Diphenyl Viologen on an HOPG Electrode Surface: Less Sharp Redox

Aug 12, 2013 - Diphenyl Viologen on an HOPG Electrode Surface: Less Sharp Redox. Wave than Dibenzyl Viologen. Tomohiro Higashi. † and Takamasa ...
0 downloads 0 Views 1MB Size
Download PDF
Article pubs.acs.org/Langmuir

Diphenyl Viologen on an HOPG Electrode Surface: Less Sharp Redox Wave than Dibenzyl Viologen Tomohiro Higashi† and Takamasa Sagara*,‡ †

Department of Science and Technology and ‡Division of Chemistry and Materials Science Graduate School of Engineering, Nagasaki University, Nagasaki, Nagasaki 852-8521, Japan S Supporting Information *

ABSTRACT: Redox behavior of diphenyl viologen (dPhV) on a basal plane of a highly oriented pyrolytic graphite (HOPG) electrode was described using the results of voltammetric and electroreflectance measurements. Its characteristics were compared to those of dibenzyl viologen (dBV), which undergoes the first-order faradaic phase transition. Unlike dBV, dPhV-dication (dPhV2+) was found to take a strongly adsorbed state on an HOPG surface. This is due to much stronger π−π interaction between phenyl rings of dPhV2+ and HOPG surface than between benzyl groups of dBV2+ and the surface. The participation of this strongly adsorbed dPhV2+ in the redox process can be avoided by (1) a shorter than ∼3 min time period elapsing from touching a freshly cleaved HOPG surface to dPhV solution until the start of potential scan, (2) complete equilibration at the electrode potentials at which superficial dPhV molecules are fully reduced, or (3) multiple cyclic potential scanning to repeat oxidation−reduction of adsorbed species. Even in such conditions, although voltammograms of thin-layer electrochemistry for the surface-confined dPhV•+/dPhV2+ couple are obtained with peak widths being as narrow as those of dBV, it is not the first-order phase transition. The participation of strongly adsorbed dPhV2+ molecules results in another new voltammetric feature with a broader peak. The film formed by strongly adsorbed dPhV2+ was hydrophilic, whereas dBV2+ does not form such a film but only a gas-like layer. Measurements using X-ray photoelectron spectroscopy confirmed that the film consists of dPhV2+ with coexistent water. These results reveal a typical case that delicate interaction balance among V2+, V•+, and electrode surface determines whether the two-dimensional first-order transition takes place or not.

■

INTRODUCTION Viologens (1,1′-disubstituent-4,4′-bipyridinium dications) are among the most extensively studied organic electroactive species. They undergo two consecutive, one-electron reductions to radical cation and then to neutral form.1,2 A number of viologens have been thoroughly examined as electrochromic materials, electron-transfer mediators, and supramolecular building blocks.1−4 Viologens form their adsorption layers on metal electrode surfaces, such as Hg, Au, Ag, Cu, and others.1,5−30 It is known that dimethyl viologen, for example, exhibits various adsorption orientations on an Hg electrode depending on both its oxidation state and electrode surface charge.1 A soluble dialkyl viologen with long chains often exhibits a spike-like voltammetric peak pair at a less negative potential than the redox potential in the solution phase.7−20 It has been concluded that the spike response originates from potential-driven phase transition associated with reduction of the viologen, leading to the formation of a two-dimensional (2D) condensed film of the reduced form. Such a transition has been observed at Hg,7−12 Au(111), 13,14 and highly oriented pyrolytic graphite (HOPG)15−20 electrodes in aqueous media. As a typical © 2013 American Chemical Society

example, heptyl viologen (HV) undergoes a faradaic phase transition of the first-order at an HOPG electrode surface from a gas-like adsorption layer of oxidized form (HV2+) to an insoluble 2D condensed monolayer of the one-electron reduced monoradical monocation form (HV•+).15−18 Arihara and coworkers claimed, using their results of IR reflection spectroscopic measurements, that 4,4′-bibyridirium moieties of HV•+ assume side-on orientation in the condensed phase on an HOPG surface to allow π−π stacking among themselves and all-trans alkyl chains assume a parallel-to-surface orientation.19 We have recently found that a dichloride salt of dibenzyl viologen (dBV), a typical viologen possessing aromatic side chains, also undergoes a faradaic phase transition of the firstorder between a gas-like adsorption layer and a 2D condensed monolayer on an HOPG electrode.20 Results of the electroreflectance (ER) measurements revealed flat-lying orientation of the longitudinal molecular axis of the side-on bipyridinium moiety of dBV•+ molecules on the electrode surface. Beside Received: April 27, 2013 Revised: August 7, 2013 Published: August 12, 2013 11516

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

paper,20 dBV2+ forms only its gas-like adsorption layer of an HOPG electrode, which undergoes a faradaic phase transition of the first-order upon one-electron reduction to form a wellordered condensed phase of dBV•+ with side-on configuration.20 Difference of intermolecular interaction may make the difference to the electrochemistry between dPhV and dBV. The behavior of dPhV is now the question to be addressed in comparison to dBV. This study may provide us with the new physical insight into the adsorption layer redox process of diaryl viologen, especially from a perspective of interaction balance, and give an impact for development of tailored fabrication of functional electrode with proper molecular design.

HOPG, Safarowsky and co-workers have observed a phase with a square-shaped assembly unit consisting of four dBV2+ molecules on a chloride modified Cu(100) electrode surface using in situ STM.21,22 They claimed that not only electrostatic attraction between dBV2+ molecules and an underlying Cl− adlayer on the Cu(100) surface but also electrostatic repulsion of dBV2+ molecules with their neighbors are origins of the square unit formation. Pham and co-workers observed a stripe pattern phase of adsorbed dBV•+ molecules on a Cl− modified Cu(100) electrode surface using in situ STM.23,24 They proposed a model in which dBV•+ molecules take the side-on configuration to adopt themselves in π-stacked polymer chains, giving the stripe with a width of the molecular length. Many studies of the redox behavior of viologen adsorption layers on metal electrodes have focused on the dialkyl viologens and their derivatives as well as viologens with aromatic side chains. However, diaryl viologens which possess aryl groups as substituents directly bound to the central bipyridinium moiety have been less well studied. It is known that both formal redox potentials and the colors of diaryl viologen radical cations are quite variable in aqueous solution depending on the structures of the aryl substituent,1,31−37 in contrast to dialkyl viologens. This variety is due to the delocalization of molecular πelectrons extended over the aryl substituent groups.1,31,32,36 The π-electron conjugation of diaryl viologen radical cation may enhance the intermolecular π−π interaction between neighbors and also with an HOPG electrode surface. Characterization of the adsorption layer redox process of diaryl viologen may enable us to shed light in depth on the roles played by the balance of π−π interactions in the monolayer redox mechanism. Herein, we examine the redox behavior of diphenyl viologen (1,1′-diphenyl-4,4′-bipyridinium, dPhV), one of the most typical diaryl viologens, on an HOPG electrode surface. We compare dPhV to dBV in terms of the redox process of the adsorption layer. We focus on the effect of π−π interaction (intermolecular, between phenyl rings of neighboring dPhV molecules, and with the HOPG surface) upon voltammetric and electroreflectance features. The extent of π-electron conjugation of bipyridinium moiety with directly connected phenyl ring is greater than that with benzyl group.29 In dBV, the methylene unit between benzene ring and bipyridinium moiety suppresses π-electron conjugation between them. Our previous DFT calculation for an optimized molecular structure of dBV20 gave rise to the dihedral angles around the central C− C bond in bipyridinium, 40.1° for dBV2+ and 2.3° dBV•+. Dihedral angles between benzene ring and the nearest-neighbor pyridinium ring are 91.0° for dBV2+ and 66.4° for dBV•+. Our DFT calculation for dPhV at the same level as for dBV gave the dihedral angles around the central C−C bond, 35.7° for dPhV2+ and 6.8° for dPhV•+, together with the dihedral angles between benzene ring and adjacent pyridinium ring, 48.1° for dPhV2+ and 47.8° for dPhV•+. We can, therefore, immediately recognize that both dPhV2+ and dPhV•+ molecules are of more plate-like structure than corresponding dBV molecules and that πelectron conjugation of bipyridinium moiety is extended to phenyl rings in dPhV. As we will touch on later, the absorption spectrum of dPhV•+ (Figure S1) is indicative of more effective π-conjugation of dPhV than dBV. These differences between dPhV and dBV may also result in distinct electrochemical behavior. For example, plate-like structure and rich π-electron conjugation may lead to enhanced adsorptivity of dPhV2+ and/ or dPhV•+ onto HOPG. As we demonstrated in previous

■

EXPERIMENTAL SECTION

Materials. Diphenyl viologen (1,1′-diphenyl-4,4′-bipyridinium dichloride: dPhV2+ 2Cl−) purchased from Tokyo Chemical Industry Co. was recrystallized from acetone + ethanol and dried in vacuo. Water was purified through a Milli-Q integral (Millipore) to a resistivity >18 MΩ cm. All other chemicals were of the highest reagent grade commercially available and used as received. A plate of HOPG (Matsushita Electric Co., Panasonic graphite, PGX 05: size 12 × 12 × 3 mm thickness) was connected perpendicularly with a copper pipe using a colloidal graphite paste. To expose a fresh basal plane, the surface of the HOPG was peeled off by the use of Scotch adhesive tape immediately before use. Voltammetric and Electroreflectance (ER) Measurements. For the voltammetric measurements, a potentiostat (HUSOU, HECS9094) was coupled with a function generator (HUSOU, HECS-321B). A ∼75 μL portion of 0.30 M KCl solution containing various concentrations of dPhV2+ 2Cl− was pipetted to place its droplet on an upward surface of a horizontally set HOPG electrode in a wetted Ar gas atmosphere. A tip of an Ag/AgCl/saturated KCl reference electrode and an Au wire counter electrode were immersed in the droplet. This “droplet configuration”16 was used in all voltammetric measurements unless otherwise stated. The active electrode area (A), which is the actual contact area of the HOPG surface to solution, was measured photographically. Measurements of ER signals were carried out by normal incidence of monochromatic light at a hanging meniscus (H-M) configuration. This configuration was set by horizontal touching a HOPG electrode to an Ar gas | dPhV solution interface from the Ar gas phase. The details of the ER instrumentation and spectroelectrochemical cell were given in our previous publication.38 Briefly, applying a sinusoidal wave potential modulation to the electrode, both real (in-phase to Eac) and imaginary (90° out-of-phase) components of the ac reflectance signal were phase-sensitively detected. They were normalized by timeaveraged reflectance to obtain real and imaginary parts of the ER signal, (ΔR/R)ER. An ER spectrum is a plot of both real and imaginary parts of (ΔR/R)ER as a function of the wavelength of the incident light (λ). All of the electrochemical control was carried out under Ar gas (>99.998%) atmosphere at room temperature 24 ± 2C°. All the potentials were referenced to an Ag/AgCl/saturated KCl electrode. XPS Measurements. XPS spectra were measured by employing Axis Urtra DLD (Shimadzu) with an Al Kα X-ray source (1486.6 eV) at a constant dwell time of 100 ms and the pass energy of 40 eV. The anode voltage 15 kV and the anode current was 10 mA. The gas pressure in the analysis chamber no higher than 5.0 × 10−7 Torr was maintained during the measurements. The HOPG sample plate was mounted on the standard sample studs using an electrically conductive double-side adhesive tape. The core-level signals were obtained by normal-to-surface takeoff under surface charge neutralization.

■

RESULTS AND DISCUSSION Voltammetric Studies in 0.3 mM dPhV Solution. First of all, we describe the voltammetric features predetermined by procedures, conditions, and history, using cyclic voltammo11517

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

In order for the history-dependent nature of voltammetric response of dPhV to be clarified, following two experiments were conducted. Experiment I. After the CV measurement for dPhV solution with two = 5 min, the electrode potential was cycled 200 times continuously between 0.05 and −0.35 V at v = 70 mV s−1. As shown in Figure 2a, continuous cycling resulted in

grams (CVs) all obtained at an HOPG electrode in contact with 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution, which is abbreviated as “dPhV solution” in this section. We will describe later the concentration dependence of CV. We found that CVs of dPhV strongly depend on the time spent after the initial contact of an HOPG surface to dPhV solution until the start of the potential scan. A freshly prepared HOPG electrode surface was touched to dPhV solution in its open circuit condition and allowed to stand for a given period of standby time. The open circuit potential was constantly 0.20 ± 0.05 V during this standby time. Then, the initial potential Ei (−0.10, 0.10, or 0.30 V) was applied. After the double-layer charging current reached steady-state typically within 1 s, the potential scan was started. The standby time elapsed from the touching of the freshly cleaved HOPG surface to dPhV solution until the start of potential scan is designated by two. The redox responses obtained in CV were independent of Ei, as far as it was in between −0.10 and 0.30 V, but dependent largely on two. When two = 5 s and potential sweep rate (v) was 50 mV s−1, we obtained Figure 1a. Both anodic and cathodic voltammetric

Figure 2. CVs for an HOPG electrode in the droplet configuration with 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution: (a) two = 5 min, multiple potential sweep (200 cycles), v = 70 mV s−1, and A = 0.79 cm2; (b) teq(−0.30 V) = 5 min, v = 70 mV s−1, and A = 0.50 cm2.

re-emergence and growth of Pc1 accompanied by decay of both Pc2 and Pa2. This time course of change, therefore, looked representing the backtracking of the change of CV with two from Figure 1a,b. Experiment II. A freshly prepared HOPG electrode surface was touched to the dPhV solution in its open circuit condition, and then −0.30 V was applied at which the electrode was equilibrated for 5 min (namely, equilibration time at −0.30 V: teq(−0.30 V) = 5 min). Then, the electrode potential was swept from Ei = −0.30 V to positive direction. Note that −0.30 V is slightly more negative to the peak potential of Pc2 (Figure 2a). Thus obtained Figure 2b exhibited both anodic and cathodic sharp voltammetric peaks Pc1 and Pa1 due to the redox couple of dPhV•+/2+. Note that the CV with teq(−0.30 V) = 5 min after the CV measurement with two = 10 min also exhibited a sharp CV response identical to Figure 2b. The midpoint-potential (E1/2) between anodic and cathodic spike-like peaks in Figure 2b is −0.18 V, which is more positive potential than E1/2 of the phase transition of dBV (−0.28 V), indicating that relative stabilization of the reduced form dPhV•+ to the oxidized form dPhV2+ is greater than that of dBV•+ relative to dBV2+. This may arise from the conjugated πelectron spreading over the entire molecular frame of radical cation of dPhV to a much greater extent than of dBV. The fullwidths at the half-height (ΔW1/2) of the sharp peaks at v = 70 mV s−1 (Figure 2b) are 15 mV for the cathodic peak and 11 mV for the anodic counterpart. These values are close to those of dBV on an HOPG electrode surface at the same v. An important finding in these experiments is that the two limiting states, one showing Figure 1a or 2b type of CV and the other showing Figure 1b type, are mutually transformable. To attain the former type with such narrow peak widths, one of the following strict initial conditions should be fulfilled: (1) two shorter than ∼3 min, (2) complete equilibration at the electrode potential ranging from −0.25 to −0.35 V, and (3) multiple cyclic potential scanning until suppressing Pc2.

Figure 1. CVs for an HOPG electrode in the droplet configuration with 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution (dPhV solution): (a) two = 5 s, v = 50 mV s−1, and A = 0.55 cm2; (b) two = 10 min, v = 70 mV s−1, and A = 0.79 cm2.

peaks due to the redox couple of dPhV•+/2+ appeared significantly sharp. We label the sharp cathodic peak at −0.19 V as Pc1 and the sharp anodic one at −0.17 V as Pa1. Note that we avoided negative potential scan to more negative than −0.40 V in order not to produce insoluble deposit by bulk reduction, which is beyond the scope of this work. When two = 10 min, Figure 1b was obtained. It exhibited one cathodic voltammetric peak (Pc2) due to one-electron reduction of dPhV2+ and two anodic peaks (Pa1 and Pa2), along with disappearance of Pc1. Peak Pc2 was characterized by its broader width and more negative peak potential as compared to Pc1. The adsorption layer of dPhV2+ molecules was apparently changed during the standby time. Appearance of two peaks for the reoxidation of dPhV•+ suggests the formation of at least two different adsorption states of dPhV•+ molecules on an HOPG surface after reduction at Pc2, and one of them is the state that originates Pa1 upon reoxidation. When 5 s < two < 3 min, Pc1 became broader with increasing of two. With further increase of two, Pc2 emerged at two = 3 min, while both Pc1 and Pa1 became smaller. Then around two = 5 min, Pc1 disappeared, and finally before two = 10 min, the waveform converged with the CV in Figure 1b. 11518

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

bipyridinium-diphenyl plate-like structure, using the values reported for asymmetric viologens6 and crystallographic data33 for dPhV2+ 2Cl−·2H2O. The experimental Γ value is well in accord with the calculated projection areas in a side-on configuration. We concluded, therefore, that dPhV•+ molecules on an HOPG electrode surface producing the Pc1−Pa1 pair assume a side-on configuration in such a way that the bipyridinium-diphenyl plate is perpendicular to the HOPG surface with its longitudinal axis parallel to the surface. It is worthwhile to note that Γ obtained from the CV peaks Pa1 + Pa2, when these two peaks are observed, agreed well with the above-mentioned value, 2.6 × 10−10 mol cm−2. This fact reveals that, despite significant difference of CV curves between Figures 1b and 2b, the amount of adsorbed dPhV•+ molecules remains unaltered, regardless of the adsorption structure before reduction. It is now the key to address the new state on a HOPG surface produced by long-time contact with 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution at an open circuit condition or at a positive potential than Pa1. Most likely, it is a more strongly adsorbed dPhV2+ layer than the gas-like adsorption layer, because the reduction potential of the state (Pc2) is more negative than the potential of Pc1, the gas-like layer reduction peak. We tried to directly observe the molecules using in situ electrochemical STM. Although we saw some images of molecular rows on scattered occasions, we could not obtain any reliable molecular level images. Therefore, we used XPS analysis to characterize the state. Dication Adsorption Film and its XPS Analysis. Aforementioned results point to the formation of a strongly adsorbed dPhV2+ layer. Our approach included both a filmtransfer voltammetric measurement and an ex situ surface analysis to analyze this more strongly adsorbed dPhV2+ film than its gas-like one. After a part HOPG surface was touched to the 0.3 mM dPhV solution droplet for 10 min without electrochemical control (as the same procedure as to produce Figure 1b), it was rinsed by water. Then, we found a newly formed hydrophilic region in sharp contrast to the surrounding bare hydrophobic surface area on the HOPG surface. It remained even after copious rinsing, as was obvious from our naked-eye observation of a contact angle of a pure water small droplet placed on the surface for test. This hardly soluble, hydrophilic film formation is likely the result of strong adsorption of dPhV2+ molecules on an HOPG surface. The formation of such a hardly soluble film of oxidized form of viologen molecules to water on an HOPG surface has never been reported so far to our best knowledge. If dPhV2+ molecules adsorbed onto HOPG surface only as a gaslike adsorption layer, just like dBV2+, dPhV2+ molecules should have been displaced from the HOPG surface by rinsing with copious amounts of water. To examine the electroactivity of the film and to analyze its composition, the film-transfer voltammetric measurements were conducted. Figure 4 shows the CV obtained through the film-transfer procedure. The above-mentioned HOPG electrode with the hardly soluble film was brought into contact (0.50 cm2) with a 0.3 M KCl solution free of viologen. After two precycles to reduce surface adsorbed O2, we obtained a steady-state voltammogram as shown. The waveforms of the peaks are rather similar to Pc2 and Pa2. The faradaic peak charge amounts to 0.58 × 10−10 mol cm−2, being much less than the saturated adsorption amount in 0.3 mM dPhV solution (2.6 × 10−10 mol cm−2). This submonolayer amount indicates

To sum up, we also found that long time contact of the HOPG surface to dPhV2+ solution produces a new state, which gives rise to Pc2 instead of Pc1, and that a part of the new state is altered through the reduction at Pc2, leading to emergence of Pa1. Sweep rate dependence of CV that gives only the Pc1−Pa1 pair is shown in Figure 3a. Both anodic and cathodic peak

Figure 3. Plot of peak currents (ip) of 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution as a function of v for the CVs with (a) teq(−0.30 V) = 5 min and Ei = −0.30 V and (b) log−log plot of ip with two = 10 min and v. A = 0.50 and 0.55 cm2 for (a) and (b), respectively.

currents (ip) were proportional to v in the v range from 10 to 100 mV s−1, indicative of a redox reaction of a surface-confined redox couple dPhV•+/2+. The redox of dPhV is ascribed to a thin-layer electrochemistry but not a phase transition of the first-order, in sharp contrast to dBV on an HOPG that showed v dependence typical to a faradaic first-order phase transition with nucleation−growth−collision (NGC) kinetics.20,39 Figure 3b shows a log−log plot of ip and v obtained in the same conditions to obtain Figure 1b. In the range of 10−150 mV s−1, ip values for peaks Pc2, Pa2, and Pa1 were proportional to v0.7, v1.0, and v0.6, respectively. Taking also the peak shape into consideration, the broad cathodic Pc2 may partially be controlled by diffusion process to give v0.7 dependence, which may assume influence on v0.6 dependence of Pa1. The adsorption amount of redox active species (Γ) was obtained from the CV peak faradaic charge. Because the double-layer charging contribution to the peak charge is negligible at a basal-plane HOPG electrode,16,20,40−42 Γ can be equated to the superficial amount of surface-confined dPhV molecules. When CV gives rise to only the Pc1−Pa1 pair, Γ was 2.6 (±0.1) × 10−10 mol cm−2 for both Pc1 and Pa1, regardless of both v and sweep direction. This Γ value is in line with one monolayer formation of dPhV with the area occupied by one dPhV molecule of 0.64 nm2. The projection areas for one flatlying dPhV molecule were estimated to be ∼1.2 and 0.67 nm2 for, respectively, face-on and side-on configurations of the 11519

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

caused by secondary electrons interacting with viologen dication, a strong electron acceptor.29 Similar multicomponent N 1s spectra involving reduction product formation were reported for various types of solid-state dichloride salts of viologens.29,30,46−48 Figure 5 also shows the O 1s core-level spectrum. Since O2, as contaminants from air, tends to desorb rapidly from HOPG surfaces under UHV conditions, the O 1s spectra might be associated with the coadsorbed water at cationic sites of dPhV2+ and dPhV•+. Taken together, we can conclude that the formation of the hydrophilic film is due to strong adsorption of dPhV2+ molecules. In the case of dPhV molecules, the benzene rings as the substituents are on the same axis of the bipyridinium moiety. In the case of dBV2+ molecules, which also have a benzene ring, the dihedral angle between benzene ring and the neighboring bipyridinium ring is almost 90° in light of our previous DFT calculations.20 Thus, the difference of intramolecular relative arrangement of benzene rings between dPhV and dBV brings about significantly more attractive interaction of dPhV to an HOPG surface than dBV. We interpreted that the π−π interaction between benzene rings of dPhV and HOPG surface worked effectively as a driving force for the dPhV2+ film formation on an HOPG surface. Electroreflectance Study. To gain deeper insight into the mechanism of surface redox reaction of dPhV, in situ ER measurements were made. The ER spectral measurements see in principle the reflectance spectral difference at two potentials. The redox response of the target should be time invariant during the measurements usually taking >10 min. Therefore, we should abandon the chance to see the Pc2−Pa2 pair. Around the potentials of this peak pair, the adsorption layer is elusive so that it partly changes into the other state with time during the ER measurements. We restricted the discussion using ER spectrum to the Pc1−Pa1 pair. Figure 6 shows ER spectrum (ERS) at Edc = E1/2 = −0.18 V after multiple potential scanning in the range from 0.00 to

Figure 4. CV of an HOPG in contact with a 0.30 M KCl solution free of dPhV (v = 50 mV s−1 and A = 0.50 cm2). The electrode was first in contact with 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution for 10 min without electrochemical control and then rinsed with water before the measurement with dPhV-free 0.30 M KCl solution.

that the adsorption is not completely an irreversible process. The strongly adsorbed state of dPhV2+ may have originated from π−π interaction between the molecules and HOPG surface. Partial desorption of strongly adsorbed dPhV2+ molecules from HOPG surface to decrease the electrostatic intermolecular repulsion may also take place. Upon reoxidation in the voltammetric measurements, neither loss of dPhV2+ nor transformation to the state to produce Pa1 was observed (Figure 4). It is important to note that when we made the same film-transfer experiments for HV and dBV, we found the complete absence of any redox active film formation after the transferral. Most likely, the bipyridinium moiety together with side phenyl groups serves strong π−π interaction of dPhV with underlying HOPG surface by totally flat orientation. Such a strong affinity to HOPG of the oxidized form of viologen lacks for HV and dBV. Figure 5 shows an XPS core-level spectra for N 1s and O 1s of the film of interest on an HOPG surface. The presence of N

Figure 5. XPS core-level spectra for (a) N 1s and (b) O 1s of dPhV strongly adsorbed film on an HOPG surface. The solid line represents the fitted curves.

and Cl in the wide scan spectrum (Figure S2) indicates existence of dichloride salt of viologen on an HOPG surface. The N 1s core-level spectrum (Figure 5) can fit with three bands as shown. Based on literature,29,30,43−48 the peak at 401.7 eV can be ascribed to the positively charged nitrogen of the dPhV2+ and that at 398.4 eV to the nitrogen of viologen radical cation (V•+) formed by X-ray excitation in the analysis chamber. The presence of the peak at 405.7 eV was also reported by Biniak and co-workers with an assignment to chemisorbed nitrogen oxide.43 The generation of reduced viologen species from initially present viologen dichloride salt has been rationalized in terms of instability of the viologen dication under X-ray irradiation.29 It is most likely that such a fractional transformation of dication into its reduced species is

Figure 6. ERS of an HOPG electrode in H-M configuration on 0.3 mM dPhV2+ 2Cl− + 0.30 M KCl solution: Edc = −0.180 V, ΔEac = 0.07 V, and f = 14 Hz. A spectral curve obtained by simulation (see text for details) was superimposed by dotted line.

−0.35 V until reaching the last cycle CV in Figure 2a. Thus, the spectrum represents the Pc1−Pa1 pair. The ER spectral feature is far different from the difference absorption spectrum between dPhV2+ (almost colorless in visible region) and dPhV•+ (see Figure S1). This reveals that the ER signal is due to the redox couple of dPhV•+/2+ with light-absorbing dPhV•+ being in direct contact with the HOPG surface, since otherwise the ERS should have been nearly the same as the solution difference absorption spectrum.38 We carried out ER spectral simulation 11520

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

by the use of solution absorption spectra of dPhV (Figure S1), Kramers−Kronig relation, and three-phase stratified optical model,38 as the same as we described previously.15 The simulated curve for the real part, the dotted line in Figure 6, is well in accord with the experimental ERS. That is, we did not find spectral change of dPhV•+ molecules due to their more significant intermolecular interaction in the adsorption layer producing the Pc1−Pa1 pair than in solution. This is probably why the Pc1−Pa1 pair is not the first-order phase transition, because strong attractive interactions between nearest neighbors are the prerequisite of the first-order transition. Concentration-Dependent Voltammograms. Figure 7 shows typical CVs at four different concentrations of dPhV2+

Figure 8. Plots of (a) adsorbed amounts of dPhV (Γanode and Γcathode), and (b) ΔW1/2 for anodic and cathodic peaks in CVs at v = 70 mV s−1 against cdPhV when measurement procedures were the same, as in Figure 7.

adsorption orientation with a side-on configuration of dPhV•+, while Γcathode at cdPhV < 70 μM amounts to submonolayer. As shown in Figure 8b, ΔW1/2 for the cathodic peak increased steeply with decreasing cdPhV < 70 μM. Equilibration at −0.30 V allowed saturation of dPhV•+ to form its complete monolayer amount at all the concentrations examined as shown in Figure 8a. Upon oxidation, the produced dPhV2+ molecules come into a gas-like layer which is in equilibrium with solution-phase molecules. At lower cdPhV, the equilibrium amount of adsorbed dPhV2+ is also lower. Therefore, at lower concentrations, Γcathode for rereduction is smaller than Γanode. That is, upon reoxidation, a part of the reoxidation product dPhV2+ molecules are displaced from HOPG surface into solution until the occurrence of following reduction. The intermolecular interaction between dPhV•+ molecules is weaker because of lower superficial density of the molecules, and this results in the increase in ΔW1/2 of the cathodic peak. During the time period in which surface viologens are in oxidized form, the gas-like adsorbed dPhV2+ gradually changes into strongly adsorbed state. But this process is so slow that its reduction is not apparent in the rereduction peak at v = 70 mV s−1. When we made this period a bit longer, we observed slight increase of Γcathode, presumably due to additional contribution to the cathodic current from the reduction of the strongly adsorbed molecules. These results revealed that the dPhV2+ gas-like adsorption layer formation is a reversible adsorption process in equilibrium with dPhV2+ molecules in the solution phase. On the other hand, dPhV•+ adsorption layer always tends to be saturated spontaneously when cdPhV > 10 μM. The formation of strongly adsorbed dPhV2+ state is a slow process. Model of Adsorption State. Taken together, all the CV features described above may come from the presence of two types of adsorption states for dPhV2+ (Scheme 1). One is a gas-

Figure 7. Collection of CVs for an HOPG electrode in the droplet configuration with 0.30 M KCl containing typical four different concentrations of dPhV2+ 2Cl− after being equilibrated at −0.3 V from where potential was scanned to positive at v = 70 mV s−1 and reversed at 0.05 V. A = 0.64 cm2 for 10 μM dPhV, 0.64 cm2 for 50 μM dPhV, 0.79 cm2 for 70 μM dPhV, and 0.79 cm2 for 500 μM dPhV.

2Cl− (cdPhV) obtained using the experimental procedure of aforementioned Experiment II with teq(−0.30 V) = 5 min. The potential scan was started from −0.30 V to positive and reversed at 0.05 V to record the rereduction peak. In the range of cdPhV up to 500 μM, we obtained a pair of redox peaks of the Pc1−Pa1 pair, but obvious asymmetry of anodic and cathodic peaks was observed, especially at lower concentrations. The value of Γ obtained from the anodic CV peak faradaic charge (Γanode) was constant at 2.5 × 10−10 mol cm−2, regardless of both cdPhV and v (Figure 8a), while cathodic counterpart (Γcathode) decreased with decreasing cdPhV. The value of Γanode corresponds to one monolayer with flat-lying 11521

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

conjugation of dPhV, especially in its reduced form, are expected to enhance intermolecular interaction as well as molecule−HOPG interaction. Although this may be an advantage for the first-order phase transition, it does not take place as a fact. Presumably, simultaneously enhanced molecule−HOPG substrate interaction may prevent dPhV•+ from formation of a well-ordered phase. In other words, Redcond phase is not enough well-ordered so as to be the first-order transition counterpart phase against Oxgas phase. The part of the modeling is speculative because we have presently no direct evidence of 2D molecular order structures of Oxstr and Redflat. We are currently making further efforts of STM and spectroelectrochemical measurements.

Scheme 1. Proposed Model Structures for Adsorption States and Redox Processes and Assignment of Voltammetric Peaks

■

CONCLUSION Adsorption state of dPhV at an HOPG electrode surface and its one-electron redox process were studied in depth using voltammetry, ER spectroscopy, and XPS. Obtained characteristics were compared to those of dBV, which also possesses benzene rings in the substituent groups. It was found that, unlike dBV, formation of strongly adsorbed state of the oxidized form, dPhV2+, takes place at positive potentials and at the open circuit potential, although it is slow process. To avoid the participation of strongly adsorbed dPhV 2+ in the voltammetric response, one of the following three strict conditions should be fulfilled: (1) a shorter than ∼3 min time period elapsing from touching a freshly cleaved HOPG surface to dPhV solution until the start of potential scan; (2) complete equilibration at the electrode potentials at which superficial dPhV molecules are fully reduced; and (3) multiple cyclic potential scanning to repeat oxidation−reduction. However, even though the participation can be excluded, the redox reaction of dPhV•+/2+ is of a thin-layer electrochemistry but not a first-order phase transition. The redox behavior was summarized in a model scheme. Two major differences of dPhV from dBV have been uncovered. First, dPhV exhibits strong π−π interaction with the HOPG surface even in the dication form presumably due to its plate-like molecular structure and extended π-conjugation in the molecule. This reflects the formation of strongly adsorbed state of dPhV2+ which was characterized by film-transfer and XPS measurements as a water-insoluble hydrophilic film on an HOPG surface. Second, even though a surface redox process between a gas-like adsorption state of dPhV2+ and a side-on configurated flat-lying oriented adsorption state of dPhV•+ undergoes, it is not the first-order phase transition, in sharp contrast to dBV. This may be a typical case that delicate interaction balance among V2+, V•+, and electrode surface determines whether the 2D first-order transition takes place or not. In order to gain deeper insight about the requirement for the first-order phase transition, we need to quantitatively evaluate the intermolecular interaction energies and to find the possible presence of obstacle states that inhibits the propagation of the phase-transition front. We are currently underway of addressing these needs.

like adsorption layer (Oxgas) that gives rise to a sharp reduction peak (Pc1). This state is found upon the initial contact of a fresh HOPG surface to dPhV2+ solution before it is transformed within 5 min of two to a strongly adsorbed state (Oxstr) that gives rise to a broad reduction peak (Pc2). As depicted in Scheme 1, dPhV2+ molecules in Oxstr layer exhibit flat-lying adsorption orientation with a face-on configuration because of their strong π−π interaction with HOPG surface and intermolecular electrostatic repulsion. Even when the electrode surface with Oxstr was rinsed by a copious amount of purified water, the adsorbed dPhV2+ molecules in Oxstr partly remained on the HOPG surface. This was demonstrated by the use of the results of a film-transfer experiment and XPS analysis. Reduction of dPhV2+ in Oxgas produces dPhV•+ in condensed phase (Redcond). The latter is in a side-on configuration with flat-lying longitudinal molecular axis with a superficial concentration of 2.6 × 10−10 mol cm−2. One-electron redox process of the Pc1−Pa1 pair is of a thin-layer electrochemistry but not as a first-order phase transition. The cathodic process of Pc2 originates from reductive formation of Oxstr phase accompanied by molecular reorientation to side-on configuration. Simultaneously, additional molecules may be supplied to the bare part of the HOPG surface from solution phase to an extent of compensating 2D density of Γ via diffusion, and they join Redcond or a new phases as dPhV•+, Redflat. Peaks Pa2 and Pa1 were ascribed, respectively, to reoxidation of adsorbed dPhV•+ in Redflat to dPhV2+ and oxidative dissolution of dPhV•+ in Redcond to dPhV2+ phase of Oxgas. Because Pa2 is a broader peak and at a more negative potential than Pa1, indicating weak intermolecular interaction, dPhV•+ molecules in Redflat are presumably flat-lying and face-on. The presence of Oxstr is due to strong π−π interaction of dPhV2+ with HOPG surface. When Oxstr phase is once reduced to be Redcond + Redflat phases and then subjected to oxidation−reduction cycle, the resulting phase can finally be Oxgas as demonstrated by CV (Figure 2a). Both equilibration at −0.30 V followed by oxidation and multiple potential scan may eventually restore the interface where the adsorption layer of dPhV2+ is exclusively composed of Oxgas only. Even though side-on orientation of dPhV•+ molecules in Redcond phase is typical in the first-order transitions of viologens on HOPG, its redox of dPhV is not of the first-order transition. On one hand, the plate-like structure and rich π-electron

■

ASSOCIATED CONTENT

S Supporting Information *

Figure S1, absorption spectra of the aqueous solutions of reduced form of dPhV and dBV in an aqueous solution. Figure S2, XPS wide scan spectrum and core-level spectrum for Cl 2p of dPhV adsorption film on an HOPG surface. This material is available free of charge via the Internet at http://pubs.acs.org. 11522

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

■

Article

(16) Sagara, T.; Miuchi, K. Double potential step chronoreflectometry approach to bi-stable potential region of the faradaic phase transition of heptyl viologen at a HOPG electrode. J. Electroanal. Chem. 2004, 567, 193−202. (17) Tanaka, Y.; Sagara, T. Formation of viologen radical cation condensed phase through two-dimensional molecular organization process on an HOPG electrode surface in binary viologen solutions. J. Electroanal. Chem. 2008, 619−620, 65−74. (18) Sagara, T.; Tanaka, S.; Miuchi, K.; Nakashima, N. Characteristics of faradaic phase transition of an adsorption layer of heptyl viologen at a basal plane HOPG electrode. J. Electroanal. Chem. 2002, 524−525, 68−76. (19) Arihara, K.; Kitamura, F.; Nukanobu, K.; Ohsaka, T.; Tokuda, K. Voltammetric and spectroscopic study of the adsorption of alkyl viologens on a HOPG electrode. J. Electroanal. Chem. 1999, 473, 138− 144. (20) Higashi, T.; Shigemitsu, Y.; Sagara, T. Faradaic phase transition of dibenzyl viologen on an HOPG electrode surface studied by in situ electrochemical STM and electroreflectance spectroscopy. Langmuir 2011, 27, 13910−13917. (21) Safarowsky, C.; Wandelt, K.; Broekmann, P. Formation of supramolecular cavitands on copper electrode surfaces. Langmuir 2004, 20, 8261−8269. (22) Safarowsky, C.; Rang, A.; Schalley, C. A.; Wandelt, K.; Broekmann, P. Formation of 2D supramolecular architectures at electrochemical solid/liquid interfaces. Electrochim. Acta 2005, 50, 4257−4268. (23) Pham, D. T.; Wandelt, K.; Broekmann, P. 2D ordering phenomena under non-equilibrium conditions: An in situ STM approach. Chem. Phys. Chem. 2007, 8, 2318−2320. (24) Pham, D. T.; Tsay, S. L.; Gentz, K.; Zoerlein, C.; Kossmann, S.; Tsay, J. S.; Kirchner, B.; Wandelt, K.; Broekmann, P. Quasi-reversible chloride adsorption/desorption through a polycationic organic film on Cu(100). J. Phys. Chem. C 2007, 111, 16428−16436. (25) Jiang, M.; Sak, E.; Gentz, K.; Krupski, A.; Wandelt, K. Redox activity and structural transition of heptyl viologen adlayers on Cu(100). Chem. Phys. Chem. 2010, 11, 1542−1549. (26) Tsay, S. L.; Tsay, J. S.; Fu, T. Y.; Broekmann, P.; Sagara, T.; Wandelt, K. Molecular structures of dicarboxylated viologens on a Cu(100) surface during an ongoing charge transfer reaction. Phys. Chem. Chem. Phys. 2010, 12, 14950−14959. (27) Röefzaad, M.; Jiang, M.; Zamlynny, V.; Wandelt, K. Potential dependent structure transitions of heptyl viologen layers on Cu(100) studied by in situ STM and IRRAS. J. Electroanal. Chem. 2011, 662, 219−228. (28) Hromadová, M.; Kolivoska, V.; Sokolova, R.; Gal, M.; Pospíšil, L.; Valásek, M. On the adsorption of extended viologens at the electrode|electrolyte interface. Langmuir 2010, 26, 17232−17236. (29) Breuer, S.; Pham, D. T.; Huemann, S.; Gentz, K.; Zoerlein, C.; Hunger, R.; Wandelt, K.; Broekmann, P. Organic layers at metal/ electrolyte interfaces: Molecular structure and reactivity of viologen monolayers. New J. Phys. 2008, 10, 125033. (30) Pham, D. T.; Keller, H.; Breuer, S.; Huemann, S.; Hai, N. T. N.; Zoerlein, C.; Wandelt, K.; Broekmann, P. Anion/cation layers at electrified interfaces: A comprehensive STM, XRD and XPS case study. Chimia 2009, 63, 115−121. (31) Nanasawa, M.; Miwa, M.; Hirai, M.; Kuwabara, T. Synthesis of viologens with extended π-conjugation and their photochromic behavior on near-IR absorption. J. Org. Chem. 2000, 65, 593−595. (32) Kamogawa, H.; Sato, S. Redox photochromism of arylviologen crystals. Bull. Chem. Soc. Jpn. 1991, 64, 321−323. (33) Porter, W. W., III; Vaid, T. P. Isolation and characterization of phenyl viologen as a radical cation and neutral molecule. J. Org. Chem. 2005, 70, 5028−5035. (34) Rosseinsky, D. R.; Monk, P. M. S.; Hann, R. A. Aniondependent aqueous electrodeposition of electrochromic 1,1′-biscyanophenyl-4,4′-bipyridilium (cyanophenylparaquat) radical cation by cyclic voltammetry and spectroelectrochemical studies. Electrochim. Acta 1990, 35, 1113−1123.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected] Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This work was financially supported in part by Grant-in-aid for Scientific Research from MEXT, Japan (no. 24550158) to T.S., by Tokyo Ohka Foundation to T.S. and by Kato Foundation for Promotion of Science to T.H. The authors deeply thank Mr. Hiroshi Furukawa at Nagasaki University for his technical assistance in XPS measurements.

■

REFERENCES

(1) Monk, P. M. S. The Viologens: Physicochemical Properties, Synthesis, and Applications of the Salts of 4,4′-bipyridine; John Wiley & Sons: Chichester, England, 1989. (2) Bird, C. L.; Kuhn, A. T. Electrochemistry of the viologens. Chem. Soc. Rev. 1981, 10, 49−82. (3) Seung, S. J.; Yun, H. J.; Kim, Y. H.; Goddard Iii, W. A.; Flood, A. H.; Laursen, B. W.; Tseng, H. R.; Stoddart, J. F.; Jeppesen, J. O.; Jang, W. C.; Steuerman, D. W.; Delonno, E.; Heath, J. R. Structures and properties of self-assembled monolayers of bistable [2]rotaxanes on Au (111) surfaces from molecular dynamics simulations validated with experiment. J. Am. Chem. Soc. 2005, 127, 1563−1575. (4) Mortimer, R. J. Organic electrochromic materials. Electrochim. Acta 1999, 44, 2971−2981. (5) Cieslinski, R.; Armstrong, N. R. Voltammetric, chronoamperometric and chronoabsorptometric studies of the nucleation of n-heptyl viologen films on SnO2, silane-modified SnO2 and ion-beam-treated ito-metallized polymer films. J. Electroanal. Chem. 1984, 161, 59−73. (6) Cotton, T. M.; Kim, J. H.; Uphaus, R. A. Spectroscopic and electrochemical studies of oriented monolayers on electrode surfaces. Microchem. J. 1990, 42, 44−71. (7) Kitamura, F.; Ohsaka, T.; Tokuda, K. Adsorption behavior of viologen derivatives on hanging mercury electrode surfaces. J. Electroanal. Chem. 1993, 347, 371−381. (8) Millán, J. I.; Rodríguez-Amaro, R.; Ruiz, J. J.; Camacho, L. Reorientation of the Cation Radical of Heptyl Viologen on Mercury in Water/DMSO Mixed Media. Langmuir 1999, 15, 618−623. (9) Millán, J. I.; Amaro, R. R.; Ruiz, J. J.; Camacho, L. Twodimensional condensation and reorientation of the bromide salt of the heptyl viologen cation radical at the Hg/DMSO interface. J. Phys. Chem. B 1999, 103, 3669−3676. (10) Arihara, K.; Kitamura, F.; Ohsaka, T.; Tokuda, K. Adsorption structure of the heptyl viologen cation radical on a mercury electrode surface: Voltammetric and in situ infrared reflection absorption spectroscopic studies. J. Electroanal. Chem. 2000, 488, 117−124. (11) Salas, R.; Sánchez-Maestre, M.; Rodríguez-Amaro, R.; Muñoz, E.; Ruiz, J. J.; Camacho, L. Voltammetric study of the two-dimensional phase formed by the cation radical of methyl viologen on mercury in the presence of iodide ions. Langmuir 1995, 11, 1791−1796. (12) Kobayashi, K.; Fujisaki, F.; Yoshimine, T.; Niki, K. An analysis of the voltammetric adsorption waves of methyl viologen. Bull. Chem. Soc. Jpn. 1986, 59, 3715−3722. (13) Arihara, K.; Kitamura, F. Adsorption states of heptyl viologen on an Au(111) electrode surface studied by infrared reflection absorption spectroscopy. J. Electroanal. Chem. 2003, 550−551, 149−159. (14) Arihara, K.; Ohsaka, T.; Kitamura, F. Characteristic cyclic voltammograms of alkyl viologens at single crystal gold electrodes. Phys. Chem. Chem. Phys. 2002, 4, 1002−1005. (15) Sagara, T.; Tanaka, S.; Fukuoka, Y.; Nakashima, N. Study of the voltammetric spike response of heptyl viologen at a HOPG electrode horizontally touched to a gas/heptyl viologen aqueous solution interface. Langmuir 2001, 17, 1620−1629. 11523

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524

Langmuir

Article

(35) Freitag, M.; Galoppini, E. Cucurbituril complexes of viologens bound to TiO2 films. Langmuir 2010, 26, 8262−8269. (36) Kamogawa, H.; Satoh, S. Organic solid photochromism by photoreduction mechanism: Aryl viologens embedded in poly(Nvinyl-2-pyrrolidone). J. Polym. Sci., Part A: Polym. Chem. 1988, 26, 653−656. (37) Freitag, M.; Gundlach, L.; Piotrowiak, P.; Galoppini, E. Fluorescence enhancement of Di-p-tolyl viologen by complexation in cucurbit[7]uril. J. Am. Chem. Soc. 2012, 134, 3358−3366. (38) Sagara, T. UV-visible Reflectance Spectroscopy of Thin Organic Films at Electrode Surfaces. In Advances in Electrochemical Science and Engineering; Alkire, R. C.; Kolb, D. M.; Lipkowski, J.; Ross, P. N., Eds; Wiley-VCH: Weinheim, 2006; Vol. 9, pp 47−95. (39) Maestre, M. S.; Rodríguez-Amaro, R.; Muñoz, E.; Ruiz, J. J.; Camacho, L. Use of cyclic voltammetry for studying two-dimensional phase transitions: Behaviour at low scan rates. J. Electroanal. Chem. 1994, 373, 31−37. (40) Cline, K. K.; McDermott, M. T.; McCreery, R. L. Anomalously slow electron transfer at ordered graphite electrodes: Influence of electronic factors and reactive sites. J. Phys. Chem. 1994, 98, 5314− 5319. (41) Gerischer, H.; McIntyre, R.; Scherson, D.; Storck, W. Density of the electronic states of graphite: Derivation from differential capacitance measurements. J. Phys. Chem. 1987, 91, 1930−1935. (42) Gerischer, H. An interpretation of the double layer capacity of graphite electrodes in relation to the density of states at the Fermi level. J. Phys. Chem. 1985, 89, 4249−4251. ́ (43) Biniak, S.; Szymański, G.; Siedlewski, J. Swiatkowski, A. The characterization of activated carbons with oxygen and nitrogen surface groups. Carbon 1997, 35, 1799−1810. (44) Masuda, T.; Irie, M.; Uosaki, K. Photoswitching of electron transfer property of diarylethene-viologen linked molecular layer constructed on a hydrogen-terminated Si(111) Surface. Thin Solid Films 2009, 518, 591−595. (45) Alvaro, M.; García, H.; García, S.; Márquez, F.; Scaiano, J. C. Intrazeolite photochemistry. 17. Zeolites as electron donors: Photolysis of methylviologen incorporated within zeolites. J. Phys. Chem. B 1997, 101, 3043−3051. (46) Liu, X.; Neoh, K. G.; Zhao, L.; Kang, E. T. Surface functionalization of glass and polymeric substrates via graft copolymerization of viologen in an aqueous medium. Langmuir 2002, 18, 2914−2921. (47) Liu, X.; Neoh, K. G.; Kang, E. T. Viologen-functionalized conductive surfaces: Physicochemical and electrochemical characteristics, and stability. Langmuir 2002, 18, 9041−9047. (48) Liu, X.; Neoh, K. G.; Cen, L.; Kang, E. T. Enzymatic activity of glucose oxidase covalently wired via viologen to electrically conductive polypyrrole films. Biosens. Bioelectron. 2004, 19, 823−834.

11524

dx.doi.org/10.1021/la401606v | Langmuir 2013, 29, 11516−11524