Paper-Based All-Solid-State Ion-Sensing Platform with a Solid Contact

Nov 21, 2017 - We report the design, structure, and performance of a planar paper-based ion-sensing platform that utilizes colloid-imprinted mesoporou...
0 downloads 0 Views 7MB Size
This article is made available for a limited time sponsored by ACS under the ACS Free to Read License, which permits copying and redistribution of the article for non-commercial scholarly purposes.

Article Cite This: ACS Appl. Nano Mater. 2018, 1, 293−301

www.acsanm.org

Paper-Based All-Solid-State Ion-Sensing Platform with a Solid Contact Comprising Colloid-Imprinted Mesoporous Carbon and a Redox Buffer Jinbo Hu, Wenyang Zhao, Philippe Bühlmann,* and Andreas Stein* Department of Chemistry, University of Minnesota, 207 Pleasant Street SE, Minneapolis, Minnesota 55455, United States S Supporting Information *

ABSTRACT: We report the design, structure, and performance of a planar paper-based ion-sensing platform that utilizes colloid-imprinted mesoporous (CIM) carbon as a solid contact, with a redox buffer as the internal reference. This device contains an all-solid-state ion-selective electrode and an all-solid-state reference electrode that are integrated into the paper substrate with a symmetrical cell design. To ensure calibration-free sensor operation, each interfacial potential within the device is well-defined by the use of a redox buffer added to the sensing and reference membranes that controls the interfacial potentials at the CIM carbon/sensing membrane and CIM carbon/reference membrane interfaces. Two types of redox buffers were evaluated for this purpose, i.e., one based on the tetrakis(pentafluorophenyl)borate salts of cobalt(II/III) tris(4,4′-dinonyl-2,2′-bipyridyl) and one consisting of 7,7,8,8-tetracyanoquinodimethane and its corresponding anion radical. The feasibility of the design was demonstrated with aqueous KCl solutions. By design, the device only needs one droplet of sample, and it does not need any supply reagents or sensor pretreatment (i.e., conditioning and calibration) to function. KEYWORDS: paper-based sensor, mesoporous carbon, all-solid-state ion-selective electrode, redox buffer, potentiometry, chloride



INTRODUCTION Because of its abundance, low cost, and suitability for mass manufacturing, paper has recently been extensively explored as a substrate to fabricate disposable analytical platforms with built-in microfluidic channels defined by hydrophobic barriers.1−5 Compared to their conventional counterparts that usually rely on expensive consumables and frequent maintenance, paper-based analytical devices are more affordable and simpler to use, making them very attractive for clinical applications, especially in developing countries with limited resources. Existing paper-based analytical devices rely on different sensing techniques for various analytes. While the colorimetric methods bear the advantage of simple data visualization and interpretation,6 electrochemical detection is insensitive to color interference and is generally more accurate in terms of quantitative analysis.7−9 Ion sensing has long been an interest in analytical chemistry and is of vital importance in many fields, such as clinical analysis, process management, and environmental analysis.10−14 With the growing demand for point-of-care and in-field testing, increased research efforts have focused on miniaturized ionsensing systems with small sample volumes, simple operation, reduced cost, and increased compatibility with mass manu© 2017 American Chemical Society

facturing techniques, such as printing. Paper and nanostructured solid-contact transducing materials are frequently used as key components in these devices.4,15−18 Depending on the specific sensor design, paper provides different functions for the ion sensors reported in the literature. Paper has long been used as a substrate to mechanically support the other components of the sensor. Disposable strip-type ion sensors with paper as a key component were reported in the 1990s, where a filter paper coated with silver was sandwiched between two insulating layers to mechanically support and electrically connect an ion-selective electrode (ISE) membrane.19,20 This sensor design was recently combined with nanostructured or conductive polymer-based solid-contact transducing materials to construct high-performance strip-type ion-sensing devices. For example, carbon nanotubes21−23 or graphene24 were deposited onto a paper substrate to serve as an ion-to-electron transducer, and Li+ measurements in whole blood were demonstrated with such a sensor design.22 With an additional hydrophobic layer of sputtered gold and poly(3Received: November 4, 2017 Accepted: November 21, 2017 Published: November 21, 2017 293

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials

mesophase pitch carbon precursor, which is essential to constructing electrodes with high electrode-to-electrode reproducibility and excellent resistance to possible oxygen interference.32,33 CIM carbon is produced in powder form, which makes CIM carbon compatible with the mass production methods that are needed to fabricate paper-based sensors on a large scale, such as by inkjet printing. The CIM carbon-based ISE and reference electrodes are integrated into the paper substrate with a symmetrical cell design. Unlike our previously reported device that employed a conventional Ag/AgCl/KCl inner reference system,29 a redox buffer is doped into the ISE and reference membranes to control the interfacial potentials at the CIM carbon/ISE membrane and CIM carbon/reference membrane interfaces, serving as an internal reference system. For this purpose, two types of redox buffers were evaluated, i.e., a cobalt(II/III) tris(4,4′-dinonyl-2,2′-bipyridyl) ([Co II/III (C 9 ,C 9 -bipy) 3 ](TPFPB)2/3)34 redox buffer and the redox couple consisting of 7,7,8,8,-tetracyanoquinodimethane (TCNQ) and the corresponding anion radical.35,36 As a proof of concept, paper-based all-solid-state Cl−-sensing devices were constructed and tested with aqueous solutions. In our previous work, we reported the concept of CIM carbon-based reference electrodes on the paper substrate with the [CoII/III(C9,C9-bipy)3](TPFPB)2/3 redox buffer.33 Here we extend our work by constructing paperbased devices that comprise both CIM carbon-based sensing and reference electrodes, evaluate a second redox buffer based on TCNQ, discuss the requirements for the redox buffer, report the characterization of the structural features of the device using scanning electron microscopy (SEM), and explore the applications of these devices in biological samples.

octylthiophene) on top of a conductive paper coated with carbon nanotubes, nanomolar detection limits for Cd2+, Ag+, and K+ were achieved.23 Besides providing mechanical support, paper can also be used as a disposable microfluidic sampling tool to replace conventional sample holders (e.g., beakers or tubing) because it transports fluids by capillary forces. It was reported that paper can be sandwiched between an all-solidstate ISE and a reference electrode to transport the sample solutions from other areas to the sensing electrodes.25,26 With the addition of a complexing agent into a pretreated microfluidic paper substrate, a separation system was achieved in which the transport of interfering ions was significantly slowed by complexation, and thus only the target ion could reach the sensing electrodes.27 For controlled microfluidic sampling, hydrophobic barriers can be deposited onto the paper substrate to define microfluidic channels. For example, a threedimensional paper-based ion sensor was developed with a conventional reusable poly(vinyl chloride) (PVC) sensing membrane sandwiched between two disposable paper substrates containing sample and reference zones defined by printed wax. With careful sensor assembly and calibration, various ions could be detected in aqueous sample solutions.28 Although attractive features, such as low detection limits and suitability for biological samples, were demonstrated with the aforementioned state-of-the-art, paper-based ion-sensing devices, those devices all require cumbersome pretreatment protocols including sensor conditioning and calibration, which may impede their practical use, especially when they are used by less skilled operators. To address this issue, we previously introduced a conditioning- and calibration-free paper-based ion-sensing platform that is designed for singleuse operation.29 The device is based on a conventional potentiometric cell that is imbedded in the paper substrate, with sample and reference zones defined by a polyurethane hydrophobic barrier. To provide well-defined reference potentials, an inner reference system composed of stencilprinted Ag/AgCl electrodes and KCl reference electrolytes was introduced. Because of the symmetrical cell design and precise control of each interfacial potential, the response of this ionsensing platform can be predicted quantitatively, and the device can be used to detect ion concentrations in undiluted blood serum without prior sensor calibration.29 The use of the Ag/ AgCl/KCl inner reference system, however, complicates the measuring protocol by requiring the supply and application of a liquid KCl reference electrolyte. Therefore, it is desirable to eliminate the KCl reference electrolyte, further simplifying the sensor operation. Herein, we report a paper-based all-solid-state ion-sensing device that only requires one droplet of sample and requires neither sensor pretreatment (such as conditioning and calibration) nor externally added reagents (such as a reference electrolyte). This device relies on an all-solid-state ISE and an all-solid-state reference electrode that are both based on colloid-imprinted mesoporous (CIM) carbon. CIM carbon is a novel solid contact material that comprises open and interconnected 20−30-nm-diameter mesopores with a bicontinuous carbon and pore space.30,31 This unique porous structure provides a large interfacial contact area between the CIM carbon and the ion-sensing phase with which the carbon is impregnated, thus creating a high double-layer interfacial capacitance and leading to high sensor potential stability. Also, CIM carbon exhibits a low number of redox-active surface functionalities because it is prepared from a high-purity



EXPERIMENTAL SECTION

Materials. Reagents were obtained from the following sources: Fumion FAA-3 ionomer anion exchanger from FuMA-Tech GmbH (Bietigheim-Bissingen, Germany), the ionic liquid 1-methyl-3octylimidazolium bis(trifluoromethylsulfonyl)imide [C 8 min + ][C1C1N−] from IoLiTec (Tuscaloosa, AL), TCNQ, tetraethylammonium tetrafluoroborate, o-nitrophenyl octyl ether (o-NPOE), highmolecular-weight PVC from Sigma-Aldrich (St. Louis, MO), Tecoflex SG-80A polyurethane from Thermedic Polymer Products (Woburn, MA), and Autonorm freeze-dried blood serum from SERO (Stasjonsveien, Norway). All chemicals were used as received without further purification. Deionized water was purified to a resistivity of 18.2 MΩ/cm with a Milli-Q PLUS reagent-grade water system (Millipore, Bedford, MA). CIM carbon,30−32 the K+ salt of the TCNQ− radical anion (KTCNQ),37 and the redox couple consisting of [CoII(C9,C9-bipy)3](TPFPB)2 and [CoIII(C9,C9-bipy)3](TPFPB)334 were prepared as previously reported. Precursor Solutions of Sensing and Reference Membranes. Fumion FAA-3 ionomer anion exchanger was used as the sensing membrane and loaded with Cl− counterions using a previously reported procedure.29 A total of 150 mg of Cl−-loaded Fumion FAA-3 ionomer was dissolved in 2 mL of methanol under magnetic stirring. Precursor solutions for reference membranes were prepared by dissolving 60 mg of the ionic liquid [C8min+][C1C1N−], 120 mg of PVC as a polymeric matrix, and 120 mg of o-NPOE as a plasticizer in 2 mL of anhydrous tetrahydrofuran. The redox buffers containing 1.4 mmol/kg of [CoII/III(C9,C9-bipy)3](TPFPB)2/3 or 3 mmol/kg of TCNQ/KTCNQ were added into both the sensing membrane and reference membrane precursor solutions. Fabrication of a Paper-Based All-Solid-State Ion-Sensing Platform. Paper-based sample zones and microfluidic channels were defined by patterning polyurethane lines that penetrated the whole thickness of filter paper pieces (Whatman grade 589/2 white ribbon, GE Healthcare, Chicago, IL). Approximately 2.5 g of polyurethane was 294

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials dissolved in 40 mL of tetrahydrofuran, and this solution was then applied to both sides of the filter paper using a glass capillary tube, forming polyurethane barriers against hydrophilic samples approximately 2 mm in width. To form sensing and reference membranes of approximately 1 mm width, a 10 μL microcapillary was used to apply the corresponding precursor solutions to the filter paper. Where necessary, the viscosity of the precursor solutions was adjusted by dilution with additional aliquots of the corresponding solvents for better penetration into the filter paper. To ensure full penetration of the membrane components through the entire thickness of the paper, the precursor solutions were applied to both sides of the paper twice each. The CIM carbon-sensing membrane suspension was prepared by ultrasonicating 75 mg of CIM carbon in 1 mL of the sensing membrane precursor solution, including the redox buffer, for 30 min. In a similar way, CIM carbon reference membrane suspensions were prepared by ultrasonicating for 30 min 60 mg of CIM carbon in 1 mL of the precursor solution containing the reference membrane components, including the redox buffer. The resulting sensing and reference suspensions were then applied to paper using a capillary to form a homogeneous mixture of CIM carbon and the corresponding membranes, with an effort to maximize the contact area between this homogeneous mixture and the membrane. Electrochemical Measurements. The electrode potentials were measured with an EMF 16 voltmeter (input impedance 10 TΩ) controlled by EMF Suite 1.03 software (Lawson Labs, Malvern, PA). Two copper alligator clips were used to connect the two CIM carbonbased sensing and reference electrodes to the voltmeter. A series of samples containing different concentrations of Cl− and K+ were obtained by the sequential dilution of a KCl solution. A series of blood serum samples with various Cl− concentrations were prepared by adding small amounts of a 0.8 M KCl solution into 10-fold-diluted blood serum. The activity coefficients were calculated according to a two-parameter Debye−Hückel approximation.38 After the samples were applied to the sensing zones, it took approximately 10 s for the solutions to be wicked onto the sensing and reference membranes. Then, the electromotive force (emf) response over the following 30 s was measured. Cyclic voltammograms of TCNQ and KTCNQ were obtained at room temperature with a CHI600C potentiostat (CH Instruments, Austin, TX). A three-electrode setup was used, with a gold electrode as the working electrode, a platinum wire as the counter electrode, and a silver wire in 10 mM AgNO3/acetonitrile as a nonaqueous reference electrode with a CoralPor glass frit for separation from the sample.39 A 0.1 M tetraethylammonium tetrafluoroborate solution with acetonitrile as the solvent was used as the supporting electrolyte solution.

Figure 1. (a) Photograph of an all-solid-state paper-based ion-sensing device with a CIM carbon-based ISE, a CIM carbon-based reference electrode, and a microfluidic sample zone defined by polyurethane. (b) Schematic representation of all relevant interfaces in a paper-based Cl− sensor with a commercial anion-exchanger Fumion FAA-3 ionomer film as the sensing membrane and an ionic-liquid-doped and plasticized PVC film as the reference membrane. Both membranes are doped with a redox buffer containing a redox couple, shown as “Red” and “Ox”. (c) Electrical potential profile across the paper-based all-solid-state ion-sensing platform, with the only sample-dependent interfacial potential being the phase-boundary potential at the sensing membrane/sample interface.

The ion-sensing platform utilized a polyurethane-based hydrophobic barrier that defined a sample zone. In this sensor design, paper served as both a microfluidic sampling tool and a supportive porous matrix into which the sensing components were embedded.42 To ensure single-use operation and calibration-free ion sensing, a high reproducibility of the standard electrode potential, E°, for multiple devices is required.15 This can be achieved by precisely defining each interfacial potential within the device (see Figure 1b). To this end, a redox buffer (i.e., [CoII/III(C9,C9-bipy)3](TPFPB)2/3 or TCNQ/KTCNQ) was doped into both the sensing and reference membranes to control the interfacial potentials at the CIM carbon/sensing membrane interface (Δφ1) and the CIM carbon/reference membrane interface (Δφ4). A redox buffer contains both the oxidized and reduced species of a redox couple. Similar to a pH buffer whose pH can be controlled by an acid and its conjugate base, a redox buffer controls the electrical potential to a value that is determined by its standard redox potential and the ratio of the oxidized and reduced species of the redox couple, as predicted by the Nernst equation:43 a RT E = E°′ + ln ox nF ared (1)



RESULTS AND DISCUSSION Design of the Paper-Based All-Solid-State Ion-Sensing Platform. The ion-sensing platform was fabricated by integrating a CIM carbon-based ISE and a CIM carbon-based reference electrode onto the paper substrate (Figure 1a). The solid contacts comprising CIM carbon and redox buffer were contacted with the corresponding sensing and reference membranes, which were applied separately and embedded in the paper. For a demonstration, the commercial high-capacity ionomer Fumion FAA-3 was used as the anion exchanger for Cl− sensing, and the ionic liquid [C8min+][C1C1N−] was doped into the reference membrane to provide a sample-independent reference potential. The hydrophilic Fumion FAA-3 sensing membranes are considered to be suitable for clinical Cl− sensing because they exhibit much lower interference from hydrophobic anions (e.g., Br− and SCN−) than more hydrophobic ion exchangers.40 Their hydrophilicity also decreases the interference from biofouling caused by lipids,40 and because of their high ion-exchange capacity, they have a high resistance to Donnan failure at high ion concentrations.41 295

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials

Figure 2. SEM images of the ion-sensing platform. Top views of (a) the paper substrate, (b) the device showing the interface between the sensing membrane and the sample zone, and (c) the device showing the interface between the CIM carbon-based sensing electrode and the sensing membrane. Magnified views of (d) the CIM carbon-based sensing electrode and (e) the uncoated CIM carbon, showing the mesopores. (f) Crosssectional view of the ion-sensing platform, showing the sensing membrane-infiltrated paper substrate with the CIM carbon-based sensing electrode.

where E is the electrical potential of the redox buffer, E°′ is the standard reduction potential of the redox couple, R is the ideal gas constant, F is the Faraday constant, T is the temperature, n is the number of moles of electrons transferred in the redox reaction, and aox and ared are the activities of the oxidized and reduced species of the redox couple, respectively. A few requirements must be fulfilled for an adequate redox buffer. First and foremost, a redox buffer should have a welldefined redox potential. Conductive polymers usually exhibit high redox capacitance, but a continuum of less well-defined redox potentials that can be related to the inhomogeneity in crystallinity, conformation, film morphology, and so on.15 Therefore, compared to molecular redox species that exhibit well-defined redox potentials, conductive polymers are less attractive for redox buffer applications. Second, both the oxidized and reduced species of the redox buffer should be chemically and electrochemically stable with reversible redox activity. In this regard, the standard reduction potential of the redox couple should be close to 0 V (vs Ag/AgCl) so that the redox couple cannot be easily oxidized or reduced by redoxactive interferents that are present in the system. For example, although ferrocene is a well-studied redox molecule with a suitable standard reduction potential of 0.40 V (vs SHE), it cannot be used for redox buffer applications because its oxidized form, the ferrocenium ion, is not stable under certain circumstances. It was reported that Cl− ions can induce the degradation of ferrocenium ions into nonreducible FeCl4− in both organic electrolytes and ferrocene-doped ISE membranes, which can lead to irreversible redox chemistry.44 Furthermore, the redox couple should also exhibit fast electron-transfer rates, thus enabling adequate ion-to-electron transduction when the redox buffer is used as the transducer layer in an all-solid-state ISE. The phase-boundary potentials across the other interfaces within the paper-based all-solid-state ion-sensing devices should

also be well-defined. In this work, the interface between the sensing membrane and the sample (Δφ2) was controlled by partitioning of the primary ion (i.e., Cl−) between the Fumion FAA-3 ionomer film and sample, resulting in an interfacial potential dependent on the Cl− concentration in the sample that can be described quantitatively by the classical phaseboundary-potential model.45 Moreover, the reference membrane was doped with an ionic liquid (i.e., [C8min+][C1C1N−]) that can leach into the sample on a slow but continuous basis, thus providing a constant but sample-independent reference potential at the sample/reference membrane interface (Δφ3).46,47 Therefore, the only sample-dependent interfacial potential within the device was the phase-boundary potential at the sensing membrane/sample interface (Δφ2), with all other interfacial potentials being sample-independent and welldefined. The measured overall emf was the electrical potential difference between the two CIM carbon-based electrodes, which was the sum of all interfacial potentials within the cell (Figure 1c). By such a design, the correlation between the measured emf and the concentration of Cl− can be established. Structural Features of the Paper-Based All-Solid-State Ion-Sensing Platform. The structural features of the ionsensing platform were characterized by SEM. Figure 2a shows the nonordered pore structure of the paper substrate made from hydrophilic cellulose fibers, which can be coated and penetrated by the sensing membranes to form well-defined sensing interfaces (Figure 2b,f). Unlike in a conventional ISE, in which an inner Ag/AgCl reference half-cell is in contact with the KCl reference electrolyte solution, the sensing membrane was contacted on the side opposite to the sample with a CIM carbon-based electrode (Figure 2c). The interfacial potential at that interface is stabilized by both the redox reaction of the redox buffer and the electrical double layer at the CIM carbon/ sensing membrane interface.32,34 A magnified image of the CIM carbon-based electrode shows a particle size of approximately 296

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials 10 μm (Figure 2d), and these CIM carbon particles were bound together by the sensing or reference membranes. The highly porous surface of CIM carbon can be observed in the high-magnification image in Figure 2e, and the pore size of CIM carbon can be tuned by using Ludox colloidal silica spheres with different particle sizes.30,31 When CIM carbon was mixed with the precursor solution of the sensing membrane under sonication, the sensing components infiltrated the highly interconnected mesopores of CIM carbon to form a bicontinuous, ionically, and electronically conducting structure with a large interfacial contact area and high double-layer capacitance.32 Therefore, effective ion-to-electron transduction was achieved. A cross-sectional view of the device is given in Figure 2f. It can be seen that the sensing membrane penetrated the entire thickness of the paper substrate, and a good contact was formed between the sensing membrane and the CIM carbon transduction layer. Incorporation of a [Co II/III(C 9,C 9-bipy) 3 ](TPFPB) 2/3 Redox Buffer. [Co I I (C 9 ,C 9 -bipy) 3 ](TPFPB) 2 and [CoIII(C9,C9-bipy)3](TPFPB)3 are hydrophobic cobalt complexes with bipyridine-based ligands. They are chemically stable under ambient conditions and exhibit a standard reduction potential of approximately −0.35 V (vs Ag/AgCl) with fast electron-transfer rates.34,48 Previously, redox buffers based on [CoII/III(C9,C9-bipy)3](TPFPB)2/3 were doped into ISE and reference membranes to construct all-solid state potentiometric ISEs and reference electrodes with high electrode-to-electrode E° reproducibility and minimal oxygen interference because of the high redox buffer capacity.34,49 The combination of [CoII/III(C9,C9-bipy)3](TPFPB)2/3 and CIM carbon led to highly reproducible all-solid-state ISEs with a standard deviation of E° of less than 1 mV.32 Therefore, the redox buffer comprising [CoII/III(C9,C9-bipy)3](TPFPB)2/3 was also doped into both the sensing and reference membranes in the current paper-based ion-sensing platform with CIM carbon electrodes. The potentiometric Cl− response of the paper-based allsolid-state ion-sensing platform was evaluated with aqueous KCl sample solutions of different concentrations. For each measurement, 20 μL of the sample was applied to the sample zone of the device, and the potential difference between the two CIM carbon-based electrodes was measured once the sample reached the sensing and reference membranes. Neither prior sensor calibration nor conditioning was needed. Because these devices were designed for single-use operations, a separate device was employed for each data point in the calibration curve. As presented in Figure 3, the paper-based allsolid-state Cl− sensors exhibited a response slope of −60.6 mV/ decade and an R2 value of 0.991, which is within error consistent with the theoretically predicted Nernstian response. The observed linear range of the device was from 10−1.1 to 10−3.1 M, with a lower detection limit of 10−3.14 M. This lower detection limit was very close to that of our previously reported paper-based Cl−-sensing device (i.e., 10−3.10 M)29 and appears to be limited by the high-capacity anion-exchange film used as the sensing membrane.41 Unlike our previously reported paperbased Cl−-sensing device that utilized an Ag/AgCl/KCl inner reference system,29 the current all-solid-state device does not need a reference electrolyte (i.e., KCl solution) to function. An initial evaluation of the new ion sensors with 10-fold-diluted blood serum samples revealed a good Nernstian response slope of −59.9 mV/decade but a relatively low linearity with an R2 value of 0.628. We assume that the observed reduced

Figure 3. Potentiometric Cl− calibration curve of the ion-sensing platform with a redox buffer consisting of [CoII/III(C9,C9-bipy)3](TPFPB)2/3. The response was obtained with KCl solutions. Each data point corresponds to one individual device.

reproducibility is caused by the accelerated loss of the redox buffer into the lipophilic biological samples,34 and we are currently developing new redox buffers that will eliminate this problem. Conventional-Size All-Solid-State ISEs with TCNQ/ KTCNQ as a Redox Buffer. Besides [CoII/III(C9,C9-bipy)3](TPFPB)2/3, a redox couple consisting of TCNQ and its ionradical salt KTCNQ was also evaluated as a redox buffer. TCNQ has long been of interest in molecular electronics because of its ability to accept electrons to form simple and stable charge-transfer complexes and ion-radical salts.37,50−53 TCNQ ion-radical salts were used in potentiometric ion sensors as early as the 1970s, where they were employed as solid-state ion-selective membranes based on the selective distribution of the primary ion (e.g., Ag+) between the sample solution and TCNQ ion-radical salts (e.g., AgTCNQ).54 Recent publications by Paczosa-Bator et al. report outstanding electrochemical performance of all-solid-state ISEs based on TCNQ, NaTCNQ, or a mixture of the two.35,36 They reported that a high electrode E° reproducibility was achieved with a standard deviation of E° as low as 1.4 mV when TCNQ was used as the ion-to-electron transducer between a glassy carbon electrode and a K+-ISE membrane.35 In follow-up work,36 TCNQ and NaTCNQ were mixed with various carbon nanomaterials (i.e., graphene, carbon nanotubes, ordered mesoporous carbon CMK-3, and carbon black) with a 1:1:1 weight ratio to serve as the solid contact. The resulting all-solidstate ISEs all exhibited good electrode E° reproducibility with a standard deviation of E° of less than 4.0 mV. The reported potential stability is remarkable as well, with a drift of less than 10 μV/h over a measuring period of 72 h.36 Notably, in these two papers, a high E° reproducibility was observed after the electrodes had been conditioned in aqueous solutions for 24 h. This is very different from the all-solid-state K+-ISEs based on the [CoII/III(C9,C9-bipy)3](TPFPB)2/3 redox buffer, which exhibited significantly decreased E° reproducibility upon electrode conditioning because of the loss of redox buffer into the aqueous solution in contact with the redox-buffercontaining membrane (i.e., the standard deviation of E° increased from 0.7 mV after 1 h of conditioning to 16.3 mV after 24 h of electrode conditioning).34 The reported data suggest that TCNQ and its ion-radical salt may not suffer as much from the leaching problem as the [CoII/III(C9,C9297

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials bipy)3](TPFPB)2/3 redox buffer does and appear to be an effective redox buffer pair. Therefore, the use of TCNQ/ TCNQ− as a redox buffer was also used tested in this work. TCNQ is known to exhibit two reversible one-electron redox processes, in which TCNQ is first reduced to TCNQ•− and then further reduced to TCNQ2−.55,56 Cyclic voltammograms for TCNQ and KTCNQ are shown in Figure S1. Because the observed standard potential of the TCNQ0/•− redox process is close to 0 V, both the oxidized and reduced species within the redox couple are expected to be quite stable under ambient conditions. The voltammograms show that the TCNQ/ KTCNQ couple is a good candidate for the preparation of a redox buffer not only because of its chemical and electrochemical stability but also because of the fast and reversible electron transfer. To further evaluate the effectiveness of E° control of the TCNQ/KTCNQ redox buffer in potentiometric ion sensing, 3 mmol/kg each of TCNQ and KTCNQ in a 1:1 molar ratio were doped into sensing membranes to construct conventionalsize (nonplanar) ISEs with a CIM carbon-based solid contact. These bulk electrodes comprised CIM carbon films separating an underlying gold electrode from a Fumion FAA-3 ionomer anion-exchanger sensing membrane (i.e., the same electrode configuration as that reported previously;32,33 see Figure S2). The sensing membranes were doped with TCNQ/KTCNQ. To study the effect of electrode conditioning on the E° reproducibility, Cl− calibrations were performed without electrode conditioning (i.e., no prior contact with aqueous solutions), and after conditioning, the electrodes were placed in a 1 mM KCl solution for 24 h. When the electrodes were tested with aqueous KCl solutions without conditioning (Figure 4a), a linear response was obtained with a close to Nernstian slope of −53.9 ± 1.2 mV/decade and a fairly reproducible E° of −47.0 ± 4.3 mV (n = 5). This E° reproducibility is in good agreement with the literature that reports the use of TCNQ/NaTCNQ with other nanostructured carbon materials to construct all-solid-state ISEs. Specifically, standard deviations of E° of 2.4 mV for graphene, 3.7 mV for carbon nanotubes, 3.2 mV for carbon black, and 3.2 mV for the ordered mesoporous carbon CMK-3 were reported.36 However, after conditioning of the electrodes for 24 h, larger electrode-to-electrode variations were observed in this work, with the standard deviation of E° increasing to 14.0 mV (Figure 4b). This observation is very similar to that of all-solid-state K+-ISEs based on the [CoII/III(C9,C9-bipy)3](TPFPB)2/3 redox buffer, whose standard deviation of E° increased from 0.7 to 16.3 mV after conditioning of the electrodes in an aqueous solution for an additional 23 h.34 The diminished electrode-to-electrode E° reducibility was also accompanied by an E° drift of 32.1 mV (i.e., from −47.0 mV to −14.9 mV), indicating that the ratio of TCNQ/KTCNQ present in the sensing membrane may have changed. The all-solid-state ISEs with the TCNQ/KTCNQ redox buffer were also tested with 10-fold-diluted blood serum samples, where similar results were observed. Compared to aqueous samples, 10-fold-diluted blood serum samples are more lipophilic, which may accelerate leaching of the redox buffer into the sample. Figure 5 presents Cl− calibration curves of CIM carbon-based electrodes with the TCNQ/KTCNQ redox buffer, as obtained in 10-fold-diluted blood serum samples. It can be seen that, without any electrode conditioning (Figure 5a), the electrodes exhibited a sub-Nernstian slope of −39.4 ± 3.4 mV/decade and a good electrode E° reducibility of

Figure 4. Potentiometric Cl− calibration curves of CIM carbon-based all-solid-state bulk electrodes with a TCNQ/KTCNQ redox buffer doped into the sensing membrane (n = 5). (a) Response to KCl solutions without electrode conditioning. (b) Response to KCl solutions after conditioning of the electrodes in a 1 mM KCl solution for 24 h.

−33.8 ± 4.4 mV. After 24 h of electrode conditioning (Figure 5b), the response slope improved to −53.8 ± 1.0 mV/decade, but the electrode E° reproducibility decreased to 10.4 ± 19.0 mV (n = 6) with a drift of 44.2 mV. It is likely that the reduced electrode-to-electrode E° reproducibility was caused by the leaching of TCNQ/KTCNQ into the sample. Although these results suggest that TCNQ/KTCNQ is not suitable for application as a redox buffer in potentiometric ion sensors for long-term measurements, it may still be used for single-use devices, where the device is typically only contacted with the sample over a very short period of time. Because the total measuring time of the paper-based all-solid-state ion-sensing devices is 30 s, a redox buffer containing TCNQ/KTCNQ can be used for such disposable paper-based devices. Paper-Based All-Solid-State Ion-Sensing Platform with a TCNQ/KTCNQ Redox Buffer. The TCNQ/KTCNQ couple was used to dope the sensing and reference membranes (3 mmol/kg) of the paper-based all-solid-state ion-sensing platform. The potentiometric Cl− calibration curve of the resulting devices is shown in Figure 6, in which a close to Nernstian response was acquired with a slope of −63.6 mV/ decade and an R2 value of 0.984. When the devices were tested with 10-fold-diluted blood serum samples, the Nernstian response appears to be preserved, but with a lower R2 value of 0.856. Combined with the results from a [CoII/III(C9,C9bipy)3](TPFPB)2/3 redox buffer, these results demonstrate the feasibility of the current sensor design but also show that a 298

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials

±5%, which corresponds to variation in the emf value of approximately 1.1 mV. Covalent attachment of the redox buffer to CIM carbon or the sensing membrane may be promising solutions to achieve such high device-to-device reproducibility.



CONCLUSIONS This paper reports the design, structure, and performance of a disposable paper-based all-solid-state ion-sensing platform that is based on a CIM carbon solid contact and a redox buffer. The transduction mechanism of the sensor is based on a design in which each phase-boundary potential is well-defined and the structural features of the device are fully characterized. Unlike previously reported paper-based ion sensors, this all-solid-state device does not require any sensor pretreatment or reference electrolyte to function. Also, this planar device is compatible with mass production techniques such as printing. Proof-ofconcept paper-based all-solid-state Cl− sensors were constructed, and the feasibility of the sensor design was demonstrated with aqueous KCl solutions with two types of redox buffers, namely, [CoII/III(C9,C9-bipy)3](TPFPB)2/3 and TCNQ/KTCNQ. For measurements in biological samples, a more robust redox buffer system is still required because it appears that the leaching of the redox buffer limits the longterm stability of the sensors. Covalent attachment of the redox buffer to the sensing membrane or CIM carbon appears to be a promising solution.



ASSOCIATED CONTENT

S Supporting Information *

Figure 5. Potentiometric Cl− calibration curves of CIM carbon-based all-solid-state bulk electrodes with a TCNQ/KTCNQ redox buffer doped into the sensing membrane (n = 6). (a) Response to 10-folddiluted blood serum samples without electrode conditioning. (b) Response to 10-fold-diluted blood serum samples after conditioning of the electrodes in a 1 mM KCl solution for 24 h.

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsanm.7b00151. Cyclic voltammograms for TCNQ and KTCNQ (Figure S1) and schematic diagram of a CIM carbon-based, allsolid-state ISE used to evaluate the effectiveness of the TCNQ/KTCNQ redox buffer (Figure S2) (PDF)



AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. ORCID

Philippe Bühlmann: 0000-0001-9302-4674 Andreas Stein: 0000-0001-8576-0727 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS The University of Minnesota Initiative for Renewable Energy and the Environment supported this work. J.H. thanks the Krogh family for a Lester C. and Joan M. Krogh Fellowship and the Graduate School of the University of Minnesota for a Doctoral Dissertation Fellowship. We thank Carl Schu from Medtronic PLC (industrial fellow of the Industrial Partnership for Research in Interfacial and Materials Engineering, IPRIME, program) for valuable discussions.

Figure 6. Potentiometric Cl− calibration curve of all-solid-state paperbased ion-sensing platform with a TCNQ/KTCNQ redox buffer doped into both the sensing and reference membranes. The response was measured with KCl solutions (solid circle) and 10-fold-diluted blood serum samples (open circle). Each data point is based on one device.



more robust redox buffer system that will not leach out is needed for measurements of the blood samples. For meaningful clinical measurements, a very high electrode reproducibility is required. According to the U.S. Code of Federal Regulations,57 in clinical laboratories the acceptable measuring error for Cl− is

REFERENCES

(1) Martinez, A. W.; Phillips, S. T.; Butte, M. J.; Whitesides, G. M. Patterned Paper as a Platform for Inexpensive, Low-Volume, Portable Bioassays. Angew. Chem., Int. Ed. 2007, 46, 1318−1320.

299

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials (2) Yamada, K.; Henares, T. G.; Suzuki, K.; Citterio, D. Paper-Based Inkjet-Printed Microfluidic Analytical Devices. Angew. Chem., Int. Ed. 2015, 54, 5294−5310. (3) Mahadeva, S. K.; Walus, K.; Stoeber, B. Paper as a Platform for Sensing Applications and Other Devices: A Review. ACS Appl. Mater. Interfaces 2015, 7, 8345−8362. (4) Yamada, K.; Shibata, H.; Suzuki, K.; Citterio, D. Toward Practical Application of Paper-Based Microfluidics for Medical Diagnostics: State-of-the-Art and Challenges. Lab Chip 2017, 17, 1206−1249. (5) Yang, Y.; Noviana, E.; Nguyen, M. P.; Geiss, B. J.; Dandy, D. S.; Henry, C. S. Paper-Based Microfluidic Devices: Emerging Themes and Applications. Anal. Chem. 2017, 89, 71−91. (6) Martinez, A. W.; Phillips, S. T.; Carrilho, E.; Thomas, S. W.; Sindi, H.; Whitesides, G. M. Simple Telemedicine for Developing Regions: Camera Phones and Paper-Based Microfluidic Devices for Real-Time, Off-Site Diagnosis. Anal. Chem. 2008, 80, 3699−3707. (7) Dungchai, W.; Chailapakul, O.; Henry, C. S. Electrochemical Detection for Paper-Based Microfluidics. Anal. Chem. 2009, 81, 5821− 5826. (8) Adkins, J.; Boehle, K.; Henry, C. Electrochemical Paper-Based Microfluidic Devices. Electrophoresis 2015, 36, 1811−1824. (9) Mettakoonpitak, J.; Boehle, K.; Nantaphol, S.; Teengam, P.; Adkins, J. A.; Srisa-Art, M.; Henry, C. S. Electrochemistry on PaperBased Analytical Devices: A Review. Electroanalysis 2016, 28, 1420− 1436. (10) Bakker, E.; Bühlmann, P.; Pretsch, E. Carrier-Based IonSelective Electrodes and Bulk Optodes. 1. General Characteristics. Chem. Rev. 1997, 97, 3083−3132. (11) Bobacka, J.; Ivaska, A.; Lewenstam, A. Potentiometric Ion Sensors. Chem. Rev. 2008, 108, 329−351. (12) Lindner, E.; Gyurcsányi, R. E. Quality Control Criteria for SolidContact, Solvent Polymeric Membrane Ion-Selective Electrodes. J. Solid State Electrochem. 2009, 13, 51−68. (13) Bühlmann, P.; Chen, L. D. Ion-Selective Electrodes With Ionophore-Doped Sensing Membranes. In Supramolecular Chemistry: From Molecules to Nanomaterials; Jonathan, W., Steed, P. A. G., Eds.; John Wiley & Sons, Ltd.: New York, 2012. (14) Bakker, E. Electroanalysis with Membrane Electrodes and Liquid−Liquid Interfaces. Anal. Chem. 2016, 88, 395−413. (15) Hu, J.; Stein, A.; Bühlmann, P. Rational Design of All-Solid-State Ion-Selective Electrodes and Reference Electrodes. TrAC, Trends Anal. Chem. 2016, 76, 102−114. (16) Ding, J.; Li, B.; Chen, L.; Qin, W. A Three-Dimensional Origami Paper-Based Device for Potentiometric Biosensing. Angew. Chem., Int. Ed. 2016, 55, 13033−13037. (17) Walcarius, A. Recent Trends on Electrochemical Sensors Based on Ordered Mesoporous Carbon. Sensors 2017, 17, 1863. (18) Walcarius, A. Mesoporous Materials-Based Electrochemical Sensors. Electroanalysis 2015, 27, 1303−1340. (19) Borchardt, M.; Dumschat, C.; Cammann, k.; Knoll, M. Disposable Ion-Selective Electrodes. Sens. Actuators, B 1995, 25, 721−723. (20) Dumschat, C.; Borchardt, M.; Diekmann, C.; Cammann, K.; Knoll, M. Potentiometric Test Strip. Sens. Actuators, B 1995, 24, 279− 281. (21) Novell, M.; Parrilla, M.; Crespo, G. A.; Rius, F. X.; Andrade, F. J. Paper-Based Ion-Selective Potentiometric Sensors. Anal. Chem. 2012, 84, 4695−4702. (22) Novell, M.; Guinovart, T.; Blondeau, P.; Rius, F. X.; Andrade, F. J. A Paper-Based Potentiometric Cell for Decentralized Monitoring of Li Levels in Whole Blood. Lab Chip 2014, 14, 1308−1314. (23) Mensah, S. T.; Gonzalez, Y.; Calvo-Marzal, P.; ChumbimuniTorres, K. Y. Nanomolar Detection Limits of Cd2+, Ag+, and K+ Using Paper-Strip Ion-Selective Electrodes. Anal. Chem. 2014, 86, 7269− 7273. (24) Ping, J.; Wang, Y.; Fan, K.; Tang, W.; Wu, J.; Ying, Y. HighPerformance Flexible Potentiometric Sensing Devices Using FreeStanding Graphene Paper. J. Mater. Chem. B 2013, 1, 4781−4791.

(25) Szű cs, J.; Gyurcsányi, R. E. Towards Protein Assays on Paper Platforms with Potentiometric Detection. Electroanalysis 2012, 24, 146−152. (26) Cui, J.; Lisak, G.; Strzalkowska, S.; Bobacka, J. Potentiometric Sensing Utilizing Paper-Based Microfluidic Sampling. Analyst 2014, 139, 2133−2136. (27) Ding, J.; He, N.; Lisak, G.; Qin, W.; Bobacka, J. Paper-Based Microfluidic Sampling and Separation of Analytes for Potentiometric Ion Sensing. Sens. Actuators, B 2017, 243, 346−352. (28) Lan, W.-J.; Zou, X. U.; Hamedi, M. M.; Hu, J.; Parolo, C.; Maxwell, E. J.; Bühlmann, P.; Whitesides, G. M. Paper-Based Potentiometric Ion Sensing. Anal. Chem. 2014, 86, 9548−9553. (29) Hu, J.; Stein, A.; Bühlmann, P. A Disposable Planar Paper-Based Potentiometric Ion-Sensing Platform. Angew. Chem., Int. Ed. 2016, 55, 7544−7547. (30) Li, Z.; Jaroniec, M. Colloidal Imprinting: A Novel Approach to the Synthesis of Mesoporous Carbons. J. Am. Chem. Soc. 2001, 123, 9208−9209. (31) Li, Z.; Jaroniec, M. Synthesis and Adsorption Properties of Colloid-Imprinted Carbons with Surface and Volume Mesoporosity. Chem. Mater. 2003, 15, 1327−1333. (32) Hu, J.; Zou, X. U.; Stein, A.; Bühlmann, P. Ion-Selective Electrodes with Colloid-Imprinted Mesoporous Carbon as Solid Contact. Anal. Chem. 2014, 86, 7111−7118. (33) Hu, J.; Ho, K. T.; Zou, X. U.; Smyrl, W. H.; Stein, A.; Bühlmann, P. All-Solid-State Reference Electrodes Based on ColloidImprinted Mesoporous Carbon and Their Application in Disposable Paper-based Potentiometric Sensing Devices. Anal. Chem. 2015, 87, 2981−2987. (34) Zou, X. U.; Zhen, X. V.; Cheong, J. H.; Bühlmann, P. Calibration-Free Ionophore-Based Ion-Selective Electrodes With a Co(II)/Co(III) Redox Couple-Based Solid Contact. Anal. Chem. 2014, 86, 8687−8692. (35) Paczosa-Bator, B.; Pięk, M.; Piech, R. Application of Nanostructured TCNQ to Potentiometric Ion-Selective K+ and Na+ Electrodes. Anal. Chem. 2015, 87, 1718−1725. (36) Pięk, M.; Piech, R.; Paczosa-Bator, B. The Complex Crystal of NaTCNQ−TCNQ Supported on Different Carbon Materials as Ionto-Electron Transducer in All-Solid-State Sodium-Selective Electrode. J. Electrochem. Soc. 2016, 163, B573−B579. (37) Melby, L. R.; Harder, R. J.; Hertler, W. R.; Mahler, W.; Benson, R. E.; Mochel, W. E. Substituted Quinodimethans. II. Anion-radical Derivatives and Complexes of 7,7,8,8-Tetracyanoquinodimethan. J. Am. Chem. Soc. 1962, 84, 3374−3387. (38) Meier, P. C. Two-Parameter Debye-Hückel Approximation for the Evaluation of Mean Activity Coefficients of 109 Electrolytes. Anal. Chim. Acta 1982, 136, 363−368. (39) Mousavi, M. P. S.; Saba, S. A.; Anderson, E. L.; Hillmyer, M. A.; Bühlmann, P. Avoiding Errors in Electrochemical Measurements: Effect of Frit Material on the Performance of Reference Electrodes with Porous Frit Junctions. Anal. Chem. 2016, 88, 8706−8713. (40) Grygolowicz-Pawlak, E.; Crespo, G. A.; Ghahraman Afshar, M.; Mistlberger, G.; Bakker, E. Potentiometric Sensors with Ion-Exchange Donnan Exclusion Membranes. Anal. Chem. 2013, 85, 6208−6212. (41) Ogawara, S.; Carey, J. L.; Zou, X. U.; Bühlmann, P. Donnan Failure of Ion-Selective Electrodes with Hydrophilic High-Capacity Ion-Exchanger Membranes. ACS Sensors 2016, 1, 95−101. (42) Hu, J.; Stein, A.; Bühlmann, P. Miniaturized Potentiometric IonSensing Systems: from Bulk Electrodes to Paper-based Ion-Sensing Devices. The Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Atlanta, GA, 2016. (43) de Levie, R. Redox Buffer Strength. J. Chem. Educ. 1999, 76, 574. (44) Cuartero, M.; Acres, R. G.; Bradley, J.; Jarolimova, Z.; Wang, L.; Bakker, E.; Crespo, G. A.; De Marco, R. Electrochemical Mechanism of Ferrocene-Based Redox Molecules in Thin Film Membrane Electrodes. Electrochim. Acta 2017, 238, 357−367. (45) Bakker, E.; Bühlmann, P.; Pretsch, E. The Phase-Boundary Potential Model. Talanta 2004, 63, 3−20. 300

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301

Article

ACS Applied Nano Materials (46) Yoshimatsu, T.; Kakiuchi, T. Ionic Liquid Salt Bridge in Dilute Aqueous Solutions. Anal. Sci. 2007, 23, 1049−1052. (47) Zhang, T.; Lai, C.-Z.; Fierke, M. A.; Stein, A.; Bühlmann, P. Advantages and Limitations of Reference Electrodes with an Ionic Liquid Junction and Three-Dimensionally Ordered Macroporous Carbon as Solid Contact. Anal. Chem. 2012, 84, 7771−7778. (48) Sapp, S. A.; Elliott, C. M.; Contado, C.; Caramori, S.; Bignozzi, C. A. Substituted Polypyridine Complexes of Cobalt(II/III) as Efficient Electron-Transfer Mediators in Dye-Sensitized Solar Cells. J. Am. Chem. Soc. 2002, 124, 11215−11222. (49) Zou, X. U.; Chen, L. D.; Lai, C.-Z.; Bühlmann, P. Ionic Liquid Reference Electrodes With a Well-Controlled Co(II)/Co(III) Redox Buffer as Solid Contact. Electroanalysis 2015, 27, 602−608. (50) Ferraris, J.; Cowan, D. O.; Walatka, V.; Perlstein, J. H. Electron Transfer in a New Highly Conducting Donor-Acceptor Complex. J. Am. Chem. Soc. 1973, 95, 948−949. (51) Jaeger, C. D.; Bard, A. J. Electrochemical Behavior of Tetrathiafulvalene-Tetracyanoquinodimethane Electrodes in Aqueous Media. J. Am. Chem. Soc. 1979, 101, 1690−1699. (52) Le, T. H.; Nafady, A.; Qu, X.; Bond, A. M.; Martin, L. L. Redox and Acid−Base Chemistry of 7,7,8,8-Tetracyanoquinodimethane, 7,7,8,8-Tetracyanoquinodimethane Radical Anion, 7,7,8,8-Tetracyanoquinodimethane Dianion, and Dihydro-7,7,8,8-Tetracyanoquinodimethane in Acetonitrile. Anal. Chem. 2012, 84, 2343−2350. (53) Talin, A. A.; Centrone, A.; Ford, A. C.; Foster, M. E.; Stavila, V.; Haney, P.; Kinney, R. A.; Szalai, V.; El Gabaly, F.; Yoon, H. P.; Léonard, F.; Allendorf, M. D. Tunable Electrical Conductivity in Metal-Organic Framework Thin-Film Devices. Science 2014, 343, 66− 69. (54) Sharp, M.; Johansson, G. Ion-Selective Electrodes Based on 7,7,8,8-Tetracyanoquinodimethane-Radical Salts. Anal. Chim. Acta 1971, 54, 13−21. (55) Lehmann, M. W.; Evans, D. H. Toward Infinite-Dilution Voltammetry. J. Phys. Chem. B 1998, 102, 9928−9933. (56) Nafady, A.; Bond, A. M.; Bilyk, A.; Harris, A. R.; Bhatt, A. I.; O’Mullane, A. P.; De Marco, R. Tuning the Electrocrystallization Parameters of Semiconducting Co[TCNQ]2-Based Materials To Yield either Single Nanowires or Crystalline Thin Films. J. Am. Chem. Soc. 2007, 129, 2369−2382. (57) Laboratory Requirements, Code of Federal Regulations, Section 493.931, Title 42, 2003.

301

DOI: 10.1021/acsanm.7b00151 ACS Appl. Nano Mater. 2018, 1, 293−301