of manufacturing processes for increasing the extraction efficiency and the microextraction efficiency (around 30%) also correlates with the improving the limit of quantification, meaning that its applicability can obtained in the section 3.2 for the same microelectrodes configuration be broaden to environmental, biological or pharmaceutical analysis. 8 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 Fig. 5. A) Scheme of the experi­ mental procedure and the electrical connections to carry out the impedi­ metric tests of the microextraction tests over time; B) Equivalent circuit of the impedance spectra when the microelectrodes 1.1 and 2.1 (left scheme) and the microelectrodes 2.1 and 2.2 (right scheme) are used; C) Nyquist plots for different micro­ extraction times obtained from the micro microelectrodes 1.1 and 2.1; D) Nyquist plots for different micro­ extraction times obtained from the microelectrodes 2.1 and 2.2. Arabic numbers represent the extraction time corresponding to each recorded plot; E) In red colour, the evolution curve obtained from the Nyquist diagram for microelectrodes 2.1 and 2.2. Each point represents the mean RA value of three replicates recorded consecu­ tively for each time, with the error bars being the corresponding standard deviation. In blue colour, the evolu­ tion curve of the experimental micro­ extraction efficiency obtained by chromatography. Each point repre­ sents the mean percentage of the microextraction efficiency of three replicates recorded consecutively for each time, with the error bars being the corresponding standard deviation. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) This is the first platform described in the literature with a real potential Data availability to be applied for in situ extraction and detection analysis. Future work will focus on the full portability of the developed proof-of-concept for Data will be made available on request. decentralized analysis, replacing the electronic analyzer and syringe pumps used here with an integrated digital impedance analyzer and Acknowledgments battery-powered electronic micropumps. This work was financially supported by the Ministry of Science, Declaration of Competing Interest Innovation and Universities - State Research Agency -AEI- and European Regional Development Fund -ERDF- (RTI2018-101974-B-C-21 and The authors declare that they have no known competing financial PID2021-127653NB-C21). E. Santigosa thanks Universitat Autònoma de interests or personal relationships that could have appeared to influence Barcelona (UAB) for her PIF fellowship supported by the Agencia de the work reported in this paper. Gestió d’Ajusts Universitaris i de Recerca (2017-SGR-329). A. Moreno is grateful to MICINN for the grant of a PhD studentship from the FPI program (PRE2019-088338). 9 E. Santigosa-Murillo et al. Microchemical Journal 193 (2023) 109044 Appendix A. Supplementary data determination of parabens in water samples, Talanta. 165 (2017) 496–501, https://doi.org/10.1016/j.talanta.2016.12.059. [18] M. Karami, Y. Yamini, Y. Abdossalami Asl, M. Rezazadeh, On-chip pulsed Supplementary data to this article can be found online at https://doi. electromembrane extraction as a new concept for analysis of biological fluids in a org/10.1016/j.microc.2023.109044. small device, J Chromatogr A. (1527 (2017)) 1–9, https://doi.org/10.1016/j. chroma.2017.10.049. [19] M. Ramos-Payan, S. Maspoch, A. Llobera, An effective microfluidic based liquid- References phase microextraction device (μLPME) for extraction of non-steroidal anti- inflammatory drugs from biological and environmental samples, Anal Chim Acta. [1] H.A. Stone, A.D. Stroock, A. Ajdari, Engineering Flows in Small Devices: 946 (2016) 56–63, https://doi.org/10.1016/j.aca.2016.09.040. Microfluidics Toward a Lab-on-a-Chip, Engineering flows in small devices: [20] Y.A. Asl, Y. Yamini, S. Seidi, M. Rezazadeh, Simultaneous extraction of acidic and Microfluidics toward a lab-on-a-chip 36 (1) (2004) 381–411. basic drugs via on-chip electromembrane extraction, Anal Chim Acta. 937 (2016) [2] D. Mark, S. Haeberle, G. Roth, F.V. Stetten, R. Zengerle, Microfluidic lab-on-a-chip 61–68, https://doi.org/10.1016/j.aca.2016.07.048. platforms: Requirements, characteristics and applications, Chem Soc Rev. 39 [21] M. Ramos Payán, E. Santigosa, R. Fernández Torres, M.Á. Bello López, A New (2010) 1153–1182, https://doi.org/10.1039/b820557b. Microchip Design. A Versatile Combination of Electromembrane Extraction and [3] S. Haeberle, R. Zengerle, Microfluidic platforms for lab-on-a-chip applications, Lab Liquid-Phase Microextraction in a Single Chip Device, Anal Chem. 90 (2018) Chip. 7 (2007) 1094–1110, https://doi.org/10.1039/b706364b. 10417–10424, https://doi.org/10.1021/acs.analchem.8b02292. [4] A. Sarafraz-Yazdi, A. Amiri, Liquid-phase microextraction, TrAC -, Trends in [22] F. Zarghampour, Y. Yamini, M. Baharfar, M. Faraji, Simultaneous extraction of Analytical Chemistry. 29 (2010) 1–14, https://doi.org/10.1016/j. acidic and basic drugs: Via on-chip electromembrane extraction using a single- trac.2009.10.003. compartment microfluidic device, Analyst. 144 (2019) 1159–1166, https://doi. [5] E. Carasek, G. Bernardi, S.N. do Carmo, C.M.S. Vieira, Alternative green extraction org/10.1039/c8an01668b. phases applied to microextraction techniques for organic compound determination, [23] E. Santigosa, S. Pedersen-Bjergaard, M. Muñoz, M. Ramos-Payán, Green Separations. 6 (3) (2019) 35. microfluidic liquid-phase microextraction of polar and non-polar acids from urine, [6] C. Huang, Z. Chen, A. Gjelstad, S. Pedersen-Bjergaard, X. Shen, Electromembrane Anal Bioanal Chem. 413 (2021) 3717–3723, https://doi.org/10.1007/s00216-021- extraction, TrAC -, Trends in Analytical Chemistry. 95 (2017) 47–56, https://doi. 03320-9. org/10.1016/j.trac.2017.07.027. [24] C. Huang, K.F. Seip, A. Gjelstad, S. Pedersen-Bjergaard, Electromembrane [7] P. Gimenez-Gomez, R. Escude-Pujol, C. Jimenez-Jorquera, M. Gutierrez-Capitan, extraction of polar basic drugs from plasma with pure bis(2-ethylhexyl) phosphite Multisensor Portable Meter for Environmental Applications, IEEE Sens J. 15 (2015) as supported liquid membrane, Anal Chim Acta. 934 (2016) 80–87, https://doi. 6517–6523, https://doi.org/10.1109/JSEN.2015.2460011. org/10.1016/j.aca.2016.06.002. [8] D. Zhang, Q. Liu, Biosensors and bioelectronics on smartphone for portable [25] M.S. Restan, H. Jensen, X. Shen, C. Huang, Ø.G. Martinsen, P. Kubáň, A. Gjelstad, biochemical detection, Biosens Bioelectron. 75 (2016) 273–284, https://doi.org/ S. Pedersen-Bjergaard, Comprehensive study of buffer systems and local pH effects 10.1016/j.bios.2015.08.037. in electromembrane extraction, Anal Chim Acta. 984 (2017) 116–123, https://doi. [9] A. Márquez, J. Aymerich, M. Dei, R. Rodríguez-Rodríguez, M. Vázquez-Carrera, org/10.1016/j.aca.2017.06.049. J. Pizarro-Delgado, P. Giménez-Gómez, Á. Merlos, L. Terés, F. Serra-Graells, [26] S.S. Nasrollahi, S.S.H. Davarani, H.R. Moazami, Impedometric investigation of salt C. Jiménez-Jorquera, C. Domínguez, X. Muñoz-Berbel, Reconfigurable multiplexed effects on electromembrane extraction: Practical hints for pH adjustment, point of Care System for monitoring type 1 diabetes patients, Biosens Bioelectron. Electrochim Acta. 296 (2019) 355–363, https://doi.org/10.1016/j. 136 (2019) 38–46. electacta.2018.11.037. [10] A.F.D. Cruz, N. Norena, A. Kaushik, S. Bhansali, A low-cost miniaturized [27] E. Santigosa-Murillo, X. Muñoz-Berbel, S. Maspoch, M. Muñoz, M. Ramos-Payán, potentiostat for point-of-care diagnosis, Biosens Bioelectron. 62 (2014) 249–254, Impedance model for voltage optimization of parabens extraction in an https://doi.org/10.1016/j.bios.2014.06.053. electromembrane millifluidic device, J Chromatogr A. 1625 (2020) 461270. [11] P. Cui, S. Wang, Application of microfluidic chip technology in pharmaceutical [28] G.D. Alkimin, A.M.V.M. Soares, C. Barata, B. Nunes, Evaluation of ketoprofen analysis: A review, J Pharm Anal. 9 (2019) 238–247, https://doi.org/10.1016/j. toxicity in two freshwater species: Effects on biochemical, physiological and jpha.2018.12.001. population endpoints, Environmental Pollution. 265 (2020) 114993. [12] M. Sonker, V. Sahore, A.T. Woolley, Recent advances in microfluidic sample [29] J.K. Gierse, C.M. Koboldt, M.C. Walker, K. Seibert, P.C. Isakson, Kinetic basis for preparation and separation techniques for molecular biomarker analysis: A critical selective inhibition of cyclo-oxygenases, Biochemical Journal. 339 (3) (1999) review, Anal Chim Acta. 986 (2017) 1–11, https://doi.org/10.1016/j. 607–614. aca.2017.07.043. [30] P. Giménez-Gómez, M. Gutiérrez-Capitán, F. Capdevila, A. Puig-Pujol, C. Jiménez- [13] A. Shahvar, D. Shamsaei, M. Saraji, N. Arab, S. Alijani, Microfluidic-based liquid- Jorquera, C. Fernández-Sánchez, Compact analytical flow system for the liquid microextraction in combination with smartphone-based on-chip detection simultaneous determination of l-lactic and l-malic in red wines, Sci Rep. 10 (2020), for the determination of copper in biological, environmental, and food samples, https://doi.org/10.1038/s41598-020-76502-7. Microchemical Journal. 160 (2021) 105655. [31] M.P. di Cagno, F. Clarelli, J. Vabenø, C. Lesley, S.D. Rahman, J. Cauzzo, [14] K. Sato, M. Tokeshi, T. Sawada, T. Kitamori, Molecular transport between two E. Franceschinis, N. Realdon, P.C. Stein, Experimental Determination of Drug phases in a microchannel, Analytical Sciences. 16 (2000) 455–456, https://doi. Diffusion Coefficients in Unstirred Aqueous Environments by Temporally Resolved org/10.2116/analsci.16.455. Concentration Measurements, Mol Pharm. 15 (2018) 1488–1494, https://doi.org/ [15] N.J. Petersen, H. Jensen, S.H. Hansen, S.T. Foss, D. Snakenborg, S. Pedersen- 10.1021/acs.molpharmaceut.7b01053. Bjergaard, On-chip electro membrane extraction, Microfluid Nanofluidics. 9 (2010) [32] F. Lagrange, F. Pehourcq, B. Bannwarth, J.J. Leng, M.C. Saux, Passage of S-(+)- and 881–888, https://doi.org/10.1007/s10404-010-0603-6. R-(-)-ketoprofen across the human isolated perfused placenta, Fundam Clin [16] M.D. Ramos Payán, H. Jensen, N.J. Petersen, S.H. Hansen, S. Pedersen-Bjergaard, Pharmacol. 12 (1998) 286–291, https://doi.org/10.1111/j.1472-8206.1998. Liquid-phase microextraction in a microfluidic-chip - High enrichment and sample tb00956.x. clean-up from small sample volumes based on three-phase extraction, Anal Chim [33] K. Krabbenhøft, J. Krabbenhøft, Application of the Poisson-Nernst-Planck Acta. 735 (2012) 46–53, https://doi.org/10.1016/j.aca.2012.05.023. equations to the migration test, Cem Concr Res. 38 (2008) 77–88, https://doi.org/ [17] M. Ramos-Payan, S. Maspoch, A. Llobera, A simple and fast Double-Flow 10.1016/j.cemconres.2007.08.006. microfluidic device-based liquid-phase microextraction (DF-µLPME) for the 10